<%BANNER%>

Quantitative Imaging of Cocaine and Its Metabolites in Brain Tissue by Matrix-Assisted Laser Desorption/Ionization Linea...

Permanent Link: http://ufdc.ufl.edu/UFE0041855/00001

Material Information

Title: Quantitative Imaging of Cocaine and Its Metabolites in Brain Tissue by Matrix-Assisted Laser Desorption/Ionization Linear Ion Trap Tandem Mass Spectrometry
Physical Description: 1 online resource (210 p.)
Language: english
Creator: Reich, Richard
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: brain, cocaine, imaging, ion, linear, maldi, mass, matrix, metabolites, quantification, spectrometry, tandem, tissue, trap
Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Quantitative Imaging of Cocaine and Its Metabolites in Brain Tissue by Matrix-Assisted Laser Desorption/Ionization Linear Ion Trap Tandem Mass Spectrometry Detection of drugs in tissue typically requires extensive sample preparation in which the tissue is first homogenized, followed by drug extraction, before the extracts are finally analyzed by liquid chromatography/mass spectrometry (LC/MS). Directly analyzing drugs in intact tissue would eliminate any complications introduced by sample preparation. A matrix-assisted laser desorption/ionization tandem mass spectrometry (MALDI-MS^n) method has been developed for the quantification of cocaine and its metabolites present in postmortem brain tissue of a chronic human cocaine user. It is shown that tandem mass spectrometry (MS^n) increases selectivity, which is critical for differentiating analyte ions from background ions such as matrix clusters and endogenous compounds found in brain tissue. It is also shown that the use of internal standards corrects for signal variability during quantitative MALDI, which can be caused by inhomogeneous crystal formation, inconsistent sample preparation, and laser shot-to-shot variability. The MALDI-MS^n method developed allows for a single MS^2 experiment that uses a wide isolation window to isolate both analyte and internal standard target ions. This method is shown to provide improved precision (~10-20 times reduction in percent relative standard deviation) for quantitative analysis compared to using two alternating MS^2 experiments that separately isolate the target analyte and internal standard ions. A wide isolation window reduces signal variability when the analyte and internal standard signals are ratioed. However, the wide isolation window not only isolates the analyte and internal standard ions, but also other ions that are not of interest. These ions fill up the finite storage capacity of the ion trap and may lead to space-charge effects, which result in reduced resolution and peak shifts that interfere with detection of the target ions. Since the current instrument software only allows for one isolation window during MS^n, a multi-notch isolation waveform that selectively isolates the analyte and internal standard ions was created to remove the effects of background interferences and boost the sensitivity for analyte and internal standard ions. A multi-notch stored waveform inverse Fourier transform (SWIFT) pulse was calculated with frequency notches corresponding to the secular frequencies of the M+H+ ions of cocaine (COC), benzoylecgonine (BE), cocaethylene (CE), and their trideuterated analogs, COC-d3, BE-d3, and CE-d3. Multi-notch SWIFT isolation was found to have lower precision than wide isolation, which may be caused by frequency shifts of the analyte and internal standard ions from space-charge effects caused by high m/z background ions from the tissue (e.g., lipids). Finally, a two-stage SWIFT isolation method was developed that uses a high-mass filter to eject high m/z background ions before the multi-notch SWIFT isolation is applied. The two-stage SWIFT isolation showed similar precision to wide isolation for the MALDI-MS^2 analysis of cocaine and its metabolites in brain tissue. The two-stage SWIFT isolation and wide isolation were used to quantitatively image cocaine and its metabolites in postmortem human brain tissue, and were compared to the quantitative analysis of human brain tissue homogenate using MALDI-MS^2.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Richard Reich.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Yost, Richard A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041855:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041855/00001

Material Information

Title: Quantitative Imaging of Cocaine and Its Metabolites in Brain Tissue by Matrix-Assisted Laser Desorption/Ionization Linear Ion Trap Tandem Mass Spectrometry
Physical Description: 1 online resource (210 p.)
Language: english
Creator: Reich, Richard
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: brain, cocaine, imaging, ion, linear, maldi, mass, matrix, metabolites, quantification, spectrometry, tandem, tissue, trap
Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Quantitative Imaging of Cocaine and Its Metabolites in Brain Tissue by Matrix-Assisted Laser Desorption/Ionization Linear Ion Trap Tandem Mass Spectrometry Detection of drugs in tissue typically requires extensive sample preparation in which the tissue is first homogenized, followed by drug extraction, before the extracts are finally analyzed by liquid chromatography/mass spectrometry (LC/MS). Directly analyzing drugs in intact tissue would eliminate any complications introduced by sample preparation. A matrix-assisted laser desorption/ionization tandem mass spectrometry (MALDI-MS^n) method has been developed for the quantification of cocaine and its metabolites present in postmortem brain tissue of a chronic human cocaine user. It is shown that tandem mass spectrometry (MS^n) increases selectivity, which is critical for differentiating analyte ions from background ions such as matrix clusters and endogenous compounds found in brain tissue. It is also shown that the use of internal standards corrects for signal variability during quantitative MALDI, which can be caused by inhomogeneous crystal formation, inconsistent sample preparation, and laser shot-to-shot variability. The MALDI-MS^n method developed allows for a single MS^2 experiment that uses a wide isolation window to isolate both analyte and internal standard target ions. This method is shown to provide improved precision (~10-20 times reduction in percent relative standard deviation) for quantitative analysis compared to using two alternating MS^2 experiments that separately isolate the target analyte and internal standard ions. A wide isolation window reduces signal variability when the analyte and internal standard signals are ratioed. However, the wide isolation window not only isolates the analyte and internal standard ions, but also other ions that are not of interest. These ions fill up the finite storage capacity of the ion trap and may lead to space-charge effects, which result in reduced resolution and peak shifts that interfere with detection of the target ions. Since the current instrument software only allows for one isolation window during MS^n, a multi-notch isolation waveform that selectively isolates the analyte and internal standard ions was created to remove the effects of background interferences and boost the sensitivity for analyte and internal standard ions. A multi-notch stored waveform inverse Fourier transform (SWIFT) pulse was calculated with frequency notches corresponding to the secular frequencies of the M+H+ ions of cocaine (COC), benzoylecgonine (BE), cocaethylene (CE), and their trideuterated analogs, COC-d3, BE-d3, and CE-d3. Multi-notch SWIFT isolation was found to have lower precision than wide isolation, which may be caused by frequency shifts of the analyte and internal standard ions from space-charge effects caused by high m/z background ions from the tissue (e.g., lipids). Finally, a two-stage SWIFT isolation method was developed that uses a high-mass filter to eject high m/z background ions before the multi-notch SWIFT isolation is applied. The two-stage SWIFT isolation showed similar precision to wide isolation for the MALDI-MS^2 analysis of cocaine and its metabolites in brain tissue. The two-stage SWIFT isolation and wide isolation were used to quantitatively image cocaine and its metabolites in postmortem human brain tissue, and were compared to the quantitative analysis of human brain tissue homogenate using MALDI-MS^2.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Richard Reich.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Yost, Richard A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0041855:00001


This item has the following downloads:


Full Text





QUANTITATIVE IMAGING OF COCAINE AND ITS METABOLITES IN
BRAIN TISSUE BY MATRIX-ASSISTED LASER DESORPTION/IONIZATION
LINEAR ION TRAP TANDEM MASS SPECTROMETRY




















By

RICHARD FRED REICH, JR.


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2010


































2010 Richard Fred Reich, Jr.

































To my best friend and wife, Erica,
who displayed incredible patience, support,
and sacrifice throughout my educational journey.
God gave me the greatest gift in the world
by placing you in my life.









ACKNOWLEDGMENTS

First, I would like to thank my research advisor, Rick Yost, who welcomed me back into

his research group again. I have learned to appreciate the free environment that you allow your

students to thrive in during graduate school. You have allowed me to think independently and to

mature with minimal guidance in your research laboratory. I know that the experiences that I

have gained under your mentorship will serve me well in the future.

I thank my USAFA co-worker and retired forensic toxicologist, Joseph Levisky, for

providing me tissue samples from the El Paso County Coroner's Office in Colorado Springs,

Colorado. I thank my undergraduate research assistants, Kasia Cudzilo and Kyle Cromwell, who

helped prepare samples and served as a soundboard for brainstorming research ideas. I would

also like to thank the Yost research group, past and present, for all their help and support. I hope

to continue to see you at future ASMS conferences. You made the hospitality suites a lot of fun.

I thank my good friend Pat Castle, who encouraged me throughout my PhD program.

Even from out-of-state, you were able to keep me physically and spiritually fit. I love you,

brother.

I thank the parishioners of St. Augustine Catholic Church, who served as my family away

from home. I must also thank my brother Knights from Council 13900. Thank you for

providing me the opportunity to serve you as Grand Knight. You have helped me to further

develop my leadership skills and provided me experiences that will last a lifetime.

Last, thanks go to my wife and my parents. I am very grateful for their patience with my

effort and support when I needed them. Erica, you are the love of my life. Thank you for

understanding who I am and allowing me to be myself. I look forward to our adventures

together at our next assignment at Patrick AFB.









TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ..............................................................................................................4

LIST OF TABLES ......... ........... .............................................. 8

LIST O F FIG U RE S ................................................................. 9

ABSTRACT ........................................... .. ...... .......... 13

CHAPTER

1 IN TRODU CTION ...................................................... ................ ........ 15

Cocaine ......................... ... ... ...................................15
Cocaine M etabolism ....................................... ................ .......... .............. 15
Neurobiological Mechanism of Cocaine......................................................................16
Analysis of D rugs of Abuse in Tissue ...................................................... ............... 16
T issue e Im aging T echniqu es ......... ........................................ ....................... .......................... 17
MALDI-MS Imaging (MALDI-MSI).............................................................. ...............19
Spatial Resolution...................... .......... .. .. .......... ........ 20
Tissue Preparation .................... .................... ...................21
E x cisio n o f T issue e ..................................................................................................... 2 2
Tissue Sectioning and M counting ............................................. ............................ 22
S am p le T ran sfer......................................................................................................... 2 3
M A L D I M atrix ................................................................24
M atrix selection.............................................. 24
Matrix deposition .......................................... ...... ............... 25
Tissue w washing .................................................... ............... .. ........ 25
Q uantitative M A L D I-M S ................................................................................................. 26
Intern al Stan dard s ................................................................................ 2 6
T andem M ass Spectrom etry ........................................ .............................................27
M A L D I-M SI Instrum entation ........................................................................ ..................27
L inear Ion Trap M ass Spectrom etry .......................................................................... .. .... 28
M ass-Selectiv e In stability ....................................................................... ..................29
Io n S to rag e ................................. ............................................................ ............... 3 0
A utom atic G ain C control .......................................................................... ................... 3 1
Helium Buffer Gas .................................. ... ...... ... .................. 31
Resonance Ejection ......................... ........... .. ........... ..... ..... 32
M a ss A n a ly sis ............................................................................................................ 3 3
Iso latio n ........................................................................................3 3
A ctivation .............................................................................................................34
Overview of Dissertation ............................................................. ................35

2 WIDE ISOLATION ..............................................48









In tro d u ctio n ....................................48.............................
E x p erim mental ....................................50.............................
C h e m ic a ls ...............................................................................5 0
T issue C collection .............................................................................. 5 1
Tissue Sectioning and Sample Preparation ......................... ........ ...............51
M ass Spectrom etry ................................................................52
R results and D discussion ........................................53
MS2 and MS3 Mass Spectra of COC and COC-d3 ....................................................53
Improving Signal Reproducibility with Internal Standards................. ....54
Increasing Analyte Selectivity with MSn ............................................ .............54
Combining Internal Standards with MSn using a Wide Isolation Window ...................56
Isolation Window Width and Automatic Gain Control ...........................................58
Quantification of Cocaine in Postmortem Human Brain Tissue .............. ............... 60
C onclu sions.......... ..........................................................63

3 SW IF T ISO L A T IO N ......................................................................... 74

In tro d u ctio n ................. ............... ....................................................................................... 7 4
E x p erim mental ................................7...................7..........
C h e m ic a ls ...............................................................................7 7
Tissue C collection ............. ....... ................. ......................................................... 77
Tissue Sectioning and Sample Preparation ........................ .............................. 78
M ass Spectrom etry ........................................................................78
SW IF T C calculation .......... ...................................................................... ....... .. .. .. 79
Inverse Fourier transform .......................................... .. ......... .................79
Quadratic phase m odulation ......................................................... ............... 80
Temporal spectral inhomogeneity ................... ....... ................................. 80
A p o d iz atio n ...........................................................................................8 2
SW IFT Application to LTQ .............. .................... ........................................83
Results and Discussion ...................................... .... ......... 83
Optimization of a Dual-Notch SWIFT ..............................................................83
Frequency optimization......................... ........... ......... 83
Burst count optim ization ......................................... .. .. ...........................84
Amplitude optimization............................................ 86
Selective Ion Isolation of Standards on MALDI Plate ................ ...... .........87
Improving MALDI Precision with SWIFT ................................... .................87
Selective Ion Isolation of Standards on Tissue................. ........................89
C onclu sions.......... ..........................................................90

4 QUANTITATIVE ANALYSIS OF DRUGS IN BRAIN TISSUE ..................................... 100

In tro d u ctio n ................... ...................1.............................0
E x p erim mental ............................................................................... 10 1
C h e m ic a ls ..............................................................................1 0 1
Tissue Collection ................................................................ ....... 101
Tissue Sectioning and Sam ple Preparation ........................................ ........ .......101
Tissue Homogenization ..................................... ............. ........ ................. 102


6









Preparation of Standard Solutions ....................................................... .... ........... 102
Preparation of Unknown Sam ple Solution................................................................ 103
Solid-P hase E extraction ......................................................................... ................... 103
M ass Spectrometry .................. ............................... .............. ........... 104
SW IF T C alcu lation ......... ...................................................................... ........ .. ....... .. 10 5
SW IFT A application to LTQ ................................................ .............................. 106
R results and D discussion ..................................... .......... ........ .............. .. 106
H exa-N otch SW IFT Isolation................................................ ............................ 106
SW IF T Isolation on T issu e............................................. ......................................... 108
Io n E je ctio n ....................... ..............................................................................1 0 9
Tw o-Stage Isolation......................... ...... ...... .... .............112
H igh M ass F ilter (H M F ).................................................................... ...................... 113
Combining HMF with Hexa-Notch SWIFT.....................................................114
M S/M S with Two-Stage Isolation............................................................ ...............116
Comparing Wide Isolation and Two-Stage SWIFT Isolation ............ ... ................. 118
Two-Stage SWIFT MALDI-MS/MS Quantification .......................................... 120
SPE-MALDI-MS/MS Quantification.............................. ...........121
C onclusions.....................................................................124

5 CONCLUSIONS AND FUTURE WORK..................................................... ................160

C o n c lu sio n s ........................................................................................................................... 1 6 0
F future W ork ......................................................164

APPENDIX

A B E T A C A L C U L A TIO N ............................................................................ ....................165

B C ++ SW IFT PR O G R A M .......................................................................... .....................168

C L T Q M O D IFIC A T IO N S ...................... .. .. ......... .. ....................... ............................... 185

L IST O F R E FE R E N C E S ...................... .. .. ......... .. ............................. .................................202

B IO G R A PH IC A L SK E T C H ...................... .. .. ......... .. ............................... .........................209
















7









LIST OF TABLES


Table page

4-1 Hexa-notch SWIFT properties based on m/z 305.8 at q = 0.830 .................................127

4-2 Hexa-notch SWIFT properties based on m/z 290.2 at q = 0.830 ...............................134

4-3 Hexa-notch SWIFT properties based on m/z 290.2 at q = 0.791 ....................................136

4-4 Quantification of BE, COC, and CE from Unspiked Human Brain Tissue................ 156









LIST OF FIGURES

Figure page

1-1 M etabolism of cocaine ........... ... ............. .... .................. ........ ... 38

1-2 C cocaine's m echanism of action ........................................ ...................... .....................39

1-3 D opam inergic pathw ay. .......................................................................... ......................40

1-4 C cocaine im aging in tissue ......................................................................... ...................4 1

1-5 Tissue preparation and MALDI-MS imaging protocol.. ................................. ...............42

1-6 Schematic of the LTQ with MALDI source. ........................................ ................43

1-7 Basic design of the two-dimensional linear ion trap................................ ....................44

1-8 Scheme for application of DC, RF trapping, and AC excitation voltages necessary
for operation of the 2D ion trap. ............................................................................ ..... 45

1-9 Mathieu stability diagram for the linear ion trap. ................................... ............... 46

1-10 A simplified scan function for the quadrupole ion trap showing the prescan and the
analytical scan which makes up one microscan ............................................................ 47

2-1 C cocaine dissociation pathw ay ............................................................................. .... .... 65

2-2 MALDI-MS signal variability with and without internal standards...............................66

2-3 Comparing mass spectra of COC and COC-d3 on MALDI plate and on brain tissue.......67

2-4 Fragmentation of the benzyldimethyldodecylammonium ion.102 ............ .................68

2-5 W ide isolation M A L D I-M S2......................................................... ...............................69

2-6 Images of standards spiked on brain tissue................................ ...................70

2-7 Calibration curves for alternating scans MS2 and wide isolation MS2 ...........................71

2-8 Mass spectrometric image of cocaine in brain tissue.............................. ...............72

2-9 C cocaine qu antification ............................................................................. .................... 73

3-1 Temporal spectral inhomogeneity of SW IFT ........................................ ............... 91

3-2 Effects of apodization on SW IFT. ............................................ ............................. 92

3-3 Optim ization of frequency notches........................................ ............... ................93









3-4 O ptim ization of burst counts...................................................................... ... ...............94

3-5 Optimization of SW IFT amplitude .........................................................................95

3-6 SWIFT isolation and wide isolation comparison .......................................................96

3-7 SWIFT isolation and wide isolation calibration curves..................................................97

3-8 Quad-notch SWIFT isolation of standards on MALDI plate..........................................98

3-9 Quad-notch SWIFT isolation of standards on brain tissue. ....................... ...............99

4-1 Solid-phase extraction schem e............................................... .............................. 126

4-2 Frequency domain of hexa-notch SWIFT isolation waveform.............. .......... 128

4-3 Relationship between secular frequency (co) and q-space. ............................................129

4-4 Variable hexa-notch SWIFT amplitude (m/z 305.8 at q = 0.830). .................................130

4-5 Mass spectra (m/z 80 to 2000) of hexa-notch SWIFT at different amplitudes..............131

4-6 Mass spectra (m/z 280 to 330) of hexa-notch SWIFT at different amplitudes.. ..............132

4-7 Pseudopotential well depth (Dx) of the ion trap..........................................................133

4-8 Isolation window width (Da) determined by the preset q of isolation.............................135

4-9 Hexa-notch SWIFT applied at variable q of isolation. .................................................137

4-10 Variable hexa-notch SWIFT amplitude (m/z 290.2 at q = 0.791)............................ 138

4-11 Mass spectra (m/z 280 to 330) of hexa-notch SWIFT at different amplitudes ............139

4-12 Mass spectra (m/z 80 to 2000) of hexa-notch SWIFT at different amplitudes.. ..............140

4-13 Frequency domain of high mass filter (HMF)......................................... ..............141

4-14 V variable am plitude of H M F............................................................................. ....... 142

4-15 Mass spectra (m/z 80 to 2000) of HMF at different amplitudes .................................... 143

4-16 Mass spectra (m/z 280 to 330) of HMF at different amplitudes.................................... 144

4-17 Frequency domain of two-stage SWIFT isolation. ................... ......................... 145

4-18 The time domain of the two-stage SW IFT isolation....................................................... 146

4-19 Variable amplitude of two-stage SWIFT isolation .................................................... 147









4-20 Mass spectra (m/z 80 to 2000) of two-stage SWIFT isolation at different amplitudes.... 148

4-21 Mass spectra (m/z 280 to 330) of two-stage SWIFT isolation at different amplitudes.... 149

4-22 MS/MS product spectra from the application of a two-stage isolation. ........................150

4-23 Comparison of MS/MS with wide isolation and two-stage SWIFT isolation..............151

4-24 Mass spectra comparison of wide isolation and two-stage SWIFT isolation ................152

4-25 BE calibration curve for BE spiked on intact brain tissue......... ...............................153

4-26 COC calibration curve for COC spiked on intact brain tissue............... ... .................154

4-27 CE calibration curve for CE spiked on intact brain tissue......... ......... .................155

4-28 BE calibration curve for BE standards spiked in blank brain tissue homogenate .........157

4-29 COC calibration curve for COC standards spiked into blank brain tissue homogenate..158

4-30 CE calibration curve for CE standards spiked into blank brain tissue homogenate. .......159

A-i LabView block diagram of subVI aqb_conf................... ...... ................165

A-2 Trap Calculator LabView program used to calculate f/ through an iterative process......166

A-3 Iterative calculation of f using the LabView program Trap Calculator........................167

B Program introductory com m ents............................................. ............................. 168

B-2 Definition of constants and declaration of variables....................................................... 169

B -3 V variable definitions.......... .................................................................... ......... .......170

B-4 Initialize variables for notches 1 through 4........................................... ............... 171

B-5 Initialize variables for notches 5 and 6. ........................................ ....................... 172

B -6 T est initialized v ariables ......................................................................... ................... 173

B-7 Setup output files. .............................. ... ...... ... .................. 174

B-8 Build components 1, 2, 3, 4, 5, 7, 9 and 11 of excitation waveform. ............................175

B-9 Build components 6, 8, 10 and 12 of excitation waveform. .........................................176

B-10 Build components 1 through 5 of isolation waveform.........................................177

B-11 Build components 6 through 10 of isolation waveform........................................178









B-12 Build components 11 through 13 of isolation waveform.....................................179

B-13 Create data array for frequency data ....................................................... .............. 180

B-14 Inverse Fourier transform, midpoint time reflection, and apodization..........................181

B-15 Normalize waveform and download to function generator. .........................................182

B-16 W rite waveform data to output files. ........................................ ......................... 183

B -17 Fast Fourier transform function. ........................................ ....................................... 184

C-1 Adding SWIFT waveform to AD734 chip (U64) on LTQ Analog PCB .........................189

C-2 Analog PCB modification to apply SWIFT waveform .................................................190

C -3 L TQ program m able trigger .............................................................................. ....... 191

C -4 Program m able trigger locations .......................................................................... ....... 192

C-5 Digital PCB modifications to access programmable trigger.................. .. ..................193

C-6 Trigger at beginning of scan. ................................................ ............................... 194

C-7 Trigger at injection .................................. .............. ............. ......... 195

C-8 Trigger at isolation .................................... ................ .......... .. ............ 196

C-9 Toggling LTQ isolation w aveform on and off ...............................................................197

C-10 Mass spectra of cocaine with isolation waveform toggled. ...........................................198

C-11 Trigger at isolation w ith isolation toggled. ............................................ .....................199

C -12 T rigger at activ ation ............................................................................... ....................200

C-13 Trigger at scan out. ......................................... .......... ............ 201









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

QUANTITATIVE IMAGING OF COCAINE AND ITS METABOLITES IN
BRAIN TISSUE BY MATRIX-ASSISTED LASER DESORPTION/IONIZATION
LINEAR ION TRAP TANDEM MASS SPECTROMETRY

By

Richard Fred Reich, Jr.

August 2010

Chair: Richard Alan Yost
Major: Chemistry

Detection of drugs in tissue typically requires extensive sample preparation in which the

tissue is first homogenized, followed by drug extraction, before the extracts are finally analyzed

by liquid chromatography/mass spectrometry (LC/MS). Directly analyzing drugs in intact tissue

would eliminate any complications introduced by sample preparation. A matrix-assisted laser

desorption/ionization tandem mass spectrometry (MALDI-MSn) method has been developed for

the quantification of cocaine and its metabolites present in postmortem brain tissue of a chronic

human cocaine user. It is shown that tandem mass spectrometry (MSn) increases selectivity,

which is critical for differentiating analyte ions from background ions such as matrix clusters and

endogenous compounds found in brain tissue. It is also shown that the use of internal standards

corrects for signal variability during quantitative MALDI, which can be caused by

inhomogeneous crystal formation, inconsistent sample preparation, and laser shot-to-shot

variability. The MALDI-MSn method developed allows for a single MS2 experiment that uses a

wide isolation window to isolate both analyte and internal standard target ions. This method is

shown to provide improved precision (-10-20 times reduction in percent relative standard









deviation) for quantitative analysis compared to using two alternating MS2 experiments that

separately isolate the target analyte and internal standard ions.

A wide isolation window reduces signal variability when the analyte and internal standard

signals are ratioed. However, the wide isolation window not only isolates the analyte and

internal standard ions, but also other ions that are not of interest. These ions fill up the finite

storage capacity of the ion trap and may lead to space-charge effects, which result in reduced

resolution and peak shifts that interfere with detection of the target ions. Since the current

instrument software only allows for one isolation window during MSn, a multi-notch isolation

waveform that selectively isolates the analyte and internal standard ions was created to remove

the effects of background interference and boost the sensitivity for analyte and internal standard

ions. A multi-notch stored waveform inverse Fourier transform (SWIFT) pulse was calculated

with frequency notches corresponding to the secular frequencies of the [M+H] ions of cocaine

(COC), benzoylecgonine (BE), cocaethylene (CE), and their trideuterated analogs, COC-d3, BE-

d3, and CE-d3. Multi-notch SWIFT isolation was found to have lower precision than wide

isolation, which may be caused by frequency shifts of the analyte and internal standard ions from

space-charge effects caused by high m/z background ions from the tissue (e.g., lipids).

Finally, a two-stage SWIFT isolation method was developed that uses a high-mass filter to

eject high m/z background ions before the multi-notch SWIFT isolation is applied. The two-

stage SWIFT isolation showed similar precision to wide isolation for the MALDI-MS2 analysis

of cocaine and its metabolites in brain tissue. The two-stage SWIFT isolation and wide isolation

were used to quantitatively image cocaine and its metabolites in postmortem human brain tissue,

and were compared to the quantitative analysis of human brain tissue homogenate using

MALDI-MS2.









CHAPTER 1
INTRODUCTION

Cocaine

Postmortem toxicology is a special field of forensic toxicology that is used to determine

whether alcohol, drugs, or other poisons may have caused or contributed to the death of a person.

Cocaine is the most frequent cause of drug-related deaths in the United States, either as the direct

cause of death or as a contributing factor.1 According to the federal Drug Abuse Warning

Network survey, 40% of the 11,942 drug-related deaths reported in 2007 involved cocaine.2

This explains why cocaine analysis is of particular interest to the field of postmortem toxicology.

Cocaine Metabolism

Understanding the metabolism of cocaine (COC) and the relative proportion of COC to its

detectable metabolites can provide valuable forensic inferences about the extent of prior abuse.

For example, an individual with a brain concentration of 8 mg/kg of COC and 0.5 mg/kg of

benzoylecgonine (BE; a metabolite of COC) must have taken the drug just before death, because

BE does not cross the blood/brain barrier as freely as its lipophilic parent compound (COC).1

Conversely, addicts who have ingested large amounts of COC over several days are usually

found to have only modest COC concentrations, but high concentration of BE.1

The major route of cocaine metabolism (Figure 1-1) is hydrolysis of COC by hepatic and

plasma esterases, with loss of a benzoyl group to give ecgonine methyl ester (EVME). The

secondary route is spontaneous hydrolysis, probably non-enzymatic, which leads to BE by

demethylation. The final degradation of COC, which is a sequel to both the principal and

secondary routes of metabolism, leads to ecgonine. N-demethylation of COC is a minor route

leading to norcocaine. The principal metabolites are therefore BE, EVME, and ecgonine itself,

which are inactive; and norcocaine which is active, and may be relevant after acute intoxication.'









In the presence of alcohol, a further active metabolite, cocaethylene (CE) is formed in the liver

by a transesterification reaction which adds an extra methyl group to COC.1

Neurobiological Mechanism of Cocaine

Under normal conditions, dopamine (DA) is released by a neuron into the synapse, where

it can bind with DA receptors on neighboring neurons3 (Figure 1-2). Normally DA is then

recycled back into the transmitting neuron by a specialized protein called the dopamine

transporter (DAT). If COC is present, it attaches to the DAT and blocks the normal recycling

process, resulting in a build-up of DA in the synapse which contributes to the pleasurable effects

of COC. DA-rich brain regions such as the ventral tegmental area (VTA), nucleus accumbens

(NAc), and prefrontal cortex are frequent targets of COC addiction research. Of particular

interest is the pathway consisting of dopaminergic neurons originating in the VTA that terminate

in the NAc (Figure 1-3). This projection may function as a "reward center", in that it seems to

show activation in response to drugs of abuse like COC in addition to natural rewards like food

or sex. Note that research was performed on tissue sections from the NAc, a DA-rich area of the

striatum, which may contain an accumulation of COC due to its affinity to bind with the DAT.

Analysis of Drugs of Abuse in Tissue

A large variety of specimens are collected in the field of postmortem toxicology including

blood, liver, brain, and urine.4 For the analysis of drugs of abuse, brain samples show several

advantages over all other specimens in postmortem toxicology.5 One advantage is due to the

brain being an isolated compartment, which delays putrefaction after death.6 Also, the metabolic

activity is lower in the brain than in other tissues or in blood, resulting in slower decomposition.7

Finally, drugs of abuse establish their effects through the central nervous system. Therefore, it

can be assumed that concentrations of drugs of abuse found in the brain better reflect drug

concentrations at their site of action at the time of death.8









Analysis of drugs of abuse in the brain has applications in forensic and postmortem

toxicology. Drug concentrations in the brain may be needed to substantiate fatal overdoses9 and

support neurotoxicity studies.10 Direct measurement of drug and metabolite concentrations in

discrete brain regions can also be used to study the mechanisms of drug action,11 regional

distribution,12 and preferential accumulation of drugs.13

Conventional drug analysis in tissue involves homogenization of the tissue prior to

subsequent chromatographic analysis.14 Such sample pretreatments are known to introduce

variation in detection due to inhomogeneity of the analyte within the sample matrix.15 Also,

homogenization of tissue eliminates the opportunity to acquire detailed anatomical and

histological information for in situ drug distribution. Imaging techniques that include mass

spectrometric imaging can help provide this information.

Tissue Imaging Techniques

A number of analytical techniques are capable of imaging drugs in vivo and ex vivo,

including positron emission tomography (PET),16 single photon emission computed

tomography,17 magnetic resonance imaging,18 x-ray computed tomography,19 optical

fluorescence imaging,20 optical bioluminescence imaging,21 ultrasound,22 whole body

autoradiography (WBA),23 infrared imaging,24 and magnetically labeled nanoparticles.25

However, disadvantages of these imaging methods include low sensitivity, low specificity,

limited functional and molecular information, poor spatial resolution, and the need for the drug

to be labeled with either a radioactive isotope or a fluorescent tag, which can be time-consuming

and costly.26 Figure 1-4 shows cocaine imaging in tissue using PET to image [11C]cocaine and

WBA to image [3H]cocaine. A specific disadvantage for those techniques that require a

chemical tag is the need to monitor the tag rather than the intact drug, and therefore, the ability to

differentiate the drug from a metabolite that may have retained the tag is difficult. In addition, a









chemical tag may alter the pharmacological properties of the compound, which could affect both

bioavailability and localization within the tissue.

Mass spectrometric imaging (MSI) has higher molecular specificity compared to other

tissue imaging techniques, particularly when used in combination with tandem mass

spectrometry (MS/MS).27 The high selectivity of the instrument eliminates the need for labeling,

because the ion (or product ion as in tandem mass spectrometry) is monitored directly and leaves

the drug molecule of interest functionally unmodified. An unmodified drug compound also

removes the potential interference of fluorescent/radioactive labels with the biological function

(e.g., when the drug must pass through the blood/brain barrier). This analyte specificity of the

instrument also provides the ability to simultaneously image drugs and their metabolites due to

the parallel detection of multiple analytes. With MSI, an image can be produced for each of the

hundreds of detected analytes within the mass spectral data set. Another advantage is its high

sensitivity. Unfortunately, MSI is a destructive imaging technique, although only a few

molecular monolayers of sample are affected by the analysis. This characteristic precludes MSI

from being used for in vivo studies.

MSI collects chemical data normally associated with mass spectrometry, but in a spatially

defined manner, and processes that information into chemical image maps. Secondary ion mass

spectrometry (SIMS)28 and matrix-assisted laser desorption/ionization (MALDI) mass

spectrometry (MS)29 are the two main techniques used with MSI. MALDI-MS has been shown

to be very effective for the direct analysis of drugs and their metabolites in tissues.30-44 MALDI-

MS is currently the most common MSI technique used for mapping pharmaceuticals in tissue,

although new MSI techniques may emerge as new surface ionization methods are developed.

Ambient ionization methods such as desorption electrospray ionization (DESI)45 and laser









ablation electrospray ionization (LAESI)46 show potential for the in vivo analysis on the surface

of skin of organisms with high specificity. DESI has been used for the in vivo detection of the

antihistamine Loratadine from the finger of a person who had taken 10 mg of the drug, 40 min

prior to analysis.45 DESI has also been used to localize clozapine directly from histological

sections of brain, lung, kidney, and testis without prior chemical treatment.47

MALDI-MS Imaging (MALDI-MSI)

The most commonly used ionization source for mass spectrometric imaging is MALDI.

Unlike SIMS and DESI, MALDI requires the addition of a matrix to the sample. An advantage

of the matrix is that the solvent used to apply the matrix is used to extract the analyte out of the

tissue, not just analyte at the surface. However, the matrix solvent allows the potential for

analyte migration. MALDI is a soft ionization technique in which laser energy is applied for an

instant to a co-crystallized mixture of a compound (called a matrix) and the analyte molecules. A

typical matrix is a small organic compound that absorbs at the wavelength of the laser and

consequently promotes desorption of the analyte. The ionization mechanisms of MALDI are not

fully understood, but have been critically reviewed.48 In brief, the chromophore of the matrix

couples with the laser energy and causes a rapid vibrational excitation that desorbs matrix and

analyte molecules from the solid solution. The photo-excited matrix molecules are then

stabilized through proton transfer to the analyte (e.g., [M+H] ) Figure 1-5 illustrates the overall

protocol of a MALDI-MSI experiment. The first step in sample preparation for MALDI-MSI

involves application of a homogeneous layer of matrix to the sample (Figure 1-5D). The sample

is then analyzed by moving it step-wise beneath a pulsed laser beam (Figure 1-5E) and MALDI

mass spectra are acquired from each point (Figure 1-5F). Two-dimensional images may then be

obtained by plotting the relative or absolute ion abundance (considered to be proportional to

analyte concentration) versus spatial dimensions of X and Y (Figure 1-5G).









Laser wavelength is an important parameter in MALDI. The most commonly used

wavelength is 337 nm from the nitrogen laser, but harmonics of the Nd:YAG laser 1065 nm

fundamental (3x, 355 nm and 4x, 266 nm), various excimer laser lines that include XeCl (308

nm), KrF (248 nm), and ArF (193 nm), and infrared lasers such as carbon dioxide (10.6 alm) and

Er:YAG (2.94 alm) lasers have been employed.48 It has been shown that MALDI mass spectra

obtained from UV and IR laser wavelengths are similar.49 However, IR MALDI requires higher

laser pulse energy due to lower MALDI matrix absorption, and the sample consumption is also

higher.50 Characteristics of IR-MALDI that have been reported include a greater tendency to

form multiply charged high-mass ions, less metastable fragmentation, and adduct ion

formation.

Spatial Resolution

Spatial resolution for MALDI-MSI experiments is limited by laser spot size, laser step

size, matrix crystal size, and analyte migration. Spatial resolution increases with decreasing laser

spot size, but MALDI mass spectrometers are usually equipped with N2 (337 nm) or tripled

Nd:YAG (355 nm) lasers having relatively large spot sizes (about 100 alm diameter). The rate of

energy redistribution rapidly increases with smaller laser spot sizes and higher laser fluences are

required for MALDI to occur.52 These higher laser fluences can cause extensive fragmentation.

Laser spot sizes focused to 7-8 alm in diameter lead to a decrease in ion yields of two orders of

magnitude compared to a normal (100 alm) laser spot,53 since the cross sectional area analyzed is

approximately 100 times smaller. In addition, high spatial resolution experiments are more

sensitive to analyte migration during the matrix application step and dramatically increase

analysis time for whole tissue section analysis.









An alternate approach to increase spatial resolution is by oversampling (using a step size

that is smaller than the laser spot width). This method involves first, the complete ablation of the

MALDI matrix coating the sample at each sample position and second, moving the sample target

a distance less than the diameter of the laser beam before repeating the process. The reported

method enabled commercial MALDI instruments with large laser spots sizes (100 am) to image

with approximately 25 am imaging spatial resolution.54

Another factor that determines spatial resolution for MALDI-MSI is the size of the

matrix crystals formed during the matrix deposition process. The size of the sample-matrix co-

crystals grown is strongly dependent on the sample-matrix solution composition and the rate at

which the crystals are grown.55 For the majority of MALDI-MSI experiments, the spot size of

the laser is such that multiple crystals are sampled in each laser shot, thus the spatial resolution is

limited by the laser spot size and not the crystals formed. However, it is still important to avoid

non-uniformities in the matrix layer (crystals), which can cause ionization yields to vary across

the sample and hinder the interpretation of spatial information. Some approaches for MALDI

matrix application, such as inkjet printing,56'57 can produce a uniform coating of small crystals.

Different approaches for the application of MALDI matrix will be discussed further in the matrix

deposition section.

Tissue Preparation

MALDI-MSI of intact tissue involves preparation procedures with minimal sample

handling, which decreases analyte losses compared to analyses that involve the preparation of

tissue homogenates followed by extraction. Nonetheless, tissue preparation for MALDI-MSI is

critical to maintain the integrity of the spatial arrangement of drug and metabolite compounds

within tissue. Mishandling or improperly storing tissue samples in the early sample preparation









steps may cause delocalization or degradation of the analytes. Experimental procedures that

should be considered include excision of tissue, tissue sectioning, sample transfer to MALDI

target plate or glass microscope slide, matrix application, and tissue storage after sectioning.

Excision of Tissue

Tissue samples should be surgically removed so that the original shape of the tissue is

retained. Immediately after removal, the tissue may be loosely wrapped in aluminum foil and

frozen in liquid nitrogen by gently lowering the tissue into the liquid nitrogen over a period of 30

- 60 seconds.58 Immersing the tissue into the liquid nitrogen too quickly can lead to cracking and

brittle edges. The foil acts to provide support for more malleable tissue and prevents adhesion of

the tissue to the sides of the liquid nitrogen dewar. Freshly excised tissue that is placed into

small plastic tubes may mold to the shape of the tube when frozen. Whole tissues may remain

frozen in a freezer at -800C for at least a year with little to no degradation of the sample.58

Tissue Sectioning and Mounting

Frozen tissue samples are cut into thin sections in a cryostat, which allows for accurate

sectioning to be accomplished at sub-freezing temperatures with minimal sample contamination.

It is recommended that tissue samples be attached to the sample stage (Figure 1-5A) of the

cryostat by freeze mounting with a few drops of deionized water at the interface between the

tissue and the stage.59 It is not advised to use an embedding medium such as agar or OCT

(optimal cutting temperature polymer) to mount the tissue to the sample stage, because these

compounds could suppress ion formation in MALDI-MS analysis.5 Tissue samples, once

mounted to the cryostat sample stage, are sliced with a stainless steel microtome blade. The

sample stage temperature is typically maintained between -5 C and -25 C, depending on the

tissue type. Tissues that have a higher fat content require lower temperatures to avoid tearing

during sectioning. Although tissue thickness is not critical for most studies, 10 20 alm thick









tissue sections are optimal for handling. Analyte signal intensity has previously been shown to

increase with increasing tissue section thickness; it was hypothesized that, during matrix

application, matrix solvent may obtain access to the interior of the tissue to extract more

analyte.35

Sample Transfer

The tissue section can be transferred with a thin artist's brush and carefully positioned onto

a cold MALDI target plate or glass microscope slide. Care should be taken during the transfer to

avoid folding or tearing the thinly sliced tissue. Tears or rips distort the tissue section and create

holes or gaps, which were not present in the native tissue. All equipment that will come into

contact with the frozen tissue including the plate or glass slides should be kept in the cold box of

the cryostat during sectioning. Once the tissue slice is positioned on the cold target plate or glass

slide, they are removed from the cold box and quickly warmed, thus thaw-mounting the tissue

onto the sample plate or slide (Figure 1-5B). Thaw-mounted tissue samples should be stored in a

freezer at -80 C until analyzed.

When tissue samples are ready to be analyzed, they are dehydrated in a vacuum

desiccator at room temperature to remove moisture and avoid lateral migration of analytes before

application of MALDI matrix. Traditional low pressure (- 10-6 Torr) MALDI requires samples

to be dried completely (- 2 hours) before exposure to vacuum conditions. This prohibits the

analysis of freshly cut tissue and reduces sample throughput. In addition, low pressure MALDI

has been shown to produce in-source fragmentation of lipids in tissue, which makes low-level

detection difficult (unpublished results). MALDI operated at an intermediate pressure (IP) of

0.17 Torr (100,000-times higher than traditional vacuum MALDI) has been shown to reduce the

degree of source fragmentation by collisional cooling.60 Tissue drying times with IP-MALDI









can be reduced to 30 minutes, which will increase sample throughput and allow for the analysis

of tissue samples shortly after dissection.

MALDI Matrix

Matrix Selection

The success of MALDI-MSI for the analysis of drugs in tissue is dependent on the choice

of matrix. The common UV-absorbing molecules used as matrices for MALDI analysis are

benzoic acid-based components with low molecular weights (< 500 Da) such as sinapinic acid

(SA, 3,5-dimethoxy-4-hydroxycinnamic acid), a-cyano-4-hydroxycinnamic acid (CHCA), and

2,5-dihydroxybenzoic acid (DHB). Various MALDI matrices, including organic, solid ionic,

liquid, and liquid/solid two-phase matrices, have been reviewed.48 Unfortunately, ions formed

from most matrix compounds dominate the low-mass range background for a typical MALDI-

MS spectrum, making MS/MS or high resolution MS critical for the analysis of small molecules.

One approach to circumvent matrix interference is to use a higher molecular weight

matrix, which does not interfere in the low mass region. To this end, some porphyrins have been

employed as MALDI matrices.61 Although the porphyrin matrices have been shown to be

valuable for the detection of low-mass analytes with minimum mass interference from matrix

signals,61, 62 poor ion production yield for drug molecules in tissues was observed when these

porphyrin matrices were employed.63

Because of the nature of biological tissues, the growth of matrix crystals is more

complicated on tissue than on an inert plate where a small volume of matrix is mixed with a neat

drug solution. For example, on tissue, the matrix solvent not only plays a role in the co-

crystallization of the matrix and analyte molecules, but the solvent composition also facilitates

the extraction of analyte molecules to the surface of the tissue. Therefore, selecting a solvent

composition that can readily dissolve the analyte is critical for crystal formation as well as









analyte extraction. Solvent composition can also play an active role in protonation of the analyte

and result in higher ionization efficiency. Strong acids such as 0.1% trifluoroacetic acid (TFA)

are normally added to the matrix solution to assist protonation of proteins, but have been found

to have a marginal effect on the ionization efficiency for small molecules.39 For small

molecules, a higher matrix concentration (matrix-to-analyte ratio) can also produce better quality

mass spectra.39

Matrix Deposition

The analyte signal intensity, suppression of the matrix signal, and laser shot-to-shot

reproducibility can be affected by the distribution of matrix and analyte during crystallization.64

Crystal irregularities can occur when the matrix/analyte mixture partitions during the slow

crystallization process;65 thus, it is very important that the solubilities of all components are

suitably matched. Many sample preparation procedures for improved co-crystallization of

matrix and analyte have been reported and include electrospraying,65 fast evaporation,66

pneumatic spraying,67 spray-droplet method,68 sublimation,69 inkjet printing,7 acoustic drop

ejection,70 and solvent-free matrix dry-coating.71 By far the most common matrix deposition

approach for MALDI-MSI of drugs in tissue is pneumatic spraying31' 32, 35-38,41,43,44,72,73 with

either CHCA, SA, or DHB matrix. Pneumatic spraying is an inexpensive and easy technique of

applying MALDI matrix that is effective at depositing a homogeneous layer of small matrix

crystals across the entire tissue sample. For the research conducted, MALDI matrix solution was

applied to tissue by an artistic airbrush (Aztek A470, Testors; Rockford, IL, USA), Figure 1-5C.

The application of MALDI matrix by airbrush has been previously published.67

Tissue Washing

To optimize matrix crystallization, a washing step is sometimes performed before matrix

deposition, which allows the majority of salts to be removed from the surface of the tissue.5









Recent studies have shown that matrix crystallization and analyte incorporation are hampered by

the presence of high concentrations of salt, which can result in an inhomogeneous sample surface

and lead to high signal variability.74 The removal of salt from tissue sections is typically

performed by rinsing in 70-80% ethanol.58 Improved peptide and protein signals were

demonstrated with tissue-washing in organic solvents traditionally used for lipid extraction (i.e.,

methylene chloride, hexane, toluene, acetone, and xylene), especially from older or even

archived tissue sections.75 However, great care must be taken to prevent migration of analyte

molecules or even the loss of analyte; thus, tissue washing is not recommended for small

molecule applications such as cocaine analysis.

Quantitative MALDI-MS

Internal Standards

Although MALDI-MS is an established method for qualitative analysis, quantitative

analysis is more difficult because MALDI exhibits irreproducible analyte signals as a result of

inhomogeneous crystal formation, inconsistent sample preparation, and laser shot-to-shot

variability.76 Indeed, relative standard deviations can be higher than 50%.62, 77 The addition of

an internal standard can compensate for several of these experimental factors that seriously

complicate quantitative MALDI-MS.77-

An appropriate internal standard for MALDI must compensate not only for any

crystallization irregularities but also for subsequent desorption and gas-phase effects. In

choosing an internal standard, the relative polarities of the analytes and internal standard as well

as their solvent solubilities should be considered.77 Structural similarities should reflect the gas-

phase behavior of the involved molecules, and extend to solubility. Naturally, an isotope-labeled

standard is the ideal choice, since its chemical behavior is nearly identical to its unlabeled

counterpart.82 Such a standard guarantees identical crystallization and gas-phase behavior of the









analyte and internal standard.83 Traditional MALDI experiments demonstrate that using the ratio

of the analyte peak intensities to those of a deuterated internal standard can improve signal

reproducibility.82

Tandem Mass Spectrometry

Another challenge for quantitative MALDI-MS, particularly for the analysis of small

molecules such as drugs of abuse, is the strong interference for m/z < 500 due to MALDI matrix

ions.84 In addition, interference can originate from a multitude of ions produced from

endogenous compounds (e.g., lipids) found in tissue sections during tissue analysis. The high

molecular specificity of tandem mass spectrometry (MSn) eliminates the problem of interfering

ions by fragmenting the desorbed ions in the mass spectrometer and matching the fragment

masses with the molecular structure of the analyte. The analytical advantage of the linear ion

trap mass spectrometer is the ability to perform multiple stages of MS, which provides an

increase in molecular specificity with each stage of mass analysis. For this reason, all the

research conducted was performed on the linear ion trap mass spectrometer and thus the

background is a focus on this instrumentation.

MALDI-MSI Instrumentation

All MSI experiments reported in this dissertation were performed on a Thermo Scientific

LTQ XL linear ion trap (LIT) mass spectrometer (Thermo Scientific; San Jose, CA, USA) with

an intermediate pressure MALDI source, as shown in Figure 1-6 and described in detail in

elsewhere.67 85 The MALDI source uses a nitrogen gas laser that fires pulses at 337.7 nm with a

frequency of 60 Hz and energy of 250 tJ per pulse at 100% laser power. An iris attenuator is

used to vary the laser power. The laser energy is directed to the MALDI source by a fiber optic

cable. It is then focused using a series of mirrors and lenses to a spot size of approximately 100

[am at an incident angle of 320.67 The LTQ XL MALDI source uses nitrogen gas to maintain a









pressure of 75 mTorr (170 mTorr for LTQ MALDI), which is considerably higher than a

standard high vacuum MALDI source ( 10-6 Torr), but substantially below that of an

atmospheric pressure MALDI source.

The sample plate consists of a bottom support plate, which attaches to either a 96- or 384-

well microtiter plate (12.7 cm x 8.6 cm) for general MALDI applications, or a microscope slide

holder (2.5 cm x 7.5 cm, 0.1 cm thick) that is designed to hold four standard microscope slides

for tissue imaging applications. Microscope slides are affixed to the slide holder with double-

sided tape (Scotch 1.27 cm wide, 3M; Minneapolis, MN, USA). The MALDI control software

automatically identifies which plate configuration is being used and calibrates the position of the

sample plate. The sample plate mounts onto an XY stage by means of spring tension clamps,

and two precision vacuum-rated stepper motors control the two-dimensional movement. These

actuators position the XY stage with an accuracy of better than 3 am. The precision in going

back to a specific location is 1 am without taking the plate out and approximately 7 am after

taking the plate out of the vacuum and putting it back in.

A modified set of quadrupole rods, which can accommodate the entrance of the laser

beam and access for camera viewing, is added to the front of the LTQ multiple arrangement

behind the MALDI sample plate, Figure 1-6. Ions produced from the MALDI process are

directed into the LIT mass analyzer through the ion optics consisting of a series of quadrupoles,

lenses, and octopole.

Linear Ion Trap Mass Spectrometry

The LIT is a two-dimensional (2D) quadrupole ion trap (QIT), which is related to the

three-dimensional (3D) QIT that was first introduced and described as a mass storage device in

1953 by Wolfgang Paul and Helmut Steinwedel.86 The LIT operates in a fashion analogous to

the QIT. However, unlike the 3D QIT which contains two end cap electrodes and a ring









electrode, the LIT is composed of a segmented hyperbolic quadrupole mass analyzer with three

sets of hyperbolic rods87 of lengths 12 mm (front), 37 mm (center), and 12 mm (back) shown in

Figure 1-7.

Ions are trapped axially by applying separate direct current (DC) voltages (100 V) to all

three sections while radial trapping is accomplished by applying an oscillating radio-frequency

(RF) voltage (+5 kV rod to ground, 1 MHz) in two phases to the X and Y rod pairs shown in

Figure 1-8. A two-phase supplemental alternating current (AC) voltage (80 V, 5-500 kHz) is

applied across the X rods for isolation, activation, and ejection of ions. Ions are ejected radially

from the trap through opposing 30-mm long slits in the center section of X rods by mass-

selective instability scanning.8

Mass-Selective Instability

Mass-selective instability scanningss is accomplished by setting the DC component of the

center section rods to zero while the amplitude of the RF resonance excitation voltage applied to

the X rods is increased. As the amplitude of the RF voltage is increased, the magnitude of the

oscillations of the trapped ions also increases so that the ions eventually develop unstable

trajectories along the X axis, and are subsequently ejected from the trap in order of increasing

mass-to-charge (m/z) value. Ions are ejected through the slits in the center X rods and strike a set

of detectors consisting of a conversion dynode and an electron multiplier situated at each slit to

catch the ejected ions.

Ions trapped inside the LIT follow trajectories described by the second-order Mathieu

differential equation.89 Solutions to the differential equation are in terms of two reduced

parameters, a and q, which can be used to determine whether an ion will have a stable or

unstable trajectory in the trap under the defined conditions of the electric field. The values of a









and q depend on the dimensions of the trap and the potentials applied according to Equations 1-1

and 1-2:89

8eU
ax =-ay = (1-1)
Smro2 2

4eV
q =-q= m- 2 (1-2)


Uis the applied DC amplitude (and is zero in the LIT), Vis the applied RF potential, e is the

charge on an ion (1.602 x 10-19 C), m is the mass of an ion, ro is the radius of the hyperbolic rod

profiles (ro = 4 mm), and Q is the angular drive frequency.

Ion Storage

From the known solutions to the Mathieu equation one can generate a stability diagram

(Figure 1-9) that shows the common region in (a, q) space for which the X and Y components of

the ion trajectory are stable simultaneously such that the ion can be confined in the trap.90 The

parameters fl and ly at any given coordinate of a and q relate to the secular frequency co of the

ion in the X and Y directions, respectively (Equation 1-3).

,u = 0.5fluQ (1-3)

As the value of f approaches zero, the ion's secular frequency approaches zero, and the ion

is not contained. When the value of f equals one, the ion's secular frequency equals half the

frequency of the RF field, and the magnitude of its oscillation increases so that the ion escapes

the trap or collides with one of the electrode surfaces. When /f has a value between zero and

unity, the ion can be trapped by the oscillating fields and will oscillate in a periodic mode at its

secular frequency in x and y.









Automatic Gain Control

The LIT has a storage capacity of approximately 107 ions;s7 however, the ion trap can

trap or hold only a certain number of ions before repulsive forces (space-charge) cause

distortions in the applied trapping field, causing a degradation in resolution, a reduction in peak

height, and a shift in mass assignments.91 At severe space-charge conditions, the mass peaks are

further broadened and reduced in peak height to the point where they disappear into the baseline.

Beyond the extreme limit of space-charge, the ion density becomes so large that additional ions

injected into the quadrupole field may not be trapped at all, or previously trapped ions may be

displaced.92 In order to control the number of ions that accumulate in the trap, a method was

developed by Finnigan MAT93 called automatic gain control (AGC). AGC quantitatively

assesses the ion generation rate by use of a prescan (Figure 1-10), and then inversely applies a

period of ionization (ion injection time) during each operational cycle of the ion trap to ensure

that the number of ions in the trap never reaches an adverse level of space-charge.

The AGC feature assists in maintaining the quality of the MALDI spectra by adjusting

the number of laser shots per laser spot to produce a similar number of ions for each scan.94 If

the number of ions produced per shot is low, more laser shots are fired. If the number of ions

produced per shot is high, fewer laser shots are fired. The spectra are normalized so that the

displayed signal reflects the actual signal level.

Helium Buffer Gas

During the ion injection time, ions transmitted from the ion optics are directed into the

LIT where they accumulate before they are scanned out and detected. The ions enter the trap

with a range of kinetic energies. Unless the ions enter the trap at the correct phase angle of the

RF drive potential, they will not have the correct combination of velocity and displacement to

remain in stable orbits and be trapped.95 Even if they meet these conditions, they still enter the









trap with too much kinetic energy to be trapped forever. To remove some of this kinetic energy,

a helium buffer gas is introduced into the trap. The flow of gas (1 mL/min) into and out of the

trap is matched so that the partial pressure of helium in the mass analyzer cavity is maintained at

approximately 1 mTorr. The ions are kinetically cooled to the center of the trap (over a period of

a few milliseconds) through collisions with the low-molecular-weight helium atoms. As a result,

mass resolution is improved, because the ions are ejected from the LIT in dense ion packets.

Resonance Ejection

One of the inherent features of the ion trap during the mass selective instability scan is that

while the ions of lower m/z are being scanned out of the ion trap to the detectors, the higher m/z

ions are still in the trap, and the space-charge that they contribute causes a broadening of the

peaks formed as the lower m/z ions are being ejected.96 This deleterious effect on peak shape can

be reduced dramatically through a technique called resonance ejection. This is performed on the

LTQ by applying a supplementary AC voltage to the X rods in dipolar fashion at a frequency of

just less than half of the RF drive frequency (500 kHz, fi = 0.843) and amplitude at a resonant

ejection qx of 0.88. As ions are scanned along the a = 0 line by ramping the RF voltage on the X

rods, ions of increasing m/z consecutively come into resonance with the resonance ejection

amplitude at qx = 0.88. As the ions come in resonance with the supplementary RF field, the ions

gain kinetic energy and are quickly ejected in a tight pack from the ion trap along the X axis.

The RF voltage at which an ion is ejected from the mass analyzer is defined as its resonance

ejection RF voltage. Without resonance ejection, or with ejection at a q > 0.88, ion motion may

grow in the Y direction, resulting in reduced ejection efficiency through the 0.25 mm-thick slot.

This use of resonance ejection greatly improves mass resolution, and enables the trapping of a

larger number of ions in the ion trap without sacrificing resolution and peak shape, since

resonant ejection is more tolerant of space-charge effects.









Mass Analysis

After ions have been successfully stored in the LIT, a number of different mass analyses

can be performed. The analyses that were used in this research project include single-stage full

scan (MS), and multi-stage full scan (MSn, n = number of stages from 2 to 10). With MS, the

ions formed in the ion source are stored in the mass analyzer, and then are sequentially scanned

out of the mass analyzer to produce a full mass spectrum.

Isolation

MS2 includes two stages of mass analysis. In the first stage, the ions formed in the ion

source are stored in the mass analyzer. The RF voltage is then increased to move the stored ions

towards higher q values until the ion of the m/z of interest (parent ion) is at a q of 0.83 (Figure 1-

10). The parent ion is then selectively isolated and all other ions are ejected from the mass

analyzer. This isolation occurs with the use of a sum-of-sines waveform consisting of many

discrete sine components (5-500 kHz) spaced every 0.5 kHz. Sine components are removed

from the isolation waveform at the secular frequency of the ions to be stored. This isolation

waveform is applied to the X rods of the center section of the LIT in dipolar fashion to isolate a

narrow m/z window (isolation window).96 The isolation waveform applies a resonance ejection

RF voltage at all frequencies corresponding to the secular frequency of the unwanted m/z ions.

The isolation waveform is applied for a period of 16 ms at an amplitude adjusted to assure all

other ions throughout the mass range are resonantly ejected from the trap. Resonance ejection of

an ion occurs when an auxiliary RF field is applied that matches its secular frequency in the X-

direction. The ion absorbs kinetic energy to the point that its magnitude of oscillation increases

along the X axis so that it escapes the trap or collides with one of the center rod surfaces. The

isolation waveform leaves a "frequency notch" around the frequencies corresponding to the m/z

of the parent ion, thus isolating the parent ion in the trap.









Activation

After isolation, the RF amplitude is decreased to move the parent ion to a q of 0.25-0.35

(Figure 1-10). This allows all the product ions formed during collision induced dissociation

(CID) of the parent ion to be trapped. The parent ion is then activated with a resonance

excitation RF voltage, which is applied across the X rods in dipolar fashion (Figure 1-8) for 30

ms at a single frequency corresponding to the secular frequency of the ion to be excited, which is

placed at a q of 0.25-0.35. The resonance excitation RF voltage has lower amplitude than the

resonance ejection RF voltage, and thus is not strong enough to eject an ion from the mass

analyzer. However, with sufficient voltage, the ion gains kinetic energy by resonant absorption,

which results in more translational motion and increased collisions with the helium buffer gas

present in the mass analyzer. After many collisions, the ion gains enough internal energy to

cause it to dissociate into product ions. These fragment ions are then confined within the ion

trap, because of the quadrupolar field except for those fragments that fall below the low mass

cutoff (right edge of the Mathieu stability diagram in Figure 1-9) or those that do not retain a

charge. This process is called collision-induced dissociation (CID).

The amount of energy imparted in the CID process significantly influences the type of

fragmentation induced; it can be increased by either increasing the resonance excitation

amplitude or the time of resonance excitation.96 Within compound classes, the amount of energy

required to fragment an ion is generally proportional to the m/z of the ion. Ions of higher m/z

generally require a greater resonance excitation amplitude or a longer period of time.97 The CID

process can be used to obtain structurally characteristic fragmentation patterns that can be used

to identify selected analytes in complex mixtures. In the second stage of mass analysis, the

product ions are stored in the mass analyzer. They are then sequentially scanned out of the mass

analyzer to produce a full product ion mass spectrum.









MS" on the LTQ can have two to ten stages of mass analysis. For MS3 and higher, the

first two stages are similar to MS2 except that the product ions are not scanned out. Instead,

product ions of one m/z are selected and all other ions are resonantly ejected from the mass

analyzer. The selected product ions now become the new parent ions for the next stage of mass

analysis. With each stage of analysis, the selected parent ions undergo CID to produce product

ions. In the nth stage of mass analysis, the final product ions are stored in the mass analyzer, and

are then sequentially scanned out to produce a full final product ion mass spectrum.

Overview of Dissertation

Quantitative imaging of drugs of abuse and their metabolites in brain tissue using MALDI-

MS could prove to be an invaluable tool in the field of postmortem forensic toxicology. The

purpose of this research was to design a quantitative mass spectrometric imaging method for

determining the regional composition of drugs and their metabolites in postmortem brain tissue.

A MALDI-MS imaging method that combined the use of internal standards for minimizing

signal variability with the high molecular specificity of MSn could provide a visual snapshot for

the forensic toxicologist that reflects the distribution and concentration of drugs of abuse near the

time of death. This information could be used to substantiate fatal overdoses as well as provide

supportive data for neurotoxicity studies.

Current LTQ software allows for only one isolation window in MSn experiments, isolating

one parent mass (or range of masses) for collision-induced dissociation. This means that MSn of

the target ions of the analyte and internal standard would typically be performed with two

separate MSn experiments. This would increase the response variability and could counteract the

signal normalizing effects of using an internal standard. Chapter 2 describes a MALDI-MSn

method developed that allows for a single MS3 experiment that uses a wide isolation window to

isolate both analyte and internal standard target ions. This method is shown to provide improved









precision (-10-20 times reduction in percent relative standard deviation) for quantitative analysis

compared to using two alternating MS3 experiments that separately isolate the target analyte and

internal standard ions. The wide isolation method was used to quantify cocaine in brain tissue of

a human cocaine user.

The wide isolation method is capable of improving precision of MALDI-MSn quantitation

by isolating the analyte and internal standard ions within a single MSn experiment, but it also

isolates other ions that are not of interest that fill up the ion trap and can interfere with detection

of the target ions. Chapter 3 describes another strategy for isolating both the analyte and internal

standard ions within a single MSn experiment without storing unwanted background ions. This

method employs a multi-notch stored waveform inverse Fourier transform (SWIFT) waveform

that is applied to the linear ion trap mass spectrometer for selectively isolating multiple pairs of

analyte and internal standard ions during a single MSn scan. The precision of the multi-notch

SWIFT isolation method for the MALDI-MS2 analysis of cocaine was compared to the

alternating MS2 scan method and the wide isolation method.

Chapter 4 further develops the SWIFT isolation method by incorporating a high mass filter

(HMF) SWIFT excitation waveform to eject high m/z background ions present from endogenous

brain tissue compounds. This two-stage SWIFT isolation method (i.e. HMF and multi-notch

SWIFT) was compared with the wide isolation method for quantification of cocaine and its

metabolites in brain tissue. Quantitative results were then compared with a more conventional

method for the quantification of cocaine in brain tissue that involves homogenate preparation,

followed by solid-phase extraction, and MALDI-MS2 analysis. Chapter 5 offers a conclusion to

the areas examined and provides ideas for future studies. Appendix A illustrates how beta is

calculated, which is used to determine the desired frequency notches of the SWIFT waveforms.









Appendix B contains the C+ program used to calculate the SWIFT waveforms. Appendix C

describes the modifications made to the LTQ instrument, which allow for the application of

SWIFT.









Minor Route:
H N-demethylation CH3
0ICH3 ,


Norcocaine o
Cocair

Principal Route:
Hydrolysis by hepatic and plasma
esterases; loss of benzoyl group
CH3
I O
SHOCH3

Ecgonine methyl ester (EME)


In presence of ethanol:
Transesterification cl


Cocaethylene (CE)


Secondary Rou
Spontaneous hyy
enzymatic); den
CH3
I o
OH

Benzoylecgonine (BE) o




/


te:
drolysis (non-
lethylation


CH3
I o

OH
Ecgonine
Final Degradation Product

Figure 1-1. Metabolism of cocaine (adapted from Goldfrank's Toxicologic Emergencies, 8th
Ed.,New York: McGraw-Hill, 2006). The major route of metabolism is hydrolysis of
cocaine by hepatic and plasma esterases, with loss of a benzoyl group to give
ecgonine methyl ester. The secondary route is spontaneous hydrolysis, probably non-
enzymatic, which leads to benzoylecgonine by demethylation. The final degradation
of cocaine, which is a sequel to both the principal and secondary routes of
metabolism, leads to ecgonine. N-demethylation of cocaine is a minor route leading
to norcocaine.













* Oopai mr*


rBupoHk,
Ir p r'f or IV r

.norm~illy


Lopi~amins
I r.-krrpDrlwJr
1b4akl] by
CA. inn


Figure 1-2. Cocaine's mechanism of action (Human Illnesses and Behavioral Health,
, web accessed on 30 June 2008). Under normal
conditions, dopamine is released by a neuron into the synapse, where it can bind with
dopamine receptors on neighboring neurons. Normally dopamine is then recycled
back into the transmitting neuron by a specialized protein called the dopamine
transporter. If cocaine is present, it attaches to the dopamine transporter and blocks
the normal recycling process, resulting in a build-up of dopamine in the synapse
which contributes to the pleasurable effects of cocaine.


H rib*4 % -,g
D Ld t I














































Figure 1-3. Dopaminergic pathway (www.humanillnesses.com, Web. Accessed on June 30,
2008.). Dopaminergic pathways are neural pathways in the brain which transmit the
neurotransmitter dopamine from one region of the brain to another (e.g., from the
ventral tegmental area to the nucleus accumbens (NAC)).








(a)









(b) ,Brain



--p



Heart
Figure 1-4. Cocaine imaging in tissue (www.invivopharm.com, web accessed on 30 June 2008).
(a) Positron emission tomography (PET) image of [11C]cocaine binding across
species. (b) Whole body autoradiography (WBA) image of rat injected with
[3H]cocaine.




































Figure 1-5. Tissue preparation and MALDI-MS imaging protocol. (A) Tissue sample is freeze-
mounted onto the crytostat stage with deionized water and then cut into 20-lm thick
slices. (B) Tissue slices are thaw-mounted onto glass microscope slide. (C)
Airbrush is used to apply a homogeneous layer of MALDI matrix (D) to the sample.
(E) Sample is then analyzed by moving it step-wise beneath a pulsed laser beam. (F)
A position-specific mass spectrum is produced from every laser spot. (G) Specific
ions are extracted from the mass spectrum using software to generate an image.









N2 Laser
Electronics


Fiber
Optic\


... m :


Sam
Laser beam





quadrupole quadrupol
Sample late


MS with MALDI source
p


ple plate loading

linear ion trap


octopole


S
*
- -
-
- -
S


Figure 1-6. Schematic of the LTQ with MALDI source.67


4111111111111W, T


Samn --le W at e


!

















Back
Section


Exit Slit
30 mm x 0.25 mm


z N,"Front
Section
X


Figure 1-7. Basic design of the two-dimensional linear ion trap.


tz








y

ILZ


DC DC 2 DC3T
Axial Trapping


Radial Quadrupolar Trapping


SGND

AC+ AC-

GND


y
L .-


Radial Dipolar Excitation


Figure 1-8. Scheme for application of DC, RF trapping, and AC excitation voltages necessary
for operation of the 2D ion trap. (A) Separate DC voltages (100 V) applied to the
separate sections of each rod produce an axial trapping field. (B) Two phases of the
primary RF voltage (5 kV rod to ground, 1 MHz) are applied to all the electrodes of
the ion trap to form the quadrupolar field. (C) Two phases of supplemental AC
voltage (+80 V, 5-500 kHz) are applied to only the X rods for isolation, activation,
and ejection of ions.87


A


y

L.











0.25


0.20 0.3
0.3
0.7


0.15 .0.6 3Y

ax 0.5
0.5

S0.6
0.4





0.1
0.0 0.9
o.oo ,i \o-
0.0 0.2 0.4 0.6 0.8
q,


Figure 1-9. Mathieu stability diagram for the linear ion trap. The lines labeled xP and 3y
describe the oscillatory characteristics of ion motion. Solutions to Equations 1-1 and
1-2 give coordinates in ax and qx space that can be mapped onto the above diagram.90













Rf
amplitude



ion
injection

resonance
excite'eject
amplitude

ion
signal


2a 4
/


analytical scan


time


Figure 1-10. A simplified scan function for the quadrupole ion trap showing the prescan and the
analytical scan which makes up one microscan. The four steps of the QIT operation:
ion injection (1), isolation (2), excitation (3), and mass analysis (4) are shown in the
scan function. The prescan mass analysis step (4*) is rapid, because only an ion-
current measurement is required. At step la, the RF amplitude is increased to move
the ions of interest to a q of 0.83 for isolation. At step 2a, the RF amplitude is
decreased to move the isolated ions to a q of 0.25-0.35 for activation.92


n I I









CHAPTER 2
WIDE ISOLATION

Introduction

Concentrations of drugs of abuse found in brain tissue better reflect drug concentrations at

their site of action at the time of death than any other type of specimen used for postmortem

forensic toxicology.5 Conventional quantification of cocaine in brain tissue involves

homogenate preparation, followed by extraction and/or derivatization.98 The extracts are then

usually analyzed by gas chromatography/mass spectrometry (GC/MS), liquid

chromatography/mass spectrometry (LC/MS), GC, or LC. Lengthy extraction procedures are

required to remove large concentrations of lipids and other endogenous materials present in the

brain, which may interfere with analysis.99 Multiple sample pretreatment steps also allow

opportunity for loss of analyte,33 and tissue homogenization eliminates spatial information,

which could provide histologically-specific drug distribution. Attempts have been made to

determine the regional distribution of cocaine in postmortem brain of chronic human cocaine

users.8, 12,98 These analyses were performed on sections of- 100-200 mg of tissue from different

regions of the brain, which were assumed to be homogeneous and accurately representative of

the drug concentration in that excised region.

Direct MALDI-MS analysis of intact tissue can provide quantitative information about

the distribution of cocaine in human brain more rapidly, with higher spatial resolution, and with

less sample loss than drug analysis methods that involve tissue homogenization. Furthermore,

the distribution of cocaine in brain tissue acquired by MALDI-MS can be directly related to the

histology. The majority of MALDI-MS instruments use a time-of-flight (TOF) mass analyzer,

which has benefits of high mass range and high throughput. Quantitative MALDI-MS is

challenging, however, because MALDI exhibits irreproducible signal intensities due to









inhomogeneous crystal formation, inconsistent sample preparation, and laser shot-to-shot

variability. A typical MALDI-TOF experiment will obtain 200-1000 consecutive mass spectra at

each sample spot (one laser shot per spectrum), which are averaged to improve the

reproducibility of the MALDI signal.27 Similarly, MALDI-MS instruments that utilize a linear

ion trap (LIT) mass analyzer can obtain multiple mass spectra at each spot and average them to

improve reproducibility.85 Here we obtain a single mass spectrum at each spot, with typically 10

laser shots used to fill the ion trap for each spectrum. Note that any ion trap has a finite ion

storage capacity before space-charge reduces resolution and causes peak shifts.96 MALDI-LIT

instruments can minimize space-charge effects by utilizing automatic gain control (AGC),96

which automatically controls the number of laser shots used to fill the trap (typically 1-20 shots)

to optimally fill the ion trap for maximum signal without loss of mass resolution. Laser power

can also be optimized along with choice of matrix compound to maximize analyte signal while

avoiding space-charge effects. The use of internal standards for quantitative MALDI-MS has

been shown to improve signal stability, if the solution-phase properties are carefully matched as

in an isotopic standard.7

Quantification of small drug molecules like cocaine using MALDI-MS is further

complicated by the presence of interfering matrix peaks in the low mass range along with ions

that may be produced from endogenous compounds present in the brain tissue.33 One of the

strengths of a linear ion trap mass spectrometer is its ability to perform multiple stages of mass

analysis (MSn) to significantly increase the selectivity for the analyte of interest. A MALDI-MSn

method could be developed to remove interference from both MALDI matrix and the complex

sample environment of brain tissue; however, a problem arises when trying to combine the use of

MSn with the use of internal standards. Instrument software allows for only one isolation









window (IW) in MSn experiments, isolating one parent mass (or range of masses) for collision-

induced dissociation (CID). This means that MSn of the target ions of the analyte and internal

standard would typically be performed with two separate MSn experiments. This would increase

the response variability and could counteract the signal normalizing effects of using an internal

standard. In contrast, using a 6-Dalton (Da)-wide IW centered at a mass-to-charge (m/z) between

the [M+H] ions of cocaine and its trideuterated analog allows for isolation and CID of both ions

during a single MSn experiment. This single isolation method reduces the signal variability

inherent with MALDI compared to isolating each ion individually with a 1-Da IW (in two

alternating MSn experiments). This method is used here to detect and quantitatively image

cocaine in postmortem human brain tissue.

This study demonstrates that MSn increases selectivity, which is critical for

differentiating analyte ions from matrix ions and endogenous compounds found in brain tissue.

It is also shown that the use of internal standards corrects for signal variability in quantitative

MALDI arising from inhomogeneous crystal formation, inconsistent sample preparation, and

laser shot-to-shot variability. Using a single MSn experiment with a wide IW to isolate both

analyte and internal standard target ions provides improved precision (10-20 times reduction in

%RSD) for quantitative imaging studies compared to using two alternating MSn experiments that

isolate the analyte and internal standard target ions separately.

Experimental

Chemicals

Cocaine (COC; MW 303.4 Da) and COC-d3 (MW 306.4 Da, 0.29% do) were purchased

from Cerilliant (Round Rock, TX, USA) at concentrations of 1 mg/mL and 100 tg/mL,

respectively, in acetonitrile. High-performance liquid chromatography (HPLC)-grade

acetonitrile, methanol, and water were purchased from Fisher Scientific (Pittsburgh, PA, USA).









Working standards of COC and COC-d3 were diluted with acetonitrile and then stored at 4 oC.

COC calibration standards were prepared in acetonitrile at concentrations of 5.0, 2.5, 1.25, 0.625,

0.312, 0.156, 0.078, 0.039, 0.020, 0.010, and 0.005 pg/mL with the COC-d3 internal standard at a

concentration of 2.0 [g/mL. Sinapinic acid (SA; MW 224.2 Da), 2,5-dihydroxybenzoic acid

(DHB; MW 154.1 Da), and a-cyano-4-hydroxycinnamic acid (CHCA; MW 189.2 Da) were

purchased from Acros Organics (Geel, Belgium). Saturated matrix solutions (40 mg/mL DHB,

10 mg/mL SA, and 10 mg/mL CHCA) were prepared in methanol/water (70:30, vol/vol) on the

day of use.

Tissue Collection

Human brain tissue samples were provided by the El Paso County Coroner's Office in

Colorado Springs, CO. Postmortem brain material was excised from the nucleus accumbens

(NAc) from case number 07A-369, whose toxicological analysis indicated the presence of

cocaine in blood at 69 ng/mL (COC concentration in the brain tissue was not quantified). The

NAc is a dopamine-rich area of the striatum, which may contain an accumulation of COC due to

its affinity to bind with the dopamine transporter.100 At autopsy, the excised tissue was

immediately snap-frozen in liquid nitrogen and then stored in a -80 C freezer until analyzed.

Tissue Sectioning and Sample Preparation

Frozen brain tissue was cut into thin sections (20 pm thickness) in a cryostat (HM 505E;

Microm International GmbH, Waldorf, Germany) at -25 C. The tissue samples were frozen to

the cryostat sample stage using distilled water. Serial brain sections were collected onto

microscope slides where they were thaw mounted and then stored at -80 C. Before mass

spectrometric analysis, the tissue sections were removed from the freezer and placed in a vacuum

dessicator for 30 min before spiking standards (1-lIL droplets by micropipet) and applying

MALDI matrix. The matrix was applied to the tissue sections using an artistic airbrush (Aztek









A470; Testors, Rockford, IL, USA). The application of MALDI matrix by airbrush has been

previously published.67 Matrix was applied using the dried-droplet method for experiments

performed on MALDI plate.

Mass Spectrometry

Mass spectra were acquired using an LTQ linear ion trap with a vMALDI ion source

(Thermo Finnigan, San Jose, CA, USA), equipped with a nitrogen laser (337 nm) at a frequency

of 20 Hz and 100-tm spot size. A more detailed description of this instrument has been

published.67 An average of 10 laser shots per scan was used to produce mass spectra, except for

experiments that used AGC, in which the number of laser shots was automatically varied to

optimally fill the trap with ions, thus avoiding space charge-related peak broadening and mass

shifts. AGC assess the ion generation rate by use of a prescan, and then adjusts the number of

laser shots per scan to produce a similar number of ions for each scan. The spectra are

normalized to the number of laser shots for each scan.

Resonance excitation is used for isolation, activation, and mass analysis. For MSn

experiments, unwanted ions are resonantly ejected from the ion trap by applying a 5-500 kHz

multi-frequency isolation waveform consisting of sine components spaced every 0.5 kHz. The

ions of interest are isolated by removing sine components from the isolation waveform that

correspond to the secular frequency of the desired ion(s). Ions are selected for isolation in the

LTQ software by entering the m/z with its IW. The mass range for the ion is defined as (m/z -

IW/2) to (m/z +IW/2). The IW should be narrow enough to minimize including interfering

peaks, but wide enough to avoid loss of sensitivity for the desired ion(s). However, it is

important to note that the activation width for resonance excitation (CID) has the same value as

the IW. Therefore, the collision energy applied during MSn is spread over the activation width.

Thus, increasing the IW decreases the true collision energy for each ion.









The tissue-mounted microscope slides were affixed to a slide holder plate with double-

sided tape. The plate was then inserted into the LTQ, and the plate was rastered beneath the laser

spot at 100-iim steps to produce position-specific mass spectra. Specific ions and the total ion

current (TIC) signal were extracted from the raw data files using ImageQuest version 1.0

(Thermo Fisher Scientific, San Jose, CA, USA), which was used to generate an image.

Results and Discussion

MS2 and MS3 Mass Spectra of COC and COC-d3

DHB was selected as the MALDI matrix in this study, as preliminary investigations

showed that it produces more efficient ionization for COC at low concentrations than SA or

CHCA. DHB was also preferred as the matrix for COC analysis due to its lack of interference

with the [M+H] ion of COC (m/z 304) and COC-d3 (m/z 307). The COC standards in

acetonitrile were characterized by MSn. The MS2 spectrum of m/z 304 and 307 (IW = 1.0 Da,

CID = 20) each show one major product ion, corresponding to a neutral loss (NL) of benzoic

acid (NL 122) at m/z 182 and 185, respectively. MS3 was performed on the product ion signal at

m/z 182 of COC (IW = 1.0 Da; CID = 30), resulting in product ions at m/z 150 (NL of 32;

CH3OH), m/z 122 (NL of 60; CH3OH + CO), m/z 119 (NL of 63; CH3OH + CH3NH2), m/z 108

(NL of 74; CH3OH + CH2CO), m/z 91 (NL of 91; CH3OH + CH3NH2 + CO), and m/z 82 (NL of

100; CH3OH + C4H40 via a 6-electron Alder ene rearrangement). The structures of the fragment

ions of the [M+H]+ ion of COC and its proposed fragmentation pathway (Figure 2-1) have been

previously published.101 MS3 was performed on the product ion signal at m/z 185 of COC-d3

(IW = 1.0 Da; CID = 30) resulting in product ions at m/z 153 (NL of 32; CH3OH), m/z 125 (NL

of 60; CH3OH + CO), m/z 119 (NL of 63; CH3OH + CH3NH2), m/z 111 (NL of 74; CH3OH +

CH2CO), m/z 91 (NL of 91; CH3OH + CH3NH2 + CO), and m/z 85 (NL of 100; CH3OH +

C4H40). The m/z values of the fragment ions of COC-d3 at m/z 91 and 119 are the same as those









for COC because these ions have lost the trideuterated tag that was originally located on the N-

methyl group.

Improving Signal Reproducibility with Internal Standards

Quantitative analysis by MALDI is challenging, because of signal irreproducibility due to

variation in sample preparation, inhomogeneous co-crystallization of analyte and MALDI

matrix, and laser shot-to-shot variability. Figure 2-2a shows the m/z 304 signal of the [M+H]

ion of COC detected from COC/COC-d3 standard solutions spotted 1 iL each in triplicate onto a

MALDI plate with 1 iL of DHB matrix pippeted on top. The COC/COC-d3 solutions were

composed of different concentrations of COC (5.0, 2.5, 1.2, and 0.63 ag/mL) mixed with 1.0

1ig/mL of COC-d3. The histogram shows the high variability in signal for each concentration

with %RSD ranging from 29 to 67%, making it difficult to distinguish signal from one

concentration to another. Figure 2-2b shows the m/z 304 signal of COC normalized to the

[M+H]+ ion signal of COC-d3 at m/z 307. Signal variability was reduced dramatically (%RSD

ranged from 0.26 to 1.33%) by normalizing the analyte signal to that of the internal standard

making quantification by MALDI possible.

Increasing Analyte Selectivity with MS"

Figure 2-3a shows a full-scan MS spectrum of a 1.25:1 mixture (by mass) of COC and

COC-d3 standards spiked (1 aIL of 1.25 1ig/mL and 1.0 1ig/mL, respectively) onto a MALDI plate

with DHB matrix. Peaks at m/z 304 and 307 represent the [M+H] ions of COC and COC-d3,

respectively. A number of cluster ions, fragment ions, and a molecular ion of DHB are also

present, including m/z 137 [DHB+H-H20] m/z 154 [DHB] m/z 177 [DHB+Na] m/z 199

[2DHB+Na] m/z 221 [DHB-2H+3Na] m/z 273 [2DHB+H-2H20] m/z 291 [2DHB+H-H20]+,

and m/z 331 [2DHB+Na]+. Figure 2-3b shows a full-scan MS spectrum of a 1:1 mixture (by









mass) of COC and COC-d3 standards spiked (1 tL of 1.0 alg/mL each) onto a 20-lam thick

human brain tissue slice with DHB airbrushed. The [M+H]+ ions of COC and COC-d3 are

observed at m/z 304 and 307, respectively. The same cluster ions, fragment ions, and molecular

ion of DHB are present, in addition to numerous ions of endogenous compounds from the brain

tissue, including the phosphocholine head group of phosphatidyl choline at m/z 184

[(CH3)3NCH2CH2PO4H]+. Identification of COC and COC-d3 on the MALDI plate and brain

tissue was confirmed by characteristic MS2 product ions at m/z 182 and 185, respectively.

MS2 spectra of m/z 304 with COC spiked at concentrations below 5 ng/mL on plate and

on tissue revealed an isobaric compound that has product ions at m/z 212 and 91. The isobaric

ion likely originates from the surfactant, benzyldimethyldodecylammonium chloride

(C12BAC).102 The widespread use of C12BAC and other BACs as disinfectants makes it a likely

trace contaminant in the laboratory. The ion at m/z 212 results from fragmentation of the carbon-

nitrogen bond between the toluyl substituent and the quaternary amine (Figure 2-4). The m/z 91

ion is a stable tropylium ion formed by fragmentation in which the toluyl substituent retains the

positive charge (Figure 2-4). MS2 of m/z 304 with COC spiked at concentrations below 5 ng/mL

also results in the detection of product ions of isobaric compounds at m/z 256 and 286. These

ions have not yet been identified, but are not present when DHB has been characterized on

MALDI plate alone.

The presence of isobaric ions in samples increases with sample complexity and may

interfere with quantification at low analyte concentrations. MSn can improve analyte selectivity

and produce higher signal-to-noise ratios, resulting in lower detection and quantification limits

for the analyte. Combining the use of MSn with internal standards is commonly performed by

alternating MSn scans of the analyte and the internal standard ions, and then ratioing the resulting









product ion signals. This method is effective for use with ionization techniques such as

electrospray and atmospheric pressure chemical ionization; however, due to the shot-to-shot

variability of MALDI, acquiring analyte and internal standard signals in alternating MSn

experiments may counteract the signal normalizing effects gained by using an internal standard.

Combining Internal Standards with MS" using a Wide Isolation Window

One method for combining the use of internal standards with MSn is to perform MS2 on

the analyte and internal standard ions separately during alternate MSn experiments illustrated in

Figure 2-5a. The [M+H] ion of COC (m/z 304) is isolated with a 1 Da window and then

collisionally activated to produce the product ion at m/z 182 shown in the MS2 spectrum in

Figure 2-5b. In a separate MS2 scan, the [M+H]+ ion of COC-d3 (m/z 307) is isolated with a 1

Da window and CID is applied, resulting in the product ion at m/z 185 shown in the MS2

spectrum in Figure 2-5c. The analyte ion signal at m/z 182 can then be normalized to the internal

standard ion signal at m/z 185. An alternative approach to using two separate MSn experiments

is to use a single wide isolation window (e.g., 6-Da) centered at m/z 305.8, shown in Figure 2-5d,

allowing the simultaneous isolation and CID of the [M+H] ion of COC (m/z 304) and COC-d3

(m/z 307). The resulting MS2 spectrum, shown in Figure 2-5e, contains the product ions of COC

and COC-d3 at m/z 182 and 185, respectively.

The performance of the MSn experiment using a single wide isolation window was

compared with that using two alternating MSn experiments by detecting COC and COC-d3

spiked on top of human brain tissue. Figure 2-6a shows a microscope image of a 20-lim thick

human brain tissue slice with COC/COC-d3 solutions spotted 1 atL each in triplicate (A, B, and

C) on the surface of the tissue and then airbrushed with DHB. The five COC/COC-d3 solutions

spotted all contained 2.0 lg/mL of COC-d3 in addition to 0.31, 0.62, 1.2, and 5.0 lg/mL of COC,









respectively. The compositions of the solutions spotted (1-5) are shown in the table below the

image. The average dried spot size was 0.25 cm in diameter. Figure 2-6b shows the MS2

product ion image of m/z 305.8 (IW = 6 Da, CID = 20) of the entire tissue slice generated from

signal extracted from the mass range m/z 182-186 and normalized to the TIC. Higher signal

intensity correlates with the darker shade of gray, illustrating how the COC and COC-d3 co-

crystallize along with the DHB towards the edge of each spot. The LTQ software was used to

outline each spot to be analyzed. Each spot was analyzed twice: first by performing MS2 ofm/z

304 (IW = 1 Da, CID = 20) followed by MS2 ofm/z 307 (IW = 1 Da, CID = 20), and then by

MS2 of m/z 305.8 (IW = 6 Da, CID = 20). For each analysis, all of the spectra (-500 scans) were

averaged for each spot, and the m/z 182 signal for COC was normalized to the m/z 185 signal for

COC-d3 and plotted against the concentration of COC spiked to produce two different calibration

curves shown in Figure 2-7. Figure 2-7a shows the average ratio of peak intensities m/z 182 to

m/z 185 as a function of the spiked COC concentration for alternating MS2 experiment (i.e., MS2

ofm/z 304 in one scan and then MS2 ofm/z 307 the following scan). The line of best fit was y =

0.68(0.07)x + 0.2(0.2) over the range 0.31-5.0 lg/mL with a standard error of the estimate

(SEE) = 0.2833; the %RSD ranged from 12% to 30%. The 95% confidence intervals for the

slope and y-intercept were 0.44 to 0.91 and -0.4 to 0.9, respectively. Figure 2-7b shows the

average ratio of peak intensities m/z 182 to 185 as a function of the spiked COC concentration

for a single MS2 experiment with a wide 6-Da isolation window centered at m/z 305.8 (i.e., MS2

ofm/z 304 and m/z 307 in one scan). The line of best fit was y = 0.492(0.001)x +

0.023(0.003) over the range 0.31 to 5.0 lg/mL with an SEE = 0.0052; the %RSD ranged from

0.50% to 5.1%. The 95% confidence intervals for the slope and y-intercept were 0.488 to 0.496

and 0.011 to 0.034, respectively. Precision was dramatically improved by using the single MS2









experiment with 6-Da wide isolation window compared with isolating each ion individually with

a 1-Da window (two alternating MS2 experiments). There was a 10-20 times reduction in %RSD

and a 50 times reduction in SEE by using the wide isolation method.

Isolation Window Width and Automatic Gain Control

Usually the smallest isolation width is desired for MSn experiments performed with an

ion trap mass spectrometer to avoid isolating unwanted background ions and reducing analytical

specificity. The minimum acceptable ion isolation width is defined as the lowest range

providing no appreciable signal attenuation of the analyte and internal standard ions when

compared to a wider setting. Signal attenuation can result either from losses during the

resonance ejection step which is used to remove masses below and above the selected m/z range,

or from decreased CID efficiency of the analyte and internal standard ions.

The effect of isolation width on the intensity of the product ions of the [M+H] ions of

COC (m/z 304) and COC-d3 (m/z 307) together in a single MS2 scan was investigated. Five

solutions of COC and COC-d3 were prepared at equal concentrations and diluted with

acetonitrile (0.12, 0.25, 0.50, 1.0, and 2.0 lg/mL). All five solutions were spotted in triplicate 1

ItL each onto a MALDI plate followed by 1 ItL of DHB matrix. For all MS2 experiments, the

parent ion was set to m/z 305.8, the center of the mass range between m/z 304.3 and m/z 307.3,

and the CID was set to 20. The size of the isolation window width centered at m/z 305.8 was

varied (4 Da, 6 Da, and 8 Da), and the ratio of the intensities of the products ions at m/z 182 and

m/z 185 for COC and COC-d3, respectively, were observed. It is important to note that MS2 of

m/z 304 (IW = 1.5 Da) produced a negligible amount of m/z 185 (<0.0005%), the product ion of

m/z 307. Also, MS2 of m/z 307 (IW = 1.5 Da) produced a negligible amount of m/z 182

(<0.002%), the product ion of m/z 304. The expected signal ratio of COC to COC-d3 is 1.02 for









equal masses based on a calculated molar ratio of 1.01 corrected for the isotopic purity of COC-

d3 (0.29% COC-do). The measured signal ratio of m/z 182 to 185 was approximately equal to 1

for concentrations below 0.50 alg/mL, but the ratio increased at concentrations above 0.50 alg/mL

(i.e., the m/z 185 signal decreased with respect to m/z 182). It was also observed that the signal

ratio of m/z 182 to 185 was higher for a 4 Da isolation window (2.07 at 1.0 alg/mL and 3.50 at

2.0 lg/mL) compared with the 6 Da (1.18 at 1.0 lg/mL and 1.68 at 2.0 lg/mL) and 8 Da (1.05 at

1.0 [tg/mL and 1.81 at 2.0 [tg/mL) isolation windows widths. This suggests that either some of

the m/z 307 ion is being lost during isolation or that the m/z 307 ion is being less efficiently

excited during the CID step when narrower IWs are used.

An effort was made to separate the isolation step from the CID step of the MS2

experiment to better understand the effect of isolation window width on the signal intensities of

the MS2 product ions of the [M+H] ions of COC and COC-d3. The above experiment was

repeated on the five COC/COC-d3 solutions, except that no CID voltage was applied so that the

ions at m/z 304 and 307 were isolated but not fragmented. The ratio of intensities of m/z 304 and

m/z 307 were then monitored for different isolation window widths (4, 6, and 8 Da). Results

showed that the signal ratio of m/z 304 and 307 remained approximately equal to 1 for

concentrations 0.12 2.0 alg/mL for isolation widths of 6 and 8 Da; however, the signal ratio

steadily increased for a 4 Da isolation window at concentrations above 0.50 alg/mL. The

increase in the signal ratio of m/z 304 to 307 (i.e., m/z 307 signal decreased with respect to m/z

304) at higher concentration is presumably due to a mass shift of m/z 307 outside the isolation

window, resulting in resonance ejection of some of the m/z 307 ions. This mass shift could be

caused by space-charge effects at higher ion populations in the ion trap, and may be corrected by

using AGC. The experiment was repeated again, comparing the signal ratio of m/z 304 to 307









with and without AGC with a 4 Da isolation window and no CID applied. Results showed that

when AGC was used, the signal ratio of m/z 304 to 307 remained approximately equal to 1 for all

concentrations analyzed (0.12 2.0 ag/mL), indicating that AGC can minimize space-charge

effects, which may lead to ejection of the higher m/z ion when a narrower isolation width is

employed.

Quantification of Cocaine in Postmortem Human Brain Tissue

The MS3 wide isolation method developed for COC was applied to human brain tissue

from a subject whose toxicology report showed the presence of COC. The MS2 product ion of

the [M+H] ion of COC at m/z 182 was not distinguishable from the background signal;

therefore, an MS3 wide isolation method was developed to increase selectivity. The MS3 wide

isolation method was evaluated by spotting 1 iL of a 4.0 ag/mL solution of COC and COC-d3

onto a MALDI plate followed by 1 iL of DHB matrix. The method involves centering a 6-Da

isolation window at m/z 305.8 and applying a CID of 20 followed by a 6-Da isolation window

centered at m/z 183.5 (between COC and COC-d3 product ions at m/z 182 and 185) with a CID

of 30. The resulting MS3 product ion spectrum revealed characteristic fragment ions of COC at

m/z 150, 82, 108, 122, 119, and 91 and for COC-d3 ions at m/z 153, 85, 111, 125, 119, and 91.

The MS3 wide isolation method was applied to unspiked brain tissue from a cocaine user, and

COC was detected and confirmed by matching all six of these MS3 ions. The relative intensities

of the five most intense fragment ions (all but m/z 91) were within 12% of the standard fragment

ion intensities.

Figure 2-8 shows the MS3 product ion image of m/z 305.8 (IW = 6 Da,CID = 20) of the

entire tissue slice generated from signal extracted from the mass range m/z 150 151 and

normalized by the TIC. The image shows no localization of COC in the section of the nucleus









accumbens analyzed. Browne et al.98 analyzed 1 g samples from different regions of 3 human

brains by solid-phase extraction (SPE) and LC. From these studies, cocaine and

benzoylecgonine were found to be distributed throughout the different regions of the brain.

However, significant differences in the concentration of cocaine were apparent in different

regions of the brain (e.g., cocaine concentration was higher in the basal ganglia than the section

of the cerebellum analyzed). These findings were consistent with other brain cocaine

distribution studies, which reported that concentrations of cocaine and of its metabolites showed

little regional heterogeneity in postmortem brain of chronic users of cocaine.8' 12 The

homogeneous distribution of cocaine and its metabolites in specific regions of the brain may be a

result of the high concentrations typical of behavioral usage.

Before quantifying unspiked COC in human brain tissue with the MS3 method, it was

necessary to show that the response factors for COC and COC-d3 were equal, so that the

calibration curve of COC-d3 could be used. A series of 1:1 solutions of COC and COC-d3 at

various concentrations (0.03, 0.06, 0.13, 0.25, 5.0, 1.0, and 2.0 alg/mL) were prepared and spiked

in triplicate, 1 pL each, on top of serial tissue sections, and then DHB matrix was airbrushed

over the tissue slices. Each spot was analyzed using the MS3 wide isolation method, and the m/z

150 signal from COC was plotted versus the m/z 153 signal from COC-d3. The slope of the plot

was 1.062 0.002 with a correlation coefficient r2 = 0.99,998 over the concentration range 0.03

to 2.0 alg/mL. The 95% confidence interval for the slope was 1.057 to 1.066. The expected

slope based on a molar ratio of 1.01 and an isotopic purity for COC-d3 of 0.29% do is 1.02,

which means that COC has a 4% higher response factor than COC-d3 over the concentration

range measured.









The MS3 wide isolation method was used to quantify the unspiked COC that was detected

in the postmortem human brain tissue. The calibration curve used for quantification (Figure 2-8)

was created by imaging three different concentrations of COC-d3 (0.06, 0.13, and 0.25 alg/mL)

were spiked (1 |iL) onto a glass slide before thaw mounting a 20 ilm-thick brain tissue slice on

top and airbrushing DHB matrix. All three spots were then analyzed using the MS3 wide

isolation method. Approximately 2000 scans were acquired to image the entire area of each of

the spots (average area = 0.17 cm2). The m/z 153 signal from each spot was used to develop a

calibration curve that resulted in a line of best fit ofy = 399(27)x 17(4) (Figure 2-8). COC-

d3 was shown to have a linear response with increasing concentrations spiked underneath tissue.

Since the MS3 wide isolation method analyzes both COC and COC-d3 simultaneously, unspiked

COC was detected from each spot analyzed at m/z 150, and the corresponding signal was plotted

(0) alongside each corresponding COC-d3 signal (o) in Figure 2-9. An area of the tissue (500

MS scans) that was not spiked with COC-d3 was analyzed using the MS3 wide isolation method

and the acquired m/z 150 signal was averaged with the m/z 150 signals from the spiked COC-d3

spots, resulting in a very trace signal of 29 1 counts (highlighted as dashed line on Figure 2-8).

Assuming that the amount of unspiked COC extracted from the tissue has a 1:1 response with the

COC-d3 spiked on top of tissue, the calibration curve for COC-d3 can be used to quantify the

amount of COC present in the analyzed tissue. From the equation of the line, it was determined

that COC was present at a level equivalent to 0.12 + 0.01 [g/mL.

Using the 1 [iL volume of COC-d3 spiked underneath tissue, it is calculated that the mass

of COC present is 1.2 x 10-4 Ig. Given that the area of an analyzed spot on tissue was 0.17 cm2

and that the tissue thickness was 20 im (2.0 x 10-3 cm), the volume of tissue from which COC

was extracted was 3.4 x 10-4 cm3. The mass of the tissue is 3.4 x 104g (density of wet tissue









-1.0 g/cm3), resulting in an absolute concentration of COC detected in this area of the

postmortem brain tissue of 0.35 pg/g (350 ppb).

The MALDI-MS method has a smaller sample requirement (-100 .g tissue) and less

sample preparation than conventional GC/MS techniques, which require 1000 to 10,000 times

more sample (0.1 to 1.0 g of brain tissue) to be homogenized before solid-phase extraction and

GC/MS analysis.8' 12 The GC/MS method developed by Kalasinsky et al.12 reported a limit of

detection of 0.1 ng/mL for the analysis of COC in brain tissue. COC was detectable at 30 ng/mL

with the MALDI-MS3 wide isolation method developed here. Although the MALDI-MS3 wide

isolation method is not as sensitive as the GC/MS method (primarily because it uses a 1000 times

smaller sample), it readily detects cocaine at a level an order of magnitude below the lowest level

(300 ng/mL) reported for COC detected by GC/MS analysis of 15 autopsied brain regions of 14

human chronic cocaine users.12

Conclusions

It has been demonstrated that MS2 and MS3 increase selectivity, which is critical for

differentiating analyte and internal standard ions from matrix ions and endogenous compounds

found in brain tissue. It has also been shown that the use of internal standards corrects for signal

variability during quantitative MALDI. A method was developed that allows for a single MS2

experiment that uses a wide isolation window to isolate both analyte and internal standard ions.

This method was shown to provide improved precision (~ 10-20 times reduction in %RSD) for

quantitative analysis of COC in postmortem brain tissue compared with using two alternating

MS2 experiments that isolate the analyte and internal standard target ions separately. When COC

concentration is too low to distinguish the MS2 product ion at m/z 182 from the background, the

MS3 wide isolation method can be applied to increase selectivity.









The wide isolation window developed for the analysis of COC could be applied to

quantitative MALDI-MSn imaging of other drugs of abuse and their metabolites in brain tissue,

which could prove to be an invaluable tool in the field of postmortem forensic toxicology. A

MALDI-MS imaging method that combined the use of internal standards for minimizing signal

variability with the high molecular specificity of MSn could provide a visual snapshot for the

forensic toxicologist that reflects the true distribution and concentration of drugs of abuse at the

time of death. This information could be used to substantiate fatal overdoses as well as provide

supportive data for neurotoxicity studies.












m/z 304


CH3 CH3
I I
SN -C4H40


m/z 82



m/z

m/z 91


m/z 119 m/z 122 m/108
m/z 122 m/z 108


Figure 2-1. Cocaine dissociation pathway.101










5.0E+06
(a) E Run 1
4.5E+06 Run 2
SRun 3
4.0E+06

O 3.5E+06
Mean MMean ean Mean
1.9E+06 2.4E+06 1.8E+06 6.0E+05
3.0E+06 RSD % RSD o RSD % RSD
32 67 29 61
2.5E+06

2.0E+06

S1.5E+06


1.OE+06
AhIdIIH -


5.00 2.50 1.25 0.63
COC Concentration (pg/mL)


(b) Run 1
(b)
E Run 2
5.00 Run 3



4.00 Mean Mean Mean Mean



3.00



2.00



1.00



0.00
5.00 2.50 1.25 0.63
COC Concentration ([pg/mL)


SMALDI-MS signal variability with and without internal standards. Signal of m/z
304, [M+H] of COC (a) and m/z 304 signal ratioed to m/z 307 signal, [M+H] of
COC-d3 (b). All solutions spotted 1 gL in triplicate on MALDI plate with DHB
matrix. The internal standard (COC-d3) was maintained at 1 ag/mL for all solutions.


5.0E+05

0.OE+00




Figure 2-2











[COC+H]
304.3


[2DHB+H-2H20]
273.2


[DHB-2H+3Na]4
[DHB] [2DHB+Na]+
154.2

[DHB+Na]+
177.2 1
199.2 221.2
1 8 4 .2 ....... ........ ........... .... ..........................
184.2 [(CH3)3NCH2CH2PO4H]+


175.2
154.2


I 291.1


[COC-d3+H]
30






-[2DHB+H-H20]


[2DHB+Na]
331.0
I I, .


231.1


222.Z


214.1

I II. .,I ,L .l .


[COC+H]
304.3
S[COC-d3+H]
307.3
273.2 296.2 3132
|240 37 3.5-2 33 4
,, ,, .... .. 1 .1 .


Figure 2-3. Comparing mass spectra of COC and COC-d3 on MALDI plate and on brain tissue.
MALDI mass spectrum of (a) a solution of COC (1.25 plg/mL) and COC-d3 (1.0
plg/mL) spotted (1 piL) with DHB matrix on MALDI plate (run 1 from Figure 1) and
(b) a solution of COC and COC-d3 (1.0 plg/mL each) spiked (1 pIL) on postmortem
human brain tissue with DHB matrix airbrushed.


100 -


80


[DHB+H-H20]
137.2


104.3


1,





1323
114 3
I ..... .


100 150 200


I1 1 1 .I11 HI IIIII I


250 300 350


II"hdll "' i '' d'J" jd"l"i"ji J'


I


III


I I









/ (CH2),CH3


CH3 -N -CH3
3 ^ v- 3


,1


4%4


z 304 (C1, 332

m/z 304 (CiI), 332 (C14)


S(CH,),CH3
CH3 -+N =CH2
m/z 212 (C,), 240 (C,)



/(CH2),CH3
CH. N CH3


CH3




+
6CH2


-6


mz 91


50 100 150 200 250 300 350
m/z
Fragmentation of the benzyldimethyldodecylammonium ion.102


Figure 2-4.












1Da
[COC+H]


1Da

[COC-d3+H]


I [B I .......
03 304 35 306 307 308 309 310
3Da 3Da
3fl72 3fl5. 8 n


182 2


185 3

(c)




307 2

100 150 200 250 300 350
m/z
(e)
1822
: 2



o -
o

o


o.^


Figure 2-5. Wide isolation MALDI-MS2. COC (1.0 lg/mL) and COC-d3 (1.0 lg/mL) spotted (1
|iL) with DHB matrix on MALDI plate. (a) Individual isolation (1 Da) and
collisional activation of m/z 304 and m/z 307 with resulting MS2 spectra of the
[M+H]+ ions of(b) COC at m/z 304 and (c) COC-d3 at m/z 307. (d) Simultaneous
isolation and CID activation ofm/z 304 and m/z 307 with a 6-Da window centered at
m/z 305.8 and (e) the resulting MS2 spectra containing both COC and COC-d3
fragment ions at m/z 182 and m/z 185, respectively.























69





















Solution # (Triplicate Runs A, B, C)


SD 10 20 30 40 SO 60 7 80 go WO







-2 3 4 5-
1 2 3 4 5


COC (pg/mL) 5.0 2.5 1.3 0.62 0.31
COC-d3(pg/mL) 2.0 2.0 2.0 2.0 2.0


Figure 2-6. Images of standards spiked on brain tissue. (a) Photomicrograph of 20 [tm thick
human brain tissue mounted on slide with COC/COC-d3 solutions spiked (1 |iL) in
triplicate (A, B, and C) on top of tissue and then airbrushed with DHB matrix. (b)
MS2 product ion image generated from signal selected from mass range m/z 182-186
and normalized by the TIC.









































00 10 20 30 40 50 60
Cocaine Concentration (tLg/mL)

(b)
30



n0 1 ._2 A 1.5 A
i 25
2 D Is2 Is5B.... ..

Ils Is C
20 -- -- \"" .....


o R- = 1 .


15 ---------------- ---- ----------
15

05" --------
'COC(' C'in ,, RSD


.. ....Y..... It.......
&I "* -r "''





00
0 0 ------- ------------------------ -------------
00 10 20 30 40 50 60
Cocaine Concentration (tg/mL)

Figure 2-7. Calibration curves for alternating scans MS2 and wide isolation MS2. Peak intensity
ratio of m/z 182 to m/z 185 versus COC concentration for two alternating MS2
experiments (a) and for a single MS2 experiment using a 6-Da isolation and activation
window.
































Figure 2-8. Mass spectrometric image of cocaine in brain tissue. The MS3 product ion image of
m/z 305.8 (IW = 6 Da, CID = 20) of the entire tissue slice generated from signal
extracted from the mass range m/z 150 151 and normalized by the TIC.
















70- O*UnspikedCOC


S60 y= 399x 17
S R2- 0.9955

3M 50
















COC-d3 Concentration (Ctg/mL)
40 .-p rtL












of COC at z 150 and plotted against the calibration curve of COC- to quantify the
20 amount of COC present in tissue.

10 -- -
-- -- -- --- --ni -
0 -- -- -- -- -- I | -- I I -- - -

0.00 0.05 0.10 0.15 0.20 0.25 0.30
COC-d3 Concentration (tg/mL)

Figure 2-9. Cocaine quantification. Calibration curve created from plotting signal of MS3
fragment ion of COC-d3 at m/z 153 versus COC-d3 concentration spiked beneath 20
1tm tissue from cocaine user. Unspiked COC was detected by the MS3 fragment ion
of COC at m/z 150 and plotted against the calibration curve of COC-d3 to quantify the
amount of COC present in tissue.









CHAPTER 3
SWIFT ISOLATION

Introduction

Direct MALDI-MS analysis of intact tissue can provide information about the

distribution of cocaine in human brain more rapidly, with higher spatial resolution, and with less

sample loss than drug analysis methods that involve tissue homogenization. Furthermore, the

distribution of cocaine in brain tissue acquired by MALDI-MS can be directly related to the

histology. Quantitative analysis by MALDI-MS is challenging, however, because MALDI

exhibits irreproducible signal intensities due to inhomogeneous crystal formation, inconsistent

sample preparation, and laser shot-to-shot variability. The use of internal standards for

quantitative MALDI-MS has been shown to improve signal stability, if the solution-phase

properties are carefully matched as in an isotopic standard.77

Quantification of small drug molecules such as COC using MALDI-MS is further

complicated by the presence of interfering matrix peaks in the low-mass range along with ions

that may be produced from endogenous compounds present in the brain tissue. One of the

strengths of a linear ion trap mass spectrometer is its ability to perform multiple stages of mass

analysis (MSn) to significantly increase the selectivity for the analyte of interest. A MALDI-MSn

method could be developed to remove interference from both MALDI matrix and the complex

sample environment of brain tissue; however, a problem arises when trying to combine the use of

MSn with the use of internal standards. Current instruments allow for only one isolation window

(IW) in MSn experiments, isolating a single parent mass (or range of masses) for collision-

induced dissociation (CID). This means that MSn of the target ions of the analyte and internal

standard typically requires two separate MSn experiments. This increases the response

variability and can counteract the signal normalizing effects of using an internal standard.









It was previously reported that using a wide IW (e.g., 6 Daltons (Da)) that simultaneously

isolates the analyte and internal standards ions in a single MS2 experiment provided improved

precision for quantitative MALDI-MSn of COC in tissue, when compared to using alternate MS2

experiments that separately isolate the target analyte and internal standard ions.103 Isolating both

analyte and internal standard with a wide IW reduces signal variability when the analyte and

internal standard signals are ratioed. However, the wide IW not only isolates the analyte and

internal standard ions, but also ions in between that are not of interest. These ions may fill up the

ion trap and may also interfere with detection of the target ions. This chapter reports a multi-

notch isolation waveform that selectively isolates the analyte and internal standard ions, reducing

the effects of background interference and boosting the sensitivity for analyte ions. SWIFT

(stored waveform inverse Fourier transform) is a broadband excitation technique that is capable

of selectively isolating multiple ions and has the potential for improving precision of quantitative

MALDI-MSn.

SWIFT was first introduced to the field of mass spectrometry by Marshall et al. in 1985

for use with the Fourier transform ion cyclotron resonance mass spectrometer (FT/ICR).104

SWIFT was later applied successfully to the quadrupole ion trap (QIT)105, 106 due to the similar

principles of operation between the FT/ICR and the QIT In 1994, Cooks and coworkers107

made improvements on the application of SWIFT to the QIT for selective isolation by employing

a two-stage course/fine procedure for isolating ions from a population of trapped ions. The

advantage of the procedure was that the coarse step removes most of the ions that contribute to

space-charging, and thereafter the frequencies of the analyte ions remain relatively constant.

Ions trapped in a linear ion trap (LIT) are stored radially in the center section of

quadrupole rods by a two-dimensional radio frequency (RF) field, and stored axially by stopping









potentials applied to the end sections of quadrupole rods. The quadrupolar field within the mass

analyzer has a voltage of constant angular frequency (e.g., Q = 1188 kHz for the LIT used here)

and variable amplitude (0 to 5 kVo-p), which drives ionic motion in both the axial and radial

directions. Ionic motion must be stable in both directions for an ion to remain trapped. Trapped

ions oscillate in the quadrupolar field with characteristic frequencies known as secular

frequencies (con). For a given set of trapping conditions, these frequencies are characteristic of

ion mass-to-charge (m/z). By subjecting ions to a signal of frequency equal to a characteristic

ion frequency, they can be radially excited and ejected from the ion trap. Secular frequencies

(wa) are given by Equation 3-1,


o,, = (2n + /7) (0 2

where n = 0, 1, 2, ... etc., Q is the RF drive frequency, and fl is a complex function of the

Mathieu parameters a and q, whose solutions classify an ion as stable or unstable.108 Since DC

voltage is not applied to the LIT quadrupole electrodes, a = 0 resulting in Equation 3-2.

(3-2)

2 2
(2+ 8 2)2- q q2 ( 2)2 q 2
(4+8)2 q 2 (f-4)2 q2
(6 + )2 q ( 6)2 -q
((8+ p)2 q 2 8)02 q 2
(10 + (12+/)2 (f -12)2

In order to solve for fl in Equation 3-2, it is first necessary to solve for the q value for the ion to

be isolated. This q value is calculated using Equation 3-3, in which (m/z)center is the ion placed at

the q of isolation (0.83).

(m / z),eter (0.83)
q m(3-3)
m/z









ft is then calculated through an iterative process starting with an initial ft value given by the

Dehmelt's approximation given in Equation 3-4 for q values less than 0.4.


P I (3-4)


This chapter introduces the use of a multi-notch SWIFT applied to the linear ion trap mass

spectrometer for selectively isolating multiple pairs of analyte and internal standard ions during a

single MSn scan to improve precision during MALDI-MSn quantification. A dual-notch SWIFT

waveform was optimized for the isolation of cocaine and its metabolite benzoylecgonine along

with their corresponding trideuterated analogs.

Experimental

Chemicals

Cocaine (COC; MW 303.4 Da), benzoylecgonine (BE; MW 289.3 Da) were purchased

from Cerilliant (Round Rock, TX, USA) at concentrations of 1 mg/mL in acetonitrile. COC-d3

(MW 306.4 Da, 0.17% do) and BE-d3 (MW 292.3 Da, 0.08% do) were also purchased from

Cerilliant at concentrations of 100 aig/mL. in acetonitrile. High-performance liquid

chromatography (HPLC)-grade acetonitrile, methanol, and water were purchased from Fisher

Scientific (Pittsburgh, PA, USA). Working standards of COC, COC-d3, BE, and BE-d3 were

diluted with acetonitrile and then stored at 4 C. MALDI matrix, 2,5-dihydroxybenzoic acid

(DHB; MW 154.1 Da), was purchased from ACROS Organics (Geel, Belgium). Saturated DHB

matrix solutions (40 mg/mL DHB) were prepared in methanol/water (70:30, vol/vol) on the day

of use.

Tissue Collection

Human brain tissue samples were provided by the El Paso County Coroner's Office in

Colorado Springs, CO. Postmortem brain material was excised from the nucleus accumbens









(NAc) from case number 07A-369, whose toxicologic analysis indicated the presence of cocaine

in blood at 69 ng/mL (COC concentration in the brain tissue was not quantified). The NAc is a

dopamine-rich area of the striatum, which may contain an accumulation of COC due to its

affinity to bind with the dopamine transporter.100 At autopsy, the excised tissue was immediately

snap-frozen in liquid nitrogen and then stored in a -80 C freezer until analyzed.

Tissue Sectioning and Sample Preparation

Frozen brain tissue was cut into thin sections (20 am thickness) in a cryostat (HM 505E;

Microm International GmbH, Waldorf, Germany) at -25 C. The tissue samples were frozen to

the cryostat sample stage using distilled water. Serial brain sections were collected onto

microscope slides where they were thaw mounted and then stored at -80 C. Before mass

spectrometric analysis, the tissue sections were removed from the freezer and placed in a vacuum

desiccator for 30 min before spiking standards (1-liL droplets by micropipette) and applying

MALDI matrix. The matrix was applied to the tissue sections using an artistic airbrush (Aztek

A470; Testors, Rockford, IL, USA). The application of MALDI matrix by airbrush has been

previously published.67

Mass Spectrometry

All experiments were performed using an LTQ linear ion trap with a vMALDI ion source

(Thermo Finnigan, San Jose, CA, USA), equipped with nitrogen laser (337 nm) at a frequency

of 20 Hz and 100-lam spot size. A more detailed description of this instrument has been

published.67 The number of laser shots was automatically varied (between 1 and 17 shots) using

automatic gain control (AGC) to optimally fill the trap with ions, thus avoiding space charge-

related peak broadening and mass shifts. AGC assesses the ion generation rate by use of a









prescan, and then adjusts the number of laser shots per scan to produce an optimal number of

ions for each scan. The spectra are normalized to the number of laser shots for each scan.

Resonance excitation is used for isolation, activation and mass analysis. For MS"

experiments, unwanted ions are resonantly ejected from the ion trap by applying a 5-500 kHz

multi-frequency isolation waveform consisting of sine components spaced every 0.5 kHz. The

ions of interest are isolated in the ion trap by removing sine components from the isolation

waveform that correspond to the secular frequency of the desired ions. Ions to be isolated are

selected in the LTQ software by entering the m/z with its IW. The mass range for the ion is

defined as (m/z IW/2) to (m/z + IW/2). The IW should be narrow enough to eliminate

interfering peaks, but wide enough to avoid loss of sensitivity for the desired ions. However, it is

important to note that the activation width for resonance excitation (CID) has the same value as

the IW. Therefore, the collision energy applied during MS" is spread over the activation width.

Thus, increasing the IW decreases the collision energy for each ion.

SWIFT Calculation

Inverse Fourier Transform

A computer program written in C+ was used to calculate the SWIFT waveform based on a

process previously described.106 Notches in the desired broadband magnitude spectrum, from

frequency 0 to 500 kHz, were calculated to have centers corresponding to the secular frequencies

of the ions to be isolated. The frequency spectrum was then transformed to the time domain

using the inverse Fourier transform (IFT), which was performed using an adaptation of the

Cooly-Tukey fast Fourier transform (FFT) algorithm.109 The algorithm generates output, which

must be midpoint reflected about the N/2 axis, where N equals the total number of points in the

SWIFT waveform. This step is similar to a time shift and therefore affects the phase, but not the

magnitude of the corresponding frequency-domain spectrum. The advantage of midpoint









reflection is that it avoids sudden voltage transients at the beginning and end of the excitation

period.

Quadratic Phase Modulation

Phase modulation of the frequency spectrum is necessary in order to reduce the dynamic

range of the time-domain waveform. The large dynamic range is caused by all of the specified

frequency components starting with the same phase at time zero. A nonlinear phase modulation

varies the phase continuously at a nonconstant rate with frequency, and results in a broader time-

domain waveform after IFT that requires less amplitude to achieve the same power. The real, R,,

and imaginary, I,, components are created from the magnitude data, Mag,, using the following

relationship:

R = Mag, cos ((35
(3-5)
I, = Mag, sin (,

Where the phase,
cp, = cO + Ji + (K / 2)i2 (3-6)

Here (po is the initial phase (zero), i is the frequency index, and J = 0.5t and K = R/N are the

quadratic terms, where N is the number of nonzero data points in the frequency spectrum.

Values of J and K are chosen to satisfy the Nyquist criterion, such that the rate of phase change

per frequency-domain data point is kept at half the Nyquist limit (n) or below, which removes

nonuniformity of the magnitude in the frequency-domain.110

Temporal Spectral Inhomogeneity

It has been shown previously that quadratic phase modulation is effective at obtaining a

more uniform excitation power spectrum.11 However, an undesirable consequence of phase

modulation is temporal spectral inhomogeneity. This means that SWIFT is essentially a

frequency scan in which the frequency content is localized in time and varies systematically with









time. Figure 3-la shows the time domain of a dual-notch SWIFT waveform with frequency

notches corresponding to the secular frequencies of the [M+H] ions of COC (m/z 304.25) and

COC-d3 (m/z 307.25). Figure 3-lb is the resulting frequency domain waveform after performing

fast Fourier transform (FFT) of the time-domain waveform in Figure 3-la. The two frequency

notches are 1.6-Da wide at 431.88477 -436.27930 kHz and 439.69727 -444.33594 kHz. FFT

of the first half (0 to 2047 ps) of the time-domain waveform in Figure 3-la is shown in Figure 3-

Ic and the FFT of the second half (2048 to 4096 ps) of the waveform is shown in Figure 3-1d.

The resulting frequency-domain spectra shown in Figures 3-1c and 3-1d illustrate that the

SWIFT waveform scans from high to low frequency and that the frequency content is localized

in time. Thus, ions of different m/z are excited at different times during the SWIFT waveform.

This may not be such a critical issue for the FT-ICR, for which SWIFT was originally designed,

but it is undesirable for LIT experiments, since many collisions with helium buffer gas (- 1

mTorr) occur during the excitation event. For the LIT, it is therefore desirable to excite ions of

all desired m/z values simultaneously.

Three solutions were previously published112 for correcting or minimizing the temporal

spectral inhomogeneity of SWIFT. One solution involves using a short-duration (lower-

resolution) SWIFT waveform that is repeated many times during the excitation event, which

serves to increase the number of times a specific frequency is represented in the time domain

during the excitation event; however, this leads to lower mass selectivity due to the lower-

resolution SWIFT waveform. A second approach (multiple foldovers) involves overmodulation

of the phase in which the quadratic term in the phase vs. frequency spectrum corresponds to a

bandwidth multiple times (number of foldovers) higher than the Nyquist limit. If the foldover

number is sufficiently high, then the SWIFT waveform effectively excites all frequency









components simultaneously. A third approach is to distribute the excitation frequency

components randomly throughout the time-domain excitation period, which has been previously

termed "filtered noise field" excitation.113

Apodization

The time-domain SWIFT waveform was multiplied by an apodization (smoothing)

function to force the time-domain signal smoothly to zero at the beginning and end of the time-

domain period. The apodization function consists of a quarter-wave sinusoid matched to the first

one-fourth of the time-domain period, followed by unit weighting for the middle half period,

followed by a quarter-wave sinusoid for the final one-fourth of the period.

It was previously shown114 that apodization can cause frequency notch distortion, which

may lead to partial ejection of ions selected to be isolated. Multiplication of the apodizing

function with the time-domain waveform corresponds to convolution of the Fourier transforms in

the frequency domain. The process of convolution widens the bases of the spectral components,

which may lead to power leakage into adjacent spectral components and loss of frequency

resolution. Notched waveforms ideally represent discontinuities in the waveform; however, the

process of apodization transforms these discontinuities or sharp edges of the notch into

continuous transitions. Thus, the edges of a narrow notch may fuse into each other, thereby

bridging the notch to some extent.

Figure 3-2a compares the time domain of a dual-notch SWIFT waveform that has been

apodized (blue) with one that has not been apodized (red). It is shown that the apodization

function only reduces the magnitude of the Gibb's oscillations (spurious oscillations that occur

when using a truncated Fourier series)15 for the first 700 ps of the time-domain waveform, and

only the magnitude of the first 200 ps have been significantly smoothed to nearly zero. Figure 3-









2b compares the frequency domain of the dual-notch SWIFT waveform that has been apodized

(blue) with one that has not been apodized (red). Apodization of the time domain results in

convolution of the edges of the frequency notches in the frequency domain, which could lead to

ejection of ions that are intended to be isolated.

SWIFT Application to LTQ

The non-apodized, time-domain, dual-notch SWIFT waveform was downloaded to the

memory of an arbitrary waveform generator (AWG) (Stanford Research Systems Model DS345,

Sunnyvale, CA, USA). The AWG has a maximum sampling rate of 40 MHz, time resolution of

16,300 data points, and a 12-bit DAC output. The LTQ has a programmable trigger that can be

used to send a TTL pulse to the AWG at a specific time during the experimental sequence (e.g.,

isolation or activation). For experiments in which the AWG was triggered during isolation, the

LTQ isolation waveform was turned off to avoid interference. Once triggered, the AWG applies

the SWIFT waveform to the LTQ analog board, where it is summed with the other waveforms

before being amplified and applied to the linear ion trap x-rods. The amplitude of the SWIFT

waveform and the number of bursts were modified manually on the AWG. A two-channel

digitizing oscilloscope (Tektronix Model TDS 540, Tektronix Inc., Beaverton, OR) was used to

observe the SWIFT waveform.

Results and Discussion

Optimization of a Dual-Notch SWIFT

Frequency optimization

A 4096-point (4k) dual-notch SWIFT was calculated from frequency 0 to 500 kHz with a

sampling frequency of 1000 kHz corresponding to a 0.244 kHz frequency step. Frequency

notches (1.6-Da wide) are centered at the secular frequencies of the [M+H] ion of COC (m/z

304.25) and COC-d3 (m/z 307.25). The theoretical secular frequencies (co) were calculated by









first calculating the q values for the ions using Equation 3-3 and the ft values using Equation 3-2.

Both q and fl were then used to calculate the corresponding secular frequencies using Equation 3-

1.

Experimental secular frequencies can shift from the theoretical secular frequencies

calculated due to space-charge (shift to lower frequencies), higher-order fields, and resonance

excitation amplitude (shift to higher frequencies).114, 116 A 1:1 solution of 1 lg/mL of COC and

COC-d3 were spotted (1 iL each) onto a MALDI plate and airbrushed with DHB matrix for

analysis. A dual-notch SWIFT waveform with 1.6-Da wide frequency notches centered at the

theoretically calculated secular frequencies of m/z 304 and m/z 307 was applied to the LTQ with

a SWIFT amplitude of 0.4 Vp-p, and a burst count of 3 set by the AWG. The center m/z 305.8

was placed at a q = 0.83. The LTQ isolation waveform was turned off and the SWIFT waveform

was triggered at isolation. The frequency notches for m/z 304 and m/z 307 were optimized

separately for maximum peak intensity by shifting the 1.6-Da window to higher frequencies a

number of frequency steps (0.244 kHz) from the starting position at a frequency lower than the

theoretically calculated secular frequency (Figure 3-3). The frequency notches were shifted by

recalculating the SWIFT waveform after each change of frequencies. The optimal frequency

notches for m/z 304 and m/z 307 was 439.69727 444.33594 kHz and 431.88477 436.27930

kHz, respectively. This corresponds to a shift to higher frequencies for m/z 304 (+ 0.488 kHz)

and m/z 307 (+ 0.732 kHz), which may be due to the amplitude of the SWIFT waveform

applied.116

Burst count optimization

Increasing the number of SWIFT frequency-domain points lengthens the duration of the

time-domain SWIFT waveform. However, the duration of the SWIFT pulse is limited by the









size of the stored-waveform data set. The AWG used for the experiments in this paper has a data

point storage limit of 16,300 points. It would be ideal to apply the SWIFT waveform during the

entire isolation event in order to maximize the opportunity for selective ejection of unwanted

ions that are in resonance with the frequencies of the SWIFT pulse. The problem with a short

SWIFT pulse duration can be overcome by applying multiple pulses (or bursts) of the same

stored waveform in a series (or train). This allows the specific frequencies of the SWIFT

waveform to be present for longer periods of time, which can lead to more efficient ejection of

selected ions. The number of SWIFT waveform bursts in a train that are applied to the LTQ can

be controlled by the AWG.

Figure 4a shows the digital scope image of a 1-burst (4.096 ms), 2-burst (8.192 ms), 3-

burst (12.288 ms), 4-burst (16.384 ms), and 5-burst (20.480 ms) train of pulses shown above the

digital scope image of the square-wave trigger during isolation (15.5 ms). Each SWIFT pulse in

a train is the same dual-notch SWIFT waveform (frequency optimized to isolate ions at m/z 304

and m/z 307) that has been merely repeated. Each train of SWIFT bursts were applied five times

individually to a 1:1 solution of COC and COC-d3 (1 alg/mL each) pipetted onto a MALDI plate

and airbrushed with DHB matrix. Figure 3-4b shows the changes in absolute peak intensities for

m/z 304 and m/z 307 with varying number of SWIFT bursts. Not shown, but also monitored,

were the absolute peak intensities of the background ions at m/z 305, 306, and 308. From 1 to 3

bursts, there was a 12% and 14% decrease in signal for m/z 304 and 307, respectively. The

signals for the background ions at m/z 305, 306, and 308 decreased by 69%, 59%, and 87%,

respectively, from 1 to 3 bursts. At 4 bursts, the intensities of the ions at m/z 304 and 307

decreased by 39% and 49%, respectively. This significant decrease in ion signal may be

attributed to the longer duration of the 4-burst train (16.384 ms) compared to the duration of the









isolation event (15.5 ms). The ions may not have enough time to relax towards the center of the

LIT before they are moved from the q of isolation (0.83) to a lower q of activation (0.25). There

is -2 ms pause after the isolation event before the RF voltage is decreased to move the ions

towards the lower q of activation. At 5 bursts (20.480 ms), all ions were ejected. The optimal

number of SWIFT bursts was determined to be 3, since there was not a significant decrease in

the signal intensity of m/z 304 and 307, and signal intensities of the background ions at m/z 305,

306, and 308 were reduced to below 3% intensity relative to the m/z 304 and 307 at 3 bursts.

Amplitude optimization

Another SWIFT parameter besides the duration of the SWIFT pulse that can affect the

ejections of ions is the amplitude. Increasing the amplitude of the SWIFT waveform in turn

increases the magnitude of the oscillations of the ions that are in resonance with the frequencies.

It has also been reported that large amplitudes distort the frequency-domain cutoffs, causing the

pulse to have an even wider frequency range, which can interfere with ions close in mass.106 The

secular frequencies of ions can shift to higher values with increased excitation amplitude,

resulting in ions absorbing energy in a range near their secular frequency and being

unintentionally ejected from the ion trap.

The amplitude of a dual-notch SWIFT waveform with a 3-burst train was optimized by

varying the amplitude on the AWG from 0 to 1.0 Vp-p and monitoring the peak intensity of m/z

304 and m/z 307. A 1:1 solution of COC and COC-d3 (1 lg/mL each) was pipetted (1 |iL) onto a

MALDI plate and airbrushed with DHB matrix. Figure 3-5 shows the absolute peak intensities

of the [M+H] ions of COC (m/z 304) and COC-d3 (m/z 307) along with the 13C isotope ions at

m/z 305 and 308. The optimal SWIFT amplitude (0.40 Vp-p) was determined to be the lowest

potential needed to maintain a 1:1 signal of COC and COC-d3, while decreasing the background









ions below 3% relative intensity, and the maximum signal for the m/z 304 and 307 ions. Notice

that at higher amplitudes above 0.50 Vp-p, more m/z 304 ions are ejected than m/z 307 ions. This

may be attributed to m/z 304 being at higher q, closer to the right-hand edge of the stability

diagram.

Selective Ion Isolation of Standards on MALDI Plate

The optimized dual-notch SWIFT was applied to COC and COC-d3 at 1 ag/mL each

pipetted onto a MALDI plate and airbrushed with DHB matrix. The amplitude on the AWG was

set to 0.4 Vp-p with a burst count of 3. Figure 3-6a compares the full scan mass spectrum (top), a

mass spectrum of a 5-Da wide isolation window centered at m/z 305.8 (middle), and the mass

spectrum from the application of the optimized dual-notch SWIFT (bottom). The 5-Da wide

isolation window is effective at eliminating the background ions outside the window, but retains

ions at m/z 305 and 306. The dual-notch SWIFT (bottom) is able to eliminate the same

background ions as well as reduce the ion intensities for m/z 305 and 306 without reducing the

signal for the desired ions at m/z 304 and 307. Figure 3-6b shows the MS2 scans resulting from

applying a 5-Da wide broadband excitation waveform (CID = 55) to the isolated ions from the 5-

Da wide isolation (top) and the dual-notch SWIFT isolation (bottom). The MS2 scan of the

isolated SWIFT ions (bottom) resulted in no background fragment ions in the product spectrum.

The only ions present are the fragment ions of m/z 304 and 307 at m/z 182 and 185, respectively,

both formed from the neutral loss of benzoic acid. In comparison, the product spectrum of the

ions isolated with the 5-Da wide isolation window (top), shows the presence of fragment ions

from the background as well as those from m/z 304 and 307.

Improving MALDI Precision with SWIFT

It was previously shown that isolating the analyte and internal standard ions in a single

MSn scan using a wide isolation window can improve precision for MALDI quantification









compared to isolating the analyte and internal standard ions separately during alternate MSn

scans.103 In order to compare the ability of a dual-notch SWIFT waveform to also improve

precision for MALDI quantification, five solutions (31, 62, 125, 250, and 500 ng/mL COC with

250 ng/mL COC-d3) were pipetted onto a MALDI plate and airbrushed with DHB matrix. The

peak intensity ofm/z 182 and m/z 185 were ratioed from three different MS2 experiments (CID =

55): alternating scans (MS2 of m/z 304 and m/z 307 during separate scans), MS2 of m/z 304 and

m/z 307 during a single scan isolated by a dual-notch SWIFT, and during a single MS2 scan

isolated by a 5-Da window. The 5-Da wide isolation experiment had the best precision (% RSD

= 1 to 9%) for isotopic ratios, followed by dual-notch SWIFT (% RSD = 5 to 23%), and then

alternating MS2 scans (% RSD = 44 to 56%). This reinforces the conclusion103 that isolating the

analyte and internal standard ions during a single MSn scan improves precision over isolating the

analyte and internals standard ions during separate scans. The dual-notch SWIFT isolation may

be less precise than wide-isolation due to irreproducible shifts in the secular frequencies of the

analyte and internal standard ions, which leads to their ejection.

Although SWIFT isolation may be less precise of a method for MALDI quantification

than wide isolation, the major advantage of SWIFT is that it can selectively isolate multiple ions

or m/z ranges. This is done simply by removing the appropriate frequencies from the SWIFT

waveform corresponding to the secular frequencies of the desired ions. A quad-notch SWIFT

was calculated with frequency notches (1.5-Da wide) corresponding to the secular frequencies of

the [M+H] ions of BE (m/z 290.25), BE-d3 (m/z 293.25), COC (m/z 304.25), and COC-d3 (m/z

307.25). The quad-notch SWIFT was applied to a solution of BE, BE-d3, COC, and COC-d3 at 1

alg/mL each that was pipette (1 |iL) onto a MALDI plate and then airbrushed with DHB matrix.

Figure 3-8a compares the full scan mass spectrum (top), the mass spectrum from a 20-Da wide









isolation window centered at m/z 298.75 (middle), and the mass spectrum from applying a quad-

notch SWIFT (bottom). The 20-Da wide isolation (middle) was effective at eliminating

background ions outside the isolation window, but because of the large width of the isolation

window necessary to isolate all of the analyte and internal standard ions present, many undesired

background ions were isolated as well. The intensities of the ions isolated by the quad-notch

SWIFT (bottom) were lower than the intensities of the same ions isolated by a 20-Da wide

isolation window centered at m/z 298.75, but the quad-notch SWIFT was able to significantly

reduce the intensities of the background ions, which should help to simplify the product spectra

from MS2 analysis.

Figure 3-8b shows the MS2 scans from applying a 20-Da broadband excitation (CID =

55) to the ions isolated from the 20-Da wide isolation (top) and a quad-notch SWIFT (bottom).

The product ion spectrum of the quad-notch SWIFT isolation (bottom) shows the presence of the

product ions of m/z 290, 293, 304, and 307 at m/z 168, 171, 182, and 185, respectively, formed

from the neutral loss of benzoic acid. Also present is the common fragment ion of both m/z 290

and 304 at m/z 150 (neutral loss of benzoic acid and methanol) and the common fragment ion of

m/z 293 and 307 at m/z 153 (neutral loss of benzoic acid and methanol). The quad-notch SWIFT

was able to reduce background ions during isolation, which resulted in the elimination of 13C

peaks and MALDI matrix ions (e.g., m/z 137, [DHB-H20+H]+) from the product ion spectrum

that are present in the product ion spectrum from the 20-Da wide isolation (top).

Selective Ion Isolation of Standards on Tissue

The quad-notch SWIFT was applied to 20-iim thick human brain tissue that was spiked

(1 aL on top of tissue) with BE, BE-d3, COC, and COC-d3 at 1 alg/mL each and then airbrushed

with DHB matrix. Figure 3-9a shows that the full mass spectrum is dominated by lipid and DHB









matrix ions and it is difficult to see the target ions at m/z 290, 293, 304, and 307. Figure 3-9b

shows an expanded view of the mass range from m/z 250 to 350 with the full scan mass spectrum

(top), mass spectrum from a 20-Da wide isolation centered at m/z 298.75 (middle), and the mass

spectrum resulting from the application of a the quad-notch SWIFT (bottom). The 20-Da wide

isolation window (middle) isolates the [M+H] ions of BE, BE-d3, COC, and COC-d3 at m/z 290,

293, 304, and 307, respectively, as well as a number of background ions. The quad-notch

SWIFT (bottom) was successful in eliminating or reducing the background ions during isolation

of the analyte and internal standard ions.

Conclusions

It was previously shown that isolating the analyte and internal standard ions during a

single MS2 scan using a wide isolation window can provide improved precision for the

quantitative analysis of COC in postmortem brain tissue compared to using two alternating MS2

scans that isolate the analyte and internal standard ions separately.103 Here, multi-notch SWIFT

waveforms were investigated as an alternative isolation technique to wide isolation for isolating

the analyte and internal standard ions during a single MS2 scan for improved MALDI-MS2

precision. It was determined that multi-notch SWIFT isolation can provide improved precision

when compared to using two alternating MS2 scans; however, it is not as precise as the wide

isolation method.

Nevertheless, SWIFT isolation offers the advantage of higher selectivity and is better

able to reduce background ions that may complicate or interfere with MS2 analysis (e.g., isobaric

product ions). Also, analysis times can be reduced as more frequency notches are added to the

SWIFT waveform. This can become very important when quantitatively imaging several

analytes from a large tissue sample.











S1.0 (a) 12 .(b)
12
Os







0 1000 2000 3000 4000 5000 o0 a son 100 m

Time microsecondss) Frequency (kHz
0.50.6













S0.4 08
4 0.40
MO2 04







0.0 0.0
02
00






0 100 2000 300 400 500 0 100 200o 300 400 500

FrequTim e microsecondss) Frequency (kHz)








--1st Half of Waveform -Znd Half of Waveforrn
Figure 3-1. Temporal spectral inhomogeneity of SWIFT1.4 (a) Time domain of dual-notch









SWIFT waveform calculated by inverse Fourier transform (IFT) of the frequency
domain. (b) Fast Fourier transform (FFT) of the full (0 to 4096 ps) time-domain
w 0.8 o0 10

0.4















SWIFT waveform. (c) FFT of the first half (0 to 2047 ps) of the time-domain SWIFT
20.2
00 0.0
0 100 200 300 400 500 0 100 2CC 300 400 500

Frequency (kI-z) Frequency (kHz)

-l1ti-Aalfof Waveform -2nd Half of Waveform


Figure 3-1. Temporal spectral inhomogeneity of SWIFT. (a) Time domain of dual-notch
SWIFT waveform calculated by inverse Fourier transform (IFT) of the frequency
domain. (b) Fast Fourier transform (FFT) of the full (0 to 4096 gs) time-domain
SWIFT waveform. (c) FFT of the first half (0 to 2047 gs) of the time-domain SWIFT
waveform. (d) FFT of the second half (2048 to 4096 ps) of the time-domain SWIFT
waveform.










0.10 __
0.06
j 0.04 -- -- -- -
O 0.02
0.00
-0.02
-0.04 ---
-0.06 ----- --- --- ---
-0.08
-0.10
0 200 400 600 800 1000

Time microsecondss)

No Apodization Apodization

1.2

0 --.3









Apodization No Apodization
0. "0.


*- 0.4
bA13 0.2


425 430 435 440 445 450

Frequency (kHz)

SApodization ....... No Apodization


Figure 3-2. Effects of apodization on SWIFT. (a) Time domain of dual-notch SWIFT waveform
that has been apodized (blue) overlapped with the time domain of the same dual-
notch SWIFT waveform without apodization (red). (b) Comparing the frequency
domain of a dual-notch SWIFT waveform with apodization (blue) and without
apodization (red).












(a)

T_-- __- ---A ---- --A



--- --- A- -- --- --- ---- -- A- A--- -- -- -
---- ----- -- ----------- t---- ---H-

---- I--- I------
S. -- ------ --- ----

----- --- ----
--- --- ----------
------!-
~~i i


5.0E+04
4.5E,04
4.0E+04
3.5E+04
3.OEt04
2.5E+04
2.OE+04
1.5E+04
1.OE+04
5,OE+03
0,0E+00


4 5 6 7 8


9 10 11 12 13


Numberof FrequencySteps


(b)
9.0Et04 (Fb


5.' 8.*4--0F r104 -r-F 01
Ui


4 J I I I I I I I I
6 2.0E+04 --------- --- --------- t-------- -----------..













Frequency notches were shifted to higher frequencies a number of frequency steps
(0.244 kHz) from the starting position to maximize the peak intensities of 304
o ...0 -k- -------f--- .... t '-t ..--------- t-
,..,2.0F+04 ---, ,t- ,t- t--C-t t-t-`t--C--- ...

S0 E 2 4 5 6 7 9 -0 L- 12 13 14



Number of Frequency Steps


Figure 3-3. Optimization of frequency notches (1.6-Da) for (a) m/z 304 and (b) m/z 307.
Frequency notches were shifted to higher frequencies a number of frequency steps
(0.244 klHz) from the starting position to maximize the peak intensities of m/z 304
and m/z 307. The optimal frequency notches for m/z 304 and m/z 307 were
439.69727 444.33594 kHz and 431.88477 436.27930 kHz, respectively.


0 1 2 3











(a)
1 burst (4.096 ms]


2 bursts (8.192 ms)

3 bursts (12.288 ms):
- ---- -- - -

4 bursts (16.384 ms):

5 bursts (20.480 m.):
-- -- Z -- --- 1-;-


i Isolation (15.5 ms) -



(b)

0_ 1.6E+05 ....... .

3 1.2E-05
S1.OE+05









abolt peak+0 finest of------- the------ [M+H],--- ionofCO-(--34)an-CC-d-(--07
( B. + 4- 1. -
Q) 4.0t+0-. *--------- --- -- --- ^---
2.0E+0--
C O.OE+O0
WU 1 2 3 4 5

Number of SWIFT Bursts

-a-m/z304 -*-m/z307


Figure 3-4. Optimization of burst counts for a dual-notch SWIFT waveform. (a) Digital scope
images of a 1 burst (4.096 ms), 2 burst (8.192 ms), 3 burst (12.288 ms), 4 burst
(16.384 ms), and 5 burst (20.480 ms) train of pulses shown above the digital scope
image of the square-wave trigger during isolation (15.5 ms). (b) The average
absolute peak intensities of the [M+H]+ ion of COC (m/z 304) and COC-d3 (m/z 307)
plotted versus the number of SWIFT bursts of a dual-notch SWIFT applied during
isolation. The error bars correspond to the standard error (5 replicates).










7.0E+06


S4.0CE+06
IMI

0
(U I







,- .I
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
Dual-Notch SWIFT Amplitude (Vp.p)

---m/z 304 --/z 305 --m/z 307 m/z 308


Figure 3-5. Optimization of SWIFT amplitude for a dual-notch SWIFT waveform. SWIFT
amplitude was optimized by varying the potential from 0 to 1.0 Vp-p and monitoring
the absolute peak intensities of the [M+H]+ ions of COC (m/z 304) and COC-d3 (m/z
307) along with the ions of their corresponding 13C isotopic peaks at m/z 305 and m/z
308. The optimal SWIFT amplitude (0.40 Vpp) was the lowest potential needed to
maintain a 1:1 signal of COC and COC-d3 (1 alg/mL each) that were pipetted (1 itL)
onto a MALDI plate and airbrushed with DHB matrix.















(a)
1000ooooo Full MS Scan
304.42 307.42
80000-
uoooo- /ooO
60000
40000-
20000 305.42 308.42
S302.50 303.42 30642 309.42
0
100000 5-Da Wide Isolation Centered atm/z 305.8
6 80000 304.25 307.33
60000-
40000
20000 305.33
0 306.33

Dual-Notch SWIFT 307.33
80000
soooo- U fo
60000
40000

O _20000- 305.42 306_42
303 304 305 306 307 308 309
m/z




(b)
5-Da Wide Isolation Centered atm/z 305.8
182.25 185.25
10000-

8000

6000

4000

2000 183.25
0 y Y ,184.25 -
0


8000-

6000-

4000

2000]

180 181 182 183 184 185 186 187 188
m/z





Figure 3-6. SWIFT isolation and wide isolation comparison. (a) Comparing full scan (top), 5-

Da wide isolation window centered at m1z 305.8 (middle), and dual-notch SWIFT

(bottom) of a 1:1 solution of COC and COC-d3 (1 pg/mL each) that was pipetted onto

a MALDI plate (1 piL) and airbrushed with DHB matrix. (b) Comparing the MS2 scan

of ions fragmented with a 5-Da broadband excitation waveform (CID = 55) applied to

ions isolated with a 5-Da wide isolation window centered at m/z 305.8 (top) and ions

isolated with a dual-notch SWIFT waveform (bottom).












3.s (pg) %RSD %RSD %RSD
31 52 5 5
3.0 62 46 17 4

125 56 21 1
4.j 2.5
C 250 50 23 2



s-- t s (..-
i F 6z 1
1.0 I -
00 ''.' ,.- 0 ,10 4 S
N 1.0
0.0 -- ------- -----I----- I-'L l '--i




0 50 100 150 200 250 300 350 400 450 500

COC Mass (pg)

Alternating MS/MS Scans U Dual-Notch SWIFT A 5-Da Wide Isolation



Figure 3-7. SWIFT isolation and wide isolation calibration curves. Five solutions (31, 62, 125,
250, and 500 ng/mL COC with 250 ng/mL COC-d3 were pipetted onto a MALDI
plate (1 piL each) and airbrushed with DHB matrix. The peak intensity ofm/z 182
and m/z 185 were ratioed from three different MS2 experiments (CID = 55):
alternating scans (MS2 ofm/z 304 and m/z 307 during separate scans), MS2 of m/z 304
and m/z 307 during a single scan isolated by a dual-notch SWIFT and during a single
MS2 scan isolated by a 5-Da window. The error bars correspond to the standard
error (3 replicates).


















Full MS Scan
[BE+H] [BE-d3+H]+
290.25 29325




287.33 _29 g9725 301o33

20-Da Wide Isolation Centered at m/z 298.75

290.33 293.33


289 33 \_
Quad-Notch SWIFT


24.33
f_\ 297.25 301.33


50000-
40000
30000
20000
10000
0
120000
100000-
eoo-
80000
60000
0-
40000
20000
-:
0
50000
40000-
30000
20000
10000
0


290


295


[COC+H]+[COC-d3+H]+
304.250





3095 313.17
304.33 307.25






04i 307S.25
304.33 307.33


| |


305


308.33
310


(b)
20-Da Wide Isolation Centered at m/z 298.75

35000 182.25 85.25
30000


35000 Quad-Notch SWIFT 12
30000
25000-
20000-
15000 171.33


5000- 150.33 --[
= *W \\


0 4 P I


1 00


150


304.33
293.33 -307.33
290.33 --j


200


250


300


Figure 3-8. Quad-notch SWIFT isolation of standards on MALDI plate. (a) Comparing full scan

(top), 20-Da wide isolation window centered at m/z 298.75 (middle), and quad-notch

SWIFT (bottom) of a solution of BE, BE-d3, COC, and COC-d3 at 1 pg/mL each that

was pipetted (1 ipL) onto a MALDI plate and then airbrushed with DHB matrix. (b)

Comparing the MS2 scan of ions fragmented with a 20-Da broadband excitation

waveform (CID = 55) applied to ions isolated with a 20-Da wide isolation window

centered at m/z 298.75 (top) and ions isolated with a quad-notch SWIFT waveform

(bottom).


(a)


290.25 293.33


315


293.25
290.25


304.25
F- 307.25


350
















(a)
18'

40000

35000

30000-

25000-

S20000-

15000-
137.42
10000

5000


4.42


798.67


78267
77267


Zoom in

273.42

231.25
313.33


760.67

73967

72367
36950 449.58 53 577.7 65167 [ d
I I I I _I L U.1 I'' 1 651'. 7 ,/ Ij


826.58
848.58

866.58


974.42


100 200 300 400 500 600 700 800 900 1000
m/z


Figure 3-9. Quad-notch SWIFT isolation of standards on brain tissue. (a) Mass spectrum of 20-

pm thick brain tissue slice that was spiked (1 tL) with a solution of BE, BE-d3, COC,

and COC-d3 at 1 plg/mL each and airbrushed with DHB matrix. (b) Expanded region

(m/z 250 350) comparing full scan (top), 20-Da wide isolation window centered at

m/z 298.75 (middle), and quad-notch SWIFT (bottom).


*04q4_1111~


.I1Uinil.lVl*L.nlI^IUlnnii~ldYlun~nm OUU I*U IU YUY









CHAPTER 4
QUANTITATIVE ANALYSIS OF DRUGS IN BRAIN TISSUE

Introduction

A large variety of specimens are collected in the field of postmortem forensic toxicology

including blood, liver, brain, and urine.4 For the analysis of drugs of abuse, brain samples show

several advantages over all other specimens in postmortem forensic toxicology.5 One advantage

is due to the brain being an isolated compartment, which delays putrefaction after death.6 Also,

the metabolic activity is lower in the brain than in other tissues or in blood, resulting in slower

decomposition.7 Finally, drugs of abuse establish their effects through the central nervous

system. Therefore, it can be assumed that concentrations of drugs of abuse found in the brain

better reflect drug concentrations at their site of action at the time of death.8

Analysis of drugs of abuse in the brain has applications in forensic and postmortem

toxicology. Drug concentrations in the brain may be needed to substantiate fatal overdoses9 and

support neurotoxicity studies.10 Direct measurement of drug and metabolite concentrations in

discrete brain regions can also be used to study the mechanisms of drug action,11 regional

distribution,12 and preferential accumulation of drugs.13

Conventional drug analysis in tissue involves tissue homogenization of the tissue prior to

subsequent chromatographic analysis.14 Such sample pretreatments are known to introduce

variation in detection due to inhomogeneity of the analyte within the sample matrix.15 Also,

homogenization of tissue eliminates the opportunity to acquire detailed anatomical and

histological information for in situ drug distribution. Imaging techniques that include mass

spectrometric imaging can help provide this information.









Experimental


Chemicals

Cocaine (COC; MW 303.4 Da), benzoylecgonine (BE; MW 289.3 Da), and cocaethylene

(CE; MW 317.4 Da) were purchased from Cerilliant (Round Rock, TX, USA) at concentrations

of 1 mg/mL in acetonitrile. COC-d3 (MW 306.4 Da, 0.17% do), BE-d3 (MW 292.3 Da, 0.08%

do), and CE-d3 (MW 320.4 Da 0.15% do) were also purchased from Cerilliant at concentrations

of 100 [g/mL in acetonitrile. High-performance liquid chromatography (HPLC)-grade

acetonitrile, methanol, and water were purchased from Fisher Scientific (Pittsburgh, PA, USA).

Working standards of COC, COC-d3, BE, BE-d3, CE, and CE-d3 were diluted with acetonitrile

and then stored at 4 C. MALDI matrix, 2,5-dihydroxybenzoic acid (DHB; MW 154.1 Da), was

purchased from ACROS Organics (Geel, Belgium). Saturated DHB matrix solutions (40 mg/mL

DHB) were prepared in methanol/water (70:30, vol/vol) on the day of use.

Tissue Collection

Human brain tissue samples were provided by the El Paso County Coroner's Office in

Colorado Springs, CO. Postmortem brain material was excised from the nucleus accumbens

(NAc) from case number 07A-369, whose toxicologic analysis indicated the presence of cocaine

in blood at 69 ng/mL (COC concentration in the brain tissue was not quantified). The NAc is a

dopamine-rich area of the striatum, which may contain an accumulation of COC due to its

affinity to bind with the dopamine transporter.100 At autopsy, the excised tissue was immediately

snap-frozen in liquid nitrogen and then stored in a -80 C freezer until analyzed.

Tissue Sectioning and Sample Preparation

Frozen brain tissue was cut into thin sections (20 [im thickness) in a cryostat (HM 505E;

Microm International GmbH, Waldorf, Germany) at -25 C. The tissue samples were frozen to

the cryostat sample stage using distilled water. Serial brain sections were collected onto









microscope slides where they were thaw mounted and then stored at -80 C. Before mass

spectrometric analysis, the tissue sections were removed from the freezer and placed in a vacuum

desiccator for 30 min before spiking standards (1-[iL droplets by micropipette) and applying

MALDI matrix. The matrix was applied to the tissue sections using an artistic airbrush (Aztek

A470; Testors, Rockford, IL, USA). The application of MALDI matrix by airbrush has been

previously published.67

Tissue Homogenization

One gram of blank human brain tissue (i.e., tissue for which toxicological analysis did not

indicate the presence of the analyte drug) was cut and weighed, and then finely minced with a

scalpel. The minced tissue was then placed into a glass tissue grinder (Duall 21; Kontes Glass

Inc., Vineland, New Jersey, USA), where it was homogenized into a liquid. The reservoir and

pestle shaft of the tissue grinder were rinsed with 3 mL of 60 [M sodium fluoride (NaF), which

serves as an inhibitor for esterases to prevent COC hydrolysis.99 The volume of NaF added was

measured to be approximately twice the mass of the corresponding tissue sample. The

homogenized tissue was then transferred to an 8-mL glass vial and placed in a sonicator for 1

min.

Preparation of Standard Solutions

Five standard solutions were prepared for spiking into tissue homogenate for generation of

calibration curves. The concentrations of these 1-mL solutions was 62, 125, 250, 500, and 1000

ng/mL each of BE, COC, and CE, as well as a mixture of their corresponding internal standards

(BE-d3, COC-d3, and CE-d3; 200 ng/mL each). The five standard solutions were dried with

nitrogen gas and then reconstituted with a 400-1aL aliquot of the sonicated homogenate. The

solutions were immediately vortex-mixed (1 min) and centrifuged at 10,000 rpm for 30 min.

The supernatants were then used for solid-phase extraction.









Preparation of Unknown Sample Solution

One gram of human brain tissue from case number 07A-369, for which toxicological

analysis indicated the presence of cocaine in blood at 69 ng/mL (COC concentration in the brain

tissue was not quantified) was cut and weighed, and then finely minced with a scalpel. The

minced tissue was then placed into a glass tissue grinder, where it was homogenized into a

liquid. The reservoir and pestle shaft of the tissue grinder were rinsed with 3 mL of 60 [iM

sodium fluoride (NaF). The homogenized tissue was then transferred to an 8-mL glass vial and

placed in a sonicator for 1 min.

A 1-mL solution of the internal standards (BE-d3, COC-d3, and CE-d3; 200 ng/mL each)

was prepared, dried with nitrogen gas and then reconstituted with a 400-liL aliquot of the

sonicated homogenate from case number 07A-369. The solution was immediately vortex-mixed

(1 min) and centrifuged at 10,000 rpm for 30 min. The supernatant was then used for solid-

phase extraction.

Solid-Phase Extraction

The extraction of cocaine and its metabolites was performed using underivatized silica (50

aim average particle size; 60 A pore size) solid-phase extraction (SPE) cartridges (HyperSep SI;

3-mL reservoir, 500-mg bed; Thermo Scientific, Bellafonte, PA, USA). The analytes in the

homogenate were separated from impurities using a selective elution scheme shown in Figure 4-

1, in which the adsorbed compounds of interest were eluted in a solvent that left the strongly

retained impurities behind on the cartridge. The SPE cartridge was first conditioned using 2 mL

of methanol followed by 2 mL of deionized water. Then a 100-l1L aliquot of the supernatant

from the centrifuged homogenate was loaded and drawn through the cartridge using low vacuum

(- 5 in. Hg; 1 in. Hg = 388.638 Pa) in a vacuum manifold (PrepSep 12-Port Vacuum Manifold;









Fisher Scientific, Pittsburgh, PA, USA). After discarding the eluent, analytes in the cartridge

were eluted using 3 mL of 5% ammonia in methanol solution. A washing step is typically

performed to remove interference in the biological matrices that may affect the assay; however,

this step was not performed to avoid loss in recoveries of the highly polar metabolite ecgonine

methyl ester. The high selectivity of MSn makes the lack of this washing step less of a concern.

The eluents from the extraction cartridge were then dried using nitrogen gas. The residue was

reconstituted in 500 piL of water/methanol (90:10, vol/vol), spotted onto a MALDI plate (1 .L)

and airbrushed with DHB matrix.

Mass Spectrometry

All experiments were performed using an LTQ linear ion trap with a vMALDI ion source

(Thermo Finnigan, San Jose, CA, USA), equipped with nitrogen laser (337 nm) at a frequency

of 20 Hz and 100-1lm spot size. A more detailed description of this instrument has been

published.67 The number of laser shots was automatically varied (between 1 and 17 shots) using

automatic gain control (AGC) to optimally fill the trap with ions, thus avoiding space charge-

related peak broadening and mass shifts. AGC assesses the ion generation rate by use of a

prescan, and then adjusts the number of laser shots per scan to produce an optimal number of

ions for each scan. The spectra are normalized to the number of laser shots for each scan.

Resonance excitation is used for isolation, activation and mass analysis. For MS"

experiments, unwanted ions are resonantly ejected from the ion trap by applying a 5-500 kHz

multi-frequency isolation waveform consisting of sine components spaced every 0.5 kHz. The

ions of interest are isolated in the ion trap by removing sine components from the isolation

waveform that correspond to the secular frequency of the desired ions. Ions to be isolated are

selected in the LTQ software by entering the m/z with its IW. The mass range for the ion is









defined as (m/z IW/2) to (m/z + IW/2). The IW should be narrow enough to eliminate including

interfering peaks, but wide enough to avoid loss of sensitivity for the desired ions. However, it is

important to note that the activation width for resonance excitation (CID) has the same value as

the IW. Therefore, the collision energy applied during MS" is spread over the activation width.

Thus, increasing the IW decreases the collision energy for each ion.

SWIFT Calculation

A computer program written in C++ was used to calculate the SWIFT waveform based on a

process previously described.106 The same procedure was used here except that the final time-

domain signal was left unapodized. Notches in the desired broadband magnitude spectrum, from

frequency 0 to 500 kHz, were calculated to have centers corresponding to the secular frequencies

of the ions to be isolated. The frequency spectrum was then transformed to the time domain

using the inverse Fourier transform (IFT), which was performed using an adaptation of the

Cooly-Tukey fast Fourier transform (FFT) algorithm.109 The algorithm generates output, which

must be midpoint reflected about the N/2 axis, where N equals the total number of points in the

SWIFT waveform. This step is similar to a time shift and therefore affects the phase, but not the

magnitude of the corresponding frequency-domain spectrum. The advantage of midpoint

reflection is that it avoids sudden voltage transients at the beginning and end of the excitation

period.

The frequency spectrum was then quadratically modulated in order to reduce the dynamic

range of the time-domain waveform. The real, R,, and imaginary, I, components are created

from the magnitude data, Mag,, using the following relationship:

R =Mag, cos (4-1)
I, =Mag, sin (,

Where the phase, ~, varies quadratically with frequency:









0, = o + Ji + (K / 2)i2


Here (po is the initial phase (zero), i is the frequency index, and J= 0.57t and K = 7t/N are the

quadratic terms, where N is the number of nonzero data points in the frequency spectrum.

Values of J and K are chosen to satisfy the Nyquist criteria, such that the rate of phase change

per frequency-domain data point is kept at half the Nyquist limit (t) or below, which removes

nonuniformity of the magnitude in the frequency-domain.110

SWIFT Application to LTQ

The resulting digitized waveform was downloaded to the memory of an arbitrary

waveform generator (AWG) (Stanford Research Systems Model DS345, Sunnyvale, CA, USA).

The AWG has a maximum sampling rate of 40 MHz, time resolution of 16,300 data points, and a

12-bit DAC output. The LTQ has a programmable trigger that can be used to send a TTL pulse

to the AWG at a specific time during the experimental sequence (e.g., isolation or activation).

For experiments in which the AWG was triggered during isolation, the LTQ isolation waveform

was turned off to avoid interference. Once triggered, the AWG applies the SWIFT waveform to

the LTQ analog board, where it is summed with the other waveforms before being amplified and

applied to the linear ion trap x-rods. The amplitude of the SWIFT waveform and the number of

bursts were modified manually on the AWG. A two-channel digitizing oscilloscope (Tektronix

Model TDS 540, Tektronix Inc., Beaverton, OR) was used to observe the SWIFT waveform.

Results and Discussion

Hexa-Notch SWIFT Isolation

One of the greatest strengths of SWIFT isolation is the ability to isolate multiple mass-to-

charge (m/z) ranges simultaneously, which allows for the selective ejection of ions that may

interfere with analysis. Multi-notch SWIFT isolation becomes a huge advantage when applied to


(4-2)









MALDI mass spectrometric imaging (MSI) when performing MSn. Typically for an MSn

experiment, only one parent ion is isolated and then activated with collision-induced dissociation

(CID) to produce product ions. MALDI MSI of a 2.0 cm by 1.0 cm tissue sample with 100-itm

laser steps (a total of 20,000 spectra) would take 5-6 hours to image, and this would have to be

repeated for every analyte analyzed by MSn. In addition, the analyte ion would normally be

analyzed separately from its internal standard ion, which has been shown to have increased

signal variability when compared to analyzing both the analyte and internal standard ions

simultaneously with a wide isolation window (Chapter 2) or by dual-notch SWIFT isolation

(Chapter 3). Multi-notch SWIFT isolation when applied to MALDI MSI can save considerable

analysis time as well as conserve laser shots, since fewer analyses will need to be performed.

A hexa-notch SWIFT isolation waveform was calculated for the [M+H]+ ions of BE (m/z

290.2), BE-d3 (m/z 293.2), COC (m/z 304.2), COC-d3 (m/z 307.2), CE (m/z 318.2), and CE-d3

(m/z 321.2). The center m/z of these ions (m/z 305.8) was placed at the q of isolation at q =

0.830. The frequencies of the 1.5-Da notches centered about each of the six ions to be isolated

were calculated using a LabView program described in Appendix A. The frequencies of the

notches calculated are listed in Table 4-1. Figure 4-2 shows the calculated hexa-notch SWIFT

isolation waveform in the frequency domain. Although the width of all six notches is maintained

at 1.5 Da, the width of the notches in the frequency domain becomes wider at higher frequencies.

This is due to the nonlinearity of frequencies in relation to q at values greater than q = 0.4

(Figure 4-3). Isolation with the LTQ typically occurs at q = 0.83, because at this high q value,

frequencies are dispersed enough in q-space to allow for selective isolation. Below q = 0.4,

frequencies are too tightly spaced to allow for selective isolation (e.g., 1 kHz = 0.002q; given an

RF drive frequency Q = 1188 kHz).









SWIFT Isolation on Tissue

The hexa-notch SWIFT isolation waveform was used to analyze BE, BE-d3, COC, COC-

d3, CE, and CE-d3 standards that were spiked (1 piL each) onto blank human brain tissue at 1

1ig/mL each, and then airbrushed with DHB matrix. Figure 4-4 shows the peak ion intensities of

the [M+H] ions of BE (m/z 290.2), BE-d3 (m/z 293.2), COC (m/z 304.2), COC-d3 (m/z 307.2),

CE (m/z 318.2), and CE-d3 (m/z 321.2) that were isolated at various hexa-notch SWIFT

amplitudes (Vp-p). There was approximately a 50% decrease in signal across the monitored ions

when the SWIFT amplitude was increased from 0.0 Vp-p to 0.1 Vp-p. There was a steady decrease

in peak ion intensities with increasing SWIFT amplitude, with m/z 290.2 decreasing most rapidly

followed by m/z 293.2. Figure 4-5a shows the mass spectrum of the spiked brain tissue with a

SWIFT amplitude of 0.0 Vp-p. The mass spectrum is dominated by endogenous lipid ions from

the brain tissue and it is difficult to distinguish the presence of any analyte ions. The most

intense ion in the mass spectrum at m/z 313 was determined to be a phthalate contaminant from a

plastic bottle that was used to store the DHB matrix solution. MSn analysis revealed product

ions at m/z 177, 149, and 121, which confirmed the phthalate contaminant. Figure 4-5b shows

the mass spectrum with a SWIFT amplitude of 0.4 Vp-p. Notice that by increasing the SWIFT

amplitude, the lower m/z background ions (< m/z 500) are ejected more efficiently than the

higher m/z ions (m/z 500-2000). The higher m/z background ions (i.e., endogenous lipid region

m/z 700 900) are not ejected until the SWIFT amplitude is at 0.8 Vp-p (Figure 4-5c). Figure 4-6

shows a more detailed view of the mass spectra that ranges from m/z 280 to m/z 330. Before

SWIFT is applied (0.0 Vp-p), m/z 313 dominates the spectrum (Figure 4-6a) and the analyte ions

are buried in the background. At a SWIFT amplitude of 0.4 Vp-p (Figure 4-6b), the m/z 313 peak

has been reduced along with other low m/z background ions, revealing the presence of the









analyte ions at m/z 290, 293, 304, 307, 318, and 321. However, notice that the intensity ofm/z

290, which should be relatively at the same intensity as its trideuterated analog at m/z 293, has

decreased. At a SWIFT amplitude of 0.8 Vp-p, which is sufficient enough to eject all background

ions including high m/z ions, the high SWIFT amplitude has also ejected ions at m/z 290, 293,

and 304 and reduced intensity of the analyte ions at m/z 307, 318, and 321 (Figure 4-6c).

Ion Ejection

To understand how ions can be ejected from the ion trap that were intended to be isolated

by a multi-notch SWIFT waveform, it is helpful to look at a diagram of the pseudopotential well

depth of the linear ion trap (Figure 4-7). Each ion confined within the ion trap is associated with

a q value, which lies on the qx-axis on the Mathieu stability diagram (Chapter 1). Ions of

relatively high m/z have q values near the left side of the stability diagram (0x = 0, qx = 0) while

ions of lower m/z have q values which extend towards the Px = 1 stability boundary, as shown

using colored circles of various sizes in Figure 4-7. At the intersection of the Px = 1 stability

boundary and the qx-axis, where q, = 0.908, the trajectories of trapped ions become unstable

along the X-axis such that ions of m/z less than the low mass cutoff (LMCO) are not stored. This

method of ion ejection, which can occur only at a boundary of the stability diagram, is referred to

as mass-selective instability.96

One other method for ions to be ejected from the ion trap is called resonant ejection,108

which is the method of ejection used by SWIFT. The advantage of resonant ejection is that it

can be carried out at any frequency. Ions are resonantly ejected from the ion trap when a

frequency is applied that is in resonance with the secular frequency (co) of the ion and has

sufficient amplitude (depth of pseudopotential well) to increase the oscillation of the excited ions

until they exit through the slits in the center X-rods of the linear ion trap.









In Figure 4-6b, it makes sense that the intensity ofm/z 290 (q = 0.875) would decrease

with increasing SWIFT amplitude, before ions at m/z 293 (q = 0.866), 304 (q = 0.835), 307 (q =

0.827), 318 (q = 0.798), and 321 (q = 0.791), because m/z 290 lies the closest to the LMCO (m/z

280, co = 500 kHz) at the mass-selective instability boundary (q = 0.908). It also has a shallower

pseudopotential well depth (Figure 4-7) than the other ions, meaning that it takes a lower SWIFT

amplitude to eject it from the ion trap. One way to correct for the instability of m/z 290 at higher

SWIFT amplitudes would be to decrease its q value and move it away from the mass-selective

instability boundary (q = 0.908). This can be accomplished by changing the m/z at the q of

isolation (q = 0.830) from the center m/z of the analytes (m/z 305.8) to the lowest m/z analyte

(m/z 290). The new q values and notch frequencies for the calculated hexa-notch SWIFT based

on m/z 290.2 at q = 0.830 are listed in Table 4-2.

One issue with not placing the center m/z ion at the q of isolation is that the LTQ software

couples the isolation window width with the activation window width. This means that in order

to perform resonance excitation (CID) on the ions at m/z 290.2, 293.2, 304.2, 307.2, 318.2, and

321.2, with m/z 290.2 placed at the q of isolation (q = 0.830), at minimum, a 62-Da wide

isolation window (m/z 259.2 m/z 321.2) centered at m/z 290.2 is required to include these ions

in activation. A 70-Da wide isolation window (m/z 255.2 m/z 325.2) centered at m/z 290.2

would be wide enough to ensure complete activation of the ion at m/z 321.2. However, in the

LTQ software, isolation windows and consequentially, activation windows that are wider than 47

Da, are assigned a q of isolation that is lower than q = 0.830 (Figure 4-8). In the LTQ software,

the q of isolation is decreased linearly with increasing isolation window width to ensure that ions

to be isolated are above the LMCO. The maximum isolation window width allowed by the LTQ

software is 100 Da. The q of isolation for a 70-Da wide isolation window centered at m/z 290.2









is q = 0.791. The q values and notch frequencies for the calculated hexa-notch SWIFT based on

m/z 290.2 at q = 0.791 are listed in Table 4-3.

Figure 4-9 compares the hexa-notch SWIFT isolation of m/z 290, 293, 304, 307, 318, and

321 at a SWIFT amplitude of 0.6 Vp-p with the m/z at q of isolation set to m/z 305.8 (q = 0.830)

(Figure 4-9a), m/z 290.2 (q = 0.830) (Figure 4-9b), and m/z 290.2 (q = 0.791) (Figure 4-9c).

Figure 4-9a shows a large decrease in signal for m/z 290 (q = 0.875) and m/z 293 (q = 0.866)

compared to Figure 4-9b when m/z 290 (q = 0.830) and m/z 293 (q = 0.822) are placed at lower q

values further away from the mass-selective instability boundary (q = 0.908). At the lower q

values, m/z 290 and m/z 293 are deeper in the pseudopotential well of the ion trap and therefore

can tolerate a higher SWIFT amplitude before they are resonantly ejected from the trap. The

analyte ions at m/z 290 (q = 0.791), 293 (q = 0.783), 304 (q = 0.755), 307 (q = 0.747), 318 (q =

0.721), and 321 (q = 0.715) are even deeper in the pseudopotential well and show an overall

increase in signal at lower q values (Figure 4-9c).

Figure 4-10 shows the peak ion intensities of the [M+H]+ ions of BE (m/z 290.2), BE-d3

(m/z 293.2), COC (m/z 304.2), COC-d3 (m/z 307.2), CE (m/z 318.2), and CE-d3 (m/z 321.2) when

the hexa-notch SWIFT described in Table 4-3 based on m/z 290.2 at q = 0.791 was applied at

various SWIFT amplitudes. The results shown in Figure 4-10 are very different than the results

shown in Figure 4-4 from the hexa-notch SWIFT described in Table 4-1 based on m/z 305.8 at q

= 0.830. Figure 4-4 showed approximately a 50% decrease in signal across the monitored ions

when the SWIFT amplitude was increased from 0.0 Vp-p to 0.1 Vp-p followed by a steady

decrease in peak ion intensities with increasing SWIFT amplitude. In contrast, when the

amplitude of the hexa-notch SWIFT (Table 4-3) is increased, the analyte ion intensities steadily

increase as well until the amplitude reaches 0.6 Vp-p. When the SWIFT amplitude is increased









from 0.6 Vp-p to 1.0 Vp-p, all of the analyte ion intensities decrease, but not in the order of low to

high m/z (e.g., m/z 304 decreases faster than m/z 293 and m/z 318 decreases faster than m/z 321),

as would be expected based on the pseudopotential well depth (Figure 4-7). This behavior may

be explained due to the fact that the energy absorption profile for ions subjected to resonance

excitation broadens with the amplitude of ion oscillation and shifts to higher frequencies.116 If

the ions to be isolated shift to higher frequencies, they will be outside the frequency notches of

the SWIFT isolation waveform and will come into resonance with the excitation frequencies thus

ejecting them from the trap. Therefore, it is beneficial to use the lowest SWIFT amplitude

possible to isolate the desired ions to avoid causing frequency shifts and sequential resonant

ejection (Figure 4-11). However, at lower SWIFT amplitudes (Figure 4-12), higher m/z

background ions will not be ejected and will remain trapped.

Two-Stage Isolation

One strategy for ejecting high m/z background ions while avoiding frequency shifts caused

from higher SWIFT amplitudes, is to perform two-stage isolation. The first stage would be a

coarse isolation that would coarsely isolate the ions of interest, while ejecting background ions.

The frequency notches during this stage would be wide enough so that if any frequency shifts did

occur, the ions of interest would not be ejected. This coarse isolation stage would be followed

by a fine isolation stage, which would utilize a higher degree of mass discrimination to eject

background ions close to the ions of interest.

This two-stage coarse/fine SWIFT isolation was first proposed by Soni and Cooks107 for

isolating ions having a single m/z value from a population of trapped ions. The coarse/fine

isolation technique used a doubly notched SWIFT pulse to perform the isolation in two steps at

two different q values. The coarse isolation step used a single notch centered at q = 0.0778 to

coarsely isolate the ion of interest. Then the RF amplitude was increased to move the trapped









ions to a higher q value in which a second narrower notch centered at q = 0.4035 was used to

finely isolate ions of a single m/z value only. The resonance frequencies of ions are more spread

out at higher q values (Figure 4-3), which allows for higher mass discrimination. The advantage

of this two-stage isolation is that the coarse step removes most of the ions that contribute to

space charging, and thereafter the frequencies of the analyte ions remain relatively constant.

However, although frequency shifts were minimized with the two-stage strategy, Soni and Cooks

still reported a 20% loss of target ion population as a result of the sharp mass discrimination of

the second fine isolation notch.107

High Mass Filter (HMF)

In order to minimize frequency shifts, a SWIFT excitation waveform was calculated that

ejects background ions higher in m/z than the highest m/z analyte (m/z 321). This SWIFT

excitation waveform was termed high mass filter (HMF), because it serves to filter out or eject

ions above a certain m/z. Figure 4-13 shows the frequency domain of a HMF that was calculated

to excite at frequencies 0 to 338 kHz. The right-hand edge of the HMF at 338 kHz corresponds

to a m/z cutoff at m/z 325.8 based on m/z 290.2 placed at a q of isolation (q = 0.791). This HMF

is designed to eject/excite ions from m/z 325.8 to greater than m/z 2000 (LTQ upper m/z limit; 48

kHz).

The HMF SWIFT excitation waveform was applied to the analysis of BE, BE-d3, COC,

COC-d3, CE, and CE-d3 standards that were spiked (1 aiL each) onto blank human brain tissue at

1 alg/mL each, and then airbrushed with DHB matrix. Figure 4-14 shows the peak ion intensities

of the [M+H]+ ions of BE (m/z 290.2), BE-d3 (m/z 293.2), COC (m/z 304.2), COC-d3 (m/z 307.2),

CE (m/z 318.2), and CE-d3 (m/z 321.2) while the HMF was applied at various SWIFT amplitudes

(Vp-p). When the HMF amplitude is increased from 0.0 Vp-p to 0.2 Vp-p there is a 19 to 43%









increase in signal for the analyte ions as the high m/z background signal begins to decrease. The

signal for the analyte ions then shows a decrease when the HMF amplitude is increased from 0.2

Vp-p to 0.4 Vp-p. At a HMF amplitude of 0.5 Vp-p, all of the high m/z background ions are

removed, resulting in a 14 to 22% increase in analyte signal from 0.4 Vp-p to 0.5 Vp-p. The

analyte signals then begin to steadily decrease as the HMF amplitude is increased from 0.5 Vp-p

to 1.0 Vp-p. At a HMF amplitude of 1.0 Vp-p the analyte signals have decreased 36 to 47% from

the amplitude of 0.5 Vp-p with the analyte ions at m/z 290.2 and 293.2 decreasing the most (47%).

Figure 4-15 shows the mass spectra of the spiked brain tissue with the HMF amplitude at

0.0 Vp-p (Figure 4-15a), 0.4 Vp-p (Figure 4-15b), and 0.5 Vp-p (Figure 4-15c). Notice the dramatic

change in the high m/z background when the HMF amplitude is increased from 0.4 Vp-p (Figure

4-15b) to 0.5 Vp-p (Figure 4-15c). Figure 4-16 shows a more detailed view of the mass spectra

from m/z 280 to m/z 330. Although the HMF SWIFT excitation waveform should only affect

ions from m/z 325.8 to m/z 2000, the signal intensities of the analyte ions steadily increase when

the HMF amplitude is changed from 0.0 Vp-p (Figure 4-16a) to 0.4 Vp-p (Figure 4-16b) and then

to 0.5 Vp-p (Figure 4-16c). This increase in analyte signal may be attributed to the reduction in

high m/z background ions, which dominate the mass spectrum. By ejecting the high m/z

background ions, space-charge effects are reduced that would normally cause frequency shifts of

the analyte ions outside the notches of the SWIFT isolation waveform resulting in some ejection

of analyte ions.

Combining HMF with Hexa-Notch SWIFT

Two-stage isolation was performed by combining the HMF SWIFT excitation waveform to

eject ions above m/z 325.8 with a hexa-notch SWIFT isolation waveform to selectively isolate

ions at m/z 290.2, 293.2, 304.2, 307.2, 318.2, and 321.2 (Figure 4-17). The frequency domain of

the two-stage isolation is shown in Figure 4-17 with the HMF waveform shown in red and the









hexa-notch waveform shown in blue. The time domain of the two-stage isolation is shown in

Figure 4-18. The HMF SWIFT excitation waveform (red) occurs from 0 to 4,096 ps and the

hexa-notch SWIFT isolation waveform (blue) occurs from 4,097 to 8,192 ps. A single burst of

the two-stage isolation pulse (8,192 ps) is triggered during the LTQ isolation event (15,500 is)

with the LTQ isolation waveform turned off.

The two-stage isolation was used to analyze BE, BE-d3, COC, COC-d3, CE, and CE-d3

standards that were spiked (1 piL each) onto blank human brain tissue at 1 alg/mL each, and then

airbrushed with DHB matrix. Figure 4-19 shows the peak ion intensities of the [M+H]+ ions of

BE (m/z 290.2), BE-d3 (m/z 293.2), COC (m/z 304.2), COC-d3 (m/z 307.2), CE (m/z 318.2), and

CE-d3 (m/z 321.2) that were isolated at various two-stage isolation amplitudes (Vp-p). Both the

HMF SWIFT excitation waveform and the hexanotch SWIFT isolation waveform were set to the

same amplitude using the function generator. The analyte ion intensities decreased steadily as

the amplitude was increased from 0.1 Vp-p to 0.4 Vp-p. The analyte ion intensities then remained

relatively constant between 0.4 Vp-p and 0.6 Vp-p. When the amplitude was increased from 0.6

Vp-p to 1.0 Vp-p, the analyte ion intensities quickly decreased with the m/z 290.2 ion decreasing

the fastest. The instability of m/z 290.2 at SWIFT amplitudes greater than 0.6 Vp-p is probably

due to the shallow position of this ion in the pseudopotential well. One strategy for correcting

this would be to move the analyte ions to lower q values further away from the LMCO during

isolation.

Figure 4-20 shows the mass spectra of the spiked blank brain tissue with the two-stage

isolation amplitude at 0.0 Vp-p (Figure 4-20a), 0.5 Vp-p (Figure 4-20b), and 0.6 Vp-p (Figure 4-

20c). When the HMF SWIFT excitation waveform was used alone without the hexa-notch

SWIFT isolation waveform, the high m/z background ions were ejected efficiently at a SWIFT









amplitude of 0.5 Vpp (Figure 4-15c); however, when the HMF SWIFT excitation waveform is

combined with the hexa-notch SWIFT isolation waveform in the two-stage isolation, the high

m/z background ions are not ejected at 0.5 Vp-p (Figure 4-20b), but requires an amplitude of 0.6

Vp-p (Figure 4-20c) for the high m/z ion background to be removed. Increasing the amplitude

from 0.5 Vp-p to 0.6 Vp-p actually results in a slight increase in the analyte ion intensities (3% to

18%), which could be due to the removal of the high m/z background ions that cause frequency

shifts of the analyte ions. Figure 4-21 shows a more detailed view of the mass spectra from m/z

280 to m/z 330. Notice how the background ion signals have been reduced around the analyte

ions.

The overall difference in the analyte ion intensities with the application of the two-stage

isolation at 0.6 Vp-p compared to no two-stage isolation (0.0 Vp-p) is 54% decrease for m/z 290.2,

37% decrease for m/z 293.2, 19% decrease for m/z 304.2, 12% decrease for m/z 307.2, 9%

decrease for m/z 318.2, and 22% decrease for m/z 321.2. The high decreases in signal for m/z

290.2 (54%) and m/z 293.2 (37%) are probably due to their proximity to the LMCO, which could

be remedied by lowering the q values of the ions during isolation. The high decrease in signal

for m/z 321.2 (22%) could be due to its proximity to the m/z cutoff (m/z 325.8) of the HMF

applied. This can be fixed by applying a different HMF that allows for more space between the

m/z 321.2 ion and the m/z cutoff, but still allows removal of the majority of the high m/z

background ions.

MS/MS with Two-Stage Isolation

The overall goal of the two-stage isolation was to provide an isolation strategy that would

allow for the isolation of the analyte and internal standard ions during a single MS/MS scan. The

MS/MS provides higher mass selectivity which is essential for distinguishing the analyte from

MALDI matrix and endogenous species (e.g., lipids) present in the brain tissue, and isolating the









analyte and internal standard ions in the same MS/MS scan has shown to improve the precision

of MALDI-MS/MS (Chapters 2 and 3).103 Figure 4-22a shows the mass spectrum of spiked

brain tissue with the two-stage SWIFT isolation waveform applied at 0.6 Vp-p. A 70-Da wide

isolation window was centered at the lowest m/z analyte ion at m/z 290.2, which placed the q of

isolation at q = 0.791. Since the LTQ software couples the isolation width with the activation

width, CID will be applied across the 70-Da wide activation window centered at m/z 290.2. This

means that ions in the mass range m/z 255.2 to 325.2 will all be activated by CID and

fragmented. It is also important to note that the collision energy applied during MS/MS is spread

over the entire activation width. Thus, increasing the isolation width decreases the collision

energy for each ion. It was determined that the CID value necessary to dissociate the analyte

ions during MS/MS and reduce the parent ions to a relative intensity of 10%, needed to be

increased with the wider activation window. It was determined that a CID of 55 was optimal for

the analytes with a 5-Da wide activation window, but the CID was increased to 90 for the 70-Da

wide activation window.

Figure 4-22b shows the MS/MS product ion spectrum of ions isolated by a two-stage

SWIFT isolation. Since the most intense fragment ion of the [M+H]+ ion of BE (m/z 290.2), BE-

d3 (m/z 293.2), COC (m/z 304.2), COC-d3 (m/z 307.2), CE (m/z 318.2), and CE-d3 (m/z 321.2) all

result from the neutral loss (NL) of benzoic acid (122 Da), it is easy to distinguish the origins of

each product ion. BE produces the product ion at m/z 168.2, BE-d3 produces the product ion at

m/z 171.2, COC produces the product ion at m/z 182.2, COC-d3 produces the product ion at m/z

185.2, CE produces the product ion at m/z 196.2, and CE-d3 produces the product ion at m/z

199.2. A common product ion of BE, COC, and CE is m/z 150.2 and a common product ion of

BE-d3, COC-d3, and CE-d3 is m/z 153.2, both of which are due to the NL of benzoic acid









(C6H5COOH) and methanol (CH30H). Common product ions were not used for quantification

due to the difficulty in determining the signal attributable to each parent ion.

Figure 4-23a shows the MS/MS product ion spectrum of ions isolated with a 40-Da wide

isolation window with CID = 55. Notice that this product ion spectrum contains the same

product ions from the MS/MS of ions isolated with a two-stage SWIFT isolation (Figure 4-23b)

with some additional ions (e.g., m/z 137 and m/z 147) produced from the MS/MS of background

ions. These background ions were isolated along with the analyte ions in the 40-Da wide

isolation (Figure 4-24b). The intense product ion at m/z 137 is [DHB+H-H20] caused by the

NL of DHB from the DHB cluster matrix ion at m/z 291, [2DHB+H-H20]+. Background ions

can complicate the product ion spectrum and might even interfere with the analyte product ion

signals if the background ions produce fragment ions that are isobaric; therefore, removal of

background ions during isolation before MS/MS is beneficial.

Comparing Wide Isolation and Two-Stage SWIFT Isolation

It was previously shown that isolating the analyte and internal standard ions in a single

MS/MS scan using a wide isolation window can improve precision for MALDI quantification

compared to isolating the analyte and internal standard ions separately during alternate MS/MS

scans (Chapters 2 and 3).103 In order to compare the ability of the two-stage SWIFT isolation

waveform to also improve precision for MALDI quantification, five solutions (mixture of BE,

COC, and CE at 62, 125, 250, 500, and 1000 ng/mL with a mixture of BE-d3, COC-d3, and CE-

d3 at 200 ng/mL) were pipetted in triplicate (1 |iL each) onto blank human brain tissue and

airbrushed with DHB matrix. The tissue was then analyzed using two different MS/MS

experiments: two-stage SWIFT isolation (0.6 Vp-p; CID = 90) and 40-Da wide isolation (CID =

55). Figure 4-25 shows the calibration curves for BE for both experiments (two-stage SWIFT









isolation and wide isolation) produced by plotting the peak intensity ratio of m/z 168 and m/z 171

versus the mass of BE standard spiked on tissue. The two-stage SWIFT isolation has

comparable precision (% RSD = 0 to 11%) for isotopic ratios compared to wide isolation (%

RSD = 2 to 11%). Figure 4-26 shows the calibration curves for COC for both experiments (two-

stage SWIFT isolation and wide isolation) produced by plotting the peak intensity ratio of m/z

182 and m/z 185 versus the mass of COC standard spiked on tissue. The two-stage SWIFT

isolation has similar precision (% RSD = 1 to 6%) for isotopic ratios compared to wide isolation

(% RSD = 3 to 5%). The precision of the two-stage SWIFT isolation for the MS/MS analysis of

COC is better than what was reported for dual-notch SWIFT isolation of COC in Chapter 3 (%

RSD = 5 to 23%). This improvement may be due to the addition of the HMF during the first

stage of the two-stage isolation, which removes the space-charge effects of the high m/z

background ions. Finally, Figure 4-27 shows the calibration curves for CE for both experiments

(two-stage SWIFT isolation and wide isolation) produced by plotting the peak intensity ratio of

m/z 196 and m/z 199 versus the mass of CE standard spiked on tissue. The two-stage SWIFT

isolation % RSD ranged from 3 to 12% compared to the % RSD for wide isolation that ranged

from 1 to 29%.

Although two-stage SWIFT isolation overall has comparable precision to that of wide

isolation, the higher mass selectivity of the two-stage SWIFT isolation affords a significant loss

in absolute signal intensity of the analyte ions when it is applied. BE and BE-d3 showed an

average percent loss in absolute signal intensity of 83% and 75%, respectively. COC and COC-

d3 showed an average percent loss in absolute signal intensity of 69% and 75%, respectively. CE

and CE-d3 showed an average percent loss in absolute signal intensity of 65% and 77%,

respectively.









Two-Stage SWIFT MALDI-MS/MS Quantification

The MS/MS two-stage SWIFT isolation method and the MS/MS 40-Da wide isolation

method were compared for the quantification of unspiked BE, COC, and CE from human brain

tissue from a subject whose toxicology report showed the presence of COC. Three different

concentrations of BE-d3, COC-d3, and CE-d3 (31, 62, and 125 ng/mL) were spiked (1 |iL) onto a

glass slide before thaw mounting a 20 iim-thick brain tissue slice on top and airbrushing DHB

matrix. All three spots were then analyzed using the MS/MS two-stage SWIFT isolation method

and then the MS/MS 40-Da wide isolation method. Approximately 2000 scans were acquired to

image the entire area of each of the spots (average area = 0.17 cm2). The m/z 171 signal from

using the MS/MS two-stage SWIFT isolation for BE-d3 from each spot was used to develop a

calibration curve that resulted in a line of best fit ofy = 0.97(0.08)x + 6(7). BE-d3 was shown

to have a linear response with increasing concentrations spiked underneath tissue. Since the

MS/MS two-stage SWIFT isolation method analyzes both BE and BE-d3 simultaneously,

unspiked BE was detected from each spot analyzed at m/z 168. An area of the tissue (500

MS/MS scans) that was not spiked with BE-d3 was analyzed using the MS/MS two-stage SWIFT

isolation method and the acquired m/z 168 signal was averaged with the m/z 168 signals from the

spiked BE-d3 spots, resulting in a very trace signal of 53 6 counts. Assuming that the amount

of unspiked BE extracted from the tissue has a 1:1 response with the BE-d3 spiked on top of

tissue, the calibration curve for BE-d3 can be used to quantify the amount of BE present in the

analyzed tissue. From the equation of the line, it was determined that BE was present at a level

equivalent to 50 ng/mL.

Using the 1 |iL volume of BE-d3 spiked underneath tissue, it is calculated that the mass of

BE present is 50 pg. Given that the area of an analyzed spot on tissue was 0.17 cm2 and that the









tissue thickness was 20 alm (2.0 x 10-3 cm), the volume of tissue from which BE was extracted

was 3.4 x 10-4 cm3. The mass of the tissue is 3.4 x 10-4 g (density of wet tissue -1.0 g/cm3),

resulting in an absolute concentration of BE detected in this area of the postmortem brain tissue

of 140 ng/g (140 ppb). Since the MS/MS two-stage SWIFT isolation method and the MS/MS

40-Da wide isolation method both acquire the analyte and internal standard ions for BE (m/z 168

and 171), COC (m/z 182 and 185), and CE (m/z 196 and 199) from each spot analyzed

simultaneously, the amount of unspiked COC and unspiked CE were also quantified from the

tissue using the same process described above. The results for the quantification of BE, COC,

and CE using both the MS/MS two-stage SWIFT isolation method and the MS/MS 40-Da wide

isolation method are summarized in Table 4-4.

SPE-MALDI-MS/MS Quantification

Drugs and their metabolites in tissue are typically quantified from tissue homogenate

instead of from intact tissue. For this reason, tissue homogenates were prepared and extracted by

solid-phase extraction (SPE) as described in the previous experimental section, and analyzed

using MALDI-MS/MS to quantify the presence of BE, COC, and CE in unspiked tissue. One

gram of blank human brain tissue (case number 07A-355) was cut, weighed (0.9830 g), and

homogenized. Then 3 mL of 60 IM NaF was added and the homogenate was sonicated. Five 1-

mL standard solutions (62, 125, 250, 500, and 1000 ng/mL each of BE, COC, and CE and 200

ng/mL each of BE-d3, COC-d3, and CE-d3) were dried with nitrogen gas and then reconstituted

with a 400-liL aliquot of the sonicated homogenate. The solutions were immediately vortex-

mixed and centrifuged. Then a 100-liL aliquot of the supernatant from the centrifuged

homogenate was loaded onto a preconditioned underivatized silica SPE cartridge. Analytes in

the cartridge were then eluted using 3 mL of 5% ammonia in methanol solution. The eluents









from the extraction cartridge were then dried using nitrogen gas, and the residue was

reconstituted in 500 pL of water/methanol (90:10, vol/vol), spotted onto a MALDI plate (1 .L)

in triplicate, and airbrushed with DHB matrix.

Each spot on the MALDI plate was analyzed using a MS/MS 5-Da wide isolation method

specific for each set of analyte and internal standard ions. The 5-Da wide isolation method was

used, because it was shown to have better precision than using alternating MS/MS scans.103 The

[M+H] ions of BE (m/z 290.2) and BE-d3 (m/z 293.2) were isolated with a 5-Da wide isolation

window centered at m/z 291.8 with CID = 55 to produce the product ions at m/z 168.2 and m/z

171.2 for BE and BE-d3, respectively. The [M+H] ions of COC (m/z 304.2) and COC-d3 (m/z

307.2) were isolated with a 5-Da wide isolation window centered at m/z 305.8 with CID = 55 to

produce the product ions at m/z 182.2 and m/z 185.2 for COC and COC-d3, respectively. The

[M+H]+ ions of CE (m/z 318.2) and CE-d3 (m/z 321.2) were isolated with a 5-Da wide isolation

window centered at m/z 319.8 with CID = 55 to produce the product ions at m/z 196.2 and m/z

199.2 for CE and CE-d3, respectively. Figure 4-28 shows the calibration curve for BE with the

peak intensity ratio of m/z 168.2 and m/z 171.2 versus the mass of BE spotted on the MALDI

plate from the tissue homogenate. The 5-Da wide isolation method was fairly precise with %

RSD ranging from 3 to 8%. The BE calibration curve has a line of best fit ofy =

0.00543(0.00005)x 0.0007(0.02). Figure 4-29 shows the calibration curve for COC with the

peak intensity ratio of m/z 182.2 and m/z 185.2 versus the mass of COC spotted on the MALDI

plate from the tissue homogenate. % RSD ranged from 4 to 12%. The COC calibration curve

showed a linear response with a line of best fit ofy = 0.00633(0.00005)x + 0.02(0.02). The

calibration curve for CE is shown in Figure 4-30 with the peak intensity ratio of m/z 196.2 and

m/z 199.2 versus the mass of CE spotted on the MALDI plate from the tissue homogenate. %









RSD ranged from 3 to 6%. The CE calibration curve showed a linear response with a line ofy =

0.00581(0.00003)x 0.02(0.02).

The equations of the calibration curves developed were used to quantify the amount of

unspiked BE, COC, and CE present in unspiked tissue homogenate from human brain tissue

(case number 07A-369), for which toxicological analysis indicated the presence of cocaine in

blood (69 ng/mL). One gram of this tissue was cut, weighed (0.9862 g), and homogenized.

Then 3 mL of 60 aM NaF was added and the homogenate was sonicated. A 1-mL solution of the

internal standards (BE-d3, COC-d3, and CE-d3; 200 ng/mL each) was prepared, dried with

nitrogen gas, and then reconstituted with a 400-liL aliquot of the sonicated homogenate from

case number 07A-369. The solution was immediately vortex-mixed and centrifuged. Then a

100-|iL aliquot of the supernatant from the centrifuged homogenate was loaded onto a

preconditioned underivatized silica SPE cartridge. Analytes in the cartridge were then eluted

using 3 mL of 5% ammonia in methanol solution. The eluents from the extraction cartridge were

then dried using nitrogen gas, and the residue was reconstituted in 500 piL of water/methanol

(90:10, vol/vol), spotted onto a MALDI plate (1 |iL) in triplicate, and airbrushed with DHB

matrix. Each spot on the MALDI plate was analyze using the MS/MS 5-Da wide isolation

method specific for each set of analyte and internal standard ions described previously. Using

the calibration curves it was determined that there was 27014 ng BE/g of tissue, 38011 ng

COC/g of tissue, and 43016 ng CE/g of tissue.

The concentrations for BE, COC, and CE are not comparable to the concentrations

determined by analyzing the tissue directly by the MS/MS two-stage SWIFT isolation method

and the MS/MS 40-Da wide isolation method (Table 4-4). One reason for this difference might

be in the sample size for the different methods. The tissue homogenate method analyzes a larger









tissue sample (- 1 g) for quantification, which averages the signal for the analytes over the entire

tissue. The intact tissue MALDI methods analyze much smaller samples (3.4 x 10-4 g) across

different regions of the tissue to develop a calibration curve for quantification. This analytical

strategy assumes that the internal standard spotted underneath the tissue will have a similar

response across the different regions of the tissue. It also assumes that the internal standard will

have similar extraction efficiencies through the tissue for all regions analyzed. In addition, since

it is difficult to spot different concentrations of the internal standard in triplicate underneath the

tissue, precision of the quantitative analysis for the intact tissue methods was not determined.

Conclusions

In Chapter 3, multi-notch SWIFT isolation waveforms were explored as a strategy for

isolating the analyte and internal standard ions during a single MS/MS scan, which has been

shown to provide improved precision for MALDI-MS compared to using two alternating MS/MS

scans that isolate the analyte and internal standard ions separately. However, it was determined

that multi-notch SWIFT isolation was not as precise as the wide isolation method. This might

have been due to frequency shifts of the analyte and internal standard ions from space-charge

effects caused by high m/z background ions.

A two-stage SWIFT isolation method was developed that utilizes a high mass filter

(HMF) SWIFT excitation waveform to remove high m/z background ions during the first stage of

isolation. This has been shown to increase isolated ion signals and improve the precision of

SWIFT when compared to the application of SWIFT without the HMF. This may suggest that

the HMF reduces the irreproducible frequency shifts of the analyte and internal standard ions by

preventing them from moving outside the notches of the multi-notch SWIFT isolation waveform

and being ejected. A hexa-notch SWIFT isolation waveform was used during the second stage

of the two-stage SWIFT isolation to mass selectively isolate the [M+H] ions of BE (m/z 168.2),









BE-d3 (m/z 171.2), COC (m/z 304.2), COC-d3 (m/z 307.2), CE (m/z 318.2), and CE-d3 (m/z

321.2). The hexa-notch SWIFT isolation waveform was able to effectively remove background

ions around the analyte and internal standard ions that may have interfered with MS/MS

analysis. The two-stage SWIFT isolation overall showed similar precision to that of wide

isolation when performing MS/MS on the analyte and internal standards spiked on blank brain

tissue. However, the higher mass selectivity of the two-stage SWIFT isolation affords a

significant loss in absolute signal intensity of the analyte and internal standard ions when it is

applied.

Two-stage SWIFT isolation was compared to wide isolation of intact tissue and SPE

extracted homogenized tissue for the MALDI-MS/MS quantification of BE, COC, and CE from

unspiked human brain tissue, whose toxicological analysis indicated the presence of cocaine in

blood. The two-stage SWIFT isolation method showed a lower analyte signal per gram of tissue

than the wide isolation method for BE, COC, and CE present in tissue. The quantification results

for the two-stage SWIFT isolation and the wide isolation of intact tissue was not comparable to

the wide isolation analysis of tissue homogenate (Table 4-4). However, the intact tissue methods

required considerably less sample preparation and smaller sample sizes than the tissue

homogenate method. There was no analysis time saved when comparing the two-stage SWIFT

isolation and wide isolation since the [M+H] ions of BE, BE-d3, COC, COC-d3, CE, and CE-d3

were all isolated simultaneously during the same MS/MS scan. In conclusion, a wide isolation

method may still be a better choice over SWIFT isolation for improving MALDI-MS/MS

precision for quantification and reducing analysis time, despite the inclusion of unwanted

background ions during isolation.















Strong
Solvent


ca
0
0 []


Key to Processes
o = Matrix

< = Impuriy

A = Compound of interest

O = Solvent A
= SolvenT B
i = Solvent C
G000019


Wash the Packing














COC0024A


0 O 0


A 0


0 0


Elutethe
Compoundsof Interest














GCo2SA


Figure 4-1. Solid-phase extraction scheme. Selective elution is the solid-phase extraction (SPE)
scheme used. Adsorbed compounds of interest are eluted in a solvent (3 mL 5% NH3
in MeOH) that leaves the strongly retained impurities behind. The SPE process
typically involves 5 steps: (1) select the proper SPE tube (HyperSep SI), (2)
condition the SPE tube (2 mL of methanol followed by 2 mL of deionized water), (3)
add the sample (100-4tL of homogenate), (4) wash the packing (not performed), and
(5) elute the compounds of interest (3 mL 5% NH3 in MeOH). Adapted from Supelco
Bulletin 910: Guide to Solid-Phase Extraction, 1998.


Selective Elution


Selec the Proper
SPE Tube or Disk


GoMi26a


CordiTion ihe SPE
Tube ow Disk













ci-'-Y :n


Add the Sample


GOMoM23A


r _


E.-'-









Table 4-1. Hexa-notch SWIFT properties based on m/z 305.8 at q = 0.830
[M+H] Notch Width Notch Frequencies
Analyte q value
Anal (m/z) qvaue (Da) (kHz)
BE 290.2 0.875 1.5 488.037109 to 494.873047

BE-d3 293.2 0.866 1.5 475.341797 to 481.445312

COC 304.2 0.835 1.5 439.208984 to 443.603516

COC-d3 307.2 0.827 1.5 431.152344 to 435.058594

CE 318.2 0.798 1.5 405.029297 to 408.203125

CE-d3 321.2 0.791 1.5 398.681641 to 401.855469








1.2


> 0.8 -

0.6
0.4'
I 0,4 -- --

0.2

0.0

380 400 420 440 460 480 500

Frequency(kHz)


Figure 4-2. Frequency domain of hexa-notch SWIFT isolation waveform. Frequency notches
correspond to the secular frequencies of the [M+H]+ ion of BE (m/z 290.2), BE-d3
(m/z 293.2), COC (m/z 304.2), COC-d3 (m/z 307.2), CE (m/z 318.2), and CE-d3 (m/z
321.2) with the center m/z 305.8 placed at the q of isolation at q = 0.83.


I i i i 1 i i i i 1










600.00


500.00


S400.00
N


CJ

I-
LL 200.00


100.00



0.00


0.10 0.20


40 0.50 0.60 0.70 0.80 0.90
-swace = secular frequency (kHz)
-space F drive frequency =1184. kH
= RF drive frequency = 1184.5 kHz


m Measured Calculated


Figure 4-3. Relationship between secular frequency (w) and q-space. The Dehmelt
approximation (red trace) states that w varies linearly with q for values of q less than
0.4. Secular frequency (w) diverts from linearity (blue trace) at q values higher than
0.4, which can be measured from the LTQ MSn diagnostic settings and verified by
calculations explained in Appendix A.


-- --------I ---- ---- -----







-I Dehmelt Approximation

i y = 424.26x 8E-14
2 m 1 w ^Hfi T < 0.4
^ ..--lrI ~ l I I --











9.0E+04

8.0E+04

7.0E+04

6.0E+04

5.0E+04

4.0E+04

3.0E+04

2.0E+04

1.0E+04

O.OE+00


0.0 0.1





-m/z 290 ni


--------------------------------------------------

.-------- ---- ---- -------- --------------------
S-----------------------------------------------

-----------------------------------------------

----, -- ---------------------------

--------------- -

- -- -- -- -- -- ----- -_- -- -- -----------

-------------------- --


0,2 0.3 0,4 0.5 0.6 0,7

Hexa-Notch SWIFTAmplitude (Vp.)


/z293 -m/z 304


0.8 0.9 1.0


m/z307 --m/z318 --m/z321


Figure 4-4. Variable hexa-notch SWIFT amplitude (m/z 305.8 at q = 0.830). Peak ion intensities
of the [M+H]+ ions of BE (m/z 290.2), BE-d3 (m/z 293.2), COC (m/z 304.2), COC-d3
(m/z 307.2), CE (m/z 318.2), and CE-d3 (m/z 321.2) at different amplitudes (Vp-p) of a
hexa-notch SWIFT isolation waveform applied to the analysis of brain tissue.











100- (a)


50-


291.04


0.0 Vp.p
NL: 1.28E5


313.15


798.59
403.34 577.61
518.38 11 I 864.73

..h l Ih 11,98263 1164.65


798.58


50 753.67
3 2 504 694.58
560 .42 69458
318.25 ., i h,


321.25


307.25


864.
r878


556.33 769.67

400 600 800


0.4 Vpp
NL: 6.02E4


75
.75

0.8 Vpp
NL: 2.45E3






1000 1200 1400 1600 1800 2000


Figure 4-5. Mass spectra (m/z 80 to 2000) of hexa-notch SWIFT at different amplitudes. Hexa-
notch SWIFT isolation waveform applied to BE, BE-d3, COC, COC-d3, CE, and CE-
d3 standards that were spiked (1 piL; 1 pg/mL each) onto blank brain tissue and
airbrushed with DHB matrix with a SWIFT amplitude of (a) 0.0 Vp-p (b) 0.4 Vp-p, and
(c) 0.8 VP-p.


100 (c)


sIM^^MM~~IIIUY MM 19 IfflnM'M~ln IW~I^N~Mkl~I IIOl~l VWWWI ~u


,llJ ll,










0.0 Vp.p
NL: 1.28E5


313.15
312.28


291.04 293.27 302.20 304.27 308.34


304.25


293.25
290.25 |


307.25



308.33 313


314.19
318.30 322.31

318.25
0.
321.33I


.17
2'% \


307.25
S308.33


329.14

4 Vpp.
L: 7.10E3


322.25
323.17 329.25
321.25 0.8 Vp
NL: 2.45E3



322.25
318.33 n


280 285 290 295 300 305
m/z


310 315 320 325 330


Figure 4-6. Mass spectra (m/z 280 to 330) of hexa-notch SWIFT at different amplitudes. Hexa-
notch SWIFT isolation waveform applied to BE, BE-d3, COC, COC-d3, CE, and CE-
d3 standards that were spiked (1 piL; 1 pg/mL each) onto blank brain tissue and
airbrushed with DHB matrix with a SWIFT amplitude of (a) 0.0 Vp-p (b) 0.4 Vp-p, and
(c) 0.8 VP-p.


100 (a)


283.35


0
Sio-5
O 100
- o

I 50-


287.25


294.17


301.17












4,a
0





c-
o
O
Q.


0








10-


q VR

8


0.30


.3 01
0.83 0.91


Figure 4-7. Pseudopotential well depth (Dx) of the ion trap. The deepest part of the
pseudopotential well is near the q of isolation at qx = 0.83. The sizes of the circles are
proportional to the m/z of the ions. Adapted from March, R.E. J. Mass Spectrom.
1997, 32, 351.









Table 4-2. Hexa-notch SWIFT properties based on m/z 290.2 at q = 0.830
[M+H] Notch Width Notch Frequencies
Analyte q value
Anal (m/z) qvaue (Da) (kHz)
BE 290.2 0.830 1.5 435.058594 to 439.453125

BE-d3 293.2 0.822 1.5 426.757812 to 430.664062

COC 304.2 0.792 1.5 400.146484 to 403.564453

COC-d3 307.2 0.784 1.5 393.798828 to 396.972656

CE 318.2 0.757 1.5 372.802734 to 375.488281

CE-d3 321.2 0.750 1.5 367.675781 to 370.117187











U.64


0.83 --


0.82 --------------------- ----- ------------

q diverges from 0.83 at
0.81 ---- isolation widths> 47 Da -----------------------------


E 0.80 ---
Smax isolation
wA windowwidth
Z- 0.79- ----
o i allowed: 100 Da

0.78 -- -


0.77 -- -


0.76-------------------------------------- -


0.75
0 10 20 30 40 50 60 70 80 90 100 110

Isolation Window Width (Da)



Figure 4-8. Isolation window width (Da) determined by the preset q of isolation. The q of
isolation diverges from q = 0.83 at isolation widths greater than 47 Da to ensure that
ions to be isolated are above the low mass cutoff (LMCO). The maximum isolation
window width allowed by the LTQ software is 100 Da.









Table 4-3. Hexa-notch SWIFT properties based on m/z 290.2 at q = 0.791
[M+H] Notch Width Notch Frequencies
Analyte q value
Anal (m/z) qvaue (Da) (kHz)
BE 290.2 0.791 1.5 399.902344 to 403.320312

BE-d3 293.2 0.783 1.5 393.066406 to 396.484375

COC 304.2 0.755 1.5 371.337891 to 374.023437

COC-d3 307.2 0.747 1.5 365.966797 to 368.652344

CE 318.2 0.721 1.5 347.656250 to 350.097656

CE-d3 321.2 0.715 1.5 343.261719 to 345.458984











304.28 307.26
304.28


293.25
287.21 290.22 294.35 298.18


100


50


0
S100
CO
-0
< 50
I 50


283.23 287.24


294.24


304.27
S307.26


33

301.21 A


318.28 321.30


m/z 305.8
q = 0.830
0.6 Vpp
NL: 4.73E3


317.15


308.32


322.33


313.18


318.28



)8.31 315.24
313.22


, 323.09 329.23

321.26m/z 290.2
q = 0.830
0.6 Vpp
NL: 5.35E3
322.26
F323.23 328.24


290.25 293.25


282.39 286.21


294.28


304.26 318.29
307.26 3:

317.17

308.29 31
302.24 / A 313.22 \


0.6 Vpp
NL: 4.37E4
322.28
A 323.38 329.18


280 285 290 295 300 305 310 315 320 325 330
m/z


Figure 4-9. Hexa-notch SWIFT applied at variable q of isolation. Hexa-notch SWIFT isolation
waveform applied to BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards that were
spiked (1 IIL; 1 pg/mL each) onto blank brain tissue and airbrushed with DHB matrix
with a SWIFT amplitude of 0.6 Vp-p and the m/z at q of isolation set to (a) m/z 305.8
(q = 0.830), (b) m/z 290.2 (q = 0.830), and (c) m/z 290.2 (q = 0.791); m/z range 280 to
330.


290.25 293.26










2.5E+05


2,5E+05 ----- -- --- -



1.OE+05 -...--- -------- ----- ----- -- V ----
4-











0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0,9 1.0

Hexa-Notch SWI FTAmplitude (Vp.p)

--m/z 290 m/z 293 m/z 304 m/z307 ---m/z 318 -0-m/z 321


Figure 4-10. Variable hexa-notch SWIFT amplitude (m/z 290.2 at q = 0.791). Peak ion
intensities of the [M+H] ions of BE (m/z 290.2), BE-d3 (m/z 293.2), COC (m/z
304.2), COC-d3 (m/z 307.2), CE (m/z 318.2), and CE-d3 (m/z 321.2) at different
amplitudes (Vpp) of a hexa-notch SWIFT isolation waveform based on m/z 290.2 at
q = 0.791.
q =0.791.











100 (a)


290.25 29


282.35
287.20
283.28
Ar JMA


290.25 293.:


co
50




-,-
0
,-

< 50-
.>


100
-Q


50



0-


290.25 29


282.36 287.21 ,


313.18
304.25 31
)323307.25


295.15 2


304.26 31
307.26

25 316.25

30829
294.27 302.23 3829 313.22

304.26 31
307.25

317.18
)3.25

94.24 301.07 308.26 33.15
294.24 301.07 A


0.0 Vp-p
NL: 3.02E5
8.28 321.27



322.23 329.16


8.29 06
321.29 06 P-P
NL: 4.47E4



322.28
323.38 329.19
8.269
321.27 2
NL: 3.11E4



322.28
A 323.23


280 285 290 295 300 305
m/z


310 315 320 325 330


Figure 4-11. Mass spectra (m/z 280 to 330) of hexa-notch SWIFT at different amplitudes. Hexa-
notch SWIFT isolation waveform based on m/z 290.2 at q = 0.791 applied to BE, BE-
d3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 IiL; 1 pg/mL each)
onto blank brain tissue and airbrushed with DHB matrix with a SWIFT amplitude of
(a) 0.0 Vp-p, (b) 0.6 Vp-p, and (c) 0.9 Vp-p


282.39 286.22

(c)










273.13
100 (a)


798.58


782.61
18 577.59


u 798.59
100 (b) 782.62

318.29 739.55
318.29
< 50 577.60 82
^ 290.25 483.62
|' 264.38
r 0 318.26
100 (c)
(C)


50 29025


0.0 Vp.p
NL: 1.15E5


848.70
864.70
I 982.62 1169.62 1321.91


0.6 Vp,
NL: 7.99E4


6.62
864.72
98259 1176.63


1542.24


0.9 Vp,
NL: 3.11E4


200 400 600 800 1000
m/z


1200 1400 1600 1800 2000


Figure 4-12. Mass spectra (m/z 80 to 2000) of hexa-notch SWIFT at different amplitudes. Hexa-
notch SWIFT isolation waveform based on m/z 290.2 at q = 0.791 applied to BE, BE-
d3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 IiL; 1 pg/mL each)
onto blank brain tissue and airbrushed with DHB matrix with a SWIFT amplitude of
(a) 0.0 Vp.p, (b) 0.6 Vp.p, and (c) 0.9 Vpp.








1.2


4~I 11 I T f-I-I


1.0 --




0 06-
^ 18-^- --^--^- ---


02
0.2 -_____--- --
0.0
0 100 200 300 400 500

Frequency(kHz)

-High Mass Filter (0 338 kHz)

Figure 4-13. Frequency domain of high mass filter (HMF). HMF is calculated to excite at
frequencies 0 to 338 kHz based on m/z 290.2 at a q of isolation (q = 0.791). This
HMF is designed to eject/excite ions from m/z 325.8 to greater than m/z 2000 (LTQ
upper m/z limit; 48 kHz).










2.3E+05
.E+0 ------ --- ----------------------
2.1E+OS --

1.9E+05 -- -

S1.7E+05

S1.5E+05

2 1.3E+05 -

1 1.1E+05 -------- --

9.OE+04- -- --

7.OE+04

5.OE+04
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

HMF SWIFTAmplitude (V.p)

-e-m/z 290 m/z293 --m/z304 m/z307 ---m/z318 .--m/z321


Figure 4-14. Variable amplitude of HMF. Peak ion intensities of the [M+H] ions of BE (m/z
290.2), BE-d3 (m/z 293.2), COC (m/z 304.2), COC-d3 (m/z 307.2) CE (m/z 318.2), and
CE-d3 (m/z 321.2) at different amplitudes (Vp-p) of a high mass filter (HMF) with m/z
cutoff at 325.8 applied to the analysis of brain tissue.











273.12
296.16


798.58


S577.59
478.42 760.63


.13


100


50


0
o 100
Sioo-
C
-a
c
<50-


0 0
r


1038.53 1305.90


258.20



273
(b)


261.17


273
(c)


848.69
87670


400 600


800 1000 1200 1400 1600 1800 2000


Figure 4-15. Mass spectra (m/z 80 to 2000) of HMF at different amplitudes. HMF SWIFT
excitation waveform applied to BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards
that were spiked (1 pIL; 1 plg/mL each) onto blank brain tissue and airbrushed with
DHB matrix with a SWIFT amplitude of(a) 0.0 Vp-p, (b) 0.4 Vp-p, and (c) 0.5 Vpp.


798.59
782.62

739.54 810.63
577.59 864.70
483.61
3


0.0 Vp.p
NL: 1.41E5


1028.61


1570.12


0.4 Vpp
NL: 3.97E4


257.16


0.5 Vpp
NL: 5.17E4


[nlJW41 U i l[qlf MIU^l L JkI IIl [ l'.ll I lP'm M I P ..











290.25


282.37
50- 287.20
283.27

5
( 100
CU (b)
-Q
S2
< 50

S282.33 287.20

100-
1007.


50-

282.32
I I I


290.25


287.19 A


293.24


295.15


304.27
S 307.26

302.18 308.30
3OQ8.3
V


304.24

I


90.23 293.25
A C


296.15
U\vU\


302.17


304.25


293.24

296.16 301.05 A


313.18


318.29 0.0 VP
S 321.27NL: 2.93E4


314.25


318.27
307.25

308.26
313.19
/315.20
318.27
307.25


308.26 315.19
313.15
/ A fA


322.25
/VrVI. V


329.14

Nk


321.280.4 Vp-
NL: 3.56E4


322.27
323.21

321.260.5 V-P
NL: 4.67E4



322.26
h 324.18


280 285 290 295 300 305
m/z


310 315 320 325 330


Figure 4-16. Mass spectra (m/z 280 to 330) of HMF at different amplitudes. HMF SWIFT
excitation waveform applied to BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards
that were spiked (1 lIL; 1 alg/mL each) onto blank brain tissue and airbrushed with
DHB matrix with a SWIFT amplitude of(a) 0.0 Vp-p, (b) 0.4 Vp-p, and (c) 0.5 Vp-p.









1.2 _

1.0

0. ---- --- --- --- ----


S0.6 --- -

0.4 t-

0.2 --- -

0.0

320 340 360 380 400 420

Frequency (kHz)


Hexa-Notch SWIFT H igh Mass Filter


Figure 4-17. Frequency domain of two-stage SWIFT isolation. The frequency domain of the
two-stage isolation performed by combining a HMF SWIFT excitation waveform
(red) to eject ions heavier than m/z 325.8 with a hexa-notch SWIFT isolation
waveform (blue) to selectively isolate ions at mz 290.2, 293.2, 304.2, 307.2, 318.2,
and 321.2.









1.0
0.8 --
0.6


0 000 000 000 000
S0.2

E- -0.2-Hh

-0.4

-D.8
-1.0

0 2000 4000 6000 3000

Time microsecondss)


-High Mass Filter -Hexa-NcItcihSWIFT


Figure 4-18. The time domain of the two-stage SWIFT isolation. The two-stage isolation is
composed of a HMF SWIFT excitation waveform (red) occurring from 0 to 4,096 ps
and the hexa-notch SWIFT isolation waveform (blue) occurring from 4,097 to 8,192
ps. A single burst of the two-stage isolation pulse (8,192 ps) is triggered during the
LTQ isolation event (15,500 ps).

















-*------r + + 4ii-+--


I-~ ~~~ -lF4I-IP


0.0 0.1 0,2 0.3 0,4 0,5 0.6 0.7 0.8 0.9 1.0

Two-Stage SWIFTAmplitude (Vpp)


-8-m/z 290


m/z 293 m/z304


m/z307 --m/z318 -*-m/z321


Figure 4-19. Variable amplitude of two-stage SWIFT isolation. Peak ion intensites of the
[M+H]+ ions of BE (m/z 290.2), BE-d3 (m/z 293.2), COC (m/z 304.2), COC-d3 (m/z
307.2) CE (m/z 318.2), and CE-d3 (m/z 321.2) at different amplitudes (Vp-p) of a two-
stage isolation composed of a HMF SWIFT excitation waveform and a hexa-notch
SWIFT isolation waveform applied to the analysis of brain tissue.


8.0E+04


7.0E+04


6.0E+04 -


5.0E+04 -


4.0E+04


3.0E+04


2.0E+04 --


1.OE+04 -


O.OE+00


---- --------------


';';';'' '';'';''


----- -------










273.13
100 0.0 Vpp
0 ( a) 291.09 0 0 P-P
2 9 NL: 1.69E5
798.60
50 577.62 79860
483.62 87762 66.68


100 0.5 VP-P
1,- (b) 782.63
S 318.27 782.63 NL: 8.84E3
S|851.69
S50- 389.54 753.65
483.62 694.60 864.71
0 916.72 1086.72 1542.24 1707.48
307.22
100 0.6 VPP
(C) NL: 5.18E3

50

264.33

500 1000 1500 2000
m/z

Figure 4-20. Mass spectra (m/z 80 to 2000) of two-stage SWIFT isolation at different
amplitudes. Two-stage isolation composed of a HMF SWIFT excitation waveform
and a hexa-notch SWIFT isolation waveform that was applied to BE, BE-d3, COC,
COC-d3, CE, and CE-d3 standards that were spiked (1 |iL; 1 ag/mL each) onto blank
brain tissue and airbrushed with DHB matrix with both SWIFT amplitudes at (a) 0.0
Vp-p, (b) 0.5 Vp-p, and (c) 0.6 Vp-p.















283.34
287.22


0.0 Vp.p
NL: 3.26E4


313.19


291.09


296.21 302.19 308.35 311.19


290.23 293.22




294.23
287.23


50


0-
0 100


0 50

50
100



100^


285.12


304.25 307.23

308.27


301.19


21 304.22 307.22




294.21 30
/ 297.16 301.16 /A


314.29 321.23 329.17


318.27



313.15
Ai


321.26
0.5 Vpp
NL: 6.55E3

Q230 0


11.18 314.26
323.4
318.24 321.24
0.
NL

8.24 313.13 322.20
31430 I A


4 329.16


5Vp-p
: 5.18E3


280 285 290 295 300 305 310 315 320 325 330
m/z


Figure 4-21. Mass spectra (m/z 280 to 330) of two-stage SWIFT isolation at different
amplitudes. Two-stage isolation composed of a HMF SWIFT excitation waveform
and a hexa-notch SWIFT isolation waveform that was applied to BE, BE-d3, COC,
COC-d3, CE, and CE-d3 standards that were spiked (1 liL; 1 alg/mL each) onto blank
brain tissue and airbrushed with DHB matrix with both SWIFT amplitudes at (a) 0.0
Vp-p, (b) 0.5 Vp-p, and (c) 0.6 Vp-p.


282.38
100 (
1(a)


290.20 293.




~ t


(b)




282.33


(c)


.


)


623.16











100l (a)
.)
o 80
CO


40

a 20-
ry


0.6 Vp-p
CID= 0
NL: 3.68E4


283.11
264.36
__ n ___


196.21


0.6 Vpp
CID = 90
NL: 1.43E3 182.18


168.18
i


108.20 139.17 150.18
~ u


199.23



200.24




214.07 245.11


304.24 318.26
2321.26
293.24 I


323.20


304.23 318.26
290.19 1 I


Figure 4-22. MS/MS product spectra from the application of a two-stage isolation. Two-stage
SWIFT is composed of a HMF SWIFT excitation waveform and a hexa-notch SWIFT
isolation waveform with both SWIFT amplitudes at 0.6 Vp-p and collision-induced
dissociation (CID) set to (a) 0 and (b) 90.










137.13


80-

60O

40

20- 122.21
104.24
100 (b)
1001 (b)


182.19


168.19


1 19.14 150.18

100 150


Wide Isolation
CID = 55
NL: 9.10E3


245.10


304.26 318.29
271.15 290.25 I 323.25


SWIFT Isolation
CID = 90
NL: 2.17E3


304.24 318.26
290.22
_LJUJU


147.07

182.21 196
156.99 200.
168.19 .2

196.21


231.05


199.24


200.24


L 214.07

200


Figure 4-23. Comparison of MS/MS with wide isolation and two-stage SWIFT isolation.
MS/MS product spectra from the application of a (a) 40-Da wide isolation with CID
= 55 and a (b) two-stage isolation composed of a HMF SWIFT excitation waveform
and a hexa-notch SWIFT isolation waveform with both SWIFT amplitudes at 0.6 Vp-p
and CID = 90.











291.05


290.30
29 0 293.24



296.19

287.22 29
AAA~lW


304.29


307.28


100
,)
o 80
CO
= 60

S40

' 20

0
100


300.97


8


304.24
I 307.


290.24 293.24






29


)4.24
297.20 302.20


318.29
321.28


311.09
313.14



314.31


318.26


25 321.26
SV
NI




308.27 315.22 32
313.18
A t.


Vide Isolation
JL: 1.63E4



2.25
323.21




VIFT Isolation
L: 3.64E4


2.24

323.20


280 285 290 295 300 305 310 315 320 325 330
m/z

Figure 4-24. Mass spectra comparison of wide isolation and two-stage SWIFT isolation.
Isolated ions from the application of a (a) 40-Da wide isolation and a (b) two-stage
isolation composed of a HMF SWIFT excitation waveform and a hexa-notch SWIFT
isolation waveform with both SWIFT amplitudes at 0.6 Vp-p.


7.1


283.11


N



32













4.5 (pg) %RSD %RSD ---

4.0 62 3 4
,0
125 11 4
3.5
S 250 5 11
S3.0 -500 0 2 .. ..-
.._. ." "' F. .; 9 9
S2.5 1000 4 6 .. ".--
SI R2 0.9957
1 2.0 -



1.0 -----------------



0.0
0,0
0 100 200 300 400 500 600 700 800 900 1D00
BE Mass (pg)

SSWIFT Isolation -BE A Wide Isolation -BE


Figure 4-25. BE calibration curve for BE spiked on intact brain tissue. Five solutions (mixture
of BE, COC, and CE at 62, 125, 250, 500, and 1000 ng/mL with a mixture of BE-d3,
COC-d3, and CE-d3 at 200 ng/mL) were pipetted in triplicate (1 pL each) onto blank
human brain tissue and airbrushed with DHB matrix. The peak intensities of the
product ions of BE and BE-d3 at m/z 168 and m/z 171, respectively, were ratioed from
two different MS/MS experiments: MS/MS of the ions isolated by a two-stage
SWIFT isolation (CID = 90) and MS/MS of the ions isolated by a 40-Da wide
isolation (CID = 55). The error bars correspond to the standard error (3 replicates).













6.0
62 2 5
125 5 4
5.0 -
S250 6 4
4.0 500 4 3

4.0
0 1000 1 4 5 6 0
3.0 ---- -------- --- 1-- R2 .



( e o BE C .C ad CE a 6, 1, 2 .., and..... n= with a8mi 0 x07





0 100 200 300 400 500 600 700 800 900 1000
COC Mass (pg)

2 SWIFT Isolation COC A Wide Isolation COC


Figure 4-26. COC calibration curve for COC spiked on intact brain tissue. Five solutions
(mixture of BE, COC, and CE at 62, 125, 250, 500, and 1000 ng/mL with a mixture
of BE-d3, COC-d3, and CE-d3 at 200 ng/mL) were pipetted in triplicate (1 aiL each)
onto blank human brain tissue and airbrushed with DHB matrix. The peak intensity
of the product ions of COC and COC-d3 at m/z 182 and m/z 185, respectively, were
ratioed from two different MS/MS experiments: MS/MS of the ions isolated by a
two-stage SWIFT isolation (CID = 90) and MS/MS of the ions isolated by a 40-Da
wide isolation (CID = 55). The error bars correspond to the standard error (3
replicates).














5.0 62 2 5

125 5 4
| 1 2 5 5 4 .... ........... .

* v i ---------,------------ ------------ ........................ .................. ... .......... ............... .
4.0 250 6 4

S500 4 3

S3.0 1000 1 4










0.0
n 2.0








0 200 400 600
CE Mass (pg)

SWIFT Isolation CE A Wide Isolation CE


800


1000


Figure 4-27. Five solutions (mixture of BE, COC, and CE at 62, 125, 250, 500, and 1000 ng/mL
with a mixture of BE-d3, COC-d3, and CE-d3 at 200 ng/mL) were pipetted in triplicate
(1 aiL each) onto blank human brain tissue and airbrushed with DHB matrix. The
peak intensity of the product ions of CE and CE-d3 at m/z 196 and m/z 199,
respectively, were ratioed from two different MS/MS experiments: MS/MS of the
ions isolated by a two-stage SWIFT isolation (CID = 90) and MS/MS of the ions
isolated by a 40-Da wide isolation (CID = 55). The error bars correspond to the
standard error (3 replicates).









Table 4-4. Quantification of BE, COC, and CE from Unspiked Human Brain Tissue
MS/MS Two-Stage SWIFT 40-Da Wide Isolation 5-Da Wide Isolation
Analyte Ion Intact Tissue Intact Tissue Tissue Homogenate
(m/z) (ng/g tissue) (ng/g tissue) (ng/g tissue)
BE 168.2 140 170 27014

COC 182.2 230 60 38011

CE 196.2 40 30 43016












Mass
(ng)


% RSD


62 3

125 6

250 3

500 3

1000 8


6.0








30
5.0




r2
" 4.0

-
-i 3.0


"T 2.0


,~ 0 o -llno 00l,. ,. 0 '007 -: 02)
R 0 9'~9


-


---- ---


0 100 200 300 400 500 600 700 800 900 1000

BE Mass (ng)

*RunA ERunB ARunC



Figure 4-28. BE calibration curve for BE standards spiked in blank brain tissue homogenate.


A0


0.0



0,0


I ililllliillliiilll ii II ii II i


...... ..









7.0
Mass
(ng)
6.0
62
125
5.0
250
500
4.0
1000

3.0


2.0




0.0
0 100 21


I


Ai


4
I 4^


' = OC' 635 310 0''005 + 0 02(0 02)
S -' = C.---) :.


30 400 500 600
COC Mass (ng)

*RunA *RunB ARunC


Figure 4-29. COC calibration curve for COC standards spiked into blank brain tissue
homogenate.


% RSD

6


8


-J
___J


1000


Sttttt


.. .










7.0
Mass % RSD ____
(ng) 0 C2 ':,
6.0 6R 0.9999 -
62 4

125 6
f 5.0
250 4

4.0 500 5
1000 3

3.0


2.0 /


1.0


0.0
0 100 200 300 400 500 600 700 800 900 1000

CE Mass (ng)

*RunA ERunB ARunC


Figure 4-30. CE calibration curve for CE standards spiked into blank brain tissue homogenate.









CHAPTER 5
CONCLUSIONS AND FUTURE WORK

Conclusions

The goal of this research was to develop a quantitative mass spectrometric imaging (MSI)

method for determining the regional composition of drugs and their metabolites in postmortem

brain tissue. This research focused on the analysis of cocaine (COC) and two of its major

metabolites, benzoylecgonine (BE) and cocaethylene (CE). COC is the most frequent cause of

drug-related deaths in the United States, so it is of particular interest to the field of postmortem

toxicology.

Conventional quantification of COC in brain tissue involves homogenate preparation,

followed by extraction and/or derivatization. The extracts are then usually analyzed by gas

chromatography/mass spectrometry (GC/MS), liquid chromatography/mass spectrometry

(LC/MS), GC, or LC. Lengthy extraction procedures are required to remove large

concentrations of lipids and other endogenous materials present in the brain, which may interfere

with the analysis. Multiple sample pretreatment steps also allow opportunity for loss of analyte,

and tissue homogenization eliminates spatial information, which could provide histologically-

specific drug distribution.

MSI using matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (MS)

could provide quantitative information about the distribution of COC and its metabolites in brain

tissue more rapidly, with higher spatial resolution, and with less sample loss than conventional

drug analysis methods that involve tissue homogenization. Quantitative MALDI-MS is

challenging, however, because MALDI exhibits irreproducible signal intensities due to

inhomogeneous crystal formation, inconsistent sample preparation, and laser shot-to-shot

variability. It has been shown though, that normalizing the analyte ion signal to that of a









structurally similar internal standard ion (e.g., [M+H] ion of COC and COC-d3) can

dramatically reduce signal variability making quantification by MALDI-MS possible.

Mass spectrometry analysis of brain tissue can be very complicated, especially without the

benefit of extraction and chromatography methods to clean up the sample and separate

compounds. The presence of isobaric ions in samples increases with sample complexity and

may interfere with quantification at low analyte concentrations. Tandem mass spectrometry

(MSn) can improve analyte selectivity and produce higher signal-to-noise ratios, resulting in

lower detection and quantification limits for the analyte. Combining the use of MSn with internal

standards is commonly performed by alternating MSn scans of the analyte and the internal

standard ions, and then ratioing the resulting product ion signals. This method is effective for

use with ionization techniques such as electrospray and atmospheric pressure chemical

ionization; however, due to the shot-to-shot variability of MALDI, acquiring analyte and internal

standard signals in alternating MSn scans may counteract the signal normalizing effects gained

by using an internal standard.

Strategies for isolating the analyte and internal standard ions during a single MSn scan

were investigated in order to improve the precision of MALDI-MSn to allow for quantification of

COC and its metabolites in brain tissue. One strategy was to use a single wide isolation window

(e.g., 5 Da) centered at a mass-to-charge (m/z) between that of the analyte and internal standard

ions. This allows for the simultaneous isolation and collision-induced dissociation (CID) of the

analyte and internal standard ions. For example, for the analysis of COC, a 5-Da isolation

window could be placed at m/z 305.8 to isolate the [M+H]+ ion of COC at m/z 304.2 and the

[M+H] ion of its trideuterated analog, COC-d3 at m/z 307.2. By applying CID across the

isolation window, a MS2 spectrum is produced containing the product ions of COC and COC-d3









at m/z 182.2 and m/z 185.2, respectively. This method was shown to provide improved precision

(- 10 to 20 times reduction in %RSD) for quantitative analysis of COC in postmortem brain

tissue compared with using two alternating MS2 scans that isolate the analyte and internal

standard ions separately. It was also shown that wide isolation can be used for multiple stages of

mass analysis (e.g., MS3) as long as the product ion derived from the deuterated internal standard

maintains the deuterated tag allowing it to be distinguished from the product ion of the analyte.

For MS3 analysis of COC, the product ion at m/z 150 for COC was ratioed with the product ion

at m/z 153 for COC-d3.

Multi-notch SWIFT isolation was investigated as an alternative isolation strategy to wide

isolation for isolating the analyte and internal standard ions during a single MS2 scan for

improved MALDI-MS2 precision. SWIFT isolation has higher mass selectivity than wide

isolation and is able to reduce background ions that may complicate or interfere with MS2

analysis (e.g., isobaric product ions). Also, analysis times and subsequently laser shots can be

reduced as more frequency notches are added to the SWIFT isolation waveform. This can

become very important when quantitatively imaging several analytes from a large tissue sample.

It was determined that multi-notch SWIFT isolation can provide improved precision when

compared to using two alternating MS2 scans that isolate the analyte and internal standard ions

separately. However, it was determined that multi-notch SWIFT isolation was not as precise as

the wide isolation method. This might be due to frequency shifts of the analyte and internal

standard ions from space-charge effects caused by high m/z background ions from the brain

tissue (e.g., lipid region at m/z 700 to 900).

A two-stage SWIFT isolation method was developed that utilizes a high mass filter (HMF)

SWIFT excitation waveform to remove high m/z background ions (e.g., m/z 325 to 2000) during









the first stage of isolation. This has been shown to reduce the frequency shifts of the analyte and

internal standard ions by preventing them from moving outside the notches of the multi-notch

SWIFT isolation waveform. This prevents the ions desired for isolation from being ejected by

the SWIFT isolation waveform and resulted in an increased signal for the isolated ions compared

to the application of a multi-notch SWIFT isolation waveform without the HMF. A hexa-notch

SWIFT isolation waveform was used during the second stage of the two-stage SWIFT isolation

to mass selectively isolate the [M+H]+ ions of BE (m/z 168.2), BE-d3 (m/z 171.2), COC (m/z

304.2), COC-d3 (m/z 307.2), CE (m/z 318.2), and CE-d3 (m/z 321.2). The hexa-notch SWIFT

isolation waveform was able to effectively remove background ions around the analyte and

internal standard ions that may have interfered with MS/MS analysis. The two-stage SWIFT

isolation overall showed similar precision to that of wide isolation when performing MS/MS on

the analyte and internal standards spiked on blank brain tissue. However, the higher mass

selectivity of the two-stage SWIFT isolation affords a significant loss in absolute signal intensity

of the analyte and internal standard ions when it is applied.

Two-stage SWIFT isolation was compared to wide isolation of intact tissue and SPE

extracted homogenized tissue for the MALDI-MS/MS quantification of BE, COC, and CE from

unspiked human brain tissue, whose toxicological analysis indicated the presence of cocaine in

blood. The two-stage SWIFT isolation method showed lower analyte signal per gram of tissue

than the wide isolation method for BE, COC, and CE. The quantification results for the two-

stage SWIFT isolation and the wide isolation of intact tissue was not comparable to that for the

wide isolation analysis of tissue homogenate (Table 4-4). However, the intact tissue methods

required considerably less sample preparation and smaller sample sizes than the tissue

homogenate method. There was no analysis time saved or fewer laser shots fired when









comparing the two-stage SWIFT isolation and wide isolation since the [M+H] ions of BE, BE-

d3, COC, COC-d3, CE, and CE-d3 were all isolated simultaneously during the same MS/MS scan.

In conclusion, a wide isolation method may still be a better choice over SWIFT isolation for

improving MALDI-MS/MS precision for quantification and reducing analysis time, despite the

inclusion of unwanted background ions during isolation.

Future Work

The linear ion trap has a limited ion storage capacity (~107 ions) before coulombic

interactions between stored ions degrade the mass resolution and reduce sensitivity (space-

charge effects). Sensitivity is reduced when space-charge, created by unwanted matrix ions,

limits the total number of analyte ions which may be trapped. Julian and Cooks first applied

SWIFT to the quadrupole ion trap during injection to resonantly eject these matrix ions and to

selectively accumulate and store analyte ions to increase sensitivity and avoid interference.106

For this research, the application of SWIFT during different periods of the LTQ scan function

was explored to include at the beginning of scan, injection period, isolation, activation, and scan

out. All of these periods of the LTQ scan function are included in the programmable trigger

provided by the LTQ software; however, application of SWIFT during the injection period was

unsuccessful and requires further investigation to exploit the sensitivity gains promised by

selective accumulation and storage of analyte ions.

Quantitative imaging of cocaine and its metabolites from brain tissue of a habitual

cocaine user showed no localization of the analytes in the section of the nucleus accumbens

analyzed. A controlled animal study involving lower doses of cocaine may show localization of

cocaine and its metabolites within specific regions of the brain, and provide more information

about the mechanisms of this drug.











APPENDIX A
BETA CALCULATION

Beta (/) is defined precisely by a continued fraction (aqb_conf) expression in terms of a

and q. Since a = 0, this expression simplifies to Equation A-i (same as Equation 3-2):


(/9 2)2


(/9 4)2


(f- 6)2


(8 + )2_
(10 +f)2_ q
(12+/ )2


(A-l)


42
( ( 1 8)2 -
(f? 10) q- -
(f? 12)2


A LabView subprogram, or subVI (VI = virtual instrument), was written to calculate aqb_conf

based on q and f inputs. Figure A-i shows the block diagram of the aqb_conf sub VI, which is a

graphical representation of Equation A-1.


Figure A-1. LabView block diagram of subVI aqb_conf, which is used to calculate the
continued fraction given q and f inputs.


aqb _conf


(4 +f)


(2+/ f)


(6 +f6)











A LabView program called Trap Calculator was written to calculate ft through an iterative

process given a specific q input, such as q = 0.83 for isolation. Trap Calculator makes a first

guess at the value for ft based on the Dehmelt approximation given in Equation A-2, same as

Equation 3-4:



a+ (A-2)

This approximation is assigned the variable xl. Ninety percent of the approximation is assigned

the variable xO. Both xO and xl are applied as ft inputs for the aqb_conf sub VI with the desired

q as the q inputs resulting in the function outputs of fx0 and fxl. The variable x2 is used to store

the next iterative approximation of f, in which x2 = xl ((x0 xl) / (fx0 / fxl 1)). Then xO is

replaced with the value of xl and xl is replaced with the value of x2. This process is repeated

until the absolute value of the difference between x0 and xl is greater than 1x10-7, and then the

variable beta is assigned the value ofxl. Figure A-2 shows the block diagram for the Trap

Calculator LabView program, which is a graphical representation of the iterative process

described to calculate ft.


Figure A-2. Trap Calculator LabView program used to calculate ft through an iterative process.












This process of calculating ft is shown graphically in Figure A-3, where ft was calculated after 5


iterations with q = 0.830. The value for f was plotted versus the number of iterations.


0.76
0.7490310890.7 361617861 0.736161658
0.74 --- ------


0.72 ._ -__ 0.735862704 _0.736161658

0.70


0.68
4a
0.66


0.64 --

0.62 --


0.60
0.586898628
0.58 ---- --- --- --- ---


Number of Iterations


Figure A-3. Iterative calculation of f using the LabView program Trap Calculator.







































167













APPENDIX B

C++ SWIFT PROGRAM




Name: REICH-HexSWIFT.cpp

Author: C++ version by Richard Reich (12 August 2009) using Microsoft Visual C++ 2008 Express Edition

Purpose: This program calculates a hextuplet-notch stored waveform inverse Fourier transform (SWIFT) pulse and
downloads the data points to an arbitrary waveform generator (DS345) to be applied to the X-rods
of the center section of a linear ion trap (LTQ) for isolation of 3 analytes and their corresponding
internal standard ions in the linear ion trap for increased precision during MALDI-MS^n quantitation.

References: This C++ program is adapted from the original C program written by P.T. Palmer (4 Sep 92). Palmer's
program was based on the paper: Chen, L.; Wang, T.C.L.; Ricca, T.L.; Marshall, A.G. "Phase-Modulated
Stored Waveform Inverse FourierTransform Excitation for Trapped Ion Mass Spectrometry"; Anal. Chem.
1987,59, 449-454. The fast Fourier transform (FFT) algorithm and required data format are described
in Vetterling, W.T.; Teukolsky, S.A.; Press, W.H.; Flannery, B.P. "Numerical Recipes in C: the Artof
Scientific Computing", 2nd ed., New York: Cambridge University Press, 1992, pp. 399-413.

Modifications: The following modifications to Peter Palmer's original C SWIFT program were made when writing program
in C++ to remove compiler warnings and errors:

-Removed "huge" from float t[N]; f[N/2+1]; r[N/2+1]; p[N/2+1]; x[N/2+1]; and y[N/2+1], because term
caused several compiling errors. (fixed errors C2146, C4430, C2086, C2371)
-Replaced "gets()" function with "gets_s()" function as recommended by compiler to avoid overrunning
buffer. "_s" suffix stands for secure. (fixed warning C4996)
Replaced "sscanfO" function with "sscanf_s()" function as recommended by compiler to avoid
overrunning buffer. (fixed warning C4996)
Replaced "strcpyO" function with "strcpy_s()" function as recommended by compiler to avoid
overruning buffer. (fixed warning C4996)
Included "cstdlib" source file to identify "exit". (fixed error C3861)
-Removed "ieeeio.h" source file and "download" function for downloading time-domain waveform over an
IEEE-488 to the DS345 waveform generator through a GPIB. Will use RS232 instead.
-Included "system("PAUSE");" at end of program before function definitions to prevent black window
from disappearing before user can read it. Requires user to hit any key to close window.

Operation: 1. Specify the following parameters:
N: number of points in time-domain SWIFT waveform
freq: sampling rate of function generator (kHz)
Ifreq: lower limit of frequency pulse (kHz)
rfreq: upper limit of frequency pulse (kHz)
mag: normalization value for time-domain waveform

2. Setup output file (ascii format) for time and frequency domain waveforms for plotting and inspection.

3. Generate desired frequency-domain waveform. Compute magnitude and phase spectra. Use a quadratic
function for phase modulation. Convert waveform from polar (magnitude and phase) to a complex
arrayof real and imaginary numbers for Fourier transform.

4. Perform inverse Fourier transform to convert SWIFT waveform from frequency domain to the time domain.

5. Reflect the time-domain waveform about its time midpoint to reduce problems with large initial
signal transients.

6. Generate the apodization or smoothing function. This is a quarter-wave sinusoid matched to one-
fourth of the time-domain period, followed by unit weighting for the next half period, followed by
a quarter-wave sinusoid for final one-fourth of the period. This function is designed to force the
time-domain signal smoothly to zero at the beginning and end of the time-domain period.

7. Multiply the time-domain waveform by the apodization function and normalize.



Figure B-1. Program introductory comments. Briefly describes purpose, references,

modifications, and operation of program.













8. Write waveform data to output file. This data includes:
frequency vs magnitude of specified waveform
frequency vs phase of specified waveform
time vs amplitude of computed waveform (reflected, apodized, and normalized)

9. Download the time-domain waveform overan RS232serial port to the function generator.

10. Write waveform data to output file.



/* insert source files from standard C library using the preprocessor #include command */
#include /*(standard input/output header); allows functions from C standard library to be used */
#include /*(standard string header); allows use of string handling and various memory handling functions */
#include /*(standard math header); allows basic mathematical operations */
#include /* added in C++ version to identify function "exit". (fixed error C3861) */

/* define symbolic constants (conventionally name is capitalized) using the preprocessor #define command */


#define SWAP(a,b)tempr = (a); (a)
#define PI 3.141592654
#define N 4096
#define MAX 2047
#define LF1 398.681641
#define RF1 401.855469
#define LF2 405.029297
#define RF2 408.203125
#define LF3 431.152344
#define RF3 435.058594
#define LF4 439.208984
#define RF4 443.603516
#define LF5 475.341797
#define RF5 481.445312
#define LF6 488.037109
#define RF6 494.873047
#define SF 1000
#define V 1.0


= (b); (b) = tempr /* function swaps values of two variables (a & b) */


/* max # points in SWIFT waveform */
/* max amplitude for SWIFT waveform */
/* default value for lower limit of first frequency band (kHZ) */
/* default value for upper limit of first frequency band (kHz) */
/* default value for lower limit of second frequency band (kHz) */
/* default value for upper limit of second frequency band (kHz) */
/* default value for lower limit of third frequency band (kHZ) */
/* default value for upper limit of third frequency band (kHz) */
/* default value for lower limit of fourth frequency band (kHz) */
/* default value for upper limit of fourth frequency band (kHz) */
/* default value for lower limit of fifth frequency band (kHz) */
/* default value for upper limit of fifth frequency band (kHz) */
/* default value for lower limit of sixth frequency band (kHz) */
/* default value for upper limit of sixth frequency band (kHz) */
/* default sampling frequency (kHz) for DS345 arbitrary waveform generator (AWG) */
/* default value for voltage of output waveform */


/* declaration of variables (property of variable and name); int=integer (whole #), float=floating point (# with fraction) */
unsigned int freq;/* DS345 sampling frequency(kHz); unsigned means no negative values */
float vpp; /* peak to peak voltage of SWIFT waveform */
float t[N]; /*time vector for plotting time domain axis in microseconds */
float f[N/2 + 1]; /* frequency vector for plotting freq domain axis in kHz */
float r[N/2 + 1]; /* magnitude spectrum of SWIFT waveform (polar coordinate) */
float p[N/2 + 1]; /* phase spectrum of SWIFT waveform (polar coordinate) */
float x[N/2 + 1]; /* real portion of SWIFT waveform (cartesian coordinate) */
float y[N/2 + 1]; /* imaginary portion of SWIFT waveform (cartesian coordinate) */
float data[2*N]; /* SWIFT waveform of N complex points (real, imag) */
float apod[N]; /* apodization or smoothing function */
intwave[N+1]; /* scaled, int format of SWIFT with checksumat end */

void fft(float data[], int n, int isign); /* void means that function doesn't return a value */

void main()


Figure B-2. Definition of constants and declaration of variables.

Figure B-2. Definition of constants and declaration of variables.














/**********************************************************************************

VARIABLE DEFINITIONS



int sflag; /* flag to indicate SWIFT waveform type */
float tempr; /* temporary variable for swap function */
char inbuf[20]; /* input buffer array for character strings */
float llfreq; /* lower limit of first frequency pulse (kHz) */
float rlfreq; /* upper limit of first frequency pulse (kHz) */
float 12freq; /* lower limit of second frequency pulse (kHz) */
float r2freq; /* upper limit of second frequency pulse (kHz) */
float 13freq; /* lower limit of third frequency pulse (kHz) */
float r3freq; /* upper limit of third frequency pulse (kHz) */
float 14freq; /* lower limit of fourth frequency pulse (kHz) */
float r4freq; /* upper limit of fourth frequency pulse (kHz) */
float 15freq; /* lower limit of fifth frequency pulse (kHz) */
float r5freq; /* upper limit of fifth frequency pulse (kHz) */
float 16freq; /* lower limit of sixth frequency pulse (kHz) */
float r6freq; /* upper limit of sixth frequency pulse (kHz) */
int errflag; /* flag to denote input errors */
float freqmax; /* Nyquist freqor maximum possible fast Fourier transform (FFT) freq (kHz) */
float deltat; /* time interval in microseconds*/
float deltaf; /* frequency interval in kHz */
int Ilindex; /* lower index of first frequency band */
int rlindex; /* upper index of first frequency band */
int 12index; /* lower index of second frequency band */
int r2index; /* upper index of second frequency band */
int 13index; /* lower index of third frequency band */
int r3index; /* upper index of third frequency band */
int 14index; /* lower index of fourth frequency band */
int r4index; /* upper index of fourth frequency band */
int ISindex; /* lower index of fifth frequency band */
int r5index; /* upper index of fifth frequency band */
int 16index; /* lower index of sixth frequency band */
int r6index; /* upper index of sixth frequency band */
float a; /* coefficientof x term for quadratic phase modulation */
float b; /* coefficientof x^2 term for quadratic phase modulation */
int i,j; /* array indices */
char fileflag; /* flag to indicate output desired */
char ascfile[20]; /* filename for ASCII file output of SWIFT waveform */
char binfile[20]; /* filename for binary file output of SWIFT waveform */
FILE *ascfp; /* file pointer for ASCII file output */
FILE *binfp; /* file pointer for binary file output */
int dlflag; /* flag to denote downloading of SWIFT waveform */
float max; /* maximum amplitude (volts) in SWIFT waveform */
float normal, norm2; /* normalization factors computed from time-domain maximum */
int checksum; /* checksum value for downloading SWIFT to AWG (DS345) */



Figure B-3. Variable definitions.














/**********************************************************************************

INITIALIZE VARIABLES



printf("This program can generate SWIFT waveforms for excitation or isolation.\n");
printf("Would you like to generate a SWIFT waveform for excitation (default = Y)? ");

gets_s(inbuf, 20);
if ((strstr(inbuf,"N") != NULL) II (strstr(inbuf,"n") != NULL)) sflag = 1; /* set flag to 1 if input is N or n */
else sflag = 0; /* if input is other than N or n, set flag to 0 */

printf("Lower limit of first frequency notch in kHz (default = %i)? ", LF1);
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
i = strlen(inbuf);/* counts length of input string from user */
if(i == 0) llfreq= LF1;/* if no input from user(i.e, string length is equal to 0), then default LF1 value is used */
else sscanf_s(inbuf, "%f", &llfreq, 20); /* read input from user, format it, and store in variable llfreq */

printf("Upper limit of first frequency notch in kHz (default = %i)? ", RF1);
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
i = strlen(inbuf); /* counts length of input string from user */
if(i ==0) rlfreq= RF1;/* if no input from user(i.e, string length is equal to 0), then default RF1 value is used */
else sscanf_s(inbuf, "%f", &rlfreq, 20);/* read input from user, format it, and store in variable rlfreq */

printf("Lower limit of second frequency notch in kHz (default = %i)? ", LF2);
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
i= strlen(inbuf); /* counts length of input string from user */
if(i ==0) 12freq = LF2;/* if no input from user (i.e, string length is equal to 0), then default LF2 value is used */
else sscanf_s(inbuf, "%f", &12freq, 20); /* read input from user, format it, and store in variable 12freq */

printf("Upper limit of second frequency notch in kHz (default = %i)? ", RF2);
gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */
i= strlen(inbuf); /* counts length of input string from user */
if (i ==0) r2freq = RF2;/* if no input from user (i.e, string length is equal to 0), then default RF2 value is used */
else sscanf_s(inbuf, "%f", &r2freq, 20); /* read input from user, format it, and store in variable r2freq */

printf("Lower limit of third frequency notch in kHz (default = %i)? ", LF3);
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
i = strlen(inbuf); /* counts length of input string from user */
if (i == 0) 13freq = LF3; /* if no input from user (i.e, string length is equal to 0), then default LF3 value is used */
else sscanf_s(inbuf, "%f", &13freq, 20); /* read input from user, format it, and store in variable 13freq */

printf("Upper limit of third frequency notch in kHz (default = %i)? ", RF3);
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
i= strlen(inbuf);/* counts length of input string from user */
if (i == 0) r3freq= RF3; /* if no input from user (i.e, string length is equal to 0), then default RF3 value is used */
else sscanf_s(inbuf, "%f", &r3freq, 20);/* read input from user, format it, and store in variable r3freq*/

printf("Lower limit of fourth frequency notch in kHz (default = %i)? ", LF4);
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
i = strlen(inbuf); /* counts length of input string from user */
if(i ==0) 14freq = LF4;/* if no input from user (i.e, string length is equal to 0), then default LF4 value is used */
else sscanf_s(inbuf, "%f", &14freq, 20); /* read input from user, format it, and store in variable 14freq */



Figure B-4. Initialize variables for notches 1 through 4.














printf("Upper limit of fourth frequency notch in kHz (default =%i)? ", RF4);
gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */
i= strlen(inbuf);/* counts length of input string from user */
if (i ==0) r4freq= RF4;/* if no input from user (i.e, string length is equal to 0), then default RF4 value is used */
else sscanf_s(inbuf, "%f", &r4freq, 20);/* read input from user, format it, and store in variable r4freq*/

printf("Lower limit of fifth frequency notch in kHz (default =%i)? ", LF5);
gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */
i = strlen(inbuf); /* counts length of input string from user */
if (i ==0) 15freq= LF5; /* if no input from user (i.e, string length is equal to 0), then default LF5 value is used */
else sscanf_s(inbuf, "%f", &15freq, 20);/* read input from user, format it, and store in variable 15freq*/

printf("Upper limit of fifth frequency notch in kHz (default = %i)? ", RF5);
gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */
i = strlen(inbuf);/* counts length of input string from user */
if (i == 0) r5freq= RF5; /* if no input from user (i.e, string length is equal to 0), then default RF5 value is used */
else sscanf_s(inbuf, "%f", &r5freq, 20);/* read input from user, format it, and store in variable r5freq */

printf("Lower limit of sixth frequency notch in kHz (default = %i)? ", LF6);
gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */
i= strlen(inbuf); /* counts length of input string from user */
if(i == 0) 16freq = LF6;/* if no input from user (i.e, string length is equal to 0), then default LF6 value is used */
else sscanf_s(inbuf, "%f", &16freq, 20);/* read input from user, format it, and store in variable 16freq*/

printf("Upper limit of sixth frequency notch in kHz (default = %i)? ", RF6);
gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */
i= strlen(inbuf); /* counts length of input string from user */
if(i ==0) r6freq= RF6;/* if no input from user(i.e, string length is equal to 0), then default RF6 value is used */
else sscanfs(inbuf, "%f", &r6freq, 20);/* read input from user, format it, and store in variable r6freq*/


Figure B-5. Initialize variables for notches 5 and 6.













if (rlfreq< llfreq I r2freq< 12freq I r3freq< 13freq I r4freq <14freq || r5freq <15freq || r6freq <16freq) /* checks to make
sure that the upper limit is not less than the lower limit of freq notch */
{
printf("Your upper limit of your frequency notch cannot be less than your lower limit.\n");
printf("This program will now terminate,\n");
exit(-1);/* causes normal program termination */
}

do /* Do-while loop for acquiring appropriate sampling frequencyfrom user */
{
errflag= 0; /* presets errflag to zero */
printf("Sampling frequencyfor DS345 in kHz (default = 1000)?", SF);
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
i= strlen(inbuf); /* counts length of input string from user */
if (i ==0) freq= SF; /* if no input from user (i.e, string length is equal to 0), then default SF value is used */
else
{
sscanf_s(inbuf, "%u", &freq, 20);/* read input from user, format it, & store in variable freq */
if(freq & 1) /* assign to freq the address of 1 */
{
printf("Number must be a multiple of 2 try again\n");
errflag= 1;
}
else if (freq >40000)
{
printf("Number must be less than or equal to 40,000 try again\n");
errflag =1;
}
}
}
while (errflag == 1);

printf("Peak-to-peak voltage of output waveform (default = %2.1f)? ", V);
gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */
i= strlen(inbuf); /* counts length of input string from user */
if (i == 0) vpp = V; /* if no input from user (i.e, string length is equal to 0), then default V value is used */
else sscanf_s(inbuf, "%f", &vpp, 20);/* read input from user, format it, and store in variable vpp */

freqmax = freq/2;/* in kHz */
deltat = 1000/freq; /* in microseconds */
deltaf= 2 freqmax/N; /* in kHz */
Ilindex= floor(llfreq/deltaf); /* floor rounds down to nearest integer; sets lower index of 1st freq band */
rlindex ceil(rlfreq/deltaf); /* ceil rounds up to nearest integer; sets uppper index of 1st freq band */
12index =floor(12freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 2nd freq band */
r2index= ceil(r2freq/deltaf); /* ceil rounds up to nearest integer; sets upper index of 2nd freq band */
13index floor(13freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 3rd freq band */
r3index= ceil(r3freq/deltaf); /* ceil rounds up to nearest integer; sets uppper index of 3rd freq band */
14index= floor(14freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 4th freq band */
r4index ceil(r4freq/deltaf); /* ceil rounds up to nearest integer; sets upper index of 4th freq band */
15index floor(15freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 5th freq band */
r5index= ceil(r5freq/deltaf); /* ceil rounds up to nearest integer; sets uppper index of 5th freq band */
16index= floor(16freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 6th freq band */
r6index ceil(r6freq/deltaf); /* ceil rounds up to nearest integer; sets upper index of 6th freq band */



Figure B-6. Test initialized variables.













/**********************************************************************************

SETUP OUTPUT FILES

********************************************************************************************************/

fileflag= 0; /* fileflag stores the type of output files desired (1 = ASCII, 2= binary) */

printf("Write SWIFT waveform data to an ASCII file (default = N)? ");
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
if( (strstr(inbuf, "Y") != NULL) II (strstr(inbuf, "y") != NULL))
{
fileflag= 1;

do
{
printf("Filename for the ASCII data file (default =SWIFT.DAT)? ");
gets_s(inbuf, 20); /* reads the input from the user & stores it in the character string "inbuf" */
i= strlen(inbuf); /* counts length of input string from user */
sscanf_s(inbuf, "%s", ascfile, 20);
if (i == 0)/* if no input from user, then ascfile set to SWIFT.DAT */
strcpy_s(ascfile, "SWIFT.DAT");
if (i > 16)
printf("Filename must be less than or equal to 16 characters- try again\n");
}
while (i > 16);

if( (strstr(ascfile, ".DAT") == NULL) && (strstr(ascfile, ".dat") == NULL) )
strcat(ascfile, ".DAT");
if( (ascfp = fopen(ascfile, "wt")) == NULL)
{
printf("Error could not open %s\n", ascfile);
fileflag= 0;
}
}

printf("Write SWIFT waveform data to a binary file (default = N)? ");
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
if( (strstr(inbuf, "Y") != NULL) II (strstr(inbuf, "y") != NULL))
{
fileflag= fileflag + 2;

do
{
printf("Filename for binary data file (default = SWIFT.ARB)? ");
gets_s(inbuf, 20);/* reads the input from the user & stores it in the
character string "inbuf" */
i = strlen(inbuf); /* counts length of input string from user */
sscanf_s(inbuf,"%s", binfile, 20);
if (i == 0) /* if no input from user, then binfile set to SWIFT.ARB */
strcpy(binfile, "SWIFT.ARB");
if(i > 16)
printf("Filename must be less than or equal to 16
characters- please try again\n");
}
while (i > 16);



Figure B-7. Setup output files.













if( (strstr(binfile, ".ARB") == NULL) && (strstr(binfile, ".arb") == NULL))
strcat(binfile, ".ARB");
if( (binfp= fopen(binfile, "wb")) == NULL)
{
printf(" Error-could not open%s\n", binfile);
fileflag= fileflag- 2;
}
}

dlflag =0;
printf("Download SWIFT waveform to the DS345 waveform generator (default = N)? ");
gets_s(inbuf, 20); /* reads the input from the user and stores it in the character string "inbuf" */
if( (strstr(inbuf, "Y") != NULL) II (strstr(inbuf, "y") != NULL) ) dlflag = 1;

/*************************************** *** *** *** *** *** *** ** ** ** ** ** ** ** ** ** ** *********..

GENERATE DESIRED FREQUENCY-DOMAIN WAVEFORM

*******.........*****.********************** *************** *** ** ** ** ** ** ** ** ** ** ** ** *******/

printf("Generating hexa-notch SWIFT waveform ...\n");

for (i = 0; i < N; i++) /* incrementi up to N (number of points in time-domain SWIFT waveform) */
t[i] =i deltat; /* initialize time axis array in microseconds */
if(sflag ==0)/* if SWIFT waveform type is excitation (sflag equal to 0) */
{
a = P/2; /* component "a" of quadratic phase modulation equation */
/*************** Building component 1, 3, 5, 7,9, and 11 of excitation SWIFT waveform******************************/

for (i =0; i <= N/2; i++)
{
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 0; /* zero fill r (magnitude, polar coordinate) array */
p[i] = 0;/* zero fill p (phase, polar coordinate) array */
x[i] = 0;/* zero fill x (real, cartesian coordinate) array */
y[i] = 0;/* zero fill y (imaginary, cartesian coordinate) array */
}

/*************** Building component 2 of excitation SWIFT waveform**************************************/
b = -Pl/(rlindex Ilindex);/* component "b" of quadratic phase modulation equation */
for (i = Ilindex; i <= rlindex; i++)/* increment i up to right-side of 1st freq notch */
{
j = i Ilindex;/* increments used in quadratic function */
r[i]= 1; /* set magnitude array to unity scale later */
p[i]= (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i]= cos(p[i]);/* set real portion of complex number */
y[i]= sin(p[i]); /* set imaginary portion of complex number */
}

/*************** Building component of excitation SWIFT waveform**************************************/
b = -Pl/(r2index 12index);/* component "b" of quadratic phase modulation equation */
for (i= 12index; i <= r2index; i++)/* increment i up to right-side of 2nd freq notch */
{
j = i 12index; /* increments (shifted from i) used in quadratic function */
r[i]= 1;/* set magnitude array to unity scale later */
p[i]= (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i]= cos(p[i]);/* set real portion of complex number */
y[i]= sin(p[i]); /* set imaginary portion of complex number */
}



Figure B-8. Build components 1, 2, 3, 4, 5, 7, 9 and 11 of excitation waveform.













/*************** Buildingcomponent6 of excitation SWIFT waveform**************************************/
b = -Pl/(r3index- 13index);/* component "b" of quadratic phase modulation equation */
for (i= 13index; i <= r3index; i++)/* increment i up to right-side of 3rd freq notch */
{
j= i 13index; /* increments (shifted from i) used in quadratic function */
r[i]= 1; /* set magnitude array to unity scale later */
p[i]= (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i]= cos(p[i]);/* set real portion of complex number */
y[i]= sin(p[i]); /* set imaginary portion of complex number */
}

/*************** Buildingcomponent8 of excitation SWIFT waveform**************************************/
b = -Pl/(r4index- 14index);/* component "b" of quadratic phase modulation equation */
for (i= 14index; i <= r4index; i++)/* increment i up to right-side of 4th freq notch */
{
j= i 14index; /* increments (shifted from i) used in quadratic function */
r[i]= 1; /* set magnitude array to unity scale later */
p[i]= (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i]= cos(p[i]);/* set real portion of complex number */
y[i]= sin(p[i]); /* set imaginary portion of complex number */
}

/*************** Buildingcomponent 10 of excitation SWIFT waveform**************************************/
b = -Pl/(r5index 15index);/* component "b" of quadratic phase modulation equation */
for (i= 15index; i <= r5index; i++)/* increment i up to right-side of 5th freq notch */
{
j= i 15index; /* increments (shifted from i) used in quadratic function */
r[i]= 1; /* set magnitude array to unity scale later */
p[i]= (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i]= cos(p[i]);/* set real portion of complex number */
y[i]= sin(p[i]); /* set imaginary portion of complex number */
}

/*************** Building component 12 of excitation SWIFT waveform**************************************/
b = -Pl/(r6index 16index);/* component "b" of quadratic phase modulation equation */
for (i = 16index; i <= r6index; i++)/* increment i up to right-side of 6th freq notch */
{
j= i 16index; /* increments (shifted from i) used in quadratic function */
r[i]= 1; /* set magnitude array to unity scale later */
p[i]= (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i]= cos(p[i]);/* set real portion of complex number */
y[i]= sin(p[i]); /* set imaginary portion of complex number */
}
}
else /* for isolation SWIFT waveform */


Figure B-9. Build components 6, 8, 10 and 12 of excitation waveform.

Figure B-9. Build components 6, 8, 10 and 12 of excitation waveform.













/*************** Building component 1 of isolation SWIFT waveform*************************************/
a= P1/2; /* component "a" of quadratic phase modulation equation */
b= -Pl/llindex; /* component "b" of quadratic phase modulation equation */
for (i= 0; i < Ilindex; i++)/* increment i up to left-side of 1st freq notch */
{
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 1; /* scale magnitude array to unity scale later */
p[i]= (a i) + (b/2 i i);/* set phase to quadratic function */
x[i] = cos(p[i]);/* set real portion of complex number */
y[i] = sin(p[i]); /* set imaginary portion of complex number */
}
/*************** Building component 2 of isolation SWIFT waveform*************************************/
for (i = Ilindex; i <= rlindex; i++)/* increment i from left-side to right-side of 1st freq notch */
{
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 0; /* set magnitude to zero */
p[i] = 0;/* set phase to zero */
x[i]= 0;/* set real portion of complex number to zero */
y[i] = 0;/* set imaginary portion of complex number to zero */
}
/*************** Buildingcomponent 3 of isolation SWIFT waveform*************************************/
b = -P/(12index rlindex); /* component "b" of quadratic phase modulation */
for (i= rlindex + 1; i <= 12index; i++)/* increment i from right of 1st freq notch to left of 2nd freq notch */
{
j = i index; /* increments (shifted from i) used in quadratic function */
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i]= 1; /* set magnitude array to unity scale later */
p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */
x[i] = cos(p[i]);/* set real portion of complex number */
y[i] = sin(p[i]); /* set imaginary portion of complex number */
}
/*************** Building component of isolation SWIFT waveform*************************************/
for (i= 12index + 1; i <= r2index; i++)/* increment i from left-side to right-side of 2nd freq notch */
{
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 0; /* set magnitude to zero */
p[i] = 0;/* set phase to zero */
x[i]= 0;/* set real portion of complex number to zero */
y[i] = 0;/* set imaginary portion of complex number to zero */
}
/*************** Buildingcomponent 5 of isolation SWIFT waveform*************************************/
b = -P/(13index r2index);/* component "b" of quadratic phase modulation */
for (i= r2index + 1; i <= 13index; i++)/* incrementi from right of 2nd freq notch to left of 3rd freq notch */
{
j= i r2index; /* increment j (shifted from i) used in quadratic function */
f[i] = deltaf i; /* initialize frequencyaxis array in kHz */
r[i] = 1; /* set magnitude array to unity scale later */
p[i] = (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i] = cos(p[i]);/* set real portion of complex number */
y[i] = sin(p[i]); /* set imaginary portion of complex number */


Figure B-1. Build components through 5 of isolation waveform.

Figure B-10. Build components 1 through 5 of isolation waveform.













/*************** Buildingcomponent6 of isolation SWIFT waveform*************************************/
for (i= 13index + 1; i <= r3index; i++)/* increment i from left-side to right-side of 3rd freq notch */
{
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 0; /* set magnitude to zero */
p[i] = 0;/* set phase to zero */
x[i]= 0;/* set real portion of complex number to zero */
y[i] = 0;/* set imaginary portion of complex number to zero */
}
/*************** Buildingcomponent7 of isolation SWIFT waveform*********************************/
b= -Pl/(14index r3index);/* component "b" of quadratic phase modulation */
for (i = r3index + 1; i <= 14index; i++)/* incrementi from right of 3rd freq notch to left of 4th freq notch */
{
j= i r3index;/* increments (shifted from i) used in quadratic function */
f[i] = deltaf i; /* initialize frequencyaxis array in kHz */
r[i] = 1; /* set magnitude array to unity scale later */
p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */
x[i] = cos(p[i]);/* set real portion of complex number */
y[i] = sin(p[i]); /* set imaginary portion of complex number */
}
/*************** Building component of isolation SWIFT waveform*********************************/
for (i= 14index + 1; i <= r4index; i++)/* increment i from left-side to right-side of 4th freq notch */
{
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 0; /* set magnitude to zero */
p[i] = 0;/* set phase to zero */
x[i]= 0;/* set real portion of complex number to zero */
y[i] = 0;/* set imaginary portion of complex number to zero */
}
/*************** Buildingcomponent9 of isolation SWIFT waveform*************************************/
b= -PI/(15index r4index);/* component "b" of quadratic phase modulation */
for (i = r4index + 1; i <= 15index; i++)/* incrementi from right of 4th freq notch to left of 5th freq notch */
{
j= i r4index;/* increments (shifted from i) used in quadratic function */
f[i] = deltaf i; /* initialize frequencyaxis array in kHz */
r[i] = 1; /* set magnitude array to unity scale later */
p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */
x[i] = cos(p[i]);/* set real portion of complex number */
y[i] = sin(p[i]); /* set imaginary portion of complex number */
}
/*************** Building component 10 of isolation SWIFT waveform**************************************/
for (i= 15index + 1; i <= r5index; i++)/* incrementi from left-side to right-side of 5th freq notch */
{
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 0; /* set magnitude to zero */
p[i] = 0;/* set phase to zero */
x[i]= 0;/* set real portion of complex number to zero */
y[i] = 0;/* set imaginary portion of complex number to zero */


Figure B- Build components 6 through 10 of isolation waveform.

Figure B-11. Build components 6 through 10 of isolation waveform.













/*************** Buildingcomponent 11 of isolation SWIFT waveform**************************************/
b = -Pl/(16index rSindex);/* component "b" of quadratic phase modulation */
for (i = r5index + 1; i <= 16index; i++) /* increment i from right of 5th freq notch to left of 6th freq notch */
{
j= i r5index;/* increments (shifted from i) used in quadratic function */
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 1; /* set magnitude array to unity scale later */
p[i] = (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i] = cos(p[i]);/* set real portion of complex number */
y[i] = sin(p[i]); /* set imaginary portion of complex number */
}
/*************** Building component 12 of isolation SWIFT waveform**************************************/
for (i = 16index + 1; i <= r6index; i++)/* increment i from left-side to right-side of 6th freq notch */
{
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 0; /* set magnitude to zero */
p[i] = 0;/* set phase to zero */
x[i] = 0; /* set real portion of complex number to zero */
y[i] = 0; /* set imaginary portion of complex number to zero */
}
/*************** Buildingcomponent 13 of isolation SWIFT waveform**************************************/
b = -PI/(N/2 r6index);/* component "b" of quadratic phase modulation */
for (i = r6index + 1; i <= N/2; i++)/* increment i from right-side of 6th freq notch to N/2 */
{
j= i r6index; /* increments (shifted from i) used in quadratic function */
f[i] = deltaf i; /* initialize frequency axis array in kHz */
r[i] = 1; /* set magnitude array to 1 */
p[i] = (a j) + (b/2 *j *j); /* set phase to quadratic function */
x[i] = cos(p[i]);/* set real portion of complex number */
y[i] = sin(p[i]); /* set imaginary portion of complex number */


Figure B-12. Build components through 13 of isolation waveform.

Figure B-12. Build components 11 through 13 of isolation waveform.














Create data array, which will be filled with frequency data.
The contents of this array are shown below for f = 100 and N =8:

f real data pts imaginary data pts

0 data[0]= x[0] data[l] = y[0]
12.5 data[2]= x[1] data[3]= y[1]
25 data[4]= x[2] data[5] = y[2]
37.5 data[6]= x[3] data[7] = y[3]
50 data[8] = x[4] data[9] = y[4]
-37.5 data[10]= x[3] data[11]= -y[3]
-25 data[12]= x[2] data[13]= -y[2]
-12.5 data[14]= x[1] data[15]= -y[1]

From this example, its obvious (hopefully) why the f, r, p, x, and y arrays
were dimensioned by N+1 (since we need values from f[0] = 0 to f[N/2 + 1] = freqmax).
However, we need only calculate values for the positive frequencies, since we can
compute the values for the negative frequencies, which are simply the complex
conjugate of the positive portion (complex conjugate of (x + yi) = (x yi)).
* ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** ** */
i=0;
for(j=0; j <=N/2;j++)
{
data[i++] = x[j];
data[i++] = y[j];
}
for (j =N/2 1;j >0; j--)

data[i++]= x[j];
data[i++]= -y[j];
}


Figure B-13. Create data array for frequency data.













/************* *** *********** ************************* *************************************************

PERFORM INVERSE FOURIER TRANSFORM TO COMPUTE TIME-DOMAIN WAVEFORM

********* .*.*.*....****************************************** ****************************************/

fft(data-l,N,-1);
for(i= 0; i<2*N; i++)/* normalize results */
data[i] = data[i]/N;

/**********************************************************************************

REFLECT TIME-DOMAIN SIGNAL ABOUT MIDPOINT

*************************************************************** ***************************************1*/

/* enclose the SWAP statement in {} since it compiles to multiple statements */

for(i =0; i< N; i++)
{
SWAP(data[i], data[N+i]);
}

/**********************************************************************************

GENERATE APODIZATION FUNCTION AND SMOOTH TIME-DOMAIN WAVEFORM

*************************************************************** ***************************************1*/

for(i =0; i< N; i++)
{
if(i < .25 N) /* set 1st quarter= 1st quarter of sine wave */
apod[i]= sin(2*PI*i/N);
else if (i < .75 N) /* set next half = unit */
apod[i]= 1;
else /* set last quarter = last quarterof sine wave */
apod[i] = apod[N-l-i]; /* compute from reflection of 1st quarter */
}

for(i=j=0;i< 2*N; i+=2,j++)
{
data[i]= data[i] *apod[j];
data[i+l] = data[i+l] apod[j];


Figure B-14. Inverse Fourier transform, midpoint time reflection, and apodization.


Figure B-14. Inverse Fourier transform, midpoint time reflection, and apodization.













/************* *** *********** ************************* *************************************************

NORMALIZE WAVEFORM AND DOWNLOAD TO FUNCTION GENERATOR

********* .*.*.*....****************************************** ****************************************/

max =0;
for(i=0; i <2*N; i++)
if(fabs(data[i]) > max) max = fabs(data[i]);
normal= MAX/max; /* calc scaling factor for wave (int data) */
norm2 = vpp/max; /* calc scaling factor for data (real data) */
checksum=0;

for (i = 0; i < N; i++) /* norm waveform to give max amp */
{
wave[i]= (int) (data[2*i] normal ;
data[2*i] = data[2*i] norm2;
checksum+= wave[i]; /* and keep running checksum */
}
wave[N] = checksum; /* tack checksum onto end of array */

if(dlflag ==1)
{
printf("Still trying to figure out how to download using RS232 instead of GPIB any suggestions? ...\n");



Figure B-15. Normalize waveform and download to function generator.


Figure B-15. Normalize waveform and download to function generator.













/************.**.**.*********************** .**.**.**.**.**.**.**.**.**.**.**.**.**.**.**.**.**.**.**.**..

WRITE WAVEFORM DATA TO OUTPUT FILES

********* .*.*.*....****************************************** ****************************************/

if (fileflag & 1)
{
printf("Outputting SWIFT data to ASCII file ...\n");

fprintf(ascfp, "\\ DATASET 1: frequency vs magnitude of SWIFT waveform\n");
fprintf(ascfp, "%i %i\n", N/2 + 1, 2); /* print number of rows and columns */
for (i = 0; i <= N/2; i++)
fprintf(ascfp, "%f %f\n", f[i], r[i]);

fprintf(ascfp, "\\ DATASET 2: frequencyvs phase of SWIFT waveform\n");
fprintf(ascfp, "%i %i\n", N/2 + 1, 2);
for (i = 0; i <= N/2; i++)
fprintf(ascfp, "%f %f\n", f[i], p[i]);

fprintf(ascfp, "\\ DATASET 3: time-domain SWIFT waveform (apodized and normalized)\n");
fprintf(ascfp, "%i %i\n", N, 2);
for (i = 0; i < N; i++)
fprintf(ascfp, "%f %f\n", t[i], data[2*i]);

for (i= 0; i < N; i++) /* normalize waveform to give original amplitude */
data[2*i] = data[2*i]/norm2;
fft(data-l,N,1); /* perform forward Fourier transform */
for (i = 0; i <= N/2; i++) /* compute magnitude */
r[i]= sqrt( pow(data[2*i],2) + pow(data[2*i+l],2) );

fprintf(ascfp, "\\ DATASET 4: frequency-domain SWIFT waveform (zero filled, mag mode)\n");
fprintf(ascfp, "%i %i\n", N/2 + 1, 2); /* print number rows and columns */
for (i = 0; i <= N/2; i++)
fprintf(ascfp, "%f %f\n", f[i], r[i]);
fclose(ascfp);
}
if (fileflag & 2)
{
printf("Outputting SWIFT data to binary file ...\n");
fwrite(wave,sizeof(int), N+1, binfp);
fclose(binfp);
}
system("PAUSE");
}


Figure B-16. Write waveform data to output files.














Function: FFT

Purpose: This function performs a discrete fast Fourier transform and is adapted frothe algorithm in "Numberical
Recipes in C", pp. 411-412. If isign is 1, it replaces data with a Fourier transform. If isign is -1,
it replaces data by nn times its inverse Fourier transform. Data is a complex array of length nn input
asa real array data[1..2*nn]. nn must be an integer power of 2 (this is not checked for).

void fft(float data[], int nn, int sign)
{
int n,mmax,m,j,istep,i;
double wtemp,wr,wpr,wpi,wi,theta;
float tempr,tempi;

n=nn<< 1;
j=1;
for(i=1; i {
if (j > i)
(datadata
SWAP(data [j],data[i+]);
SWAP(data[j+1],data[i+1]);
}
m=n>> 1;
while (m >=2 &&j > m)


m >= 1;
}
j+=m;
}
mmax=2;
while (n > mmax) /* perform Danielson-Lanczos transform */
{
istep=2*mmax;
theta=6.28318530717959/(isign*mmax);
wtemp=sin(0.5*theta);
wpr=-2.0*wtemp*wtemp;
wpi=sin(theta);
wr=1.0;
wi=0.0;
for (m=l; m<=mmax; m+=2)
{
for (i=m; i<=n; i+=istep)
{
j=i+mmax;
tempr=wr*data[j]-wi*data[j+1];
tempi=wr*data[j+l]+wi*data[j];
data[j]=data[i]-tempr;
data[j+1]=data[i+1]-tempi;
data[i] += tempr;
data[i+1] += tempi;
}
wr=(wtemp=wr)*wpr-wi*wpi+wr;
wi=wi*wpr+wtemp*wpi+wi;
}
mmax=istep;



FigureB-17. Fast Fourier transform function.


Figure B-17. Fast Fourier transform function.









APPENDIX C
LTQ MODIFICATIONS

The SWIFT waveform is calculated from the C" program in Appendix B and stored in an

arbitrary waveform generator (AWG) (Stanford Research Systems Model DS345, Sunnyvale,

CA, USA). The SWIFT output of the AWG is then connected to the LTQ analog printed circuit

board (PCB). The LTQ analog PCB contains a pair of AD734 multiplier/divider microchips

shown in Figure C-1. Chip U46 multiplies the LTQ isolation waveform signal (X1 X2) by the

isolation waveform gain (Yi Y2). This product is then divided by the denominator interface

(U1 U2) and then the results of chip U64 (Z1 Z2) are subtracted from this quotient. Chip U64

multiplies the resonant ejection/excitation (Res Ej/Ex) waveform (X1 X2) by the Res Ej/Ex

waveform gain (Y1 Y2). This product is then divided by the denominator interface (U1 U2).

The resulting quotient is subtracted by (Z1 Z2), which is usually set to ground. Z2 corresponds

to pin 10 of chip U64. The pin was lifted from the PCB and was wired to the center contact of

the BNC (Bayonet Neill-Concelman) cable connected to the output of the AWG (Figure C-2).

The grounding sheath of the AWG BNC cable was wired to a ground pin on the analog PCB.

When SWIFT is not being applied to the LTQ, a BNC grounding cap is placed on the BNC

connector wired to pin 10 of chip U64.

The LTQ software has a programmable trigger that allows an external waveform to be

triggered during a designated location of the scan function. The location of the programmable

trigger in the LTQ software is in the Diagnostics menu underneath the Tools list. By clicking on

Triggers, the window shown in Figure C-3 appears, which allows the user to input two

arguments (ARG1 and ARG2), which defines the ion trap control language (ITCL) trigger

function, trig(ARG1, ARG2). ARG1 is the trigger location type (e.g., -1 = all triggers off, 0 =

beginning of scan, 1 = injection period, 2 = isolation, 3 = activation, and 4 = scan out), and









ARG2 is the nth position of that type. If no ARG1 or ARG2 is given, the trigger is set at the start

of the analytical scan. If no ARG2 is given, it is set to 0 (the first position of that location type).

The available values for ARG2 will depend on the mode and the trigger location type (Figure C-

3). If ARG2 = -1, then triggers are on for all periods of that trigger location type. Figure C-4

shows a diagram of the mass spectrometer scan function with the different locations labeled with

the corresponding trigger functions values.

The programmable trigger was accessed by connecting a wire to pin 14 of the J1 connector

of the LTQ digital PCB (Figure C-5). The other end of the wire was connected to the center

contact of the BNC cable leading to the trigger input on the back of the AWG. The grounding

sheath of the AWG BNC cable was wired to a ground pin on the digital PCB.

The locations of the programmable trigger were tested by using a two-channel digitizing

oscilloscope with screen capture capability (Model TDS 540, Tektronix Inc., Beaverton, OR,

USA). The trigger pulse from the LTQ digital PCB was connected to channel 1 of the

oscilloscope, and the waveform output after amplification was connected to channel 2 of the

oscilloscope through an RCA (Radio Corporation of America) connector on the LTQ analog

PCB. Figure C-6 shows the oscilloscope image of channels 1 and 2 with the trigger set to the

beginning of the scan [trig(0,0)] with automatic gain control (AGC) on. Note that there are two

different triggers shown for channel 1, one during the beginning of the prescan and the other one

during the beginning of the analytical scan. The end of the prescan is indicated by the large

AGC signal from channel 2, where the radiofrequency (RF) voltage is ramped up to scan out the

trapped ions from the ion trap. Channel 2 also shows two signals, one for the prescan and one

for the analytical scan, that correspond to the LTQ isolation (ISO) waveform.









Figure C-7 shows the oscilloscope image of channels 1 and 2 with the trigger set to the

injection period [trig(1,0)] and AGC on. Channel 1 shows no visible trigger pulses, which

indicate that the programmable trigger is not sending a measurable trigger pulse during the

injection period highlighted in Figure C-4. Channel 2 however does indicate the presence of the

prescan and analytical scan each containing an isolation waveform signal and a scan out signal.

Figure C-8 shows the oscilloscope image of channels 1 and 2 with the trigger set to the

isolation period [trig(2,0)] and AGC on. Channel 1 shows visible trigger pulses that line up with

the isolation waveform signals of the prescan and analytical scan shown from channel 2. Before

the SWIFT isolation waveform can be triggered during the isolation period, it is important to turn

off the LTQ isolation waveform to avoid interference. Under the Diagnostics Tools menu in the

LTQ software, Toggles can be selected (Figure C-9). Then the Isolation waveform can be

highlighted and the off or on radio buttons selected before the Set button is pressed. It is

important to note that if the Define Scan window is opened in the LTQ software, the toggles

return to their default factory settings (i.e., LTQ isolation waveform on).

Figure C-10 shows the mass spectra of cocaine (COC) with the isolation waveform toggled

on (Figure C-lOa) and off (Figure C-lOb). COC standard was spotted (1 pL, 1 ng/iL) onto a

MALDI plate and airbrushed with 2,5-dihydroxybenzoic acid as the matrix. The [M+H] ion of

cocaine at m/z 304 was isolated with a 1-Da isolation window centered at m/z 304. With the

LTQ isolation waveform on (Figure C-lOa), the background ions are ejected leaving a visible

[COC+H]+ ion at m/z 304; however, with the LTQ isolation waveform turned off (Figure C-10b)

the background ions overwhelm the m/z 304 ion and it is not visible. The low mass cutoff

(LMCO) is visible with the LTQ isolation waveform turned off (Figure C-lOb), which is

calculated to be m/z 277 based on m/z 304 being isolated at a q of 0.83, [LMCO =









(0.83)(304)/(0.91) = 277]. Figure C-11 shows the oscilloscope images of channel 1 and 2 with

the trigger at isolation, AGC on, and the LTQ isolation waveform toggled on (Figure C-1 la) and

off (Figure C-11b). Channel 2 of Figure C-1 lb clearly shows that the isolation waveforms

disappear from the prescan and the analytical scan when the LTQ isolation waveform is toggled

off.

Figure C-12 shows the oscilloscope image of channels 1 and 2 with the trigger set at

activation [trig(3,0)] and AGC on. Channel 1 shows that activation occurs between the isolation

event and the scan out event of both the prescan and the analytical scan. Figure C-13 shows the

oscilloscope image of channels 1 and 2 with the trigger set at scan out [trig(4,0)] and AGC on.

Channel 1 shows only a single trigger pulse during the scan out of the analytical scan.












AD734 Multiplier/Divider Chips


Ixi 5
U46 X IU X2
[o
Isolation 1 P Waveform DENOMINATOR U0 E
Waveform X2 DD Output U2 I
SUO W 1I y- i-
_4_ U|1 7Z1-1- YINPUT Y -
Isolation U2 Z2 Y2
Waveform Y1 ER
Gain Y2 VN X ,

2.49K.n 0 =
IIRA W


Res Ej/Ex
Waveform


Res Ej/Ex
Waveform {
Gain


AD734
TOP VIEW
(Not to Scale)


14 VP POSITIVE SUPPLY
13DD DENOMINATOR DISABLE
12W OUTPUT
-1 Zl INPUT
iZ2
9 ER REFERENCE VOLTAGE
8 VN NEGATIVE SUPPLY


(z2 -Z)


Waveform
Output Use BNC grounding cap to ground

) 0- I Pin 10 when SWIFT is not applied.

0-- SWIFT


Figure C-1. Adding SWIFT waveform to AD734 chip (U64) on LTQ Analog PCB.

































Figure C-2. Analog PCB modification to apply SWIFT waveform.













I Tols Tasls
Fkic rea-bxck

FF I.r.e

Di.pla.. .e'Iigs,
Tc-g3es
-EZ TcQc Locatiion: J|I T.egeg: Dir
Cy.>^, ^h^ StI al P'orSa i
SMAlDI 1



Trigger Location Type Full Scan SIM (ZoomScan MS) MS/MS (Zoom SIRM)

All Triggers Off

Beginning of Scan 0 0-9 0

Injection Period 0, 1, 2, or 3 injection periods 0-9 0

Isolation No Trigger 0-9 0-2

Activation No Trigger No Trigger 0-2

Scan Out 0 0-9 SIM Windows 0-9 SRM Windows


Figure C-3. LTQ programmable trigger.











Injection Period Trig(1,0)
Trigger Location: All Triggers Off I P
Scan Position: BIEIII Beginning of Scan Trig(0,0)
Beginning of Scan
Injection Period Isolation- Trig(2,0)
Isolation
Activation Activation Scan Out- Trig(4,0)
Scan Iut Trig(3,0)







Radio Frequency (RF) Voltage

Pre-Isolation Waveform

Broadband Waveform -

Collision Induced Dissociation (CID) .
Waveform
Axial Modulation MassAnalysis
Time (ms) '


Figure C-4. Programmable trigger locations.


































192

































Figure C-5. Digital PCB modifications to access programmable trigger.










Trigger at Beginning of Scan
Trigger at beginning of scan; trig (0,0); AGC = on.
rL Stop M Pos 140ms
prescan r analyticalscan


ISO Scan Out

11


CH1+20.0V CH2 51JO M OIiIs
22-Jam-09 1442


Figure C-6. Trigger at beginning of scan.


CH2


LUWIJIII rw

- BW Lirnit
100MHz
Volts/Div

Probe
10X
Voltage
Invert
I0l


CH1 .- 1-i
<10HI















iJL


Trigger at Injection
Trigger at injection period; trig (1,0); AGC = on.
SStop M Pos: 144
4


no visible trigger pulses on scope


Scan Out


ISO

1. J.ht JI


Ms aV aa P r ar ap sa r V


OV CH2 5.00V M 100 i
22-Jan-09 14:53


ms VCH2

C'w iig




100MHz

Vofts/Oiv


lox







<10Hz


Figure C-7. Trigger at injection.


kTr



STrigger


1 lip U


ISO

RC/ I


CH1 +20.


II


*I













Tek


Trigger
1 -., 1



1st trigger
ISO

RCA




CH1+20.0V


Trigger at Isolation
Trigger at isolation; trig (2,0); AGC= on.
Stop M Pos 1441
4-


2nd trigger
Scan


CH2 51ON M l11ksi
22-Jam-09 1560


]ms CH2
Cou~rrl

S BW Limit
EU
100MHz
Volts/Div
Out

Probe
10X
4 Voltage

Invert

C1 \ 1.64
<10HI


Figure C-8. Trigger at isolation.














Took Tesft
Plot ieaback
Set device
-RF tune
Deeice calibratio
Display :erngs

Triggers
Man cabatain
Sy tem evatluaon
v MALDI


Reaback

EIiuse ili r
E ,:, II.) .:l.j I
E. d ..n 'r I -
I. Qe plBowJ'l
IsdLsai e t4',
14 Jicn RF [ I



F.F -. ,. x n
. .I'-,, r',.o .Ik ,p." r

S,.celID esbl* V


Figure C-9. Toggling LTQ isolation waveform on and off


C Off

r DOn


I %rj


SI Camol | Pr | Hop


















Isolation Waveform On
Readback
AGC
Board ID check
Electron multiplier
on gauge check ( Of
Isolation wavelorm
Main RF loop n
Multplier protect
Normalized cll energy
Ready out
RF system
Sheath gas low protect
Show AGC scan
Source ID enable





Isolation Waveform Off

Readbak
AGC
Board ID check
Electron mul tipie
Ion gauge check I T
Isolation waveform
Man RF loop n
Mutiplier protect
Normalized cll energy
Ready out
RF system
Sheath gas flow protect
Show AGC scan
Source ID enable


107.25 123.37 146.39 172.43 190.40 212.23 232.52


NL: 5.37E4 [COC+H]+
NL: 5. 4 304.17


272.25 286.25

NL: 9.83E5


321.99 339.65
317.36


335.40


LMCO






289.45

28 46 |


m/z


Figure C-10. Mass spectra of cocaine with isolation waveform toggled: (a) on and (b) off


C-

60


S40


20











"TeBk stop Ml Pos: 144tms C
(a) Isolation on co r.
Trigger W
-" "i------I --- &W Lr^ "
2nd trigger iSM rz
1sttrigger AGC ISO Scan Out V
ISO
RCA
SVetta


CHM-t--2ul1rV 0c-32 SugOV M 14OOms CHl-l 1.6DV
2--Jan- oS 1SOgl -I(niHz
FTe It. TJ gl Sltop wt Pos: 14(a mans Cal2
\(b) *' Isolationoff cosna a
Trigger

2nd trigger f-B z
Isolation disappears
st trigger AGC from RCA output. -cIl
RCA woes



CHtni--aLOJiV CI2 SAXOV M1 4lOms C-f1 N 1.1W.v
22-Jan-oS 19S2f89 I<0Hrfl


Figure C-lI. Trigger at isolation with isolation toggled: (a) on and (b) off













Tek


Trigger
la-
T1 I

1st trigger
ISO

RCA




CH1+20.0V


Figure C-12.


X-L


Trigger at Activatior
Trigger at activation; trig (3,0); AGC = on.
Stop M Pos 1441
4-


J 2nd tri
Scan Ou


CH2 51.00 M Ik0ns
22-Jamn-0 15a8


1


]ms CH2
Cou~lirr

SBW Limit
ME
100MHz
gger ltstDiv
t

Probe
10X
1f* ~Voltage
Invert

CH1 1. 1.6
<10Hl


Trigger at activation.


--










Trigger at Scan Out
Trigger at scan out; trig (4,0); AGC = on.
Stop M Po 144.0ms


CH2 1SW0 M 10mln
22-Jan-0o 15:16


CH1 \ 1J.6
<10HE


Figure C-13. Trigger at scan out.


Jit


Tek



Trigger




ISO

RCA|
2-*4


CH2
Coupling



100MHz

Iolts/Div

Probe

Voltage
Invert
isa


CH1+20.0V









LIST OF REFERENCES

(1) Karch, S. B. Karch's Pathology of Drug Abuse, 3rd ed.; CRC Press LLC: Boca Raton,
2002.

(2) Substance Abuse and Mental Health Services Administration, Office of Applied Studies
Drug Abuse Warning Network, 2007: Area Profiles of Drug-Related Mortality. HHS
Publication No. SMA 09-4407, DAWN Series D-31. Rockville, MD, 2009.

(3) Human Illnesses and Behavioral Health. http://www.humanillnesses.com (accessed 30
June 2008).

(4) Drummer, O. H.; Gerstamoulos, J. Therap. Drug Monitor. 2002, 24, 199-209.

(5) Stimpfl, T.; Reichel, S. Forensic Sci. Int. 2007, 170, 179-182.

(6) Moriya, F.; Hashimoto, Y. J. Forensic Sci. 1996, 41, 612-616.

(7) Moriya, F.; Hashimoto, Y. J. Forensic Sci. 1996, 41, 129-133.

(8) Spiehler, V. R.; Reed, D. J. Forensic Sci. 1985, 30, 1003-1011.

(9) Giroud, C.; Michaud, K.; Sporkert, F.; Eap, C.; Augsburger, M.; Cardinal, P.; Mangin, P.
J. Anal. Toxicol. 2004, 28, 464-474.

(10) In National Institute on Drug Abuse Research Monograph 163; Majewska, M. D., Ed.,
1996, pp 1-340.

(11) Ritz, M. C.; Lamb, R. T.; Goldberg, S. R.; Kuhar, M. J. Science 1987, 237, 1219-1223.

(12) Kalasinsky, K. S.; Bosy, T. Z.; Schmunk, G. A.; Ang, L.; Adams, V.; Gore, S. B.;
Smialek, J.; Furukawa, Y.; Guttman, M.; Kish, S. J. J. Forensic Sci. 2000, 45, 1041-1048.

(13) Alburges, M. E.; Crouch, D. J.; Andrenyak, D. M.; Wamsley, J. K. Neurochem. Int. 1996,
28, 51-57.

(14) James, C. A.; Breda, M.; Baratte, S.; Casati, M.; Grassi, S.; Pellegatta, B.; Sarati, S.;
Frigerio, E. Chromatographia 2004, 59, S149-S156.

(15) Kusumoto, M.; Ikeda, K.; Nishiya, Y.; Kawamura, Y. Anal. Biochem. 2001, 294, 185-
186.

(16) Schiffer, W. K.; Liebling, C. N. B.; Patel, V.; Dewey, S. L. Nucl. Med. Biol. 2007, 34,
833-847.

(17) Gatley, S. J.; Volkow, N. D. Drug Alcohol Depend. 1998, 51, 97-108.









(18) Sosnovik, D. E.; Weissleder, R. Curr. Opin. Biotechnol. 2007, 18, 4-10.

(19) Wang, G.; Yu, H.; De Man, B. Med. Phys. 2008, 35, 1051-1064.

(20) Gumbleton, M.; Stephens, D. J. Adv. Drug Delivery Rev. 2005, 57, 5-15.

(21) Contag, C. H.; Ross, B. D. J. Magn. Reson. Imaging 2002, 16, 378-387.

(22) Denoyer, A.; Ossant, F.; Arbeille, B.; Fetissof, F.; Patat, F.; Pourcelot, L.; Pisella, P.-J.
Ophthalmic Res. 2008, 40, 298-308.

(23) Som, P.; Oster, Z. H.; Wang, G.-J.; Volkow, N. D.; Sacker, D. F. Life Sci. 1994, 55,
1375-1382.

(24) Garidel, P.; Boese, M. Microsc. Res. Tech. 2007, 70, 336-349.

(25) Mahmoudi, M.; Simchi, A.; Imani, M.; Hafeli, U. O. J. Phys. Chem. C 2009, 113, 8124-
8131.

(26) Niu, G.; Chen, X. Drugs R&D 2008, 9, 351-368.

(27) Chaurand, P.; Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. Toxicol. Pathol. 2005, 33,
92-101.

(28) Pacholski, M. L.; Winograd, N. Chem. Rev. 1999, 99, 2977-3005.

(29) Caprioli, R. M.; Farmer, T. B.; Gile, J. Anal. Chem. 1997, 69, 4751-4760.

(30) Troendle, F. J.; Reddick, C. D.; Yost, R. A. J. Am. Soc. Mass Spectrom. 1999, 10, 1315-
1321.

(31) Reyzer, M. L.; Hsieh, Y.; Ng, K.; Korfmacher, W. A.; Caprioli, R. M. J. Mass Spectrom.
2003, 38, 1081-1092.

(32) Bunch, J.; Clench, M. R.; Richards, D. S. Rapid Commun. Mass Spectrom. 2004, 18,
3051-3060.

(33) Wang, H.-Y. J.; Jackson, S. N.; McEuen, J.; Woods, A. S. Anal. Chem. 2005, 77, 6682-
6686.

(34) Cristoni, S.; Brioschi, M.; Rizzi, A.; Sironi, L.; Gelosa, P.; Tremoli, E.; Bernardi, L. R.;
Banfi, C. Rapid Commun. Mass Spectrom. 2006, 20, 3483-3487.

(35) Crossman, L.; McHugh, N. A.; Hsieh, Y.; Korfmacher, W. A.; Chen, J. Rapid Commun.
Mass Spectrom. 2006, 20, 284-290.









(36) Hsieh, Y.; Casale, R.; Fukuda, E.; Chen, J.; Knemeyer, I.; Wingate, J.; Morrison, R.;
Korfmacher, W. Rapid Commun. Mass Spectrom. 2006, 20, 965-972.

(37) Khatib-Shahidi, S.; Andersson, M.; Herman, J. L.; Gillespie, T. A.; Caprioli, R. M. Anal.
Chem. 2006, 78, 6448-6456.

(38) Drexler, D. M.; Garrett, T. J.; Cantone, J. L.; Diters, R. W.; Mitroka, J. G.; Prieto
Conaway, M. C.; Adams, S. P.; Yost, R. A.; Sanders, M. J. Pharmacol. Toxicol. Methods
2007, 55, 279-288.

(39) Hsieh, Y.; Chen, J.; Korfmacher, W. A. J. Pharmacol. Toxicol. Methods 2007, 55, 193-
200.

(40) Stoeckli, M.; Staab, D.; Schweitzer, A. Int. J. Mass Spectrom. 2007, 260, 195-202.

(41) Chen, J.; Hsieh, Y.; Knemeyer, I.; Crossman, L.; Korfmacher, W. A. Drug Metab. Lett.
2008, 2, 1-4.

(42) Cornett, D. S.; Frappier, S. L.; Caprioli, R. M. Anal. Chem. 2008, 80, 5648-5653.

(43) Hopfgartner, G.; Varesio, E.; Stoeckli, M. Rapid Commun. Mass Spectrom. 2009, 23,
733-736.

(44) Li, F.; Hsieh, Y.; Kang, L.; Sondey, C.; Lachowicz, J.; Korfmacher, W. A. Bioanalysis
2009, 1, 299-307.

(45) Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Science 2004, 306, 471-473.

(46) Nemes, P.; Vertes, A. Anal. Chem. 2007, 79, 8098-8106.

(47) Cooks, R. G.; Ouyang, Z.; Takats, Z.; Wiseman, J. M. Science 2006, 311, 1566-1570.

(48) Zenobi, R.; Knochenmuss, R. Mass Spectrom. Rev. 1998, 17, 337-366.

(49) Niu, S.; Zhang, W.; Chait, B. T. J. Am. Soc. Mass Spectrom. 1998, 9, 1-7.

(50) Berkenkamp, S.; Menzel, C.; Karas, M.; Hillenkamp, F. Rapid Commun. Mass Spectrom.
1997, 11, 1399-1406.

(51) Zhang, W.; Niu, S.; Chait, B. T. J. Am. Soc. Mass Spectrom. 1998, 9, 879-884.

(52) Dreisewerd, K. Chem. Rev. 2003, 103, 395-425.

(53) Schriver, K. E.; Chaurand, P.; Caprioli, R. M. Proceedings of the 51st ASMS Conference
on Mass Spectrometry andAllied Topics: Montreal, Canada, 2003.









(54) Jurchen, J. C.; Rubakhin, S. S.; Sweedler, J. V. J. Am. Soc. Mass Spectrom. 2005, 16,
1654-1659.

(55) Cohen, S. L.; Chait, B. T. Anal. Chem. 1996, 68, 31-37.

(56) Nakanishi, T.; Ohtsu, I.; Furuta, M.; Ando, E.; Nishimura, O. J. Proteome Res. 2005, 4,
743-747.

(57) Baluya, D. L.; Garrett, T. J.; Yost, R. A. Anal. Chem. 2007, 79, 6862-6867.

(58) Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. J. Mass Spectrom. 2003, 38, 699-708.

(59) Landgraf, R. R., Analysis of Lipids in Nerve Tissue by MALDI Tandem Mass
Spectrometric Imaging. Ph.D. Dissertation, University of Florida, Gainesville, FL, 2009.

(60) Garrett, T. J.; Yost, R. A. Anal. Chem. 2006, 78, 2465-2469.

(61) Ayorinde, F. O.; Hambright, P.; Porter, T. N.; Keith, Q. L., Jr. Rapid Commun. Mass
Spectrom. 1999, 13, 2474-2479.

(62) Cohen, L. H.; Gusev, A. I. Anal. Bioanal. Chem. 2002, 373, 571-586.

(63) Karas, M.; Kruger, R. Chem. Rev. 2003, 103, 427-439.

(64) Chapman, J. R. Practical Organic Mass Spectrometry: A Guide for Chemical and
Biochemical Analysis, 2nd ed.; John Wiley and Sons: New York, 1998.

(65) Hensel, R. R.; King, R. C.; Owens, K. G. Rapid Commun. Mass Spectrom. 1997, 11,
1785-1793.

(66) Nicola, A. J.; Gusev, A. I.; Proctor, A.; Jackson, E. K.; Hercules, D. M. Rapid Commun.
Mass Spectrom. 1995, 9, 1164-1171.

(67) Garrett, T. J.; Prieto-Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford, G.;
Yost, R. A. Int. J Mass Spectrom. 2007, 260, 166-176.

(68) Sugiura, Y.; Shimma, S.; Setou, M. Anal. Chem. 2006, 78, 8227-8235.

(69) Hankin, J. A.; Barkley, R. M.; Murphy, R. C. J. Am. Soc. Mass Spectrom. 2007, 18,
1646-1652.

(70) Aerni, H.-R.; Cornett, D. S.; Caprioli, R. M. Anal. Chem. 2006, 78, 827-834.

(71) Puolitaival, S. M.; Burnum, K. E.; Cornett, D. S.; Caprioli, R. M. J. Am. Soc. Mass
Spectrom. 2008, 19, 882-886.









(72) Atkinson, S. J.; Loadman, P. M.; Sutton, C.; Patterson, L. H.; Clench, M. R. Rapid
Commun. Mass Spectrom. 2007, 21, 1271-1276.

(73) Trim, P. J.; Henson, C. M.; Avery, J. L.; McEwen, A.; Snel, M. F.; Claude, E.; Marshall,
P. S.; West, A.; Princivalle, A. P.; Clench, M. R. Anal. Chem. 2008, 80, 8628-8634.

(74) Luxembourg, S. L.; McDonnell, L. A.; Duursma, M. C.; Guo, X.; Heeren, R. M. A. Anal.
Chem. 2003, 75, 2333-2341.

(75) Lemaire, R.; Tabet, J.; Ducoroy, P.; Hendra, J. B.; Salzet, M.; Fournier, I. Anal. Chem.
2006, 78, 809-819.

(76) Onnerfjord, P.; Ekstrom, S.; Bergquist, J.; Nilsson, J.; Laurell, T.; Marko-Varga, G.
Rapid Commun. Mass Spectrom. 1999, 13, 315-322.

(77) Sleno, L.; Volmer, D. A. Rapid Commun. Mass Spectrom. 2006, 20, 1517-1524.

(78) Nicola, A. J.; Gusev, A. I.; Hercules, D. M. Appl. Spectrosc. 1996, 50, 1479-1482.

(79) Ling, Y.-C.; Lin, L.; Chen, Y.-T. Rapid Commun. Mass Spectrom. 1998, 12, 317-327.

(80) Hatsis, P.; Brombacher, S.; Corr, J.; Kovarik, P.; Volmer, D. A. Rapid Commun. Mass
Spectrom. 2003, 17, 2303-2309.

(81) Cui, M.; McCooeye, M. A.; Fraser, C.; Mester, Z. Anal. Chem. 2004, 76, 7143-7148.

(82) Gusev, A. I.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M. Fresenius. J. Anal. Chem.
1996, 354, 455-463.

(83) Kang, M.-J.; Tholey, A.; Heinzle, E. Rapid Commun. Mass Spectrom. 2001, 15, 1327-
1333.

(84) Krutchinsky, A. N.; Chait, B. T. J. Am. Soc. Mass Spectrom. 2002, 129-134.

(85) FinniganTm vMALDI Source Hardware Manual; Thermo Electron Corporation: San Jose,
CA, 2003.

(86) Paul, W.; Steinwedel, H. Apparatus for Separating Charged Particles of Different
Specific Charges. U.S. Patent 2,939,952, June 7, 1960.

(87) Schwartz, J. C.; Senko, M. W.; Syka, J. E. P. J. Am. Soc. Mass Spectrom. 2002, 13, 659-
669.

(88) Stafford, G. C.; Kelley, P. E.; Syka, J. E. P.; Reynolds, W. E.; Todd, J. F. J. Int. J. Mass
Spectrom. Ion Processes 1984, 60, 85-98.









(89) March, R. E.; Todd, J. F. J., Eds. Quadrupole Ion Trap Mass Spectrometry, 2nd ed.; John
Wiley & Sons, Inc.: Hoboken, New Jersey, 2005.

(90) Douglas, D. J.; Frank, A. J.; Mao, D. Mass Spectrom. Rev. 2005, 24, 1-29.

(91) Cox, K. A.; Cleven, C. D.; Cooks, R. G. Int. J Mass Spectrom. Ion Processes 1995, 144,
47-65.

(92) Cole, R., Ed. Electrospray Ionization Mass Spectrometry; John Wiley & Sons, Inc.: New
York, New York, 1997.

(93) Stafford, G.; Taylor, D.; Bradshaw, S.; Syka, J., Proceedings of the 35th ASMS
Conference on Mass Spectrometry and Allied Topics: Denver, CO, 1987.

(94) FinniganTM vMALDI Getting Started; Thermo Electron Corporation: San Jose, CA,
2004.

(95) Kishore, M.; Ghosh, P. Int. J. Mass Spectrom. Ion Processes 1979, 29, 345-350.

(96) March, R.; Todd, T., Eds. Practical Aspects of Ion Trap Mass Spectrometry; CRC Press:
New York, NY, 1995; Vol. 1.

(97) Bier, M.; Schwartz, J.; Zhou, J.; Taylor, D.; Syka, J.; James, M.; Fies, W.; Stafford, G.,
Proceedings of the 43rd ASMS Conference on Mass Spectrometry and Allied Topics:
Atlanta, GA, 1995.

(98) Browne, S. P.; Moore, C. M.; Scheurer, J.; Tebbett, I. R.; Logan, B. K. J. Forensic Sci.
1991, 36, 1662-1665.

(99) Srinivasan, K.; Wang, P.; Eley, A. T.; White, C. A.; Bartlett, M. G. J. Chromatogr. B
2000, 745, 287-303.

(100) Sershen, H.; Reith, M. E. A.; Lajtha, A. Neuropharmacology 1980, 19, 1145-1148.

(101) Cognard, E.; Bouchonnet, S.; Staub, C. J. Pharm. Biomed. Anal. 2006, 41, 925-934.

(102) Ferrer, I.; Furlong, E. T. Environ. Sci. Technol. 2001, 35, 2583-2588.

(103) Reich, R. F.; Cudzilo, K.; Levisky, J. A.; Yost, R. A. J. Am. Soc. Mass Spectrom. 2010,
21, 564-571.

(104) Marshall, A. G.; Wang, T. C. L.; Ricca, T. L. J. Am. Chem. Soc. 1985, 107, 7893-7897.

(105) Guan, S.; Marshall, A. G. Anal. Chem. 1993, 65, 1288-1294.

(106) Julian, R. K., Jr.; Cooks, R. G. Anal. Chem. 1993, 65, 1827-1833.










(107) Soni, M. H.; Cooks, R. G. Anal. Chem. 1994, 66, 2488-2496.

(108) March, R. E. J. Mass Spectrom. 1997, 32, 351-369.

(109) Embree, P. M.; Kimbel, B. C. Language Algorithms for Digital Signal Processing;
Prentice Hall: Englewood Cliffs, NJ, 1991.

(110) Chen, L.; Wang, T. C. L.; Ricca, T. L.; Marshall, A. G. Anal. Chem. 1987, 59, 449-454.

(111) Wang, T. C. L.; Ricca, T. L.; Marshall, A. G. Anal. Chem. 1985, 107, 7893-7897.

(112) Guan, S.; Marshall, A. G. Int. J. Mass Spectrom. Ion Processes 1996, 157/158, 5-37.

(113) Kelley, P. E. Mass Spectrometry Method Using Notch Filter. U.S. Patent 5,134,286, July
28, 1992.

(114) Nappi, M.; Frankevich, V.; Soni, M. H.; Cooks, R. G. Int. J. Mass Spectrom. 1998, 177,
91-104.

(115) Popoulis, A. The Fourier Integral and Its Applications; McGraw-Hill: New York, 1962.

(116) Williams, J. D.; Cox, K. A.; Cooks, R. G. Anal. Chem. 1994, 66, 725-729.









BIOGRAPHICAL SKETCH

Richard Fred Reich, Jr., graduated from the University of

Arizona in December, 1995, with a B.S. in chemistry and a B.A. in

German, and was commissioned in the United States Air Force.

Richard has served 14 years active duty in the Air Force as a

chemical research officer. At his first assignment, he served as an

explosives chemist at the Energetic Materials Branch, Munitions

Directorate, Air Force Research Laboratory at Eglin Air Force Base

in Fort Walton Beach, Florida, from June 1996 July 1999. Richard was the principal

investigator of melt-cast energetic and Deputy Program Manager for the in-house High Energy

Explosives Development Program. He was responsible for formulating and characterizing the

sensitivity and performance of eleven explosive formulations to qualify their use in advanced Air

Force munitions.

Richard was competitively selected by the Air Force to attend the Air Force Institute of

Technology (AFIT) at the University of Florida in August, 1999, to pursue a Master's degree in

analytical chemistry. His graduate thesis involved the trace detection of explosives using

atmospheric pressure chemical ionization tandem mass spectrometry (APCI-MS/MS). He

graduated from the University of Florida in February, 2001, with an M.S. in analytical chemistry.

Richard was then assigned to the Combustion Branch, Propulsion Directorate, Air Force

Research Laboratory at Wright-Patterson Air Force Base in Dayton, Ohio. From February 2001

to June 2004, he served as the Deputy Branch Chief of the Combustion Branch in which he

assisted the Branch Chief in leading 25 scientists and engineers in the development of state-of-

the-art combustor technology for legacy and future Air Force turbine engines. He also served as









the principal chemist responsible for characterization of particulate matter from research

combustors and development of fuel additives designed to mitigate soot production.

From June, 2004, to July, 2007, Richard was assigned to the Department of Chemistry at

the U.S. Air Force Academy in Colorado Springs, Colorado. As Assistant Professor of

chemistry, he taught general chemistry, analytical chemistry, and chemistry of weapons. He was

the course director for analytical chemistry and chemistry of weapons. He also served as the

executive officer of the Department of Chemistry, assisting the Department Head, Colonel Van

Valkenburg, in leading a department of 65 faculty and staff.

In 2007, Richard was competitively selected to attend AFIT at the University of Florida to

pursue his PhD in analytical chemistry with a completion date of August 2010. Upon

graduation, Richard will be assigned to the Air Force Technical Applications Center (AFTAC) at

Patrick AFB, Florida, as the Deputy Chief of the Verification Sciences Division. There he will

be responsible for establishing a new radiochemistry effluent laboratory. Richard is then

scheduled to return to the U.S. Air Force Academy to teach chemistry in 2013. Richard plans to

retire from the Air Force in 2016 after 20 years of service.





PAGE 1

1 QUANTITATIVE IMAGING OF COCAIN E AND ITS METABOLITES IN BRAIN TISSUE BY MATRIX-ASSISTED LASER DESORPTI ON/IONIZATION LINEAR ION TRAP TANDEM MASS SPECTROMETRY By RICHARD FRED REICH, JR. A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Richard Fred Reich, Jr.

PAGE 3

3 To my best friend and wife, Erica, who displayed incredible patience, support, and sacrifice throughout my educational journey. God gave me the greates t gift in the world by placing you in my life.

PAGE 4

4 ACKNOWLEDGMENTS First, I would like to thank my research advi sor, Rick Yost, who welcomed me back into his research group again. I have learned to appreciate the free environment that you allow your students to thrive in during gra duate school. You have allowed me to think independently and to mature with minimal guidance in your research laboratory. I know that the experiences that I have gained under your mentorship will serve me well in the future. I thank my USAFA co-worker and retired forensic toxicologist, Joseph Levisky, for providing me tissue samples from the El Paso County Coroner’s Office in Colorado Springs, Colorado. I thank my undergraduate research as sistants, Kasia Cudzilo and Kyle Cromwell, who helped prepare samples and serv ed as a soundboard for brainstorm ing research ideas. I would also like to thank the Yost research group, past a nd present, for all their help and support. I hope to continue to see you at future ASMS conferences. You made the hospitality suites a lot of fun. I thank my good friend Pat Castle, who encouraged me throughout my PhD program. Even from out-of-state, you were able to keep me physically a nd spiritually fit. I love you, brother. I thank the parishioners of St. Augustine Cathol ic Church, who served as my family away from home. I must also thank my brot her Knights from Council 13900. Thank you for providing me the opportunity to serve you as Gra nd Knight. You have helped me to further develop my leadership skills and provided me experiences that will last a lifetime. Last, thanks go to my wife and my parents. I am very grateful for their patience with my effort and support when I needed them. Erica, you are the love of my life. Thank you for understanding who I am and allowing me to be myself. I look forward to our adventures together at our next assi gnment at Patrick AFB.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ................................................................................................................ ...........8LIST OF FIGURES ............................................................................................................... ..........9ABSTRACT ...................................................................................................................... .............13 CHAPTER 1 INTRODUCTION ................................................................................................................ ..15Cocaine ....................................................................................................................... ............15Cocaine Metabolism ........................................................................................................15Neurobiological Mechanism of Cocaine .........................................................................16Analysis of Drugs of Abuse in Tissue ....................................................................................16Tissue Imaging Techniques ....................................................................................................1 7MALDI-MS Imaging (MALDI-MSI) .....................................................................................19Spatial Resolution ............................................................................................................ 20Tissue Preparation ...........................................................................................................2 1Excision of Tissue ...........................................................................................................2 2Tissue Sectioning and Mounting .....................................................................................22Sample Transfer ............................................................................................................... 23MALDI Matrix ................................................................................................................24Matrix selection ........................................................................................................24Matrix deposition .....................................................................................................25Tissue washing .........................................................................................................25Quantitative MALDI-MS .......................................................................................................26Internal Standards ............................................................................................................ 26Tandem Mass Spectrometry ............................................................................................27MALDI-MSI Instrumentation ................................................................................................27Linear Ion Trap Mass Spectrometry .......................................................................................28Mass-Selective Instability ...............................................................................................29Ion Storage ................................................................................................................... ....30Automatic Gain Control ..................................................................................................31Helium Buffer Gas ..........................................................................................................31Resonance Ejection .........................................................................................................32Mass Analysis ................................................................................................................. .33Isolation ....................................................................................................................3 3Activation .................................................................................................................34Overview of Dissertation ...................................................................................................... ..352 WIDE ISOLATION .............................................................................................................. ..48

PAGE 6

6 Introduction .................................................................................................................. ...........48Experimental .................................................................................................................. .........50Chemicals ..................................................................................................................... ...50Tissue Collection ............................................................................................................. 51Tissue Sectioning and Sample Preparation .....................................................................51Mass Spectrometry ..........................................................................................................52Results and Discussion ........................................................................................................ ...53MS2 and MS3 Mass Spectra of COC and COC-d3 ...........................................................53Improving Signal Reproducibility with Internal Standards .............................................54Increasing Analyte Selectivity with MSn ........................................................................54Combining Internal Standards with MSn using a Wide Isolation Window .....................56Isolation Window Width and Automatic Gain Control ...................................................58Quantification of Cocaine in Po stmortem Human Brain Tissue .....................................60Conclusions ................................................................................................................... ..........633 SWIFT ISOLATION ............................................................................................................. .74Introduction .................................................................................................................. ...........74Experimental .................................................................................................................. .........77Chemicals ..................................................................................................................... ...77Tissue Collection ............................................................................................................. 77Tissue Sectioning and Sample Preparation .....................................................................78Mass Spectrometry ..........................................................................................................78SWIFT Calculation ..........................................................................................................79Inverse Fourier transform .........................................................................................79Quadratic phase modulation .....................................................................................80Temporal spectral inhomogeneity ............................................................................80Apodization ..............................................................................................................82SWIFT Application to LTQ ............................................................................................83Results and Discussion ........................................................................................................ ...83Optimization of a Dual-Notch SWIFT ............................................................................83Frequency optimization ............................................................................................83Burst count optimization ..........................................................................................84Amplitude optimization ............................................................................................86Selective Ion Isolation of Standards on MALDI Plate ....................................................87Improving MALDI Precision with SWIFT .....................................................................87Selective Ion Isolation of Standards on Tissue ................................................................89Conclusions ................................................................................................................... ..........904 QUANTITATIVE ANALYSIS OF DRUGS IN BRAIN TISSUE ......................................100Introduction .................................................................................................................. .........100Experimental .................................................................................................................. .......101Chemicals ..................................................................................................................... .101Tissue Collection ...........................................................................................................10 1Tissue Sectioning and Sample Preparation ...................................................................101Tissue Homogenization .................................................................................................102

PAGE 7

7 Preparation of Standard Solutions .................................................................................102Preparation of Unknown Sample Solution ....................................................................103Solid-Phase Extraction ..................................................................................................103Mass Spectrometry ........................................................................................................104SWIFT Calculation ........................................................................................................105SWIFT Application to LTQ ..........................................................................................106Results and Discussion ........................................................................................................ .106Hexa-Notch SWIFT Isolation ........................................................................................106SWIFT Isolation on Tissue ............................................................................................108Ion Ejection .................................................................................................................. .109Two-Stage Isolation .......................................................................................................112High Mass Filter (HMF) ................................................................................................113Combining HMF with Hexa-Notch SWIFT ..................................................................114MS/MS with Two-Stage Isolation .................................................................................116Comparing Wide Isolation and Two-Stage SWIFT Isolation .......................................118Two-Stage SWIFT MALDI-MS/MS Quantification ....................................................120SPE-MALDI-MS/MS Quantification ............................................................................121Conclusions ................................................................................................................... ........1245 CONCLUSIONS AND FUTURE WORK ...........................................................................160Conclusions ................................................................................................................... ........160Future Work ................................................................................................................... .......164 APPENDIX A BETA CALCULATION ......................................................................................................165B C++ SWIFT PROGRAM .....................................................................................................168C LTQ MODIFICATIONS ......................................................................................................185LIST OF REFERENCES ............................................................................................................ .202BIOGRAPHICAL SKETCH .......................................................................................................209

PAGE 8

8 LIST OF TABLES Table page 4-1 Hexa-notch SWIFT properties based on m/z 305.8 at q = 0.830 .....................................1274-2 Hexa-notch SWIFT properties based on m/z 290.2 at q = 0.830 .....................................1344-3 Hexa-notch SWIFT properties based on m/z 290.2 at q = 0.791 .....................................1364-4 Quantification of BE, COC, and CE from Unspiked Human Brain Tissue .....................156

PAGE 9

9 LIST OF FIGURES Figure page 1-1 Metabolism of cocaine ..................................................................................................... ..381-2 Cocaine’s mechanism of action .........................................................................................391-3 Dopaminergic pathway. ..................................................................................................... 401-4 Cocaine imaging in tissue ................................................................................................. .411-5 Tissue preparation and MA LDI-MS imaging protocol.. ...................................................421-6 Schematic of the LTQ with MALDI source. .....................................................................431-7 Basic design of the twodimensional linear ion trap. .........................................................441-8 Scheme for application of DC, RF tra pping, and AC excitation voltages necessary for operation of the 2D ion trap. ........................................................................................451-9 Mathieu stability diagra m for the linear ion trap. ..............................................................461-10 A simplified scan function for the quadr upole ion trap showing the prescan and the analytical scan which makes up one microscan. ................................................................472-1 Cocaine dissociation pathway. ...........................................................................................652-2 MALDI-MS signal variability with and without internal standards.. ................................662-3 Comparing mass spectra of COC and COC-d3 on MALDI plate and on brain tissue. ......672-4 Fragmentation of the benzyldimethyldodecylammonium ion.102 ......................................682-5 Wide isolation MALDI-MS2..............................................................................................692-6 Images of standards spiked on brain tissue. .......................................................................702-7 Calibration curves for alternating scans MS2 and wide isolation MS2 ..............................712-8 Mass spectrometric image of cocaine in brain tissue. ........................................................722-9 Cocaine quantification .................................................................................................... ...733-1 Temporal spectral i nhomogeneity of SWIFT ....................................................................913-2 Effects of apodization on SWIFT. .....................................................................................923-3 Optimization of frequency notches ....................................................................................93

PAGE 10

10 3-4 Optimization of burst counts .............................................................................................. 943-5 Optimization of SWIFT amplitude ....................................................................................953-6 SWIFT isolation and wide isolation comparison. ..............................................................963-7 SWIFT isolation and wide isolation calibration curves. ....................................................973-8 Quad-notch SWIFT isolation of standards on MALDI plate. ............................................983-9 Quad-notch SWIFT isolation of standards on brain tissue. ...............................................994-1 Solid-phase extraction scheme .........................................................................................1264-2 Frequency domain of hexa-not ch SWIFT isolation waveform ........................................1284-3 Relationship between secular frequency ( ) and q -space. ..............................................1294-4 Variable hexa-notch SWIFT amplitude ( m/ z 305.8 at q = 0.830). ..................................1304-5 Mass spectra ( m/z 80 to 2000) of hexa-notch SWIF T at different amplitudes. ...............1314-6 Mass spectra ( m/z 280 to 330) of hexa-notch SWIF T at different amplitudes.. ..............1324-7 Pseudopotential well depth ( Dx) of the ion trap. ..............................................................1334-8 Isolation window width (Da) determined by the preset q of isolation. ............................1354-9 Hexa-notch SWIFT applied at variable q of isolation. ....................................................1374-10 Variable hexa-notch SWIFT amplitude ( m/z 290.2 at q = 0.791). ...................................1384-11 Mass spectra ( m/z 280 to 330) of hexa-notch SWIFT at different amplitudes. ...............1394-12 Mass spectra ( m/z 80 to 2000) of hexa-notch SWIF T at different amplitudes.. ..............1404-13 Frequency domain of high mass filter (HMF) .................................................................1414-14 Variable amplitude of HMF. ............................................................................................1424-15 Mass spectra ( m/z 80 to 2000) of HMF at different amplitudes. .....................................1434-16 Mass spectra ( m/z 280 to 330) of HMF at different amplitudes.. ....................................1444-17 Frequency domain of tw o-stage SWIFT isolation. ..........................................................1454-18 The time domain of the two-stage SWIFT isolation. .......................................................1464-19 Variable amplitude of two-stage SWIFT isolation. .........................................................147

PAGE 11

11 4-20 Mass spectra ( m/z 80 to 2000) of two-stage SWIFT is olation at different amplitudes ....1484-21 Mass spectra ( m/z 280 to 330) of two-stage SWIFT isol ation at different amplitudes.. ..1494-22 MS/MS product spectra from the app lication of a two-st age isolation. ..........................1504-23 Comparison of MS/MS with wide isol ation and two-stage SWIFT isolation .................1514-24 Mass spectra comparison of wide isol ation and two-stage SWIFT isolation.. ................1524-25 BE calibration curve for BE spiked on intact brain tissue. ..............................................1534-26 COC calibration curve for COC sp iked on intact brain tissue. ........................................1544-27 CE calibration curve for CE spiked on intact brain tissue. ..............................................1554-28 BE calibration curve for BE standards sp iked in blank brain tissue homogenate. ..........1574-29 COC calibration curve for COC standards sp iked into blank brain tissue homogenate. .1584-30 CE calibration curve for CE standards spiked into blank brain tissue homogenate. .......159A-1 LabView block diagram of subVI aqb_conf ....................................................................165A-2 Trap Calculator LabView program used to calculate through an iterative process. .....166A-3 Iterative calculation of using the LabView program Trap Calculator. .........................167B-1 Program introductory comments. .....................................................................................168B-2 Definition of constants and declaration of variables. .......................................................169B-3 Variable definitions. ..................................................................................................... ....170B-4 Initialize variables for notches 1 through 4 ......................................................................171B-5 Initialize variables for notches 5 and 6. ...........................................................................172B-6 Test initialized variables ................................................................................................ ..173B-7 Setup output files. ....................................................................................................... .....174B-8 Build components 1, 2, 3, 4, 5, 7, 9 and 11 of excitation waveform. ..............................175B-9 Build components 6, 8, 10 and 12 of excitation waveform. ............................................176B-10 Build components 1 through 5 of isolation waveform.....................................................177B-11 Build components 6 through 10 of isolation waveform. ..................................................178

PAGE 12

12 B-12 Build components 11 through 13 of isolation waveform.................................................179B-13 Create data array for frequency data. ...............................................................................180B-14 Inverse Fourier transform, midpoi nt time reflection, and apodization. ...........................181B-15 Normalize waveform and download to function generator. ............................................182B-16 Write waveform data to output files. ...............................................................................183B-17 Fast Fourier tr ansform function. ......................................................................................184C-1 Adding SWIFT waveform to AD734 chip (U64) on LTQ Analog PCB. ........................189C-2 Analog PCB modification to apply SWIFT waveform....................................................190C-3 LTQ programmable trigger. .............................................................................................191C-4 Programmable trigger locations. ......................................................................................192C-5 Digital PCB modifications to access programmable trigger. ...........................................193C-6 Trigger at be ginning of scan. ...........................................................................................19 4C-7 Trigger at injection. ..................................................................................................... .....195C-8 Trigger at isolation. ..................................................................................................... .....196C-9 Toggling LTQ isolation waveform on and off. ................................................................197C-10 Mass spectra of cocaine with isolation waveform toggled. .............................................198C-11 Trigger at isolation w ith isolation toggled. ......................................................................199C-12 Trigger at activation. ................................................................................................... .....200C-13 Trigger at scan out. ..................................................................................................... .....201

PAGE 13

13 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy QUANTITATIVE IMAGING OF COCAIN E AND ITS METABOLITES IN BRAIN TISSUE BY MATRIX-ASSISTED LASER DESORPTI ON/IONIZATION LINEAR ION TRAP TANDEM MASS SPECTROMETRY By Richard Fred Reich, Jr. August 2010 Chair: Richard Alan Yost Major: Chemistry Detection of drugs in tissue typically requires extensive samp le preparation in which the tissue is first homogenized, follo wed by drug extraction, before the extracts are finally analyzed by liquid chromatography/mass spectrometry (LC/MS). Directly analyzing drugs in intact tissue would eliminate any complications introduced by sample preparat ion. A matrix-assisted laser desorption/ionization tandem mass spectrometry (MALDI-MSn) method has been developed for the quantification of cocaine and its metabolites pr esent in postmortem brain tissue of a chronic human cocaine user. It is shown that tandem mass spectrometry (MSn) increases selectivity, which is critical for differentia ting analyte ions from b ackground ions such as matrix clusters and endogenous compounds found in brain tissue. It is al so shown that the use of internal standards corrects for signal variability during quant itative MALDI, which can be caused by inhomogeneous crystal formation, inconsistent sample preparation, a nd laser shot-to-shot variability. The MALDI-MSn method developed allows for a single MS2 experiment that uses a wide isolation window to isolate both analyte and in ternal standard target ions. This method is shown to provide improved precision (~10-20 ti mes reduction in percen t relative standard

PAGE 14

14 deviation) for quantitative analysis compared to using two alternating MS2 experiments that separately isolate the target anal yte and internal standard ions. A wide isolation window reduces signal variability when the an alyte and internal standard signals are ratioed. However, the wide isol ation window not only isolates the analyte and internal standard ions, but also other ions that are not of interest. These ions fill up the finite storage capacity of the ion trap and may lead to space-charge effects, which result in reduced resolution and peak shifts that interfere with de tection of the target ions. Since the current instrument software only allows for one isolation window during MSn, a multi-notch isolation waveform that selectively isolates the analyte an d internal standard ions was created to remove the effects of background interferences and boost the sensitivity for analyte and internal standard ions. A multi-notch stored waveform inverse Fourier transform (SWIFT) pulse was calculated with frequency notches corresponding to the secular frequencies of the [M+H]+ ions of cocaine (COC), benzoylecgonine (BE), cocaethylene (C E), and their trideuterated analogs, COC-d3, BEd3, and CE-d3. Multi-notch SWIFT isolation was f ound to have lower precision than wide isolation, which may be caused by frequency shifts of the analyte and internal standard ions from space-charge effects caused by high m/z background ions from th e tissue (e.g., lipids). Finally, a two-stage SWIFT isolation method wa s developed that uses a high-mass filter to eject high m/z background ions before the multi-notch SWIFT isolation is applied. The twostage SWIFT isolation showed similar prec ision to wide isolation for the MALDI-MS2 analysis of cocaine and its metabolites in brain tissue. The two-stage SWIFT isola tion and wide isolation were used to quantitatively image cocaine and its metabolites in postmortem human brain tissue, and were compared to the quantitative anal ysis of human brain tissue homogenate using MALDI-MS2.

PAGE 15

15 CHAPTER 1 INTRODUCTION Cocaine Postmortem toxicology is a special field of fore nsic toxicology that is used to determine whether alcohol, drugs, or other poisons may have caused or contributed to the death of a person. Cocaine is the most frequent cause of drug-related deaths in the Un ited States, either as the direct cause of death or as a contributing factor.1 According to the fe deral Drug Abuse Warning Network survey, 40% of the 11,942 drug-relate d deaths reported in 2007 involved cocaine.2 This explains why cocaine analysis is of particul ar interest to the field of postmortem toxicology. Cocaine Metabolism Understanding the metabolism of cocaine (COC ) and the relative proportion of COC to its detectable metabolites can provide valuable forensic inferences about the extent of prior abuse. For example, an individual with a brain con centration of 8 mg/kg of COC and 0.5 mg/kg of benzoylecgonine (BE; a metabolite of COC) must have taken the drug just before death, because BE does not cross the blood/brain barrier as fr eely as its lipophilic parent compound (COC).1 Conversely, addicts who have ingested large am ounts of COC over several days are usually found to have only modest COC concentr ations, but high concentration of BE.1 The major route of cocaine metabolism (Figur e 1-1) is hydrolysis of COC by hepatic and plasma esterases, with loss of a benzoyl group to give ecgonine methyl ester (EME). The secondary route is spontaneous hydrolysis, probably non-enzymatic, wh ich leads to BE by demethylation. The final degradation of COC, wh ich is a sequel to both the principal and secondary routes of metabolism, leads to ecgon ine. N-demethylation of COC is a minor route leading to norcocaine. The principal metabolite s are therefore BE, EME, and ecgonine itself, which are inactive; and norcocaine which is active, and may be re levant after acute intoxication.1

PAGE 16

16 In the presence of alcohol, a further active metabo lite, cocaethylene (CE) is formed in the liver by a transesterification r eaction which adds an extra methyl group to COC.1 Neurobiological Mechanism of Cocaine Under normal conditions, dopamine (DA) is rele ased by a neuron into the synapse, where it can bind with DA receptors on neighboring neurons3 (Figure 1-2). Normally DA is then recycled back into the transmitting neuron by a specialized protein called the dopamine transporter (DAT). If COC is present, it att aches to the DAT and blocks the normal recycling process, resulting in a build-up of DA in the syna pse which contributes to the pleasurable effects of COC. DA-rich brain regions such as the ve ntral tegmental area (VTA), nucleus accumbens (NAc), and prefrontal cortex are frequent targ ets of COC addiction rese arch. Of particular interest is the pathway consisti ng of dopaminergic neurons originating in the VTA that terminate in the NAc (Figure 1-3). This projection may fu nction as a “reward center” in that it seems to show activation in response to drugs of abuse like COC in addition to na tural rewards like food or sex. Note that research was performed on ti ssue sections from the NAc, a DA-rich area of the striatum, which may contain an accumulation of CO C due to its affinity to bind with the DAT. Analysis of Drugs of Abuse in Tissue A large variety of specimens are collected in the field of postmortem toxicology including blood, liver, brain, and urine.4 For the analysis of drugs of abuse, brain samples show several advantages over all other specimens in postmortem toxicology.5 One advantage is due to the brain being an isolated compartment, which delays putrefaction after death.6 Also, the metabolic activity is lower in the brain than in other tissu es or in blood, resulting in slower decomposition.7 Finally, drugs of abuse establish their effects th rough the central nervous system. Therefore, it can be assumed that concentrations of drugs of abuse found in the brain better reflect drug concentrations at their site of action at the time of death.8

PAGE 17

17 Analysis of drugs of abuse in the brain ha s applications in forensic and postmortem toxicology. Drug concentr ations in the brain may be needed to substantiate fatal overdoses9 and support neurotoxicity studies.10 Direct measurement of drug a nd metabolite concentrations in discrete brain regions can also be used to study the mechanisms of drug action,11 regional distribution,12 and preferential accumulation of drugs.13 Conventional drug analysis in tissue involve s homogenization of the tissue prior to subsequent chromatographic analysis.14 Such sample pretreatments are known to introduce variation in detection due to inhomogeneity of the analyte within the sample matrix.15 Also, homogenization of tissue elimin ates the opportunity to acqui re detailed anatomical and histological information for in situ drug distribution. Imaging t echniques that include mass spectrometric imaging can help provide this information. Tissue Imaging Techniques A number of analytical techni ques are capable of imaging drugs in vivo and ex vivo including positron emission tomography (PET),16 single photon emission computed tomography,17 magnetic resonance imaging,18 x-ray computed tomography,19 optical fluorescence imaging,20 optical bioluminescence imaging,21 ultrasound,22 whole body autoradiography (WBA),23 infrared imaging,24 and magnetically labeled nanoparticles.25 However, disadvantages of these imaging me thods include low sensitivity, low specificity, limited functional and molecular information, poor spatial resolution, and the need for the drug to be labeled with either a radioactive isotop e or a fluorescent tag, which can be time-consuming and costly.26 Figure 1-4 shows cocaine imagi ng in tissue using PET to image [11C]cocaine and WBA to image [3H]cocaine. A specific disadvantage fo r those techniques that require a chemical tag is the need to monitor the tag rather than the intact drug, and th erefore, the ability to differentiate the drug from a metabolite that may ha ve retained the tag is difficult. In addition, a

PAGE 18

18 chemical tag may alter the pharmacological prope rties of the compound, wh ich could affect both bioavailability and localization within the tissue. Mass spectrometric imaging (MSI) has higher molecular specificity compared to other tissue imaging techniques, particularly wh en used in combination with tandem mass spectrometry (MS/MS).27 The high selectivity of the instrume nt eliminates the need for labeling, because the ion (or product ion as in tandem mass spectrometry) is monitored directly and leaves the drug molecule of interest functionally unmodified. An unmodified drug compound also removes the potential interference of fluorescent/r adioactive labels with the biological function (e.g., when the drug must pass through the blood/brain barrier). This analyte specificity of the instrument also provides the ability to simultane ously image drugs and their metabolites due to the parallel detection of multiple analytes. With MSI, an image can be produced for each of the hundreds of detected analytes within the mass spect ral data set. Another advantage is its high sensitivity. Unfortunately, MSI is a destru ctive imaging technique, although only a few molecular monolayers of sample are affected by th e analysis. This charac teristic precludes MSI from being used for in vivo studies. MSI collects chemical data normally associat ed with mass spectrometry, but in a spatially defined manner, and processes that information into chemical image maps. Secondary ion mass spectrometry (SIMS)28 and matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (MS)29 are the two main techniques used with MSI. MALDI-MS has been shown to be very effective for the direct analys is of drugs and their metabolites in tissues.30-44 MALDIMS is currently the most common MSI techniqu e used for mapping pharmaceuticals in tissue, although new MSI techniques may emerge as ne w surface ionization methods are developed. Ambient ionization methods such as de sorption electrospray ionization (DESI)45 and laser

PAGE 19

19 ablation electrospray ionization (LAESI)46 show potential for the in vivo analysis on the surface of skin of organisms with high specif icity. DESI has been used for the in vivo detection of the antihistamine Loratadine from the finger of a person who had taken 10 mg of the drug, 40 min prior to analysis.45 DESI has also been used to locali ze clozapine directly from histological sections of brain, lung, kidney, and tes tis without prior chemical treatment.47 MALDI-MS Imaging (MALDI-MSI) The most commonly used ionization source fo r mass spectrometric imaging is MALDI. Unlike SIMS and DESI, MALDI requires the addition of a matrix to the sample. An advantage of the matrix is that the solvent used to apply th e matrix is used to extract the analyte out of the tissue, not just analyte at the surface. Howeve r, the matrix solvent allows the potential for analyte migration. MALDI is a soft ionization te chnique in which laser en ergy is applied for an instant to a co-crystallized mixt ure of a compound (called a matrix ) and the analyte molecules. A typical matrix is a small organic compound that absorbs at the wavelength of the laser and consequently promotes desorption of the analyte. The ionization mechanisms of MALDI are not fully understood, but have b een critically reviewed.48 In brief, the chromophore of the matrix couples with the laser energy a nd causes a rapid vibrational exc itation that desorbs matrix and analyte molecules from the solid solution. Th e photo-excited matrix molecules are then stabilized through proton transf er to the analyte (e.g., [M+H]+) Figure 1-5 illustrates the overall protocol of a MALDI-MSI experiment. The firs t step in sample preparation for MALDI-MSI involves application of a homogene ous layer of matrix to the sample (Figure 1-5D). The sample is then analyzed by moving it step-wise beneat h a pulsed laser beam (Figure 1-5E) and MALDI mass spectra are acquired from each point (Figure 1-5F). Two-dimensional images may then be obtained by plotting the relative or absolute ion abundance (cons idered to be proportional to analyte concentration) vers us spatial dimensions of X and Y (Figure 1-5G).

PAGE 20

20 Laser wavelength is an important paramete r in MALDI. The most commonly used wavelength is 337 nm from the nitrogen laser, but harmonics of the Nd:YAG laser 1065 nm fundamental (3x, 355 nm and 4x, 266 nm), various excimer laser lines that include XeCl (308 nm), KrF (248 nm), and ArF (193 nm), and infr ared lasers such as carbon dioxide (10.6 m) and Er:YAG (2.94 m) lasers have been employed.48 It has been shown that MALDI mass spectra obtained from UV and IR lase r wavelengths are similar.49 However, IR MALDI requires higher laser pulse energy due to lower MALDI matrix absorpti on, and the sample consumption is also higher.50 Characteristics of IR-MALDI that have b een reported include a greater tendency to form multiply charged high-mass ions, less metastable fragmentation, and adduct ion formation.51 Spatial Resolution Spatial resolution for MALDI-MSI experiments is limited by laser spot size, laser step size, matrix crystal size, and an alyte migration. Spatial resoluti on increases with decreasing laser spot size, but MALDI mass spectromet ers are usually equipped with N2 (337 nm) or tripled Nd:YAG (355 nm) lasers having rela tively large spot sizes (about 100 m diameter). The rate of energy redistribution rapidly increa ses with smaller laser spot sizes and higher laser fluences are required for MALDI to occur.52 These higher laser fluences can cause extensive fragmentation. Laser spot sizes focused to 7-8 m in diameter lead to a decrease in ion yields of two orders of magnitude compared to a normal (100 m) laser spot,53 since the cross secti onal area analyzed is approximately 100 times smaller. In addition, hi gh spatial resolution experiments are more sensitive to analyte migration during the matrix application step and dramatically increase analysis time for whole tissue section analysis.

PAGE 21

21 An alternate approach to increase spatial re solution is by oversampling (using a step size that is smaller than the laser spot width). This method involves first, the complete ablation of the MALDI matrix coating the sample at each sample position and second, moving the sample target a distance less than the diameter of the laser b eam before repeating the process. The reported method enabled commercial MALDI instrume nts with large laser spots sizes (100 m) to image with approximately 25 m imaging spatial resolution.54 Another factor that determines spatial resolution for MALDI-MSI is the size of the matrix crystals formed during the matrix depositi on process. The size of the sample-matrix cocrystals grown is strongly dependent on the sa mple-matrix solution composition and the rate at which the crystals are grown.55 For the majority of MALDI-MSI experiments, the spot size of the laser is such that multiple crystals are sample d in each laser shot, thus the spatial resolution is limited by the laser spot size and not the crystals fo rmed. However, it is still important to avoid non-uniformities in the matrix laye r (crystals), which can cause i onization yields to vary across the sample and hinder the interp retation of spatial information. Some approaches for MALDI matrix application, such as inkjet printing,56, 57 can produce a uniform coating of small crystals. Different approaches for the application of MALDI matrix will be discussed further in the matrix deposition section. Tissue Preparation MALDI-MSI of intact tissue involves prep aration procedures with minimal sample handling, which decreases analyte losses compared to analyses that involve the preparation of tissue homogenates followed by extraction. Nonethel ess, tissue preparation for MALDI-MSI is critical to maintain the inte grity of the spatial arrangement of drug and metabolite compounds within tissue. Mishandling or improperly storin g tissue samples in the early sample preparation

PAGE 22

22 steps may cause delocalization or degradation of the analytes. Experimental procedures that should be considered include excision of tissu e, tissue sectioning, samp le transfer to MALDI target plate or glass microscope slide, matrix application, and tissue st orage after sectioning. Excision of Tissue Tissue samples should be surgically removed so that the original sh ape of the tissue is retained. Immediately after removal, the tissu e may be loosely wrapped in aluminum foil and frozen in liquid nitrogen by gently lowering the tissue into the liquid nitrogen over a period of 30 – 60 seconds.58 Immersing the tissue into the liquid nitr ogen too quickly can lead to cracking and brittle edges. The foil acts to provide support fo r more malleable tissue an d prevents adhesion of the tissue to the sides of the liq uid nitrogen dewar. Freshly exci sed tissue that is placed into small plastic tubes may mold to the shape of th e tube when frozen. Whole tissues may remain frozen in a freezer at -80oC for at least a year with little to no degradation of the sample.58 Tissue Sectioning and Mounting Frozen tissue samples are cut into thin secti ons in a cryostat, which allows for accurate sectioning to be accomplished at sub-freezing temper atures with minimal sample contamination. It is recommended that tissue samples be attach ed to the sample stage (Figure 1-5A) of the cryostat by freeze mounting with a few drops of de ionized water at the interface between the tissue and the stage.59 It is not advised to use an em bedding medium such as agar or OCT (optimal cutting temperature polymer) to mount the tissue to the sample stage, because these compounds could suppress ion formation in MALDI-MS analysis.58 Tissue samples, once mounted to the cryostat sample stage, are sliced with a stainless steel microtome blade. The sample stage temperature is typically maintained between -5 oC and -25 oC, depending on the tissue type. Tissues that have a higher fat content require lower temperatures to avoid tearing during sectioning. Although tissue thickness is not critical for most studies, 10 – 20 m thick

PAGE 23

23 tissue sections are optimal for handling. Analyte signal intens ity has previously been shown to increase with increasing tissue section thic kness; it was hypothesized that, during matrix application, matrix solvent may obtain access to the interior of the tissue to extract more analyte.35 Sample Transfer The tissue section can be transferred with a th in artist’s brush and carefully positioned onto a cold MALDI target plate or glass microscope slid e. Care should be take n during the transfer to avoid folding or tearing the thinly sliced tissue. Tears or rips distort the tissue section and create holes or gaps, which were not present in the nativ e tissue. All equipment that will come into contact with the frozen tissue incl uding the plate or glass slides s hould be kept in the cold box of the cryostat during sectioning. Once the tissue slice is pos itioned on the cold target plate or glass slide, they are removed from the cold box and quickly warmed, thus thaw-mounting the tissue onto the sample plate or slide (F igure 1-5B). Thaw-mounted tissue samples should be stored in a freezer at -80 oC until analyzed. When tissue samples are ready to be an alyzed, they are dehyd rated in a vacuum desiccator at room temperature to remove moisture and avoid latera l migration of analytes before application of MALDI matrix. Traditional low pressure (~ 10-6 Torr) MALDI requires samples to be dried completely (~ 2 hours) before expos ure to vacuum conditions. This prohibits the analysis of freshly cut tissue and reduces sa mple throughput. In addition, low pressure MALDI has been shown to produce in-source fragmentatio n of lipids in tissue, which makes low-level detection difficult (unpublished re sults). MALDI operated at an intermediate pressure (IP) of 0.17 Torr (100,000-times higher than traditional vac uum MALDI) has been shown to reduce the degree of source fragmentation by collisional cooling.60 Tissue drying times with IP-MALDI

PAGE 24

24 can be reduced to 30 minutes, which will increase sample throughput and allow for the analysis of tissue samples shortly after dissection. MALDI Matrix Matrix Selection The success of MALDI-MSI for th e analysis of drugs in tissu e is dependent on the choice of matrix. The common UV-abso rbing molecules used as matr ices for MALDI analysis are benzoic acid-based components with low molecula r weights (< 500 Da) such as sinapinic acid (SA, 3,5-dimethoxy-4-hydroxycinnamic acid), -cyano-4-hydroxycinnamic acid (CHCA), and 2,5-dihydroxybenzoic acid (DHB). Various MALDI matrices, including organic, solid ionic, liquid, and liquid/solid two-phase matrices, have been reviewed.48 Unfortunately, ions formed from most matrix compounds dominate the lo w-mass range background for a typical MALDIMS spectrum, making MS/MS or high resolution MS cr itical for the analysis of small molecules. One approach to circumvent matrix interf erence is to use a higher molecular weight matrix, which does not interfere in the low mass region. To this end, some porphyrins have been employed as MALDI matrices.61 Although the porphyrin matrices have been shown to be valuable for the detection of low-mass analyt es with minimum mass interference from matrix signals,61, 62 poor ion production yield fo r drug molecules in tissues was observed when these porphyrin matrices were employed.63 Because of the nature of biological tissues the growth of matrix crystals is more complicated on tissue than on an inert plate where a small volume of matrix is mixed with a neat drug solution. For example, on tissue, the matr ix solvent not only plays a role in the cocrystallization of the matrix a nd analyte molecules, but the solv ent composition also facilitates the extraction of analyte molecule s to the surface of the tissue. Therefore, selecting a solvent composition that can readily dissolve the analyte is critical for crystal formation as well as

PAGE 25

25 analyte extraction. Solvent composition can also play an active role in protonation of the analyte and result in higher ionization e fficiency. Strong acids such as 0.1% trifluoroacetic acid (TFA) are normally added to the matrix solution to assi st protonation of protei ns, but have been found to have a marginal effect on the ionization efficiency for small molecules.39 For small molecules, a higher matrix concentration (matrixto-analyte ratio) can also produce better quality mass spectra.39 Matrix Deposition The analyte signal intensity, s uppression of the matrix signa l, and laser shot-to-shot reproducibility can be affected by the distributio n of matrix and analyte during crystallization.64 Crystal irregularities can occur when the matr ix/analyte mixture part itions during the slow crystallization process;65 thus, it is very important that th e solubilities of all components are suitably matched. Many sample preparation pr ocedures for improved co-crystallization of matrix and analyte have been repo rted and include electrospraying,65 fast evaporation,66 pneumatic spraying,67 spray-droplet method,68 sublimation,69 inkjet printing,57 acoustic drop ejection,70 and solvent-free matrix dry-coating.71 By far the most common matrix deposition approach for MALDI-MSI of drugs in tissue is pneumatic spraying31, 32, 35-38, 41, 43, 44, 72, 73 with either CHCA, SA, or DHB matrix. Pneumatic spra ying is an inexpensive and easy technique of applying MALDI matrix that is effective at depositing a homogeneous la yer of small matrix crystals across the entire tissue sample. For th e research conducted, MA LDI matrix solution was applied to tissue by an artistic ai rbrush (Aztek A470, Testors; Rockford, IL, USA), Figure 1-5C. The application of MALDI ma trix by airbrush has b een previously published.67 Tissue Washing To optimize matrix crystallization, a washing step is sometimes performed before matrix deposition, which allows the majority of salts to be removed from the surface of the tissue.58

PAGE 26

26 Recent studies have shown that matrix crystalliz ation and analyte incorporation are hampered by the presence of high concentratio ns of salt, which can result in an inhomogeneous sample surface and lead to high signal variability.74 The removal of salt from tissue sections is typically performed by rinsi ng in 70-80% ethanol.58 Improved peptide and protein signals were demonstrated with tissue-washing in organic solvents traditionally used for lipid extraction (i.e., methylene chloride, hexane, toluene, acetone, an d xylene), especially from older or even archived tissue sections.75 However, great care must be take n to prevent migration of analyte molecules or even the loss of analyte; thus tissue washing is not recommended for small molecule applications such as cocaine analysis. Quantitative MALDI-MS Internal Standards Although MALDI-MS is an established method for qualitative analys is, quantitative analysis is more difficult because MALDI exhibits irreproducible analyte si gnals as a result of inhomogeneous crystal formation, inconsistent sample preparation, a nd laser shot-to-shot variability.76 Indeed, relative standard devi ations can be higher than 50%.62, 77 The addition of an internal standard can compensate for severa l of these experimental factors that seriously complicate quantitative MALDI-MS.77-81 An appropriate internal standard for MALDI must compensate not only for any crystallization irregularities but also for subs equent desorption and ga s-phase effects. In choosing an internal standard, the relative polarities of the analytes and internal standard as well as their solvent solubiliti es should be considered.77 Structural similariti es should reflect the gasphase behavior of the involved molecules, and exte nd to solubility. Naturally, an isotope-labeled standard is the ideal choice, si nce its chemical behavior is ne arly identical to its unlabeled counterpart.82 Such a standard guarantees identical crys tallization and gas-phase behavior of the

PAGE 27

27 analyte and internal standard.83 Traditional MALDI experiments demonstrate that using the ratio of the analyte peak intensities to those of a deuter ated internal standa rd can improve signal reproducibility.82 Tandem Mass Spectrometry Another challenge for quantitative MALDI-MS, particularly for the analysis of small molecules such as drugs of abuse, is the strong in terferences for m/z < 500 due to MALDI matrix ions.84 In addition, interferences can originate from a multitude of ions produced from endogenous compounds (e.g., lipids) found in tissue sections during tissue analysis. The high molecular specificity of tandem mass spectrometry (MSn) eliminates the problem of interfering ions by fragmenting the desorbed ions in th e mass spectrometer and matching the fragment masses with the molecular structure of the analyt e. The analytical advantage of the linear ion trap mass spectrometer is the ability to perf orm multiple stages of MS, which provides an increase in molecular specificity with each st age of mass analysis. For this reason, all the research conducted was performed on the lin ear ion trap mass spectrometer and thus the background is a focus on this instrumentation. MALDI-MSI Instrumentation All MSI experiments reported in this dissert ation were performed on a Thermo Scientific LTQ XL linear ion trap (LIT) mass spectrometer (Thermo Scientific; San Jose, CA, USA) with an intermediate pressure MALDI source, as s hown in Figure 1-6 and described in detail in elsewhere.67, 85 The MALDI source uses a nitrogen gas laser that fires pulses at 337.7 nm with a frequency of 60 Hz and energy of 250 J per pulse at 100% laser power An iris attenuator is used to vary the laser power. The laser energy is directed to the MALDI source by a fiber optic cable. It is then focused using a series of mirro rs and lenses to a spot size of approximately 100 m at an incident angle of 32o.67 The LTQ XL MALDI source uses nitrogen gas to maintain a

PAGE 28

28 pressure of 75 mTorr (170 mTorr for LTQ MALD I), which is considerably higher than a standard high vacuum MALDI source (~10-6 Torr), but substantially below that of an atmospheric pressure MALDI source. The sample plate consists of a bottom support plate, which attaches to either a 96or 384well microtiter plate (12.7 cm x 8.6 cm) for gene ral MALDI applications, or a microscope slide holder (2.5 cm x 7.5 cm, 0.1 cm thick) that is desi gned to hold four standard microscope slides for tissue imaging applications. Microscope sl ides are affixed to the slide holder with doublesided tape (Scotch 1.27 cm wide, 3M; Minneapo lis, MN, USA). The MALDI control software automatically identifies which plate configuratio n is being used and calib rates the position of the sample plate. The sample plate mounts onto an XY stage by means of spring tension clamps, and two precision vacuum-rated stepper motors control the two-dimensional movement. These actuators position the XY stage with an accuracy of better than 3 m. The precision in going back to a specific location is 1 m without taking the plate out and approximately 7 m after taking the plate out of the v acuum and putting it back in. A modified set of quadrupole rods, which can accommodate the entrance of the laser beam and access for camera viewing, is added to the front of the LTQ multipole arrangement behind the MALDI sample plate, Figure 1-6. Ions produced from the MALDI process are directed into the LIT mass analyzer through the ion optics consisting of a series of quadrupoles, lenses, and octopole. Linear Ion Trap Mass Spectrometry The LIT is a two-dimensional (2D) quadrupole ion trap (QIT), which is related to the three-dimensional (3D) QIT that was first introd uced and described as a mass storage device in 1953 by Wolfgang Paul and Helmut Steinwedel.86 The LIT operates in a fashion analogous to the QIT. However, unlike the 3D QIT which contains two end cap electrodes and a ring

PAGE 29

29 electrode, the LIT is co mposed of a segmented hyperbolic qu adrupole mass analyzer with three sets of hyperbolic rods87 of lengths 12 mm (front), 37 mm ( center), and 12 mm (back) shown in Figure 1-7. Ions are trapped axially by appl ying separate direct current (DC) voltages (100 V) to all three sections while radial trapping is accomp lished by applying an oscillating radio-frequency (RF) voltage (5 kV rod to ground, 1 MHz) in two phases to the X and Y rod pairs shown in Figure 1-8. A two-phase supplemen tal alternating current (AC) voltage (80 V, 5-500 kHz) is applied across the X rods for isolation, activation, and ejection of ions. I ons are ejected radially from the trap through opposing 30-mm long slits in the center section of X rods by massselective instability scanning.87 Mass-Selective Instability Mass-selective instability scanning88 is accomplished by setting the DC component of the center section rods to zero while the amplitude of the RF resona nce excitation voltage applied to the X rods is increased. As the amplitude of th e RF voltage is increased, the magnitude of the oscillations of the trapped ions also increases so that the i ons eventually develop unstable trajectories along the X axis, and ar e subsequently ejected from th e trap in order of increasing mass-to-charge ( m/z ) value. Ions are ejected through the slits in the center X rods and strike a set of detectors consisting of a conversion dynode and an electron multiplier s ituated at each slit to catch the ejected ions. Ions trapped inside the LIT follow traject ories described by the second-order Mathieu differential equation.89 Solutions to the differential eq uation are in terms of two reduced parameters, a and q which can be used to determine whet her an ion will have a stable or unstable trajectory in the trap under the defined c onditions of the electric field. The values of a

PAGE 30

30 and q depend on the dimensions of the trap and th e potentials applied according to Equations 1-1 and 1-2:89 2 28 o y xmr eU a a (1-1) 2 24 o y xmr eV q q (1-2) U is the applied DC amplitude (and is zero in the LIT), V is the applied RF potential, e is the charge on an ion (1.602 x 10-19 C), m is the mass of an ion, ro is the radius of the hyperbolic rod profiles ( ro = 4 mm), and is the angular drive frequency. Ion Storage From the known solutions to the Mathieu eq uation one can generate a stability diagram (Figure 1-9) that show s the common region in ( a q ) space for which the X and Y components of the ion trajectory are stable simultaneously such that the ion can be confined in the trap.90 The parameters x and y at any given coordinate of a and q relate to the secular frequency of the ion in the X and Y directions respectively (Equation 1-3). u = 0.5 u (1-3) As the value of approaches zero, the ion’s secular frequency approaches zero, and the ion is not contained. When the value of equals one, the ion’s secular frequency equals half the frequency of the RF field, and the magnitude of its oscillation increases so that the ion escapes the trap or collides with one of the electrode surfaces. When has a value between zero and unity, the ion can be trapped by the oscillating fields and will oscill ate in a periodic mode at its secular frequency in x and y

PAGE 31

31 Automatic Gain Control The LIT has a storage cap acity of approximately 107 ions;87 however, the ion trap can trap or hold only a certain nu mber of ions before repulsive forces (space-charge) cause distortions in the applied trapping field, causing a degradation in resolu tion, a reduction in peak height, and a shift in mass assignments.91 At severe space-charge conditions, the mass peaks are further broadened and reduced in peak height to th e point where they disapp ear into the baseline. Beyond the extreme limit of space-charge, the ion de nsity becomes so large that additional ions injected into the quadrupole field may not be trap ped at all, or previously trapped ions may be displaced.92 In order to control the number of ions that accumulate in the trap, a method was developed by Finnigan MAT93 called automatic gain contro l (AGC). AGC quantitatively assesses the ion generation rate by use of a pres can (Figure 1-10), and th en inversely applies a period of ionization (ion injection time) during each operational cycle of the ion trap to ensure that the number of ions in the trap never re aches an adverse level of space-charge. The AGC feature assists in maintaining th e quality of the MALDI spectra by adjusting the number of laser shots per laser spot to pr oduce a similar number of ions for each scan.94 If the number of ions produced per shot is low, more laser shots are fired. If the number of ions produced per shot is high, fewer laser shots are fired. The spectra are normalized so that the displayed signal reflects the actual signal level. Helium Buffer Gas During the ion injection time, ions transmitte d from the ion optics are directed into the LIT where they accumulate before they are scanne d out and detected. The ions enter the trap with a range of kinetic energies. Unless the ions enter the trap at the correct phase angle of the RF drive potential, they will not have the correct combination of velocity and displacement to remain in stable orbits and be trapped.95 Even if they meet these conditions, they still enter the

PAGE 32

32 trap with too much kinetic energy to be trapped forever. To rem ove some of this kinetic energy, a helium buffer gas is introduced into the trap. Th e flow of gas (1 mL/min) into and out of the trap is matched so that the partial pressure of he lium in the mass analyzer cavity is maintained at approximately 1 mTorr. The ions are kinetically c ooled to the center of th e trap (over a period of a few milliseconds) through collisions with the lowmolecular-weight helium atoms. As a result, mass resolution is improved, because the ions are ejected from the LIT in dense ion packets. Resonance Ejection One of the inherent features of the ion trap dur ing the mass selective inst ability scan is that while the ions of lower m/z are being scanned out of the ion trap to the detectors, the higher m/z ions are still in the trap, and the space-charge that they contribute causes a broadening of the peaks formed as the lower m/z ions are being ejected.96 This deleterious effect on peak shape can be reduced dramatically through a technique calle d resonance ejection. Th is is performed on the LTQ by applying a supplementary AC voltage to the X rods in dipolar fashion at a frequency of just less than half of the RF drive frequency (500 kHz, x = 0.843) and amplitude at a resonant ejection qx of 0.88. As ions are scanned along the ax = 0 line by ramping the RF voltage on the X rods, ions of increasing m/z consecutively come into resona nce with the resonance ejection amplitude at qx = 0.88. As the ions come in resonance w ith the supplementary RF field, the ions gain kinetic energy and are quickly ejected in a tight pack from the ion trap along the X axis. The RF voltage at which an ion is ejected from the mass analyzer is defined as its resonance ejection RF voltage. Without resonan ce ejection, or with ejection at a q > 0.88, ion motion may grow in the Y direction, resulti ng in reduced ejection efficiency through the 0.25 mm-thick slot. This use of resonance ejection greatly improves mass resolution, and enab les the trapping of a larger number of ions in the ion trap without sacrificing resolution and peak shape, since resonant ejection is more tole rant of space-charge effects.

PAGE 33

33 Mass Analysis After ions have been successfully stored in the LIT, a number of different mass analyses can be performed. The analyses that were used in this research project include single-stage full scan (MS), and multi-stage full scan (MSn, n = number of stages from 2 to 10). With MS, the ions formed in the ion source are stored in th e mass analyzer, and then are sequentially scanned out of the mass analyzer to produce a full mass spectrum. Isolation MS2 includes two stages of mass analysis. In the first stage, the ions formed in the ion source are stored in the mass analyzer. The RF volta ge is then increased to move the stored ions towards higher q values until the ion of the m/z of interest (parent ion) is at a q of 0.83 (Figure 110). The parent ion is then se lectively isolated and all other ions are ejected from the mass analyzer. This isolation occurs with the use of a sum-of-sines waveform consisting of many discrete sine components (5-500 kHz) spaced ev ery 0.5 kHz. Sine components are removed from the isolation waveform at the secular freque ncy of the ions to be stored. This isolation waveform is applied to the X rods of the center s ection of the LIT in dipolar fashion to isolate a narrow m/z window (isolation window).96 The isolation waveform applies a resonance ejection RF voltage at all frequencies corresponding to the secular frequency of the unwanted m/z ions. The isolation waveform is applie d for a period of 16 ms at an am plitude adjusted to assure all other ions throughout the mass range are resonantly ejected from th e trap. Resonance ejection of an ion occurs when an auxiliary RF field is ap plied that matches its secular frequency in the Xdirection. The ion absorbs kinetic energy to the point that its magnitude of oscillation increases along the X axis so that it escapes the trap or collides with one of the center rod surfaces. The isolation waveform leaves a “frequency not ch” around the frequencies corresponding to the m/z of the parent ion, thus isolati ng the parent ion in the trap.

PAGE 34

34 Activation After isolation, the RF amplitude is d ecreased to move the parent ion to a q of 0.25-0.35 (Figure 1-10). This allows all the product ions formed duri ng collision induced dissociation (CID) of the parent ion to be trapped. The parent ion is then activated with a resonance excitation RF voltage, which is applied across the X rods in dipolar fash ion (Figure 1-8) for 30 ms at a single frequency corresponding to the secula r frequency of the ion to be excited, which is placed at a q of 0.25-0.35. The resonance excitation RF voltage has lower amplitude than the resonance ejection RF voltage, and thus is not strong enough to eject an ion from the mass analyzer. However, with sufficient voltage, the ion gains kinetic energy by resonant absorption, which results in more translational motion and increased collisions w ith the helium buffer gas present in the mass analyzer. After many coll isions, the ion gains e nough internal energy to cause it to dissociate into product ions. These fr agment ions are then c onfined within the ion trap, because of the quadrupolar field except for those fragments that fall below the low mass cutoff (right edge of the Mathieu stability diagra m in Figure 1-9) or thos e that do not retain a charge. This process is called col lision-induced dissoci ation (CID). The amount of energy imparted in the CID pro cess significantly influences the type of fragmentation induced; it can be increased by either increasing the resonance excitation amplitude or the time of resonance excitation.96 Within compound classes, the amount of energy required to fragment an ion is generally proportional to the m/z of the ion. Ions of higher m/z generally require a greater resonance exci tation amplitude or a longer period of time.97 The CID process can be used to obtain structurally char acteristic fragmentation patterns that can be used to identify selected analytes in complex mixtur es. In the second stage of mass analysis, the product ions are stored in the ma ss analyzer. They are then seque ntially scanned out of the mass analyzer to produce a full product ion mass spectrum.

PAGE 35

35 MSn on the LTQ can have two to ten stages of mass analysis. For MS3 and higher, the first two stages are similar to MS2 except that the pr oduct ions are not scanned out. Instead, product ions of one m/z are selected and all other ions ar e resonantly ejected from the mass analyzer. The selected product i ons now become the new parent i ons for the next stage of mass analysis. With each stage of analysis, the selected parent ions undergo CID to produce product ions. In the nth stage of mass analysis, the final product i ons are stored in the mass analyzer, and are then sequentially scanned out to produ ce a full final product ion mass spectrum. Overview of Dissertation Quantitative imaging of drugs of abuse and th eir metabolites in brain tissue using MALDIMS could prove to be an invaluable tool in th e field of postmortem forensic toxicology. The purpose of this research was to design a qua ntitative mass spectrometric imaging method for determining the regional composition of drugs and th eir metabolites in postmortem brain tissue. A MALDI-MS imaging method that combined the use of internal standards for minimizing signal variability with the high molecular specificity of MSn could provide a visual snapshot for the forensic toxicologist that reflects the distribut ion and concentration of drugs of abuse near the time of death. This information could be used to substantiate fatal overdos es as well as provide supportive data for neurotoxicity studies. Current LTQ software allows for only one isolation window in MSn experiments, isolating one parent mass (or range of masses) for collis ion-induced dissociation. This means that MSn of the target ions of the analyte and internal standard would t ypically be performed with two separate MSn experiments. This would increase the re sponse variability and could counteract the signal normalizing effects of us ing an internal standard. Chapter 2 describes a MALDI-MSn method developed that allows for a single MS3 experiment that uses a wide isolation window to isolate both analyte and internal standard target ions. This me thod is shown to provide improved

PAGE 36

36 precision (~10-20 times reduction in percent relative standard deviation) fo r quantitative analysis compared to using two alternating MS3 experiments that separately isolate the target analyte and internal standard ions. The wide isolation met hod was used to quantify cocaine in brain tissue of a human cocaine user. The wide isolation method is capable of improving precision of MALDI-MSn quantitation by isolating the analyte and internal standard ions within a single MSn experiment, but it also isolates other ions that are not of interest that fill up the ion trap and can interfere with detection of the target ions. Chapter 3 describes another strategy for isolating both the analyte and internal standard ions within a single MSn experiment without storing unwanted background ions. This method employs a multi-notch stored waveform inverse Fourier transform (SWIFT) waveform that is applied to the linear i on trap mass spectrometer for selectively isolating multiple pairs of analyte and internal standard ions during a single MSn scan. The precision of the multi-notch SWIFT isolation method for the MALDI-MS2 analysis of cocaine was compared to the alternating MS2 scan method and the wide isolation method. Chapter 4 further develops the SWIFT isola tion method by incorporating a high mass filter (HMF) SWIFT excitation waveform to eject high m/z background ions present from endogenous brain tissue compounds. This tw o-stage SWIFT isolation method (i.e. HMF and multi-notch SWIFT) was compared with the wide isolati on method for quantificatio n of cocaine and its metabolites in brain tissue. Quantitative results were then compared with a more conventional method for the quantification of cocaine in brai n tissue that involves ho mogenate preparation, followed by solid-phase extraction, and MALDI-MS2 analysis. Chapter 5 offers a conclusion to the areas examined and provides ideas for future studies. Appendix A illustrates how beta is calculated, which is used to determine the desire d frequency notches of the SWIFT waveforms.

PAGE 37

37 Appendix B contains the C++ program used to calculate the SWIFT waveforms. Appendix C describes the modifications made to the LTQ in strument, which allow for the application of SWIFT.

PAGE 38

38 Figure 1-1. Metabolism of cocaine (adapted from Goldfrank’s Toxicologic Emergencies 8th Ed.,New York: McGraw-Hill, 2006). The major route of metabolism is hydrolysis of cocaine by hepatic and plasma esterases, with loss of a benzoyl group to give ecgonine methyl ester. The secondary r oute is spontaneous hydrolysis, probably nonenzymatic, which leads to benzoylecgonine by demethylation. The final degradation of cocaine, which is a sequel to both the principal and secondary routes of metabolism, leads to ecgonine. N-demethyla tion of cocaine is a minor route leading to norcocaine. Cocaine (COC) Ecgoninemethyl ester (EME) Principal Route: Hydrolysis by hepatic and plasma esterases; loss of benzoylgroup Benzoylecgonine(BE) Secondary Route: Spontaneous hydrolysis (nonenzymatic); demethylation In presence of ethanol: Transesterification Cocaethylene(CE) Ecgonine Norcocaine Minor Route: N -demethylation Final Degradation Product

PAGE 39

39 Figure 1-2. Cocaine’s mechanism of action ( Human Illnesses and Behavioral Health , web accesse d on 30 June 2008). Under normal conditions, dopamine is released by a neuron in to the synapse, where it can bind with dopamine receptors on neighboring neurons. Normally dopamine is then recycled back into the transmittin g neuron by a specialized protein called the dopamine transporter. If cocaine is present, it attaches to the dopamine transporter and blocks the normal recycling process, resulting in a build-up of dopamine in the synapse which contributes to the pleas urable effects of cocaine.

PAGE 40

40 Figure 1-3. Dopaminergic pathway ( www.humanillnesses.com, Web. Accessed on June 30, 2008.). Dopaminergic pathways are neural pathways in the brain which transmit the neurotransmitter dopamine from one regi on of the brain to another (e.g., from the ventral tegmental area to th e nucleus accumbens (NAC)).

PAGE 41

41 Figure 1-4. Cocaine imaging in tissue (www. invivopharm.com, web accessed on 30 June 2008). (a) Positron emission tomography (PET) image of [11C]cocaine binding across species. (b) Whole body autoradiography (WBA) image of ra t injected with [3H]cocaine. (a) (b)

PAGE 42

42 Figure 1-5. Tissue preparation a nd MALDI-MS imaging protocol. (A) Tissue sample is freezemounted onto the crytostat stage with de ionized water and then cut into 20 m thick slices. (B) Tissue slices are thaw-mounted onto glass mi croscope slide. (C) Airbrush is used to apply a homogeneous la yer of MALDI matrix (D) to the sample. (E) Sample is then analyzed by moving it step-wise beneath a pulsed laser beam. (F) A position-specific mass spectrum is produced from every laser spot. (G) Specific ions are extracted from the mass spectrum using software to generate an image.

PAGE 43

43 Figure 1-6. Schematic of the LTQ with MALDI source.67

PAGE 44

44 Figure 1-7. Basic design of the two-dimensional linear ion trap.87 Exit Slit30 mm x 0.25 mm

PAGE 45

45 Figure 1-8. Scheme for applica tion of DC, RF trapping, and AC excitation voltages necessary for operation of the 2D ion trap. (A) Se parate DC voltages (100 V) applied to the separate sections of each r od produce an axial trapping fi eld. (B) Two phases of the primary RF voltage (5 kV rod to ground, 1 MH z) are aplied to all the electrodes of the ion trap to form the quadrupolar fi eld. (C) Two phases of supplemental AC voltage (80 V, 5-500 kHz) are applied to only the X rods for isolation, activation, and ejection of ions.87 A B C

PAGE 46

46 Figure 1-9. Mathieu stability diagram for the linear ion trap. The lines labeled x and y describe the oscillatory char acteristics of ion motion. So lutions to Equations 1-1 and 1-2 give coordinates in ax and qx space that can be mapped onto the above diagram.90

PAGE 47

47 Figure 1-10. A simplified scan function for the quadrupole ion trap showi ng the prescan and the analytical scan which makes up one microsca n. The four steps of the QIT operation: ion injection (1), isolation (2), excitation (3), and mass analysis (4) are shown in the scan function. The prescan mass analysis st ep (4*) is rapid, because only an ioncurrent measurement is required. At step 1a the RF amplitude is increased to move the ions of interest to a q of 0.83 for isolation. At st ep 2a, the RF amplitude is decreased to move the isolated ions to a q of 0.25-0.35 for activation.92 1a 2a

PAGE 48

48 CHAPTER 2 WIDE ISOLATION Introduction Concentrations of drugs of abuse found in brai n tissue better reflect drug concentrations at their site of action at the time of death than any other type of specimen used for postmortem forensic toxicology.5 Conventional quantification of cocaine in brain tissue involves homogenate preparation, followed by extraction and/or derivatization.98 The extracts are then usually analyzed by gas chromatogra phy/mass spectrometry (GC/MS), liquid chromatography/mass spectrometry (LC/MS), GC, or LC. Lengthy extraction procedures are required to remove large concentrations of lip ids and other endogenous materials present in the brain, which may interfere with analysis.99 Multiple sample pretr eatment steps also allow opportunity for loss of analyte,33 and tissue homogenization elim inates spatial information, which could provide histologically -specific drug distribution. A ttempts have been made to determine the regional distribu tion of cocaine in postmortem brain of chronic human cocaine users.8, 12, 98 These analyses were performed on sectio ns of ~ 100-200 mg of tissue from different regions of the brain, which were assumed to be homogeneous and accurately representative of the drug concentration in that excised region. Direct MALDI-MS analysis of intact tissu e can provide quantitative information about the distribution of cocaine in human brain more rapidly, with higher spatial resolution, and with less sample loss than drug analysis methods th at involve tissue homogenization. Furthermore, the distribution of cocaine in br ain tissue acquired by MALDI-MS ca n be directly related to the histology. The majority of MA LDI-MS instruments use a time-o f-flight (TOF) mass analyzer, which has benefits of high mass range a nd high throughput. Quantitative MALDI-MS is challenging, however, because MALDI exhibits irreproducible signal intensities due to

PAGE 49

49 inhomogeneous crystal formation, inconsistent sample preparation, a nd laser shot-to-shot variability. A typical MALDI-TOF experiment will obtain 200-1000 consecutive mass spectra at each sample spot (one laser shot per spect rum), which are averaged to improve the reproducibility of the MALDI signal.27 Similarly, MALDI-MS instru ments that utilize a linear ion trap (LIT) mass analyzer can obtain multiple mass spectra at each spot and average them to improve reproducibility.85 Here we obtain a single mass spect rum at each spot, with typically 10 laser shots used to fill the ion trap for each spect rum. Note that any ion trap has a finite ion storage capacity before space-charge redu ces resolution and causes peak shifts.96 MALDI-LIT instruments can minimize space-charge effects by utilizing automatic gain control (AGC),96 which automatically controls the number of laser shots used to fill the trap (typically 1-20 shots) to optimally fill the ion trap for maximum signal without loss of mass resolution. Laser power can also be optimized along with choice of matrix compound to maximize analyte signal while avoiding space-charge effects. The use of in ternal standards for quantitative MALDI-MS has been shown to improve signal stability, if the so lution-phase properties ar e carefully matched as in an isotopic standard.77 Quantification of small drug molecules like cocaine using MALDI-MS is further complicated by the presence of interfering matrix peaks in the low mass range along with ions that may be produced from endogenous co mpounds present in the brain tissue.33 One of the strengths of a linear ion trap mass spectrometer is its ability to perform multiple stages of mass analysis (MSn) to significantly increase the selectivity for the analyte of interest. A MALDI-MSn method could be developed to remove interfer ences from both MALDI matrix and the complex sample environment of brain tissu e; however, a problem arises when trying to combine the use of MSn with the use of internal sta ndards. Instrument software allows for only one isolation

PAGE 50

50 window (IW) in MSn experiments, isolating one parent ma ss (or range of masses) for collisioninduced dissociation (CID). This means that MSn of the target ions of the analyte and internal standard would typically be pe rformed with two separate MSn experiments. This would increase the response variability and could counteract the si gnal normalizing effects of using an internal standard. In contrast, usi ng a 6-Dalton (Da)-wide IW centered at a mass-to-charge ( m/z ) between the [M+H]+ ions of cocaine and its trideuterated anal og allows for isolation and CID of both ions during a single MSn experiment. This single isolati on method reduces the signal variability inherent with MALDI compared to isolating eac h ion individually with a 1-Da IW (in two alternating MSn experiments). This method is used he re to detect and quantitatively image cocaine in postmortem human brain tissue. This study demonstrates that MSn increases selectivity, which is critical for differentiating analyte ions from matrix ions and endogenous co mpounds found in brain tissue. It is also shown that the use of internal standards corrects for signal variability in quantitative MALDI arising from inhomogeneous crystal form ation, inconsistent sample preparation, and laser shot-to-shot variab ility. Using a single MSn experiment with a wide IW to isolate both analyte and internal standard target ions provides improved pr ecision (10-20 times reduction in %RSD) for quantitative imaging studies compared to using two alternating MSn experiments that isolate the analyte and internal stan dard target ions separately. Experimental Chemicals Cocaine (COC; MW 303.4 Da) and COC-d3 (MW 306.4 Da, 0.29% do) were purchased from Cerilliant (Round Rock, TX, USA) at concentrations of 1 mg/mL and 100 g/mL, respectively, in acetonitrile. High-perfo rmance liquid chromatography (HPLC)-grade acetonitrile, methanol, and water were purchased from Fisher Scientific (Pittsburgh, PA, USA).

PAGE 51

51 Working standards of COC and COC-d3 were diluted with acetonitr ile and then stored at 4 oC. COC calibration standards were prepared in acetonitrile at concentrations of 5.0, 2.5, 1.25, 0.625, 0.312, 0.156, 0.078, 0.039, 0.020, 0.010, and 0.005 g/mL with the COC-d3 internal standard at a concentration of 2.0 g/mL. Sinapinic acid (SA; MW 224.2 Da), 2,5-dihydroxybenzoic acid (DHB; MW 154.1 Da), and -cyano-4-hydroxycinnamic acid (CHCA; MW 189.2 Da) were purchased from Acros Organics (Geel, Belgium) Saturated matrix solutions (40 mg/mL DHB, 10 mg/mL SA, and 10 mg/mL CHCA) were prepared in methanol/water (70:30, vol/vol) on the day of use. Tissue Collection Human brain tissue samples were provided by the El Paso County Coroner’s Office in Colorado Springs, CO. Postmortem brain materi al was excised from the nucleus accumbens (NAc) from case number 07A-369, whose toxicologi cal analysis indicated the presence of cocaine in blood at 69 ng/mL (COC concentration in the brain tissue was not quantified). The NAc is a dopamine-rich area of the striatum, whic h may contain an accumulation of COC due to its affinity to bind with the dopamine transporter.100 At autopsy, the excised tissue was immediately snap-frozen in liquid n itrogen and then stored in a -80 oC freezer until analyzed. Tissue Sectioning and Sample Preparation Frozen brain tissue was cut into thin sections (20 m thickness) in a cryostat (HM 505E; Microm International GmbH, Waldorf, Germany) at -25 oC. The tissue samples were frozen to the cryostat sample stage using distilled water. Serial brain sections were collected onto microscope slides where they were th aw mounted and then stored at -80 oC. Before mass spectrometric analysis, the tissue sections were removed from the freezer and placed in a vacuum dessicator for 30 min before spiking standards (1L droplets by micropi pet) and applying MALDI matrix. The matrix was applied to the ti ssue sections using an artistic airbrush (Aztek

PAGE 52

52 A470; Testors, Rockford, IL, USA). The applic ation of MALDI matrix by airbrush has been previously published.67 Matrix was applied using the dr ied-droplet method for experiments performed on MALDI plate. Mass Spectrometry Mass spectra were acquired using an LTQ linear ion trap with a vMALDI ion source (Thermo Finnigan, San Jose, CA, USA), equipped w ith a nitrogen laser (337 nm) at a frequency of 20 Hz and 100m spot size. A more detailed descri ption of this instrument has been published.67 An average of 10 laser shots per scan wa s used to produce mass spectra, except for experiments that used AGC, in which the number of laser shots was automatically varied to optimally fill the trap with ions, thus avoidi ng space charge-related peak broadening and mass shifts. AGC assess the ion genera tion rate by use of a prescan, a nd then adjusts the number of laser shots per scan to produce a similar numbe r of ions for each scan. The spectra are normalized to the number of laser shots for each scan. Resonance excitation is used for isola tion, activation, and mass analysis. For MSn experiments, unwanted ions are resonantly ej ected from the ion trap by applying a 5-500 kHz multi-frequency isolation waveform consisting of sine components spaced every 0.5 kHz. The ions of interest are isolated by removing sine components from the is olation waveform that correspond to the secular frequency of the desired i on(s). Ions are selected for isolation in the LTQ software by entering the m/z with its IW. The mass range for the ion is defined as ( m/z – IW/2) to ( m/z +IW/2). The IW should be narrow enough to minimize including interfering peaks, but wide enough to avoid loss of sensitiv ity for the desired ion(s). However, it is important to note that the activat ion width for resonance excitati on (CID) has the same value as the IW. Therefore, the collision energy applied during MSn is spread over the activation width. Thus, increasing the IW decreases the true collision energy for each ion.

PAGE 53

53 The tissue-mounted microscope slides were affixed to a slide holder plate with doublesided tape. The plate was then inserted into th e LTQ, and the plate was rastered beneath the laser spot at 100m steps to produce position-specific mass spectra. Specific ions and the total ion current (TIC) signal were extracted from the raw data files using ImageQuest version 1.0 (Thermo Fisher Scientific, San Jose, CA, USA) which was used to generate an image. Results and Discussion MS2 and MS3 Mass Spectra of COC and COC-d3 DHB was selected as the MA LDI matrix in this study, as preliminary investigations showed that it produces more efficient ionizatio n for COC at low concentrations than SA or CHCA. DHB was also preferred as the matrix fo r COC analysis due to its lack of interference with the [M+H]+ ion of COC ( m/ z 304) and COC-d3 ( m/z 307). The COC standards in acetonitrile were ch aracterized by MSn. The MS2 spectrum of m/z 304 and 307 (IW = 1.0 Da, CID = 20) each show one major product ion, corresponding to a neutral loss (NL) of benzoic acid (NL 122) at m/z 182 and 185, respectively. MS3 was performed on the product ion signal at m/z 182 of COC (IW = 1.0 Da; CID = 30), resulting in product ions at m/z 150 (NL of 32; CH3OH), m/z 122 (NL of 60; CH3OH + CO), m/z 119 (NL of 63; CH3OH + CH3NH2), m/z 108 (NL of 74; CH3OH + CH2CO), m/z 91 (NL of 91; CH3OH + CH3NH2 + CO), and m/z 82 (NL of 100; CH3OH + C4H4O via a 6-electron Alder ene rearrangement). The structures of the fragment ions of the [M+H]+ ion of COC and its propo sed fragmentation pathway (Figure 2-1) have been previously published.101 MS3 was performed on the product ion signal at m/z 185 of COC-d3 (IW = 1.0 Da; CID = 30) resu lting in product ions at m/z 153 (NL of 32; CH3OH), m/z 125 (NL of 60; CH3OH + CO), m/z 119 (NL of 63; CH3OH + CH3NH2), m/z 111 (NL of 74; CH3OH + CH2CO), m/z 91 (NL of 91; CH3OH + CH3NH2 + CO), and m/z 85 (NL of 100; CH3OH + C4H4O). The m/ z values of the fragment ions of COC-d3 at m/z 91 and 119 are the same as those

PAGE 54

54 for COC because these ions have lost the trideuterated tag that was originally located on the N methyl group. Improving Signal Reproducibili ty with Internal Standards Quantitative analysis by MALDI is challenging, because of signal irreproducibility due to variation in sample preparat ion, inhomogeneous co-crystalli zation of analyte and MALDI matrix, and laser shot-to-shot va riability. Figure 2-2a shows the m/z 304 signal of the [M+H]+ ion of COC detected from COC/COC-d3 standard solutions spotted 1 L each in triplicate onto a MALDI plate with 1 L of DHB matrix pippeted on top. The COC/COC-d3 solutions were composed of different concentrations of COC (5.0, 2.5, 1.2, and 0.63 g/mL) mixed with 1.0 g/mL of COC-d3. The histogram shows the high variab ility in signal for each concentration with %RSD ranging from 29 to 67%, making it difficult to distinguish signal from one concentration to another. Figure 2-2b shows the m/z 304 signal of COC normalized to the [M+H]+ ion signal of COC-d3 at m/z 307. Signal variability was reduced dramatically (%RSD ranged from 0.26 to 1.33%) by normalizing the analyte signal to that of the internal standard making quantification by MALDI possible. Increasing Analyte Selectivity with MSn Figure 2-3a shows a full-scan MS spectru m of a 1.25:1 mixture (by mass) of COC and COC-d3 standards spiked (1 L of 1.25 g/mL and 1.0 g/mL, respectively) onto a MALDI plate with DHB matrix. Peaks at m/z 304 and 307 represent the [M+H]+ ions of COC and COC-d3, respectively. A number of clus ter ions, fragment ions, and a molecular ion of DHB are also present, including m/z 137 [DHB+H-H2O]+, m/z 154 [DHB]+, m/z 177 [DHB+Na]+, m/z 199 [2DHB+Na]+, m/z 221 [DHB-2H+3Na]+, m/z 273 [2DHB+H-2H2O]+, m/z 291 [2DHB+H-H2O]+, and m/z 331 [2DHB+Na]+. Figure 2-3b shows a full-scan MS spectrum of a 1:1 mixture (by

PAGE 55

55 mass) of COC and COC-d3 standards spiked (1 L of 1.0 g/mL each) onto a 20m thick human brain tissue slice with DHB airbrushed. The [M+H]+ ions of COC and COC-d3 are observed at m/z 304 and 307, respectively. The same clus ter ions, fragment ions, and molecular ion of DHB are present, in addition to numer ous ions of endogenous compounds from the brain tissue, including the phosphocholine h ead group of phosphatidyl choline at m/z 184 [(CH3)3NCH2CH2PO4H]+. Identification of COC and COC-d3 on the MALDI plate and brain tissue was confirmed by characteristic MS2 product ions at m/z 182 and 185, respectively. MS2 spectra of m/z 304 with COC spiked at concentr ations below 5 ng/mL on plate and on tissue revealed an isobaric co mpound that has product ions at m/z 212 and 91. The isobaric ion likely originates from the surfactant benzyldimethyldodecylammonium chloride (C12BAC).102 The widespread use of C12BAC and other BACs as disi nfectants makes it a likely trace contaminant in the laboratory. The ion at m/z 212 results from fragmentation of the carbonnitrogen bond between the toluyl substituent and the quaternary amine (Figure 2-4). The m/z 91 ion is a stable tropylium ion form ed by fragmentation in which the toluyl substituent retains the positive charge (Figure 2-4). MS2 of m/z 304 with COC spiked at c oncentrations below 5 ng/mL also results in the detection of pr oduct ions of isobaric compounds at m/z 256 and 286. These ions have not yet been identif ied, but are not present when DHB has been characterized on MALDI plate alone. The presence of isobaric ions in samples increases with sample complexity and may interfere with quantification at low analyte concentrations. MSn can improve analyte selectivity and produce higher signal-to-noise ratios, resulting in lower dete ction and quantification limits for the analyte. Combining the use of MSn with internal standard s is commonly performed by alternating MSn scans of the analyte and the internal sta ndard ions, and then ratioing the resulting

PAGE 56

56 product ion signals. This method is effective for use with ionization techniques such as electrospray and atmospheric pressure chemical ionization; however, due to the shot-to-shot variability of MALDI, acquiri ng analyte and internal standa rd signals in alternating MSn experiments may counteract the signal normalizing effects gained by using an internal standard. Combining Internal Standards with MSn using a Wide Isolation Window One method for combining the use of internal standards with MSn is to perform MS2 on the analyte and internal standard ions separately during alternate MSn experiments illustrated in Figure 2-5a. The [M+H]+ ion of COC ( m/z 304) is isolated with a 1 Da window and then collisionally activ ated to produce the product ion at m/z 182 shown in the MS2 spectrum in Figure 2-5b. In a separate MS2 scan, the [M+H]+ ion of COC-d3 ( m/z 307) is isolated with a 1 Da window and CID is applied, re sulting in the product ion at m/z 185 shown in the MS2 spectrum in Figure 2-5c. The analyte ion signal at m/ z 182 can then be normalized to the internal standard ion signal at m/ z 185. An alternative approach to using two separate MSn experiments is to use a single wide isolati on window (e.g., 6-Da) centered at m/z 305.8, shown in Figure 2-5d, allowing the simultaneous isol ation and CID of the [M+H]+ ion of COC ( m/z 304) and COC-d3 ( m/z 307). The resulting MS2 spectrum, shown in Figure 2-5e, contains the product ions of COC and COC-d3 at m/z 182 and 185, respectively. The performance of the MSn experiment using a single wide isolation window was compared with that using two alternating MSn experiments by detecting COC and COC-d3 spiked on top of human brain tissue. Figure 2-6a shows a microscope image of a 20m thick human brain tissue sl ice with COC/COC-d3 solutions spotted 1 L each in triplicate (A, B, and C) on the surface of the tissue and then ai rbrushed with DHB. The five COC/COC-d3 solutions spotted all contained 2.0 g/mL of COC-d3 in addition to 0.31, 0.62, 1.2, and 5.0 g/mL of COC,

PAGE 57

57 respectively. The compositions of the solutions spotted (1-5) are shown in the table below the image. The average dried spot size was 0.25 cm in diameter. Figure 2-6b shows the MS2 product ion image of m/z 305.8 (IW = 6 Da, CID = 20) of the entire tissue slice generated from signal extracted from the mass range m/z 182-186 and normalized to the TIC. Higher signal intensity correlates with the darker shad e of gray, illustrating how the COC and COC-d3 cocrystallize along with the DHB towards the edge of each spot. The LTQ software was used to outline each spot to be analyzed. Each spot was analyzed twice: first by performing MS2 of m/z 304 (IW = 1 Da, CID = 20) followed by MS2 of m/z 307 (IW = 1 Da, CID = 20), and then by MS2 of m/z 305.8 (IW = 6 Da, CID = 20). For each analys is, all of the spectra (~500 scans) were averaged for each spot, and the m/z 182 signal for COC was normalized to the m/z 185 signal for COC-d3 and plotted against the conc entration of COC spiked to produce two different calibration curves shown in Figure 2-7. Figure 2-7a s hows the average ratio of peak intensities m/z 182 to m/z 185 as a function of the spiked CO C concentration for alternating MS2 experiment (i.e., MS2 of m/z 304 in one scan and then MS2 of m/z 307 the following scan). The line of best fit was y = 0.68( 0.07)x + 0.2( 0.2) over the range 0.31-5.0 g/mL with a standard error of the estimate (SEE) = 0.2833; the %RSD ranged from 12% to 30%. The 95% confidence intervals for the slope and y-intercept were 0.44 to 0.91 and 0.4 to 0.9, respectively. Figure 2-7b shows the average ratio of peak intensities m/z 182 to 185 as a function of th e spiked COC concentration for a single MS2 experiment with a wide 6-Da isolation window centered at m/z 305.8 (i.e., MS2 of m/z 304 and m/z 307 in one scan). The line of best fit was y = 0.492( 0.001)x + 0.023( 0.003) over the range 0.31 to 5.0 g/mL with an SEE = 0.0052; the %RSD ranged from 0.50% to 5.1%. The 95% confidence intervals for the slope and y-intercept were 0.488 to 0.496 and 0.011 to 0.034, respectively. Precision was dr amatically improved by using the single MS2

PAGE 58

58 experiment with 6-Da wide isolation window comp ared with isolating each ion individually with a 1-Da window (two alternating MS2 experiments). There was a 10-20 times reduction in %RSD and a 50 times reduction in SEE by us ing the wide isolation method. Isolation Window Width and Automatic Gain Control Usually the smallest isola tion width is desired for MSn experiments performed with an ion trap mass spectrometer to avoid isolating unwanted background ions and reducing analytical specificity. The minimum acceptable ion isola tion width is defined as the lowest range providing no appreciable signal at tenuation of the analyte and in ternal standard ions when compared to a wider setting. Signal attenua tion can result either from losses during the resonance ejection step which is used to remove masses below and above the selected m/z range, or from decreased CID efficiency of th e analyte and internal standard ions. The effect of isolation width on the inte nsity of the product ions of the [M+H]+ ions of COC ( m/z 304) and COC-d3 ( m/z 307) together in a single MS2 scan was investigated. Five solutions of COC and COC-d3 were prepared at equal con centrations and diluted with acetonitrile (0.12, 0.25, 0.50, 1.0, and 2.0 g/mL). All five solutions were spotted in triplicate 1 L each onto a MALDI plate followed by 1 L of DHB matrix. For all MS2 experiments, the parent ion was set to m/z 305.8, the center of the mass range between m/z 304.3 and m/z 307.3, and the CID was set to 20. The size of the isolation window width centered at m/z 305.8 was varied (4 Da, 6 Da, and 8 Da), and the ratio of the intensities of the products ions at m/z 182 and m/z 185 for COC and COC-d3, respectively, were observed. It is important to note that MS2 of m/z 304 (IW = 1.5 Da) produced a negligible amount of m/z 185 (<0.0005%), the product ion of m/z 307. Also, MS2 of m/z 307 (IW = 1.5 Da) produced a negligible amount of m/z 182 (<0.002%), the product ion of m/z 304. The expected signal ratio of COC to COC-d3 is 1.02 for

PAGE 59

59 equal masses based on a calculated molar ratio of 1.01 corrected for the isotopic purity of COCd3 (0.29% COC-do). The measured signal ratio of m/z 182 to 185 was approximately equal to 1 for concentrations below 0.50 g/mL, but the ratio increased at concentrations above 0.50 g/mL (i.e., the m/z 185 signal decreased with respect to m/z 182). It was also obse rved that the signal ratio of m/z 182 to 185 was higher for a 4 Da isolation window (2.07 at 1.0 g/mL and 3.50 at 2.0 g/mL) compared with the 6 Da (1.18 at 1.0 g/mL and 1.68 at 2.0 g/mL) and 8 Da (1.05 at 1.0 g/mL and 1.81 at 2.0 g/mL) isolation windows widths. Th is suggests that either some of the m/z 307 ion is being lost duri ng isolation or that the m/z 307 ion is being less efficiently excited during the CID step when narrower IWs are used. An effort was made to separate the is olation step from the CID step of the MS2 experiment to better understand th e effect of isolation window wi dth on the signal intensities of the MS2 product ions of the [M+H]+ ions of COC and COC-d3. The above experiment was repeated on the five COC/COC-d3 solutions, except that no CID vo ltage was applied so that the ions at m/z 304 and 307 were isolated but not frag mented. The ratio of intensities of m/z 304 and m/z 307 were then monitored for different isolati on window widths (4, 6, and 8 Da). Results showed that the signal ratio of m/z 304 and 307 remained approximately equal to 1 for concentrations 0.12 2.0 g/mL for isolation widths of 6 and 8 Da; however, the signal ratio steadily increased for a 4 Da isolatio n window at concentrations above 0.50 g/mL. The increase in the signal ratio of m/z 304 to 307 (i.e., m/z 307 signal decreased with respect to m/z 304) at higher concentration is pr esumably due to a mass shift of m/z 307 outside the isolation window, resulting in resonance ejection of some of the m/z 307 ions. This mass shift could be caused by space-charge effects at higher ion popula tions in the ion trap, and may be corrected by using AGC. The experiment was repeated again, comparing the signal ratio of m/z 304 to 307

PAGE 60

60 with and without AGC with a 4 Da isolation wi ndow and no CID applied. Results showed that when AGC was used, the signal ratio of m/z 304 to 307 remained approximately equal to 1 for all concentrations analyzed (0.12 – 2.0 g/mL), indicating that AGC can minimize space-charge effects, which may lead to ejection of the higher m/z ion when a narrower isolation width is employed. Quantification of Cocaine in Postmortem Human Brain Tissue The MS3 wide isolation method developed for COC was applied to human brain tissue from a subject whose toxicology report showed the presence of COC. The MS2 product ion of the [M+H]+ ion of COC at m/z 182 was not distinguishable from the background signal; therefore, an MS3 wide isolation method was develope d to increase sel ectivity. The MS3 wide isolation method was evaluated by spotting 1 L of a 4.0 g/mL solution of COC and COC-d3 onto a MALDI plate followed by 1 L of DHB matrix. The met hod involves centering a 6-Da isolation window at m/z 305.8 and applying a CID of 20 follo wed by a 6-Da isolation window centered at m/z 183.5 (between COC and COC-d3 product ions at m/z 182 and 185) with a CID of 30. The resulting MS3 product ion spectrum revealed charact eristic fragment ions of COC at m/z 150, 82, 108, 122, 119, and 91 and for COC-d3 ions at m/z 153, 85, 111, 125, 119, and 91. The MS3 wide isolation method was applied to unsp iked brain tissue from a cocaine user, and COC was detected and confirmed by matching all six of these MS3 ions. The relative intensities of the five most intense fragment ions (all but m/z 91) were within 12% of the standard fragment ion intensities. Figure 2-8 shows the MS3 product ion image of m/z 305.8 (IW = 6 Da,CID = 20) of the entire tissue slice generated from signal extracted from the mass range m/z 150 – 151 and normalized by the TIC. The image shows no localiz ation of COC in the section of the nucleus

PAGE 61

61 accumbens analyzed. Browne et al.98 analyzed 1 g samples from different regions of 3 human brains by solid-phase extraction (SPE) a nd LC. From these studies, cocaine and benzoylecgonine were found to be distributed th roughout the different regions of the brain. However, significant differences in the concentr ation of cocaine were apparent in different regions of the brain (e.g., cocaine concentration was higher in the basal ganglia th an the section of the cerebellum analyzed). These findings were consistent with other brain cocaine distribution studies, which reported that concentrations of cocaine and of its metabolites showed little regional heterogeneity in postmortem brain of chronic users of cocaine.8, 12 The homogeneous distribution of cocaine and its metabolit es in specific regions of the brain may be a result of the high concentrations typical of behavioral usage. Before quantifying unspiked COC in human brain tissue with the MS3 method it was necessary to show that the re sponse factors for COC and COC-d3 were equal, so that the calibration curve of COC-d3 could be used. A series of 1:1 solutions of COC and COC-d3 at various concentrations (0.03, 0.06, 0.13, 0.25, 5.0, 1.0, and 2.0 g/mL) were prepared and spiked in triplicate, 1 L each, on top of serial tissue sections, and then DHB matrix was airbrushed over the tissue slices. Each spot was analyzed using the MS3 wide isolation method, and the m/z 150 signal from COC was plotted versus the m/z 153 signal from COC-d3. The slope of the plot was 1.062 0.002 with a correlation coefficient r2 = 0.99,998 over the con centration range 0.03 to 2.0 g/mL. The 95% confidence interval for the slope was 1.057 to 1.066. The expected slope based on a molar ratio of 1.01 and an isotopic purity for COC-d3 of 0.29% do is 1.02, which means that COC has a 4% hi gher response factor than COC-d3 over the concentration range measured.

PAGE 62

62 The MS3 wide isolation method was used to quan tify the unspiked COC that was detected in the postmortem human brain tissu e. The calibration curve used for quantification (Figure 2-8) was created by imaging three diffe rent concentrations of COC-d3 (0.06, 0.13, and 0.25 g/mL) were spiked (1 L) onto a glass slide be fore thaw mounting a 20 m-thick brain tissue slice on top and airbrushing DHB matrix. All thr ee spots were then analyzed using the MS3 wide isolation method. Approximately 2000 scans were acquired to image the entire area of each of the spots (average area = 0.17 cm2). The m/z 153 signal from each spot was used to develop a calibration curve that resulted in a line of best fit of y = 399( 27)x – 17( 4) (Figure 2-8). COCd3 was shown to have a linear re sponse with increasing concentra tions spiked underneath tissue. Since the MS3 wide isolation method analyzes both COC and COC-d3 simultaneously, unspiked COC was detected from each spot analyzed at m/z 150, and the correspondi ng signal was plotted ( ) alongside each corresponding COC-d3 signal ( ) in Figure 2-9. An area of the tissue (500 MS scans) that was not spiked with COC-d3 was analyzed using the MS3 wide isolation method and the acquired m/ z 150 signal was averaged with the m/ z 150 signals from the spiked COC-d3 spots, resulting in a very trace signal of 29 1 counts (highlighted as dashed line on Figure 2-8). Assuming that the amount of unspi ked COC extracted from the tissue has a 1:1 response with the COC-d3 spiked on top of tissue, th e calibration curve for COC-d3 can be used to quantify the amount of COC present in the anal yzed tissue. From the equation of the line, it was determined that COC was present at a level equivalent to 0.12 0.01 g/mL. Using the 1 L volume of COC-d3 spiked underneath tissue, it is calculated that the mass of COC present is 1.2 x 10-4 g. Given that the area of an analyzed spot on tissue was 0.17 cm2 and that the tissue thic kness was 20 m (2.0 x 10-3 cm), the volume of tissue from which COC was extracted was 3.4 x 10-4 cm3. The mass of the tissue is 3.4 x 10-4g (density of wet tissue

PAGE 63

63 ~1.0 g/cm3), resulting in an absolute concentratio n of COC detected in this area of the postmortem brain tissue of 0.35 g/g (350 ppb). The MALDI-MS method has a smaller sample requirement (~100 g tissue) and less sample preparation than conventional GC/MS techniques, which require 1000 to 10,000 times more sample (0.1 to 1.0 g of brain tissue) to be homogenized before solid-phase extraction and GC/MS analysis.8, 12 The GC/MS method devel oped by Kalasinsky et al.12 reported a limit of detection of 0.1 ng/mL for the anal ysis of COC in brai n tissue. COC was de tectable at 30 ng/mL with the MALDI-MS3 wide isolation method devel oped here. Although the MALDI-MS3 wide isolation method is not as sensitive as the GC/M S method (primarily because it uses a 1000 times smaller sample), it readily detects cocaine at a le vel an order of magnitude below the lowest level (300 ng/mL) reported for COC detected by GC/MS an alysis of 15 autopsied brain regions of 14 human chronic cocaine users.12 Conclusions It has been demonstrated that MS2 and MS3 increase selectivity, which is critical for differentiating analyte and internal standard ions from matrix ions and endogenous compounds found in brain tissue. It has also been shown that the use of internal stan dards corrects for signal variability during quantitative MALDI. A met hod was developed that allows for a single MS2 experiment that uses a wide isolation window to isolate both analyte and internal standard ions. This method was shown to provide improved pr ecision (~ 10-20 times reduction in %RSD) for quantitative analysis of COC in postmortem brain tissue compared with using two alternating MS2 experiments that isolate the analyte and internal standard target ions separately. When COC concentration is too low to distinguish the MS2 product ion at m/z 182 from the background, the MS3 wide isolation method can be applied to incr ease selectivity.

PAGE 64

64 The wide isolation window developed for the analysis of COC could be applied to quantitative MALDI-MSn imaging of other drugs of abuse a nd their metabolites in brain tissue, which could prove to be an inva luable tool in the field of pos tmortem forensic toxicology. A MALDI-MS imaging method that combined the use of internal standards for minimizing signal variability with the high molecular specificity of MSn could provide a visual snapshot for the forensic toxicologist that reflects the true distri bution and concentration of drugs of abuse at the time of death. This information could be used to substantiate fatal overdos es as well as provide supportive data for neurotoxicity studies.

PAGE 65

65 Figure 2-1. Cocaine dissociation pathway.101 m/z 304 m/z 304 m/z 182 m/z 150 m/z 108 m/z 122 m/z 119 m/z 91 m/z 82

PAGE 66

66 Figure 2-2. MALDI-MS signal variability with and wit hout internal standards. Signal of m/z 304, [M+H]+ of COC (a) and m/z 304 signal ratioed to m/z 307 signal, [M+H]+ of COC-d3 (b). All solutions spotted 1 L in triplicate on MALDI plate with DHB matrix. The internal standard (COC-d3) was maintained at 1 g/mL for all solutions. 0.0E+00 5.0E+05 1.0E+06 1.5E+06 2.0E+06 2.5E+06 3.0E+06 3.5E+06 4.0E+06 4.5E+06 5.0E+06 5.002.501.250.63 Run 1 Run 2 Run 3 0.00 1.00 2.00 3.00 4.00 5.00 6.00 5.002.501.250.63 Run 1 Run 2 Run 3 (a) (b)Peak Intensity m/z 304 Peak Intensity Ratio m/z 304/307COC Concentration ( g/mL)COC Concentration ( g/mL) Mean 1.9E+06 % RSD 32 Mean 2.4E+06 % RSD 67 Mean 1.8E+06 % RSD 29 Mean 6.0E+05 % RSD 61 Mean 5.21 % RSD 0.59 Mean 2.57 % RSD 0.27 Mean 1.23 % RSD 1.20 Mean 0.61 % RSD 1.33

PAGE 67

67 Figure 2-3. Comparing mass spectra of COC and COC-d3 on MALDI plate and on brain tissue. MALDI mass spectrum of (a ) a solution of COC (1.25 g/mL) and COC-d3 (1.0 g/mL) spotted (1 L) with DHB matrix on MALDI plate (run 1 from Figure 1) and (b) a solution of COC and COC-d3 (1.0 g/mL each) spiked (1 L) on postmortem human brain tissue with DHB matrix airbrushed. [COC+H]+[COC-d3+H]+[DHB+H-H2O]+[2DHB+H-2H2O]+[2DHB+Na]+[2DHB+Na]+[DHB]+[2DHB+H-H2O]+(a) (b)[DHB+Na]+[DHB-2H+3Na]+ 100 150 200 250 300 350 m/z 0 20 40 60 80 100 0 20 40 60 80 100Relative Abundance 304.3 307.3 137.2 273.2 154.2 177.2 331.0 291.1 199.2 221.2 104.3 184.2 231.1 137.2 222.2 175.2 154.2 307.3 296.2 273.2 132.3 214.1 313.2 275.2 240.3 332.4 114.3 304.3[(CH3)3NCH2CH2PO4H]+ [COC-d3+H]+[COC+H]+

PAGE 68

68 Figure 2-4. Fragmentation of the benzyldimethyldodecylammonium ion.102

PAGE 69

69 Figure 2-5. Wide isolation MALDI-MS2. COC (1.0 g/mL) and COC-d3 (1.0 g/mL) spotted (1 L) with DHB matrix on MALDI plate. (a) Individual isolation (1 Da) and collisional activation of m/z 304 and m/z 307 with resulting MS2 spectra of the [M+H]+ ions of (b) COC at m/z 304 and (c) COC-d3 at m/z 307. (d) Simultaneous isolation and CID activation of m/z 304 and m/z 307 with a 6-Da window centered at m/z 305.8 and (e) the resulting MS2 spectra containing both COC and COC-d3 fragment ions at m/z 182 and m/z 185, respectively. [COC+H]+[COC-d3+H]+ 1 Da 1 Da [COC-d3+H]+305.8 302.8 308.8 3 Da3 Da [COC+H]+ (a)(b) (c) (d) (e) 301 302 303 304 305 306 307 308 309 310 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 304.2 307.2 80 100 120 140 160 180 200 220 240 260 280 300 320 340 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 182.2 185.2 100 150 200 250 300 350 m/z 0 20 40 60 80 100 0 20 40 60 80 100Relative Abundance 182.2 304.3 185.3 307.2 301 302 303 304 305 306 307 308 309 310 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 304.2 307.2

PAGE 70

70 Figure 2-6. Images of standards spiked on brain tissue. (a) Photomicrograph of 20 m thick human brain tissue mounted on slide with COC/COC-d3 solutions spiked (1 L) in triplicate (A, B, and C) on top of tissue a nd then airbrushed with DHB matrix. (b) MS2 product ion image generated from signal selected from mass range m/z 182-186 and normalized by the TIC. 1A 1B 1C 2A 2B 3B 3A 2C 3C 4A 5A 4B 4C 5B Samples excluded 2.0 cm(a)(b) Solution# (Triplicate Runs A, B, C)12345 COC ( g/mL)5.02.51.30.620.31 COC-d3( g/mL)2.02.02.02.02.0

PAGE 71

71 Figure 2-7. Calibration curves for alternating scans MS2 and wide isolation MS2. Peak intensity ratio of m/z 182 to m/z 185 versus COC concentration for two alternating MS2 experiments (a) and for a single MS2 experiment using a 6-Da isolation and activation window. y = 0.68x + 0.2 R2 = 0.96510.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 0.01.02.03.04.05.06.0Peak Intensity Ratio m/z 182/185Cocaine Concentration ( g/mL) 182/185 A 182/185 B 182/185 C COC Conc ( g/mL) % RSD 0.3112 0.6212 1.316 2.530 5.015(a) y = 0.492x + 0.023 R = 0.999980.0 0.5 1.0 1.5 2.0 2.5 3.0 0.01.02.03.04.05.06.0Peak Intensity Ratio m/z 182/185Cocaine Concentration ( g/mL) 182/185 A 182/185 B 182/185 C (b) COC Conc ( g/mL) % RSD 0.315.1 0.622.4 1.31.3 2.52.4 5.00.50

PAGE 72

72 Figure 2-8. Mass spectrometric image of cocaine in brain tissue. The MS3 product ion image of m/z 305.8 (IW = 6 Da, CID = 20) of the entire tissue slice generated from signal extracted from the mass range m/z 150 – 151 and normalized by the TIC. ( m/z 150 151)/TIC (Counts)

PAGE 73

73 Figure 2-9. Cocaine quantifica tion. Calibration curve created from plotting signal of MS3 fragment ion of COC-d3 at m/z 153 versus COC-d3 concentration spiked beneath 20 m tissue from cocaine user. Unsp iked COC was detected by the MS3 fragment ion of COC at m/z 150 and plotted against the calibration curve of COC-d3 to quantify the amount of COC present in tissue. y = 399x -17 R = 0.99550 10 20 30 40 50 60 70 80 90 0.000.050.100.150.200.250.30Peak Intensity (counts) COC-d3Concentration ( g/mL) Spiked COC-d3 Unspiked COC 0.13 g/mL 0.25 g/mL 0.06 g/mL 28 counts 30 counts 32 counts 25 counts

PAGE 74

74 CHAPTER 3 SWIFT ISOLATION Introduction Direct MALDI-MS analysis of intact ti ssue can provide information about the distribution of cocaine in human brain more rapidly, with higher spatial resolution, and with less sample loss than drug analysis methods that i nvolve tissue homogenizatio n. Furthermore, the distribution of cocaine in brai n tissue acquired by MALDI-MS can be directly related to the histology. Quantitative analysis by MALDI-M S is challenging, however, because MALDI exhibits irreproducible signal in tensities due to inhomogeneous crystal formation, inconsistent sample preparation, and laser shot-to-shot vari ability. The use of internal standards for quantitative MALDI-MS has been shown to improve signal stab ility, if the solution-phase properties are carefully matched as in an isotopic standard.77 Quantification of small drug molecules su ch as COC using MALDI-MS is further complicated by the presence of interfering matrix peaks in the low-mass range along with ions that may be produced from endogenous compounds present in the brain tissue. One of the strengths of a linear ion trap mass spectrometer is its ability to perform multiple stages of mass analysis (MSn) to significantly increase the selectivity for the analyte of interest. A MALDI-MSn method could be developed to remove interfer ences from both MALDI matrix and the complex sample environment of brain tissu e; however, a problem arises when trying to combine the use of MSn with the use of internal standards. Current instruments allow for only one isolation window (IW) in MSn experiments, isolating a single parent mass (or range of masses) for collisioninduced dissociation (CID). This means that MSn of the target ions of the analyte and internal standard typically requires two separate MSn experiments. This increases the response variability and can count eract the signal normalizing effects of using an internal standard.

PAGE 75

75 It was previously reported that using a wi de IW (e.g., 6 Daltons (Da)) that simultaneously isolates the analyte and internal standards ions in a single MS2 experiment provided improved precision for quantitative MALDI-MSn of COC in tissue, when compared to using alternate MS2 experiments that separately isolate the ta rget analyte and inte rnal standard ions.103 Isolating both analyte and internal standard with a wide IW reduces signal variability when the analyte and internal standard signals are ra tioed. However, the wide IW no t only isolates the analyte and internal standard ions, but also ions in between that are not of interest. These ions may fill up the ion trap and may also interfere with detection of the target ions. This chapter reports a multinotch isolation waveform that sel ectively isolates the analyte and internal standard ions, reducing the effects of background interferences and boosti ng the sensitivity for analyte ions. SWIFT (stored waveform inverse Fourier transform) is a broadband excitation tec hnique that is capable of selectively isolating multiple ions and has th e potential for improving precision of quantitative MALDI-MSn. SWIFT was first introduced to the field of mass spectrometry by Marshall et al. in 1985 for use with the Fourier transform ion cycl otron resonance mass spectrometer (FT/ICR).104 SWIFT was later applied successfully to the quadrupole ion trap (QIT)105, 106 due to the similar principles of operation between the FT/ICR and the QIT In 1994, Cooks and coworkers107 made improvements on the application of SWIFT to the QIT for selective isolation by employing a two-stage course/fine procedure for isolating ions from a population of trapped ions. The advantage of the procedure was that the coarse st ep removes most of the ions that contribute to space-charging, and thereafter the frequencies of th e analyte ions remain relatively constant. Ions trapped in a linear ion trap (LIT) ar e stored radially in the center section of quadrupole rods by a two-dimensiona l radio frequency (RF) field, and stored axially by stopping

PAGE 76

76 potentials applied to the end s ections of quadrupole rods. The qua drupolar field within the mass analyzer has a voltage of c onstant angular frequency (e.g., = 1188 kHz for the LIT used here) and variable amplitude (0 to 5 kV0-p), which drives ionic motion in both the axial and radial directions. Ionic motion must be stable in both direct ions for an ion to remain trapped. Trapped ions oscillate in the quadrupol ar field with characteristic frequencies known as secular frequencies ( n). For a given set of trapping conditions, these frequencies are characteristic of ion mass-to-charge ( m/z ). By subjecting ions to a signal of frequency equal to a characteristic ion frequency, they can be radially excited and ejected from the ion trap. Secular frequencies ( n) are given by Equation 3-1, 2 ) 2 ( nn (0 1) (3-1) where n = 0, 1, 2, … etc., is the RF drive frequency, and is a complex function of the Mathieu parameters a and q whose solutions classify an ion as stable or unstable.108 Since DC voltage is not applied to the LIT quadrupole electrodes, a = 0 resulting in Equation 3-2. 2 2 2 2 2 2 2 2 2 2 2 2 2) 12 ( ) 10 ( ) 8 ( ) 6 ( ) 4 ( ) 2 ( q q q q q q 2 2 2 2 2 2 2 2 2 2 2 2) 12 ( ) 10 ( ) 8 ( ) 6 ( ) 4 ( ) 2 ( q q q q q q(3-2) In order to solve for in Equation 3-2, it is first necessary to solve for the q value for the ion to be isolated. This q value is calculated usi ng Equation 3-3, in which ( m/z )center is the ion placed at the q of isolation (0.83). z m z m qcenter/ ) 83 0 ( ) / ( (3-3)

PAGE 77

77 is then calculated through an iterati ve process starting with an initial value given by the Dehmelt’s approximation given in Equation 3-4 for q values less than 0.4. 22q (3-4) This chapter introduces the use of a multi-notc h SWIFT applied to the linear ion trap mass spectrometer for selectively isolating multiple pairs of analyte and internal standard ions during a single MSn scan to improve precision during MALDI-MSn quantification. A dual-notch SWIFT waveform was optimized for the isolation of cocaine and its metabolite benzoylecgonine along with their corresponding trideuterated analogs. Experimental Chemicals Cocaine (COC; MW 303.4 Da), benzoylecgon ine (BE; MW 289.3 Da) were purchased from Cerilliant (Round Rock, TX, USA) at concen trations of 1 mg/mL in acetonitrile. COC-d3 (MW 306.4 Da, 0.17% d0) and BE-d3 (MW 292.3 Da, 0.08% d0) were also purchased from Cerilliant at concentrations of 100 g/mL. in acetonitrile. High-performance liquid chromatography (HPLC)-grade acetonitrile, metha nol, and water were purchased from Fisher Scientific (Pittsburgh, PA, USA). Working standards of COC, COC-d3, BE, and BE-d3 were diluted with acetonitrile and then stored at 4 oC. MALDI matrix, 2,5-dihydroxybenzoic acid (DHB; MW 154.1 Da), was purchased from ACROS Organics (Geel, Belgium). Saturated DHB matrix solutions (40 mg/mL DHB) were prepared in methanol/water (70:30, vol/vol) on the day of use. Tissue Collection Human brain tissue samples were provided by the El Paso County Coroner’s Office in Colorado Springs, CO. Postmortem brain materi al was excised from the nucleus accumbens

PAGE 78

78 (NAc) from case number 07A-369, whose toxicologic analysis indicated the presence of cocaine in blood at 69 ng/mL (COC concentration in the brain tissue was not quan tified). The NAc is a dopamine-rich area of the striatum, which may contain an accumulation of COC due to its affinity to bind with the dopamine transporter.100 At autopsy, the exci sed tissue was immediately snap-frozen in liquid nitrogen and then stored in a -80 oC freezer until analyzed. Tissue Sectioning and Sample Preparation Frozen brain tissue was cut into thin sections (20 m thickness) in a cryostat (HM 505E; Microm International GmbH, Waldorf, Germany) at -25 oC. The tissue samples were frozen to the cryostat sample stage using distilled water. Serial brain sections were collected onto microscope slides where they were th aw mounted and then stored at -80 oC. Before mass spectrometric analysis, the tissue sections were removed from the freezer and placed in a vacuum desiccator for 30 min befo re spiking standards (1 L droplets by micropi pette) and applying MALDI matrix. The matrix was applied to the ti ssue sections using an artistic airbrush (Aztek A470; Testors, Rockford, IL, USA). The applic ation of MALDI matrix by airbrush has been previously published.67 Mass Spectrometry All experiments were performed using an LTQ linear ion trap with a vMALDI ion source (Thermo Finnigan, San Jose, CA, USA), equipped w ith nitrogen laser (337 nm) at a frequency of 20 Hz and 100 m spot size. A more detailed descri ption of this instrument has been published.67 The number of laser shots was automatica lly varied (between 1 and 17 shots) using automatic gain control (AGC) to optimally fill th e trap with ions, thus avoiding space chargerelated peak broadening and mass shifts. AGC assesses the ion genera tion rate by use of a

PAGE 79

79 prescan, and then adjusts the number of laser s hots per scan to produce an optimal number of ions for each scan. The spectra are normalized to the number of laser shots for each scan. Resonance excitation is used for isola tion, activation and mass analysis. For MSn experiments, unwanted ions are resonantly ej ected from the ion trap by applying a 5-500 kHz multi-frequency isolation waveform consisting of sine components spaced every 0.5 kHz. The ions of interest are isolated in the ion trap by removing sine compone nts from the isolation waveform that correspond to the secular frequency of the desired ions. Ions to be isolated are selected in the LTQ software by entering the m/z with its IW. The mass range for the ion is defined as ( m/z IW/2) to ( m/z + IW/2). The IW should be narrow enough to eliminate interfering peaks, but wide enough to avoid loss of sensitivity for the desired ions. However, it is important to note that the activat ion width for resonance excitati on (CID) has the same value as the IW. Therefore, the collision energy applied during MSn is spread over the activation width. Thus, increasing the IW decreases th e collision energy for each ion. SWIFT Calculation Inverse Fourier Transform A computer program written in C++ was used to calculate the SWIFT waveform based on a process previously described.106 Notches in the desired broa dband magnitude spectrum, from frequency 0 to 500 kHz, were calculated to have centers corresponding to the secular frequencies of the ions to be isolated. The frequency sp ectrum was then transformed to the time domain using the inverse Fourie r transform (IFT), which was perf ormed using an adaptation of the Cooly-Tukey fast Fourier transform (FFT) algorithm.109 The algorithm generates output, which must be midpoint reflected about the N/2 axis, where N equals the total number of points in the SWIFT waveform. This step is similar to a time shift and therefore affect s the phase, but not the magnitude of the corresponding frequency-domai n spectrum. The advantage of midpoint

PAGE 80

80 reflection is that it avoids sudde n voltage transients at the be ginning and end of the excitation period. Quadratic Phase Modulation Phase modulation of the frequency spectrum is necessary in order to reduce the dynamic range of the time-domain waveform. The large dy namic range is caused by all of the specified frequency components starting with the same pha se at time zero. A nonl inear phase modulation varies the phase continuously at a nonconstant rate with frequenc y, and results in a broader timedomain waveform after IFT that requires less amplitude to achieve the same power. The real, Ri, and imaginary, Ii, components are created from the magnitude data, Magi, using the following relationship: i i i i i iMag I Mag R sin cos (3-5) Where the phase, i, varies quadratically with frequency: 2 0) 2 / ( i K Jii (3-6) Here 0 is the initial phase (zero), i is the frequency index, and J = 0.5 and K = / N are the quadratic terms, where N is the number of nonzero data points in the frequency spectrum. Values of J and K are chosen to satisfy the Nyquist criterion, such that the rate of phase change per frequency-domain data point is kept at half the Nyquist limit ( ) or below, which removes nonuniformity of the magnitude in the frequency-domain.110 Temporal Spectral Inhomogeneity It has been shown previously that quadratic phase modulati on is effective at obtaining a more uniform excitation power spectrum.111 However, an undesira ble consequence of phase modulation is temporal spectral inhomogeneity. This means that SWIFT is essentially a frequency scan in which the frequency content is localized in time and vari es systematically with

PAGE 81

81 time. Figure 3-1a shows the time domain of a dual-notch SWIFT waveform with frequency notches corresponding to the secu lar frequencies of the [M+H]+ ions of COC ( m/z 304.25) and COC-d3 ( m/z 307.25). Figure 3-1b is the re sulting frequency domain waveform after performing fast Fourier transform (FFT) of the time-domain waveform in Figure 3-1a. The two frequency notches are 1.6-Da wide at 431.88477 – 436.27930 kHz and 439.69727 – 444.33594 kHz. FFT of the first half (0 to 2047 s) of the time-domain waveform in Figure 3-1a is shown in Figure 31c and the FFT of the second half (2048 to 4096 s) of the waveform is shown in Figure 3-1d. The resulting frequency-domain spectra shown in Figures 3-1c and 31d illustrate that the SWIFT waveform scans from high to low frequenc y and that the frequency content is localized in time. Thus, ions of different m/z are excited at different times during the SWIFT waveform. This may not be such a critical issue for the FT-ICR, for which SWIFT was originally designed, but it is undesirable for LIT experiments, sin ce many collisions with helium buffer gas (~ 1 mTorr) occur during the excitation event. For the LI T, it is therefore desirable to excite ions of all desired m/z values simultaneously. Three solutions were previously published112 for correcting or minimizing the temporal spectral inhomogeneity of SWIFT. One solu tion involves using a short-duration (lowerresolution) SWIFT waveform that is repeated many times during the excitation event, which serves to increase the number of times a specific frequency is represented in the time domain during the excitation event; however, this leads to lower mass selectivity due to the lowerresolution SWIFT waveform. A second approach (multiple foldovers) involves overmodulation of the phase in which the quadratic term in th e phase vs. frequency spectrum corresponds to a bandwidth multiple times (number of foldovers) hi gher than the Nyquist limit. If the foldover number is sufficiently high, then the SWIFT waveform effectively excites all frequency

PAGE 82

82 components simultaneously. A third approach is to distribute the excitation frequency components randomly throughout the time-domain ex citation period, which has been previously termed “filtered noise field” excitation.113 Apodization The time-domain SWIFT waveform was multiplied by an apodization (smoothing) function to force the time-domain signal smoothly to zero at the beginnin g and end of the timedomain period. The apodization function consists of a quarter-wave sinusoid matched to the first one-fourth of the time-domain period, followe d by unit weighting for the middle half period, followed by a quarter-wave sinusoid for th e final one-fourth of the period. It was previously shown114 that apodization can cause fr equency notch distortion, which may lead to partial ejection of ions selected to be isolated. Multip lication of the apodizing function with the time-domain waveform corresponds to convolution of the Fourier transforms in the frequency domain. The process of convolutio n widens the bases of the spectral components, which may lead to power leakage into adjacent spectral components and loss of frequency resolution. Notched waveforms ideally represent discontinuities in the waveform; however, the process of apodization transforms these discon tinuities or sharp edge s of the notch into continuous transitions. Thus, the edges of a na rrow notch may fuse into each other, thereby bridging the notch to some extent. Figure 3-2a compares the time domain of a dual-notch SWIFT waveform that has been apodized (blue) with one that has not been a podized (red). It is shown that the apodization function only reduces the magnitude of the Gibb’s os cillations (spurious os cillations that occur when using a truncated Fourier series)115 for the first 700 s of the time-domain waveform, and only the magnitude of the first 200 s have been significantly smoot hed to nearly zero. Figure 3-

PAGE 83

83 2b compares the frequency domain of the dual-no tch SWIFT waveform that has been apodized (blue) with one that has not been apodized (r ed). Apodization of the time domain results in convolution of the edges of the fr equency notches in the frequenc y domain, which could lead to ejection of ions that are intended to be isolated. SWIFT Application to LTQ The non-apodized, time-domain, dual-notch SWIFT waveform was downloaded to the memory of an arbitrary waveform generator (AWG) (Stanford Research Systems Model DS345, Sunnyvale, CA, USA). The AWG has a maximum sa mpling rate of 40 MHz, time resolution of 16,300 data points, and a 12-bit DAC output. The LTQ has a programmable trigger that can be used to send a TTL pulse to the AWG at a spec ific time during the experimental sequence (e.g., isolation or activation). For experiments in which the AWG wa s triggered during isolation, the LTQ isolation waveform was turned off to avoi d interference. Once triggered, the AWG applies the SWIFT waveform to the LTQ analog board, wh ere it is summed with the other waveforms before being amplified and applied to the linear ion trap x-rods. The amplitude of the SWIFT waveform and the number of bursts were modified manually on the AWG. A two-channel digitizing oscilloscope (Tektr onix Model TDS 540, Tektronix Inc., Beaverton, OR) was used to observe the SWIFT waveform. Results and Discussion Optimization of a Dual-Notch SWIFT Frequency optimization A 4096-point (4k) dual-notch SWIFT was calcula ted from frequency 0 to 500 kHz with a sampling frequency of 1000 kHz corresponding to a 0.244 kHz frequency step. Frequency notches (1.6-Da wide) are centered at the secular frequencies of the [M+H]+ ion of COC ( m/z 304.25) and COC-d3 ( m/z 307.25). The theoretical secular frequencies ( ) were calculated by

PAGE 84

84 first calculating the q values for the ions using Equation 3-3 and the values using Equation 3-2. Both q and were then used to calc ulate the corresponding secular frequencies using Equation 31. Experimental secular frequencies can shif t from the theoretical secular frequencies calculated due to space-charge (shift to lower frequencies), higher-order fields, and resonance excitation amplitude (shift to higher frequencies).114, 116 A 1:1 solution of 1 g/mL of COC and COC-d3 were spotted (1 L each) onto a MALDI plate and airbrushed with DHB matrix for analysis. A dual-notch SWIFT waveform with 1.6-Da wide frequency notches centered at the theoretically calculated secular frequencies of m/z 304 and m/z 307 was applied to the LTQ with a SWIFT amplitude of 0.4 Vp-p, and a burst count of 3 set by the AWG. The center m/z 305.8 was placed at a q = 0.83. The LTQ isolation waveform was turned off and the SWIFT waveform was triggered at isolation. The frequency notches for m/z 304 and m/z 307 were optimized separately for maximum peak intensity by shif ting the 1.6-Da window to higher frequencies a number of frequency steps (0.244 kHz) from the starting position at a frequency lower than the theoretically calculated secular frequency (Figure 3-3). The fr equency notches were shifted by recalculating the SWIFT waveform after each cha nge of frequencies. The optimal frequency notches for m/z 304 and m/z 307 was 439.69727 – 444.33594 kHz and 431.88477 – 436.27930 kHz, respectively. This corresponds to a shift to higher frequencies for m/z 304 (+ 0.488 kHz) and m/z 307 (+ 0.732 kHz), which may be due to the amplitude of the SWIFT waveform applied.116 Burst count optimization Increasing the number of SWIFT frequency-dom ain points lengthens the duration of the time-domain SWIFT waveform. However, the duration of the SWIFT pulse is limited by the

PAGE 85

85 size of the stored-waveform data se t. The AWG used for the experi ments in this paper has a data point storage limit of 16,300 points. It would be ideal to apply the SWIFT waveform during the entire isolation event in order to maximize th e opportunity for selectiv e ejection of unwanted ions that are in resonance with the frequencies of the SWIFT pulse. The problem with a short SWIFT pulse duration can be overcome by applying multiple pulses (or bursts) of the same stored waveform in a series (or train). This allows the specific frequencies of the SWIFT waveform to be present for longer periods of time, which can lead to more efficient ejection of selected ions. The number of SWIFT waveform bur sts in a train that are applied to the LTQ can be controlled by the AWG. Figure 4a shows the digital scope image of a 1-burst (4.096 ms), 2-burst (8.192 ms), 3burst (12.288 ms), 4-burst (16.384 ms), and 5-bu rst (20.480 ms) train of pulses shown above the digital scope image of the square-wave trigger du ring isolation (15.5 ms). Each SWIFT pulse in a train is the same dual-notch SWIFT waveform (frequency optimized to isolate ions at m/z 304 and m/z 307) that has been merely repeated. Each train of SWIFT bursts were applied five times individually to a 1:1 so lution of COC and COC-d3 (1 g/mL each) pipetted onto a MALDI plate and airbrushed with DHB matrix. Figure 3-4b shows the changes in absolute peak intensities for m/z 304 and m/z 307 with varying number of SWIFT bur sts. Not shown, but also monitored, were the absolute peak intensi ties of the background ions at m/z 305, 306, and 308. From 1 to 3 bursts, there was a 12% and 14% decrease in signal for m/z 304 and 307, respectively. The signals for the background ions at m/z 305, 306, and 308 decreased by 69%, 59%, and 87%, respectively, from 1 to 3 bursts. At 4 bursts, the intensities of the ions at m/z 304 and 307 decreased by 39% and 49%, respectively. Th is significant decrease in ion signal may be attributed to the longer duration of the 4-burst train (16.384 ms) co mpared to the duration of the

PAGE 86

86 isolation event (15.5 ms). The ions may not ha ve enough time to relax towards the center of the LIT before they are moved from the q of isolation (0.83) to a lower q of activation (0.25). There is ~2 ms pause after the isolation event before the RF voltage is decreased to move the ions towards the lower q of activation. At 5 bursts (20.480 ms), all ions were ejected. The optimal number of SWIFT bursts was determined to be 3, since there was not a si gnificant decrease in the signal intensity of m/z 304 and 307, and signal intensities of the background ions at m/z 305, 306, and 308 were reduced to below 3% intensity relative to the m/z 304 and 307 at 3 bursts. Amplitude optimization Another SWIFT parameter besides the duration of the SWIFT pulse that can affect the ejections of ions is the amplit ude. Increasing the amplitude of the SWIFT waveform in turn increases the magnitude of the oscillations of the i ons that are in resonance with the frequencies. It has also been reported that large amplitudes distort the fr equency-domain cutoffs, causing the pulse to have an even wider frequency range, which can interfere with ions close in mass.106 The secular frequencies of ions can shift to higher values with increased excitation amplitude, resulting in ions absorbing energy in a range near their secular frequency and being unintentionally ejected from the ion trap. The amplitude of a dual-notch SWIFT waveform with a 3-burst train was optimized by varying the amplitude on the AWG from 0 to 1.0 Vp-p and monitoring the peak intensity of m/z 304 and m/z 307. A 1:1 solution of COC and COC-d3 (1 g/mL each) was pipetted (1 L) onto a MALDI plate and airbrushed with DHB matrix. Figure 3-5 shows the absolute peak intensities of the [M+H]+ ions of COC ( m/z 304) and COC-d3 ( m/z 307) along with the 13C isotope ions at m/z 305 and 308. The optimal SWIFT amplitude (0.40 Vp-p) was determined to be the lowest potential needed to maintain a 1:1 signal of COC and COC-d3, while decreasing the background

PAGE 87

87 ions below 3% relative intensit y, and the maximum signal for the m/z 304 and 307 ions. Notice that at higher amplitudes above 0.50 Vp-p, more m/z 304 ions are ejected than m/z 307 ions. This may be attributed to m/z 304 being at higher q closer to the right-ha nd edge of the stability diagram. Selective Ion Isolation of Standards on MALDI Plate The optimized dual-notch SWIFT was applied to COC and COC-d3 at 1 g/mL each pipetted onto a MALDI plate and airbrushed wi th DHB matrix. The amplitude on the AWG was set to 0.4 Vp-p with a burst count of 3. Figure 3-6a compares the full scan mass spectrum (top), a mass spectrum of a 5-Da wide isolation window centered at m/z 305.8 (middle), and the mass spectrum from the application of the optimized dual-notch SWIFT (bottom). The 5-Da wide isolation window is effective at eliminating the background ions outside the window, but retains ions at m/z 305 and 306. The dual-notch SWIFT (bo ttom) is able to eliminate the same background ions as well as re duce the ion intensities for m/z 305 and 306 without reducing the signal for the desired ions at m/z 304 and 307. Figure 3-6b shows the MS2 scans resulting from applying a 5-Da wide broadband excitation waveform (CID = 55) to the isolated ions from the 5Da wide isolation (top) and the dual-not ch SWIFT isolation (bottom). The MS2 scan of the isolated SWIFT ions (bottom) resulted in no b ackground fragment ions in the product spectrum. The only ions present are the fragment ions of m/z 304 and 307 at m/z 182 and 185, respectively, both formed from the neutral loss of benzoic ac id. In comparison, the pr oduct spectrum of the ions isolated with the 5-Da wide isolation wi ndow (top), shows the presence of fragment ions from the background as well as those from m/z 304 and 307. Improving MALDI Precision with SWIFT It was previously shown that isolating the analyte and internal standard ions in a single MSn scan using a wide isolation window can improve precision for MALDI quantification

PAGE 88

88 compared to isolating the analyte and internal standard ions separately during alternate MSn scans.103 In order to compare the ability of a dual-notch SWIFT waveform to also improve precision for MALDI quantification, five solu tions (31, 62, 125, 250, a nd 500 ng/mL COC with 250 ng/mL COC-d3) were pipetted onto a MALDI plate an d airbrushed with DHB matrix. The peak intensity of m/z 182 and m/z 185 were ratioed from three different MS2 experiments (CID = 55): alternating scans (MS2 of m/z 304 and m/z 307 during separate scans), MS2 of m/z 304 and m/z 307 during a single scan isolated by a dual-notch SWIFT, and during a single MS2 scan isolated by a 5-Da window. The 5-Da wide isolation experiment had the best precision (% RSD = 1 to 9%) for isotopic ratios, followed by dual-notch SWIFT (% RSD = 5 to 23%), and then alternating MS2 scans (% RSD = 44 to 56%). This reinforces the conclusion103 that isolating the analyte and internal standard ions during a single MSn scan improves precision over isolating the analyte and internals standard ions during separa te scans. The dual-notch SWIFT isolation may be less precise than wide-isolation due to irreprod ucible shifts in the secular frequencies of the analyte and internal standard ions which leads to their ejection. Although SWIFT isolation may be less prec ise of a method for MALDI quantification than wide isolation, the major advantage of SWIFT is that it can selectivel y isolate multiple ions or m/z ranges. This is done simply by removing the appropriate frequencies from the SWIFT waveform corresponding to the secular frequenc ies of the desired ions. A quad-notch SWIFT was calculated with frequency notches (1.5-Da wide) corresponding to the secular frequencies of the [M+H]+ ions of BE ( m/z 290.25), BE-d3 ( m/z 293.25), COC ( m/z 304.25), and COC-d3 ( m/z 307.25). The quad-notch SWIFT was a pplied to a solution of BE, BE-d3, COC, and COC-d3 at 1 g/mL each that was pipette (1 L) onto a MALDI plate and then airbrushed with DHB matrix. Figure 3-8a compares the full scan mass spectru m (top), the mass spectrum from a 20-Da wide

PAGE 89

89 isolation window centered at m/z 298.75 (middle), and the mass spectrum from applying a quadnotch SWIFT (bottom). The 20-Da wide isolat ion (middle) was effective at eliminating background ions outside the isolation window, but because of the large wi dth of the isolation window necessary to isolate all of the analyte an d internal standard ions present, many undesired background ions were isolated as well. The intensities of the ions isolated by the quad-notch SWIFT (bottom) were lower than the intensitie s of the same ions isolated by a 20-Da wide isolation window centered at m/z 298.75, but the quad-notch SWIFT was able to significantly reduce the intensities of the background ions, wh ich should help to simplify the product spectra from MS2 analysis. Figure 3-8b shows the MS2 scans from applying a 20-Da broadband excitation (CID = 55) to the ions isolated from the 20-Da wide isolation (top) and a quad-notch SWIFT (bottom). The product ion spectrum of the quad-notch SWIFT isolation (botto m) shows the presence of the product ions of m/z 290, 293, 304, and 307 at m/z 168, 171, 182, and 185, respectively, formed from the neutral loss of benzoic acid. Also present is the common fragment ion of both m/z 290 and 304 at m/z 150 (neutral loss of benzoic acid and me thanol) and the common fragment ion of m/z 293 and 307 at m/z 153 (neutral loss of benzoic acid and methanol). The quad-notch SWIFT was able to reduce background ions during isol ation, which resulted in the elimination of 13C peaks and MALDI matrix ions (e.g., m/z 137, [DHB-H2O+H]+) from the product ion spectrum that are present in the pr oduct ion spectrum from the 20Da wide isolation (top). Selective Ion Isolation of Standards on Tissue The quad-notch SWIFT was applied to 20 m thick human brain tissue that was spiked (1 L on top of tissue) with BE, BE-d3, COC, and COC-d3 at 1 g/mL each and then airbrushed with DHB matrix. Figure 3-9a shows that the full mass spectrum is dominated by lipid and DHB

PAGE 90

90 matrix ions and it is difficult to see the target ions at m/z 290, 293, 304, and 307. Figure 3-9b shows an expanded view of the mass range from m/z 250 to 350 with the full scan mass spectrum (top), mass spectrum from a 20-Da wide isolation centered at m/z 298.75 (middle), and the mass spectrum resulting from the application of a the quad-notch SWIFT (bottom). The 20-Da wide isolation window (middle) isolates the [M+H]+ ions of BE, BE-d3, COC, and COC-d3 at m/z 290, 293, 304, and 307, respectively, as well as a number of background ions. The quad-notch SWIFT (bottom) was successful in eliminating or reducing the background ions during isolation of the analyte and inte rnal standard ions. Conclusions It was previously shown that isolating the analyte and inte rnal standard ions during a single MS2 scan using a wide isolation window can provide improved precision for the quantitative analysis of COC in postmortem brain tissue compared to using two alternating MS2 scans that isolate the analyte and in ternal standard ions separately.103 Here, multi-notch SWIFT waveforms were investigated as an alternative is olation technique to wide isolation for isolating the analyte and internal sta ndard ions during a single MS2 scan for improved MALDI-MS2 precision. It was determined that multi-notch SWIFT isolation can provide improved precision when compared to using two alternating MS2 scans; however, it is not as precise as the wide isolation method. Nevertheless, SWIFT isolation offers the a dvantage of higher sele ctivity and is better able to reduce background ions that may complicate or interfere with MS2 analysis (e.g., isobaric product ions). Also, analysis times can be redu ced as more frequency notches are added to the SWIFT waveform. This can become very important when quantitatively imaging several analytes from a large tissue sample.

PAGE 91

91 Figure 3-1. Temporal spectral inhomogeneity of SWIFT. (a) Time domain of dual-notch SWIFT waveform calculated by inverse Fourier transform (IFT) of the frequency domain. (b) Fast Fourier transform (FFT ) of the full (0 to 4096 s) time-domain SWIFT waveform. (c) FFT of the first half (0 to 2047 s) of the time-domain SWIFT waveform. (d) FFT of the second half (2048 to 4096 s) of the time-domain SWIFT waveform. (a) (b) (c) (d)

PAGE 92

92 Figure 3-2. Effects of ap odization on SWIFT. (a) Time domain of dual-notch SWIFT waveform that has been apodized (blue) overlapped with the time domain of the same dualnotch SWIFT waveform without apodizati on (red). (b) Comparing the frequency domain of a dual-notch SWIFT waveform with apodization (blue) and without apodization (red). (a) (b)

PAGE 93

93 Figure 3-3. Optimization of frequency notches (1.6-Da) for (a) m/z 304 and (b) m/z 307. Frequency notches were shifted to higher frequencies a number of frequency steps (0.244 kHz) from the starting position to maximize the peak intensities of m/z 304 and m/z 307. The optimal frequency notches for m/z 304 and m/z 307 were 439.69727 – 444.33594 kHz and 431.88477 – 436.27930 kHz, respectively. Theoretical(a) Theoretical(b)

PAGE 94

Figure 3 ( b 4. Optimiz a images o f (16.384 m image of t absolute p plotted v e isolation. b ) a tion of bur s f a 1 burst ( 4 m s), and 5 b u t he squarew p eak intensi t e rsus the nu m The error b s t counts fo r 4 .096 ms), 2 u rst (20.480 w ave trigger t ies of the [ M m ber of SW I b ars corresp o 94 r a dual-notc h burst (8.19 2 ms) train o f during isol a M +H]+ ion o I FT bursts o o nd to the h SWIFT w a 2 ms), 3 bur s f pulses sho w a tion (15.5 m f COC ( m/z f a dual-not c standard err o a veform. ( a s t (12.288 m w n above th e m s). (b) The 304) and C O c h SWIFT a p o r (5 replic a a ) Digital sc o m s), 4 burst e digital sco p average O C-d3 ( m/z 3 p plied duri n a tes). o pe p e 3 07) n g

PAGE 95

95 Figure 3-5. Optimization of SWIFT amplitude for a dual-notch SWIFT waveform. SWIFT amplitude was optimized by varying the poten tial from 0 to 1.0 Vp-p and monitoring the absolute peak intensities of the [M+H]+ ions of COC ( m/z 304) and COC-d3 ( m/z 307) along with the ions of their corresponding 13C isotopic peaks at m/z 305 and m/z 308. The optimal SWIFT amplitude (0.40 Vp-p) was the lowest potential needed to maintain a 1:1 signal of COC and COC-d3 (1 g/mL each) that were pipetted (1 L) onto a MALDI plate and airb rushed with DHB matrix.

PAGE 96

96 Figure 3-6. SWIFT isolation and wide isolation comp arison. (a) Comparing full scan (top), 5Da wide isolation window centered at m/z 305.8 (middle), and dual-notch SWIFT (bottom) of a 1:1 solution of COC and COC-d3 (1 g/mL each) that was pipetted onto a MALDI plate (1 L) and airbrushed with DHB matrix. (b) Comparing the MS2 scan of ions fragmented with a 5-Da broadband excitation waveform (C ID = 55) applied to ions isolated with a 5-Da wi de isolation window centered at m/z 305.8 (top) and ions isolated with a dual-notch SWIFT waveform (bottom). 303 304 305 306 307 308 309 m/z 0 20000 40000 60000 80000 100000 0 20000 40000 60000 80000 100000 In te n sity 0 20000 40000 60000 80000 100000 304.42 307.42 305.42 308.42 309.42 306.42 302.50 303.42 307.33 304.25 305.33 306.33 304.42 307.33 305.42306.42 Full MS Scan 5 Da Wide Isolation Centered at m/z 305.8 Dual Notch SWIFT(a) 180 181 182 183 184 185 186 187 188 m/z 0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000 In te n sity 182.25 185.25 183.25 184.25 182.25 185.255 Da Wide Isolation Centered at m/z 305.8 Dual Notch SWIFT(b)

PAGE 97

97 Figure 3-7. SWIFT isolation and wide isolation calib ration curves. Five solutions (31, 62, 125, 250, and 500 ng/mL COC with 250 ng/mL COC-d3 were pipetted onto a MALDI plate (1 L each) and airbrushed with DHB ma trix. The peak intensity of m/z 182 and m/z 185 were ratioed from three different MS2 experiments (CID = 55): alternating scans (MS2 of m/z 304 and m/z 307 during separate scans), MS2 of m/z 304 and m/z 307 during a single scan isolated by a dual-notch SWIFT and during a single MS2 scan isolated by a 5-Da window The error bars correspond to the standard error (3 replicates). Mass (pg) Alt Scan %RSD Dual Notch %RSD Wide Iso %RSD 31 52 5 5 62 46 17 4 125 56 21 1 250 50 23 2 500 44 21 9

PAGE 98

98 Figure 3-8. Quad-notch SWIFT isolation of standards on MALDI plate. (a) Comparing full scan (top), 20-Da wide isolat ion window centered at m/z 298.75 (middle), and quad-notch SWIFT (bottom) of a solution of BE, BE-d3, COC, and COC-d3 at 1 g/mL each that was pipetted (1 L) onto a MALDI plate and then airb rushed with DHB matrix. (b) Comparing the MS2 scan of ions fragmented with a 20-Da broadband excitation waveform (CID = 55) applied to ions isol ated with a 20-Da wide isolation window centered at m/z 298.75 (top) and ions isolated wi th a quad-notch SWIFT waveform (bottom). [BE+H]+[BE d3+H]+[COC d3+H]+[COC+H]+ 290 295 300 305 310 315 m/z 0 10000 20000 30000 40000 50000 0 20000 40000 60000 80000 100000 120000 In te n sity 0 10000 20000 30000 40000 50000 304.25 307.25 290.25 293.25 308.25 294.25 313.17 309.25 287.33301.33 297.25 307.25 304.33 290.33 293.33 294.33 301.33 297.25 289.33 308.25 307.33 304.33 293.33 290.25 308.33 Full MS Scan 20 Da Wide Isolation Centered at m/z 298.75 Quad Notch SWIFT(a) 100 150 200 250 300 350 m/z 0 5000 10000 15000 20000 25000 30000 35000 0 5000 10000 15000 20000 25000 30000 35000 In te n sity 20 Da Wide Isolation Centered at m/z 298.75 Quad Notch SWIFT168.25 182.25 185.25 182.33 168.25 137.17 307.25 307.33 304.33 150.25 150.33 153.33 171.33 185.33 290.33 293.33 171.25 153.25 304.25 290.25 293.25 [DHB+H H2O]+(b)

PAGE 99

99 Figure 3-9. Quad-notch SWIFT isolation of standards on brain tissue. (a) Mass spectrum of 20 m thick brain tissue slice that was spiked (1 L) with a solution of BE, BE-d3, COC, and COC-d3 at 1 g/mL each and airbrushed with DHB matrix. (b) Expanded region ( m/z 250 – 350) comparing full scan (top), 20 -Da wide isolation window centered at m/z 298.75 (middle), and quad-notch SWIFT (bottom). p[] 100 200 300 400 500 600 700 800 900 1000 m/z 0 5000 10000 15000 20000 25000 30000 35000 40000 Intensity 184.42 798.67 782.67 772.67 760.67 137.42 739.67 273.42 826.58 848.58 723.67 231.25 313.33 577.75 449.58 866.58 369.50 534.50 651.67 974.42 Zoom in(a) [BE+H]+[BE d3+H]+[COC d3+H]+[COC+H]+ 260 280 300 320 340 m/z 0 200 400 600 800 1000 0 200 400 600 800Intensity 0 2000 4000 6000 8000 273.42 313.33 275.42 307.42 264.50 291.25 331.33 279.42 339.50 257.42 323.42 296.17 307.25 289.17 307.42 304.42 293.42 341.75 313.50 Full MS Scan 20 Da Wide Isolation Centered at m/z 298.75 Quad Notch SWIFT290.42(b)

PAGE 100

100 CHAPTER 4 QUANTITATIVE ANALYSIS OF DRUGS IN BRAIN TISSUE Introduction A large variety of specimens are collected in the field of postmortem forensic toxicology including blood, liver, brain, and urine.4 For the analysis of drugs of abuse, brain samples show several advantages over a ll other specimens in postm ortem forensic toxicology.5 One advantage is due to the brain being an isolated compartment, which de lays putrefaction after death.6 Also, the metabolic activity is lower in the brain than in other tissues or in blood, resul ting in slower decomposition.7 Finally, drugs of abuse establish their effects through the central nervous system. Therefore, it can be assumed that con centrations of drugs of abuse found in the brain better reflect drug concentrations at thei r site of action at the time of death.8 Analysis of drugs of abuse in the brain ha s applications in forensic and postmortem toxicology. Drug concentr ations in the brain may be needed to substantiate fatal overdoses9 and support neurotoxicity studies.10 Direct measurement of drug a nd metabolite concentrations in discrete brain regions can also be used to study the mechanisms of drug action,11 regional distribution,12 and preferential accumulation of drugs.13 Conventional drug analysis in tissue involves tissue homogenization of the tissue prior to subsequent chromatographic analysis.14 Such sample pretreatments are known to introduce variation in detection due to inhomogeneity of the analyte within the sample matrix.15 Also, homogenization of tissue elimin ates the opportunity to acqui re detailed anatomical and histological information for in situ drug distribution. Imaging t echniques that include mass spectrometric imaging can help provide this information.

PAGE 101

101 Experimental Chemicals Cocaine (COC; MW 303.4 Da), benzoylecgonin e (BE; MW 289.3 Da), and cocaethylene (CE; MW 317.4 Da) were purchased from Cerilli ant (Round Rock, TX, USA) at concentrations of 1 mg/mL in acetonitrile. COC-d3 (MW 306.4 Da, 0.17% d0), BE-d3 (MW 292.3 Da, 0.08% d0), and CE-d3 (MW 320.4 Da 0.15% d0) were also purchased from Cerilliant at concentrations of 100 g/mL in acetonitrile. High-perfor mance liquid chromatography (HPLC)-grade acetonitrile, methanol, and water were purchased from Fisher Scientific (Pittsburgh, PA, USA). Working standards of COC, COC-d3, BE, BE-d3, CE, and CE-d3 were diluted with acetonitrile and then stored at 4 oC. MALDI matrix, 2,5-dihydroxybenz oic acid (DHB; MW 154.1 Da), was purchased from ACROS Organics (Geel, Belgium). Saturated DHB matrix solutions (40 mg/mL DHB) were prepared in methanol/water (70:30, vol/vol) on the day of use. Tissue Collection Human brain tissue samples were provided by the El Paso County Coroner’s Office in Colorado Springs, CO. Postmortem brain materi al was excised from the nucleus accumbens (NAc) from case number 07A-369, whose toxicologic analysis indicated the presence of cocaine in blood at 69 ng/mL (COC concentration in the brain tissue was not quan tified). The NAc is a dopamine-rich area of the striatum, which may contain an accumulation of COC due to its affinity to bind with the dopamine transporter.100 At autopsy, the exci sed tissue was immediately snap-frozen in liquid nitrogen and then stored in a -80 oC freezer until analyzed. Tissue Sectioning and Sample Preparation Frozen brain tissue was cut into thin sectio ns (20 m thickness) in a cryostat (HM 505E; Microm International GmbH, Waldorf, Germany) at -25 oC. The tissue samples were frozen to the cryostat sample stage using distilled water. Serial brain sections were collected onto

PAGE 102

102 microscope slides where they were th aw mounted and then stored at -80 oC. Before mass spectrometric analysis, the tissue sections were removed from the freezer and placed in a vacuum desiccator for 30 min before spiking standa rds (1-L droplets by mi cropipette) and applying MALDI matrix. The matrix was applied to the ti ssue sections using an artistic airbrush (Aztek A470; Testors, Rockford, IL, USA). The applic ation of MALDI matrix by airbrush has been previously published.67 Tissue Homogenization One gram of blank human brain tissue (i.e., tis sue for which toxicological analysis did not indicate the presence of the analyte drug) was cut and weighed, and then finely minced with a scalpel. The minced tissue was then placed into a glass tissue grinder (Duall 21; Kontes Glass Inc., Vineland, New Jersey, USA), where it was ho mogenized into a liquid. The reservoir and pestle shaft of the tissue grinde r were rinsed with 3 mL of 60 M sodium fluoride (NaF), which serves as an inhibitor for este rases to prevent COC hydrolysis.99 The volume of NaF added was measured to be approximately twice the ma ss of the corresponding tissue sample. The homogenized tissue was then transferred to an 8mL glass vial and placed in a sonicator for 1 min. Preparation of Standard Solutions Five standard solutions were prepared for sp iking into tissue homoge nate for generation of calibration curves. The concentrations of these 1-mL solutions was 62, 125, 250, 500, and 1000 ng/mL each of BE, COC, and CE, as well as a mi xture of their corresponding internal standards (BE-d3, COC-d3, and CE-d3; 200 ng/mL each). The five standard solutions were dried with nitrogen gas and then reconstituted with a 400 L aliquot of the sonicated homogenate. The solutions were immediately vortex-mixed (1 min) and centrifuged at 10,000 rpm for 30 min. The supernatants were then used for solid-phase extraction.

PAGE 103

103 Preparation of Unknown Sample Solution One gram of human brain tissue from cas e number 07A-369, for which toxicological analysis indicated the presence of cocaine in bl ood at 69 ng/mL (COC conc entration in the brain tissue was not quantified) was cut and weighed, a nd then finely minced with a scalpel. The minced tissue was then placed into a glass ti ssue grinder, where it was homogenized into a liquid. The reservoir and pestle shaft of the tissue gr inder were rinsed with 3 mL of 60 M sodium fluoride (NaF). The homogenized tissue was then transferred to an 8-mL glass vial and placed in a sonicator for 1 min. A 1-mL solution of the internal standards (BE-d3, COC-d3, and CE-d3; 200 ng/mL each) was prepared, dried with nitrogen ga s and then reconstituted with a 400 L aliquot of the sonicated homogenate from case number 07A369. The solution was immediately vortex-mixed (1 min) and centrifuged at 10,000 rpm for 30 min. The supernatant was then used for solidphase extraction. Solid-Phase Extraction The extraction of cocaine and its metabolites was performed using underivatized silica (50 m average particle size; 60 pore size) solid-p hase extraction (SPE) cartridges (HyperSep SI; 3-mL reservoir, 500-mg bed; Thermo Scientific Bellafonte, PA, USA). The analytes in the homogenate were separated from impurities usin g a selective elution scheme shown in Figure 41, in which the adsorbed compounds of interest we re eluted in a solvent that left the strongly retained impurities behind on the cartridge. The SPE cartridge was first conditioned using 2 mL of methanol followed by 2 mL of deionized water. Then a 100 L aliquot of the supernatant from the centrifuged homogenate was loaded an d drawn through the cartridge using low vacuum (~ 5 in. Hg; 1 in. Hg = 388.638 Pa) in a vacuum manifold (PrepSep 12-Port Vacuum Manifold;

PAGE 104

104 Fisher Scientific, Pittsburgh, PA, USA). After discarding the elue nt, analytes in the cartridge were eluted using 3 mL of 5% ammonia in me thanol solution. A washing step is typically performed to remove interferences in the biological matrices that may affect the assay; however, this step was not performed to avoid loss in re coveries of the highly polar metabolite ecgonine methyl ester. The high selectivity of MSn makes the lack of this washing step less of a concern. The eluents from the extraction ca rtridge were then dried using n itrogen gas. The residue was reconstituted in 500 L of water/methanol (90:10, vol/vol ), spotted onto a MALDI plate (1 L) and airbrushed with DHB matrix. Mass Spectrometry All experiments were performed using an LTQ linear ion trap with a vMALDI ion source (Thermo Finnigan, San Jose, CA, USA), equipped w ith nitrogen laser (337 nm) at a frequency of 20 Hz and 100 m spot size. A more detailed descri ption of this instrument has been published.67 The number of laser shots was automatica lly varied (between 1 and 17 shots) using automatic gain control (AGC) to optimally fill th e trap with ions, thus avoiding space chargerelated peak broadening and mass shifts. AGC assesses the ion genera tion rate by use of a prescan, and then adjusts the number of laser s hots per scan to produce an optimal number of ions for each scan. The spectra are normalized to the number of laser shots for each scan. Resonance excitation is used for isola tion, activation and mass analysis. For MSn experiments, unwanted ions are resonantly ej ected from the ion trap by applying a 5-500 kHz multi-frequency isolation waveform consisting of sine components spaced every 0.5 kHz. The ions of interest are isolated in the ion trap by removing sine compone nts from the isolation waveform that correspond to the secular frequency of the desired ions. Ions to be isolated are selected in the LTQ software by entering the m/z with its IW. The mass range for the ion is

PAGE 105

105 defined as ( m/z IW/2) to ( m/z + IW/2). The IW should be na rrow enough to eliminate including interfering peaks, but wide enough to avoid loss of sensitivity for the desired ions. However, it is important to note that the activat ion width for resonance excitati on (CID) has the same value as the IW. Therefore, the collision energy applied during MSn is spread over the activation width. Thus, increasing the IW decreases th e collision energy for each ion. SWIFT Calculation A computer program written in C++ was used to calculate the SWIFT waveform based on a process previously described.106 The same procedure was used he re except that the final timedomain signal was left unapodi zed. Notches in the desired br oadband magnitude spectrum, from frequency 0 to 500 kHz, were calculated to have centers corresponding to the secular frequencies of the ions to be isolated. The frequency sp ectrum was then transformed to the time domain using the inverse Fourie r transform (IFT), which was perf ormed using an adaptation of the Cooly-Tukey fast Fourier transform (FFT) algorithm.109 The algorithm generates output, which must be midpoint reflected about the N/2 axis, where N equals the total number of points in the SWIFT waveform. This step is similar to a time shift and therefore affect s the phase, but not the magnitude of the corresponding frequency-domai n spectrum. The advantage of midpoint reflection is that it avoids sudde n voltage transients at the be ginning and end of the excitation period. The frequency spectrum was then quadratically modulated in order to reduce the dynamic range of the time-domain waveform. The real, Ri, and imaginary, Ii, components are created from the magnitude data, Magi, using the following relationship: i i i i i iMag I Mag R sin cos (4-1) Where the phase, i, varies quadratically with frequency:

PAGE 106

106 2 0) 2 / ( i K Jii (4-2) Here 0 is the initial phase (zero), i is the frequency index, and J = 0.5 and K = / N are the quadratic terms, where N is the number of nonzero data points in the frequency spectrum. Values of J and K are chosen to satisfy the Nyquist criteria, such that the rate of phase change per frequency-domain data point is kept at half the Nyquist limit ( ) or below, which removes nonuniformity of the magnitude in the frequency-domain.110 SWIFT Application to LTQ The resulting digitized waveform was downl oaded to the memory of an arbitrary waveform generator (AWG) (Stanford Resear ch Systems Model DS345, Sunnyvale, CA, USA). The AWG has a maximum sampling rate of 40 MHz, time resolution of 16,300 data points, and a 12-bit DAC output. The LTQ has a programmable tr igger that can be used to send a TTL pulse to the AWG at a specific time during the experi mental sequence (e.g., isol ation or activation). For experiments in which the AWG was triggere d during isolation, the LTQ isolation waveform was turned off to avoid interference. Once tr iggered, the AWG applies the SWIFT waveform to the LTQ analog board, where it is summed with th e other waveforms before being amplified and applied to the linear ion trap x-rods. The amp litude of the SWIFT waveform and the number of bursts were modified manually on the AWG. A two-channel digi tizing oscilloscope (Tektronix Model TDS 540, Tektronix Inc., Beaverton, OR) was used to observe the SWIFT waveform. Results and Discussion Hexa-Notch SWIFT Isolation One of the greatest strengths of SWIFT isolati on is the ability to isolate multiple mass-tocharge ( m/z ) ranges simultaneously, which allows for th e selective ejection of ions that may interfere with analysis. Multinotch SWIFT isolation becomes a huge advantage when applied to

PAGE 107

107 MALDI mass spectrometric imagi ng (MSI) when performing MSn. Typically for an MSn experiment, only one parent ion is isolated and then activated w ith collision-induced dissociation (CID) to produce product ions. MALDI MSI of a 2.0 cm by 1.0 cm tissue sample with 100-m laser steps (a total of 20,000 spectra ) would take 5-6 hours to image, and this would have to be repeated for every analyte analyzed by MSn. In addition, the analyt e ion would normally be analyzed separately from its internal standard ion, which has been shown to have increased signal variability when compared to analyzing both the analyte and in ternal standard ions simultaneously with a wide isolation window (C hapter 2) or by dual-notch SWIFT isolation (Chapter 3). Multi-notch SWIF T isolation when applied to MA LDI MSI can save considerable analysis time as well as conserve laser shots, since fewer analyses will need to be performed. A hexa-notch SWIFT isolation waveform was calculated for the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2). The center m/z of these ions ( m/z 305.8) was placed at the q of isolation at q = 0.830. The frequencies of the 1.5-Da notches centered about each of the six ions to be isolated were calculated using a LabView program descri bed in Appendix A. The frequencies of the notches calculated are listed in Table 4-1. Figure 4-2 shows th e calculated hexa-notch SWIFT isolation waveform in the freque ncy domain. Although the width of all six notches is maintained at 1.5 Da, the width of the notches in the frequency domain become s wider at higher frequencies. This is due to the nonlinearity of frequencies in relation to q at values greater than q = 0.4 (Figure 4-3). Isolation with the LTQ typically occurs at q = 0.83, because at this high q value, frequencies are dispersed enough in q -space to allow for sele ctive isolation. Below q = 0.4, frequencies are too tightly spaced to allo w for selective isolation (e.g., 1 kHz = 0.002 q ; given an RF drive frequency = 1188 kHz).

PAGE 108

108 SWIFT Isolation on Tissue The hexa-notch SWIFT isolation waveform was used to analyze BE, BE-d3, COC, COCd3, CE, and CE-d3 standards that were spiked (1 L each) onto blank human brain tissue at 1 g/mL each, and then airbrushed with DHB matrix. Figure 4-4 shows the peak ion intensities of the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2) that were isolated at various hexa-notch SWIFT amplitudes (Vp-p). There was approximately a 50% decr ease in signal across the monitored ions when the SWIFT amplitude was increased from 0.0 Vp-p to 0.1 Vp-p. There was a steady decrease in peak ion intensities with increasing SWIFT amplitude, with m/z 290.2 decreasing most rapidly followed by m/z 293.2. Figure 4-5a shows the mass spectrum of the spiked brain tissue with a SWIFT amplitude of 0.0 Vp-p. The mass spectrum is dominate d by endogenous lipid ions from the brain tissue and it is difficult to distinguish the presence of any analyte ions. The most intense ion in the mass spectrum at m/z 313 was determined to be a phthalate contaminant from a plastic bottle that was used to store the DHB matrix solution. MSn analysis revealed product ions at m/z 177, 149, and 121, which confirmed the phthala te contaminant. Figure 4-5b shows the mass spectrum with a SWIFT amplitude of 0.4 Vp-p. Notice that by increasing the SWIFT amplitude, the lower m/z background ions (< m/z 500) are ejected more efficiently than the higher m/z ions ( m/z 500-2000). The higher m/z background ions (i.e., endogenous lipid region m/z 700 – 900) are not ejected until the SWIFT amplitude is at 0.8 Vp-p (Figure 4-5c). Figure 4-6 shows a more detailed view of th e mass spectra that ranges from m/z 280 to m/z 330. Before SWIFT is applied (0.0 Vp-p), m/z 313 dominates the spectrum (Figur e 4-6a) and the analyte ions are buried in the background. At a SWIFT amplitude of 0.4 Vp-p (Figure 4-6b), the m/z 313 peak has been reduced along with other low m/z background ions, revealing the presence of the

PAGE 109

109 analyte ions at m/z 290, 293, 304, 307, 318, and 321. However, notice that the intensity of m/z 290, which should be relatively at the same intensity as its trideuterated analog at m/z 293, has decreased. At a SWIFT amplitude of 0.8 Vp-p, which is sufficient enough to eject all background ions including high m/z ions, the high SWIFT amplitude has also ejected ions at m/z 290, 293, and 304 and reduced intensity of the analyte ions at m/z 307, 318, and 321 (Figure 4-6c). Ion Ejection To understand how ions can be ejected from the i on trap that were intended to be isolated by a multi-notch SWIFT waveform, it is helpful to look at a diagram of the pseudopotential well depth of the linear ion trap (Figure 4-7). Each ion confined within the ion trap is associated with a q value, which lies on the qx-axis on the Mathieu stability di agram (Chapter 1). Ions of relatively high m/z have q values near the left side of the stability diagram ( x = 0, qx = 0) while ions of lower m/z have q values which extend towards the x = 1 stability boundary, as shown using colored circles of va rious sizes in Figure 4-7. At the intersection of the x = 1 stability boundary and the qx-axis, where qx = 0.908, the trajectories of tr apped ions become unstable along the X-axis such that ions of m/z less than the low mass cutoff (LMCO) are not stored. This method of ion ejection, which can o ccur only at a boundary of the stab ility diagram, is referred to as mass-selective instability.96 One other method for ions to be ejected from the ion trap is called resonant ejection,108 which is the method of ejection used by SWIFT. The advantage of resona nt ejection is that it can be carried out at any frequency. Ions are resonantly ejected from the ion trap when a frequency is applied that is in re sonance with the secular frequency ( ) of the ion and has sufficient amplitude (depth of pseudopotential well) to increase the oscillation of the excited ions until they exit through the slits in the center X-rods of the linear ion trap.

PAGE 110

110 In Figure 4-6b, it makes sense that the intensity of m/z 290 ( q = 0.875) would decrease with increasing SWIFT amplitude, before ions at m/ z 293 ( q = 0.866), 304 ( q = 0.835), 307 ( q = 0.827), 318 ( q = 0.798), and 321 ( q = 0.791), because m/z 290 lies the closest to the LMCO ( m/z 280, = 500 kHz) at the mass-sele ctive instability boundary ( q = 0.908). It also has a shallower pseudopotential well depth (Figure 47) than the other ions, meani ng that it takes a lower SWIFT amplitude to eject it from the ion trap. One way to correct for the instability of m/z 290 at higher SWIFT amplitudes would be to decrease its q value and move it away from the mass-selective instability boundary ( q = 0.908). This can be accomplished by changing the m/z at the q of isolation ( q = 0.830) from the center m/z of the analytes ( m/z 305.8) to the lowest m/z analyte ( m/z 290). The new q values and notch frequencies for th e calculated hexa-notch SWIFT based on m/z 290.2 at q = 0.830 are listed in Table 4-2. One issue with not placing the center m/ z ion at the q of isolation is that the LTQ software couples the isolation window width with the acti vation window width. This means that in order to perform resonance excita tion (CID) on the ions at m/z 290.2, 293.2, 304.2, 307.2, 318.2, and 321.2, with m/ z 290.2 placed at the q of isolation ( q = 0.830), at minimum, a 62-Da wide isolation window ( m/z 259.2 – m/z 321.2) centered at m/z 290.2 is required to include these ions in activation. A 70-Da wide isolation window ( m/z 255.2 – m/z 325.2) centered at m/z 290.2 would be wide enough to ensure co mplete activation of the ion at m/z 321.2. However, in the LTQ software, isolation windows and consequentia lly, activation windows th at are wider than 47 Da, are assigned a q of isolation that is lower than q = 0.830 (Figure 4-8). In the LTQ software, the q of isolation is decreased linearly with increa sing isolation window width to ensure that ions to be isolated are above the LMCO. The ma ximum isolation window width allowed by the LTQ software is 100 Da. The q of isolation for a 70-Da wide isolation window centered at m/z 290.2

PAGE 111

111 is q = 0.791. The q values and notch frequencies for th e calculated hexa-notch SWIFT based on m/z 290.2 at q = 0.791 are listed in Table 4-3. Figure 4-9 compares the hexa -notch SWIFT isolation of m/z 290, 293, 304, 307, 318, and 321 at a SWIFT amplitude of 0.6 Vp-p with the m/z at q of isolation set to m/z 305.8 ( q = 0.830) (Figure 4-9a), m/z 290.2 ( q = 0.830) (Figure 4-9b), and m/z 290.2 ( q = 0.791) (Figure 4-9c). Figure 4-9a shows a large decrease in signal for m/z 290 ( q = 0.875) and m/z 293 ( q = 0.866) compared to Figure 4-9b when m/z 290 ( q = 0.830) and m/z 293 ( q = 0.822) are placed at lower q values further away from the ma ss-selective instability boundary ( q = 0.908). At the lower q values, m/z 290 and m/z 293 are deeper in the pseudopotential well of the ion trap and therefore can tolerate a higher SWIFT amplit ude before they are resonantly ejected from the trap. The analyte ions at m/z 290 ( q = 0.791), 293 ( q = 0.783), 304 ( q = 0.755), 307 ( q = 0.747), 318 ( q = 0.721), and 321 ( q = 0.715) are even deeper in the pseudopotential we ll and show an overall increase in signal at lower q values (Figure 4-9c). Figure 4-10 shows the peak i on intensities of the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2) when the hexa-notch SWIFT described in Table 4-3 based on m/z 290.2 at q = 0.791 was applied at various SWIFT amplitudes. The results shown in Figure 4-10 are very different than the results shown in Figure 4-4 from the hexa-notch SWIFT described in Table 4-1 based on m/z 305.8 at q = 0.830. Figure 4-4 showed approximately a 50% decrease in signal across the monitored ions when the SWIFT amplitude was increased from 0.0 Vp-p to 0.1 Vp-p followed by a steady decrease in peak ion intensities with increa sing SWIFT amplitude. In contrast, when the amplitude of the hexa-notch SWIFT (Table 4-3) is increased, the analyte ion intensities steadily increase as well until the amplitude reaches 0.6 Vp-p. When the SWIFT amplitude is increased

PAGE 112

112 from 0.6 Vp-p to 1.0 Vp-p, all of the analyte ion intensities decr ease, but not in the order of low to high m/z (e.g., m/z 304 decreases faster than m/z 293 and m/z 318 decreases faster than m/z 321), as would be expected based on the pseudopotential well depth (Figure 4-7). This behavior may be explained due to the fact that the energy ab sorption profile for ions subjected to resonance excitation broadens with the amplitude of ion oscillation and shifts to higher frequencies.116 If the ions to be isolated shift to higher frequencies, they will be outside the frequency notches of the SWIFT isolation waveform and will come into resonance with the excitation frequencies thus ejecting them from the trap. Therefore, it is beneficial to use the lowest SWIFT amplitude possible to isolate the desired ions to avoid causing frequency shifts and sequential resonant ejection (Figure 4-11). However, at lower SWIFT amplitudes (Figure 4-12), higher m/z background ions will not be ejec ted and will remain trapped Two-Stage Isolation One strategy for ejecting high m/ z background ions while avoiding frequency shifts caused from higher SWIFT amplitudes, is to perform tw o-stage isolation. The first stage would be a coarse isolation that would coarse ly isolate the ions of interest while ejecting background ions. The frequency notches during this stage would be wi de enough so that if a ny frequency shifts did occur, the ions of interest woul d not be ejected. This coarse isolation stage would be followed by a fine isolation stage, which would utilize a higher degree of mass di scrimination to eject background ions close to th e ions of interest. This two-stage coarse/fine SWIFT isolat ion was first proposed by Soni and Cooks107 for isolating ions having a single m/z value from a population of trapped ions. The coarse/fine isolation technique used a doubly notched SWIFT pulse to perform the isolation in two steps at two different q values. The coarse isolation step used a single notch centered at q = 0.0778 to coarsely isolate the ion of interest. Then the RF amplitude was increased to move the trapped

PAGE 113

113 ions to a higher q value in which a second narrower notch centered at q = 0.4035 was used to finely isolate ions of a single m/z value only. The resonance freque ncies of ions are more spread out at higher q values (Figure 4-3), which allows for higher mass discrimination. The advantage of this two-stage isolation is that the coarse st ep removes most of the i ons that contribute to space charging, and thereafter the frequencies of th e analyte ions remain relatively constant. However, although frequency shifts were minimized with the two-stage st rategy, Soni and Cooks still reported a 20% loss of targ et ion population as a result of the sharp mass discrimination of the second fine isolation notch.107 High Mass Filter (HMF) In order to minimize frequency shifts, a SW IFT excitation waveform was calculated that ejects background ions higher in m/z than the highest m/z analyte ( m/z 321). This SWIFT excitation waveform was termed high mass filter (HMF ), because it serves to filter out or eject ions above a certain m/z Figure 4-13 shows the frequency do main of a HMF that was calculated to excite at frequencies 0 to 338 kHz. The right-hand edge of the HMF at 338 kHz corresponds to a m/z cutoff at m/z 325.8 based on m/z 290.2 placed at a q of isolation ( q = 0.791). This HMF is designed to eject/excite ions from m/z 325.8 to greater than m/z 2000 (LTQ upper m/z limit; 48 kHz). The HMF SWIFT excitation waveform was applied to the analysis of BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 L each) onto blank human brain tissue at 1 g/mL each, and then airbrushed with DHB matri x. Figure 4-14 shows the peak ion intensities of the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2) while the HMF was applied at various SWIFT amplitudes (Vp-p). When the HMF amplitude is increased from 0.0 Vp-p to 0.2 Vp-p there is a 19 to 43%

PAGE 114

114 increase in signal for the analyte ions as the high m/z background signal begins to decrease. The signal for the analyte ions then shows a decrease when the HMF amplitude is increased from 0.2 Vp-p to 0.4 Vp-p. At a HMF amplitude of 0.5 Vp-p, all of the high m/z background ions are removed, resulting in a 14 to 22% incr ease in analyte signal from 0.4 Vp-p to 0.5 Vp-p. The analyte signals then begin to steadily decrease as the HMF amplitude is increased from 0.5 Vp-p to 1.0 Vp-p. At a HMF amplitude of 1.0 Vp-p the analyte signals have decreased 36 to 47% from the amplitude of 0.5 Vp-p with the analyte ions at m/z 290.2 and 293.2 decreasing the most (47%). Figure 4-15 shows the mass spectra of the spik ed brain tissue with the HMF amplitude at 0.0 Vp-p (Figure 4-15a), 0.4 Vp-p (Figure 4-15b), and 0.5 Vp-p (Figure 4-15c). Notice the dramatic change in the high m/z background when the HMF amplitude is increased from 0.4 Vp-p (Figure 4-15b) to 0.5 Vp-p (Figure 4-15c). Figure 4-16 shows a mo re detailed view of the mass spectra from m/z 280 to m/z 330. Although the HMF SWIFT excita tion waveform should only affect ions from m/z 325.8 to m/z 2000, the signal intensities of the an alyte ions steadily increase when the HMF amplitude is changed from 0.0 Vp-p (Figure 4-16a) to 0.4 Vp-p (Figure 4-16b) and then to 0.5 Vp-p (Figure 4-16c). This increase in analyte si gnal may be attributed to the reduction in high m/z background ions, which dominate the mass spectrum. By ejecting the high m/z background ions, space-charge effect s are reduced that would normally cause frequency shifts of the analyte ions outside the notches of the SWIF T isolation waveform resulting in some ejection of analyte ions. Combining HMF with Hexa-Notch SWIFT Two-stage isolation was performed by combin ing the HMF SWIFT excitation waveform to eject ions above m/z 325.8 with a hexa-notch SWIFT isolati on waveform to selectively isolate ions at m/z 290.2, 293.2, 304.2, 307.2, 318.2, and 321.2 (Figure 4-17) The frequency domain of the two-stage isolation is show n in Figure 4-17 with the HMF waveform shown in red and the

PAGE 115

115 hexa-notch waveform shown in blue. The time do main of the two-stage isolation is shown in Figure 4-18. The HMF SWIFT excitation waveform (red) occurs from 0 to 4,096 s and the hexa-notch SWIFT isolation waveform (blue) occurs from 4,097 to 8,192 s. A single burst of the two-stage isolation pulse (8,192 s) is triggered during th e LTQ isolation event (15,500 s) with the LTQ isolation waveform turned off. The two-stage isolation was used to analyze BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards that we re spiked (1 L each) onto blank human brain tissue at 1 g/mL each, and then airbrushed with DHB matrix. Figure 4-19 sh ows the peak ion intensities of the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2) that were isolated at vari ous two-stage isolation amplitudes (Vp-p). Both the HMF SWIFT excitation waveform and the hexanotc h SWIFT isolation waveform were set to the same amplitude using the function generator. Th e analyte ion intensities decreased steadily as the amplitude was increased from 0.1 Vp-p to 0.4 Vp-p. The analyte ion inte nsities then remained relatively constant between 0.4 Vp-p and 0.6 Vp-p. When the amplitude was increased from 0.6 Vp-p to 1.0 Vp-p, the analyte ion intensities quickly decreased with the m/z 290.2 ion decreasing the fastest. The instability of m/z 290.2 at SWIFT amplitudes greater than 0.6 Vp-p is probably due to the shallow position of th is ion in the pseudopotential we ll. One strategy for correcting this would be to move the analyte ions to lower q values further away from the LMCO during isolation. Figure 4-20 shows the mass spectra of the spik ed blank brain tissue with the two-stage isolation amplitude at 0.0 Vp-p (Figure 4-20a), 0.5 Vp-p (Figure 4-20b), and 0.6 Vp-p (Figure 420c). When the HMF SWIFT excitation waveform was used alone without the hexa-notch SWIFT isolation waveform, the high m/z background ions were ejected efficiently at a SWIFT

PAGE 116

116 amplitude of 0.5 Vp-p (Figure 4-15c); however, when the HMF SWIFT excitation waveform is combined with the hexa-notch SWIFT isolation waveform in the two-stage isolation, the high m/z background ions are not ejected at 0.5 Vp-p (Figure 4-20b), but requires an amplitude of 0.6 Vp-p (Figure 4-20c) for the high m/ z ion background to be removed. Increasing the amplitude from 0.5 Vp-p to 0.6 Vp-p actually results in a slight increase in the analyte ion intensities (3% to 18%), which could be due to the removal of the high m/z background ions that cause frequency shifts of the analyte ions. Figure 4-21 shows a more detailed view of the mass spectra from m/z 280 to m/z 330. Notice how the background ion signals have been reduced around the analyte ions. The overall difference in the an alyte ion intensities with the application of the two-stage isolation at 0.6 Vp-p compared to no two-stage isolation (0.0 Vp-p) is 54% decrease for m/z 290.2, 37% decrease for m/z 293.2, 19% decrease for m/z 304.2, 12% decrease for m/z 307.2, 9% decrease for m/z 318.2, and 22% decrease for m/z 321.2. The high decreases in signal for m/z 290.2 (54%) and m/z 293.2 (37%) are probably due to their pr oximity to the LMCO, which could be remedied by lowering the q values of the ions during isolat ion. The high decrease in signal for m/z 321.2 (22%) could be due to its proximity to the m/z cutoff ( m/ z 325.8) of the HMF applied. This can be fixed by applying a differe nt HMF that allows for more space between the m/z 321.2 ion and the m/z cutoff, but still allows removal of the majority of the high m/z background ions. MS/MS with Two-Stage Isolation The overall goal of the two-stage isolation was to provide an isolati on strategy that would allow for the isolation of the analyte and internal standard ions during a single MS/MS scan. The MS/MS provides higher mass selectivity which is essential for distinguishing the analyte from MALDI matrix and endogenous spec ies (e.g., lipids) present in the brain tissue, and isolating the

PAGE 117

117 analyte and internal standard ions in the same MS/MS scan has shown to improve the precision of MALDI-MS/MS (Chapters 2 and 3).103 Figure 4-22a shows the mass spectrum of spiked brain tissue with the two-stage SWIFT isolation waveform applied at 0.6 Vp-p. A 70-Da wide isolation window was centered at the lowest m/z analyte ion at m/z 290.2, which placed the q of isolation at q = 0.791. Since the LTQ software couples the isolation width with the activation width, CID will be applied across the 70-Da wide activation window centered at m/z 290.2. This means that ions in the mass range m/z 255.2 to 325.2 will all be activated by CID and fragmented. It is also important to note that the collision energy applied during MS/MS is spread over the entire activation width. Thus, increasi ng the isolation width decreases the collision energy for each ion. It was determined that th e CID value necessary to dissociate the analyte ions during MS/MS and reduce the parent ions to a relative inte nsity of 10%, needed to be increased with the wider activation window. It wa s determined that a CID of 55 was optimal for the analytes with a 5-Da wide activation window, but the CID was increased to 90 for the 70-Da wide activation window. Figure 4-22b shows the MS/MS pr oduct ion spectrum of ions isolated by a two-stage SWIFT isolation. Since the most intense fragment ion of the [M+H]+ ion of BE ( m/z 290.2), BEd3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2) all result from the neutral loss (NL) of benzoic acid (122 Da), it is easy to di stinguish the origins of each product ion. BE produces the product ion at m/z 168.2, BE-d3 produces the product ion at m/z 171.2, COC produces the product ion at m/z 182.2, COC-d3 produces the product ion at m/z 185.2, CE produces the product ion at m/z 196.2, and CE-d3 produces the product ion at m/z 199.2. A common product ion of BE, COC, and CE is m/z 150.2 and a common product ion of BE-d3, COC-d3, and CE-d3 is m/z 153.2, both of which are due to the NL of benzoic acid

PAGE 118

118 (C6H5COOH) and methanol (CH3OH). Common product ions were not used for quantification due to the difficulty in determining the signal attributable to each parent ion. Figure 4-23a shows the MS/MS pr oduct ion spectrum of ions isolated with a 40-Da wide isolation window with CID = 55. Notice that this product ion spectrum contains the same product ions from the MS/MS of ions isolated with a two-stage SWIFT isolation (Figure 4-23b) with some additional ions (e.g., m/z 137 and m/z 147) produced from the MS/MS of background ions. These background ions were isolated along with the analyte ions in the 40-Da wide isolation (Figure 4-24b). The intense product ion at m/z 137 is [DHB+H-H2O]+ caused by the NL of DHB from the DHB cluster matrix ion at m/z 291, [2DHB+H-H2O]+. Background ions can complicate the product ion spectrum and might even interfere with the analyte product ion signals if the background ions produce fragment i ons that are isobaric; therefore, removal of background ions during isolation before MS/MS is beneficial. Comparing Wide Isolation and Two-Stage SWIFT Isolation It was previously shown that is olating the analyte and internal standard ions in a single MS/MS scan using a wide isolation window can improve precision for MALDI quantification compared to isolating the analyte and internal standard ions separately during alternate MS/MS scans (Chapters 2 and 3).103 In order to compare the ability of the two-stage SWIFT isolation waveform to also improve precision for MALDI qua ntification, five solutions (mixture of BE, COC, and CE at 62, 125, 250, 500, and 1000 ng/mL with a mixture of BE-d3, COC-d3, and CEd3 at 200 ng/mL) were pipetted in triplicate (1 L each) onto blank human brain tissue and airbrushed with DHB matrix. The tissue wa s then analyzed using two different MS/MS experiments: two-stage SWIFT isolation (0.6 Vp-p; CID = 90) and 40-Da wide isolation (CID = 55). Figure 4-25 shows the calibration curves fo r BE for both experiments (two-stage SWIFT

PAGE 119

119 isolation and wide isolation) produced by plotting the peak intensity ratio of m/z 168 and m/z 171 versus the mass of BE standard spiked on tissue. The two-stage SWIFT isolation has comparable precision (% RSD = 0 to 11%) for isot opic ratios compared to wide isolation (% RSD = 2 to 11%). Figure 4-26 shows the calibra tion curves for COC for both experiments (twostage SWIFT isolation and wide isolation) pr oduced by plotting the peak intensity ratio of m/z 182 and m/z 185 versus the mass of COC standard sp iked on tissue. The two-stage SWIFT isolation has similar precision (% RSD = 1 to 6%) for isotopic ratios compared to wide isolation (% RSD = 3 to 5%). The precision of the two-st age SWIFT isolation for the MS/MS analysis of COC is better than what was reported for dualnotch SWIFT isolation of COC in Chapter 3 (% RSD = 5 to 23%). This improvement may be du e to the addition of the HMF during the first stage of the two-stage isol ation, which removes the space-charge effects of the high m/z background ions. Finally, Figure 4-27 shows the cal ibration curves for CE for both experiments (two-stage SWIFT isolation and wide isolation) produced by plotting the peak intensity ratio of m/z 196 and m/z 199 versus the mass of CE standard sp iked on tissue. The two-stage SWIFT isolation % RSD ranged from 3 to 12% compared to the % RSD for wide isolation that ranged from 1 to 29%. Although two-stage SWIFT isolation overall ha s comparable precision to that of wide isolation, the higher mass selectiv ity of the two-stage SWIFT isol ation affords a significant loss in absolute signal intensity of the analyt e ions when it is applied. BE and BE-d3 showed an average percent loss in absolute signal intens ity of 83% and 75%, respectively. COC and COCd3 showed an average percent loss in absolute si gnal intensity of 69% and 75%, respectively. CE and CE-d3 showed an average percent loss in abso lute signal intensity of 65% and 77%, respectively.

PAGE 120

120 Two-Stage SWIFT MALDI-MS/MS Quantification The MS/MS two-stage SWIFT isolation method and the MS/MS 40-Da wide isolation method were compared for the quantification of unspiked BE, COC, and CE from human brain tissue from a subject whose toxi cology report showed the presence of COC. Three different concentrations of BE-d3, COC-d3, and CE-d3 (31, 62, and 125 ng/mL) were spiked (1 L) onto a glass slide before thaw mounting a 20 m-thick brain tissue slice on top and airbrushing DHB matrix. All three spots were then analyzed using the MS/MS two-stage SWIFT isolation method and then the MS/MS 40-Da wide isolation met hod. Approximately 2000 scans were acquired to image the entire area of each of the spots (average area = 0.17 cm2). The m/z 171 signal from using the MS/MS two-stage SWIFT isolation for BE-d3 from each spot was used to develop a calibration curve that resulted in a line of best fit of y = 0.97( 0.08)x + 6( 7). BE-d3 was shown to have a linear response with increasing concen trations spiked underneath tissue. Since the MS/MS two-stage SWIFT isolation method analyzes both BE and BE-d3 simultaneously, unspiked BE was detected from each spot analyzed at m/z 168. An area of the tissue (500 MS/MS scans) that was not spiked with BE-d3 was analyzed using the MS/MS two-stage SWIFT isolation method and the acquired m/z 168 signal was averaged with the m/z 168 signals from the spiked BE-d3 spots, resulting in a very trace signal of 53 6 counts. Assuming that the amount of unspiked BE extracted from the ti ssue has a 1:1 response with the BE-d3 spiked on top of tissue, the calibration curve for BE-d3 can be used to quantify the amount of BE present in the analyzed tissue. From the equation of the line, it was determined that BE was present at a level equivalent to 50 ng/mL. Using the 1 L volume of BE-d3 spiked underneath tissue, it is calculated that the mass of BE present is 50 pg. Given that the area of an analyzed spot on tissue was 0.17 cm2 and that the

PAGE 121

121 tissue thickness was 20 m (2.0 x 10-3 cm), the volume of tissue from which BE was extracted was 3.4 x 10-4 cm3. The mass of the tissue is 3.4 x 10-4 g (density of wet tissue ~1.0 g/cm3), resulting in an absolute concentr ation of BE detected in this ar ea of the postmortem brain tissue of 140 ng/g (140 ppb). Since the MS/MS two-st age SWIFT isolation method and the MS/MS 40-Da wide isolation method both acquire the an alyte and internal standard ions for BE ( m/z 168 and 171), COC ( m/z 182 and 185), and CE ( m/z 196 and 199) from each spot analyzed simultaneously, the amount of unspiked COC and unspiked CE we re also quantified from the tissue using the same process described above. The results for the quantification of BE, COC, and CE using both the MS/MS two-stage SWIFT isolation method and th e MS/MS 40-Da wide isolation method are summa rized in Table 4-4. SPE-MALDI-MS/MS Quantification Drugs and their metabolites in tissue are ty pically quantified from tissue homogenate instead of from intact tissue. For this reas on, tissue homogenates were prepared and extracted by solid-phase extraction (SPE) as described in th e previous experimental section, and analyzed using MALDI-MS/MS to quantify the presence of BE, COC, and CE in unspiked tissue. One gram of blank human brain tissue (case num ber 07A-355) was cut, weighed (0.9830 g), and homogenized. Then 3 mL of 60 M NaF was added and the homoge nate was sonicated. Five 1mL standard solutions (62, 125, 250, 500, and 1000 ng/mL each of BE, COC, and CE and 200 ng/mL each of BE-d3, COC-d3, and CE-d3) were dried with nitrogen gas and then reconstituted with a 400 L aliquot of the sonicated homogenate. The solutions were immediately vortexmixed and centrifuged. Then a 100 L aliquot of the supernatant from the centrifuged homogenate was loaded onto a preconditioned unde rivatized silica SPE cartridge. Analytes in the cartridge were then eluted using 3 mL of 5% ammonia in methanol solution. The eluents

PAGE 122

122 from the extraction cartridge were then drie d using nitrogen gas, and the residue was reconstituted in 500 L of water/methanol (90:10, vol/vol ), spotted onto a MALDI plate (1 L) in triplicate, and airbrush ed with DHB matrix. Each spot on the MALDI plate was analyzed using a MS/MS 5-Da wide isolation method specific for each set of analyte a nd internal standard ions. The 5-Da wide isolation method was used, because it was shown to have better pr ecision than using alternating MS/MS scans.103 The [M+H]+ ions of BE ( m/ z 290.2) and BE-d3 ( m/z 293.2) were isolated with a 5-Da wide isolation window centered at m/z 291.8 with CID = 55 to pr oduce the product ions at m/z 168.2 and m/z 171.2 for BE and BE-d3, respectively. The [M+H]+ ions of COC ( m/z 304.2) and COC-d3 ( m/z 307.2) were isolated with a 5-Da wide isolation window centere d at m/z 305.8 with CID = 55 to produce the product ions at m/z 182.2 and m/z 185.2 for COC and COC-d3, respectively. The [M+H]+ ions of CE ( m/ z 318.2) and CE-d3 ( m/z 321.2) were isolated with a 5-Da wide isolation window centered at m/z 319.8 with CID = 55 to pr oduce the product ions at m/z 196.2 and m/z 199.2 for CE and CE-d3, respectively. Figure 4-28 shows the calibration curve for BE with the peak intensity ratio of m/z 168.2 and m/z 171.2 versus the mass of BE spotted on the MALDI plate from the tissue homogenate. The 5-Da wi de isolation method was fairly precise with % RSD ranging from 3 to 8%. The BE calibrati on curve has a line of best fit of y = 0.00543( 0.00005)x – 0.0007( 0.02). Figure 4-29 shows the calib ration curve for COC with the peak intensity ratio of m/z 182.2 and m/z 185.2 versus the mass of COC spotted on the MALDI plate from the tissue homogenate. % RSD range d from 4 to 12%. The COC calibration curve showed a linear response with a line of best fit of y = 0.00633( 0.00005)x + 0.02( 0.02). The calibration curve for CE is shown in Figur e 4-30 with the peak intensity ratio of m/z 196.2 and m/z 199.2 versus the mass of CE spotted on the MALDI plate from the tissue homogenate. %

PAGE 123

123 RSD ranged from 3 to 6%. The CE calibration cu rve showed a linear response with a line of y = 0.00581( 0.00003)x – 0.02( 0.02). The equations of the calibrati on curves developed were used to quantify the amount of unspiked BE, COC, and CE present in unspike d tissue homogenate from human brain tissue (case number 07A-369), for which toxicological an alysis indicated the presence of cocaine in blood (69 ng/mL). One gram of this tissue was cut, weighed (0.9862 g), and homogenized. Then 3 mL of 60 M NaF was added and the homogenate was sonicated. A 1-mL solution of the internal standards (BE-d3, COC-d3, and CE-d3; 200 ng/mL each) was prepared, dried with nitrogen gas, and then reconstituted with a 400 L aliquot of the sonicat ed homogenate from case number 07A-369. The solution was immediat ely vortex-mixed and centrifuged. Then a 100 L aliquot of the supernatant from the centrifuged homogenate was loaded onto a preconditioned underivatized silica SPE cartridge. An alytes in the cartridge were then eluted using 3 mL of 5% ammonia in me thanol solution. The eluents from the extraction cartridge were then dried using nitrogen gas, and the residue was r econstituted in 500 L of water/methanol (90:10, vol/vol), spotted onto a MALDI plate (1 L) in triplicate, and airbrushed with DHB matrix. Each spot on the MALDI plate was an alyze using the MS/MS 5-Da wide isolation method specific for each set of analyte and internal standard ions described previously. Using the calibration curves it was determined that there was 270 14 ng BE/g of tissue, 380 11 ng COC/g of tissue, and 430 16 ng CE/g of tissue. The concentrations for BE, COC, and CE ar e not comparable to the concentrations determined by analyzing the ti ssue directly by the MS/MS twostage SWIFT isolation method and the MS/MS 40-Da wide isolation method (Table 4-4). One reason for this difference might be in the sample size for the different methods. The tissue homogenate method analyzes a larger

PAGE 124

124 tissue sample (~ 1 g) for quantific ation, which averages the signal for the analytes over the entire tissue. The intact tissue MALDI methods analyze much smaller samples (3.4 x 10-4 g) across different regions of the tissue to develop a calib ration curve for quantificat ion. This analytical strategy assumes that the intern al standard spotted underneath the tissue will have a similar response across the different regions of the tissue. It also assumes that the internal standard will have similar extraction efficienci es through the tissue for all regions analyzed. In addition, since it is difficult to spot different concentrations of the internal standard in triplicate underneath the tissue, precision of the quantitative analysis fo r the intact tissue methods was not determined. Conclusions In Chapter 3, multi-notch SWIFT isolation wa veforms were explored as a strategy for isolating the analyte and intern al standard ions during a singl e MS/MS scan, which has been shown to provide improved precision for MALDI-M S compared to using two alternating MS/MS scans that isolate the analyte and internal standard ions separately. However, it was determined that multi-notch SWIFT isolation was not as prec ise as the wide isolation method. This might have been due to frequency shifts of the analyt e and internal standard ions from space-charge effects caused by high m/z background ions. A two-stage SWIFT isolation method was de veloped that utilizes a high mass filter (HMF) SWIFT excitation wa veform to remove high m/z background ions during the first stage of isolation. This has been show n to increase isolated ion sign als and improve the precision of SWIFT when compared to the application of SW IFT without the HMF. This may suggest that the HMF reduces the irreproducible frequency shifts of the analyte and internal standard ions by preventing them from moving outside the notches of the multi-notch SWIFT isolation waveform and being ejected. A hexa-notch SWIFT isolat ion waveform was used during the second stage of the two-stage SWIFT isolation to mass selectively isolate the [M+H]+ ions of BE ( m/z 168.2),

PAGE 125

125 BE-d3 ( m/z 171.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2). The hexa-notch SWIFT isolation waveform was able to effectively remove background ions around the analyte and internal standard ions that may have in terfered with MS/MS analysis. The two-stage SWIFT isolation overall showed similar precision to that of wide isolation when performing MS/MS on the analyte and internal standards spiked on blank brain tissue. However, the higher mass selectivity of the two-stage SWIFT isolation affords a significant loss in absolute signal intensity of the analyte and inte rnal standard ions when it is applied. Two-stage SWIFT isolation was compared to wide isolation of intact tissue and SPE extracted homogenized tissue for the MALDI-MS/MS quantification of BE, COC, and CE from unspiked human brain tissue, whose toxicological analysis indicated the presence of cocaine in blood. The two-stage SWIFT isolation method show ed a lower analyte signal per gram of tissue than the wide isolation method for BE, COC, and CE present in tissue. Th e quantification results for the two-stage SWIFT isolation and the wide is olation of intact tissue was not comparable to the wide isolation analysis of tissue homogenate (Table 4-4). However, the intact tissue methods required considerably less sample preparation and smaller sample sizes than the tissue homogenate method. There was no analysis tim e saved when comparing the two-stage SWIFT isolation and wide isolation since the [M+H]+ ions of BE, BE-d3, COC, COC-d3, CE, and CE-d3 were all isolated simultaneously during the same MS/MS scan. In conclusion, a wide isolation method may still be a better choice over SWIFT isolation for improving MALDI-MS/MS precision for quantification and reducing analys is time, despite the inclusion of unwanted background ions during isolation.

PAGE 126

Figure 4 1. Solid-p h scheme u in MeOH typically i condition add the s a (5) elute t Bulletin 9 h ase extracti o s ed. Adsor b ) that leave s i nvolves 5 s t the SPE tu b a mple (100 t he compou n 9 10: Guide t o n scheme. b ed compou n s the strongl y t eps: (1) se l b e (2 mL of m L of homo g n ds of intere t o Soli d -Ph a 126 Selective el u n ds of inter e y retained i m l ect the pro p m ethanol fo l g enate), (4) w st (3 mL 5 % a se Extracti o u tion is the s e st are elute d m purities be h p er SPE tube l lowed by 2 w ash the pa c % NH3 in M e o n, 1998. s oli d -phase e d in a solven t h ind. The S P (HyperSep mL of deio n c king (not p e OH). Adap t e xtraction ( S t (3 mL 5% P E process SI), (2) n ized water ) erformed), a t ed from Su p S PE) NH3 ) (3) a nd p elco

PAGE 127

127 Table 4-1. Hexa-notch SW IFT properties based on m/z 305.8 at q = 0.830 Analyte [M+H]+ ( m/z ) q value Notch Width (Da) Notch Frequencies (kHz) BE 290.2 0.875 1.5 488.037109 to 494.873047 BE-d3 293.2 0.866 1.5 475.341797 to 481.445312 COC 304.2 0.835 1.5 439.208984 to 443.603516 COC-d3 307.2 0.827 1.5 431.152344 to 435.058594 CE 318.2 0.798 1.5 405.029297 to 408.203125 CE-d3 321.2 0.791 1.5 398.681641 to 401.855469

PAGE 128

128 Figure 4-2. Frequency domain of hexa-notch SWIFT isolation waveform. Frequency notches correspond to the secular frequencies of the [M+H]+ ion of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2) with the center m/z 305.8 placed at the q of isolation at q = 0.83.

PAGE 129

129 Figure 4-3. Relationship be tween secular frequency ( ) and q -space. The Dehmelt approximation (red trace) states that varies linearly with q for values of q less than 0.4. Secular frequency ( ) diverts from linearity (blue trace) at q values higher than 0.4, which can be measured from the LTQ MSn diagnostic settings and verified by calculations explaine d in Appendix A.

PAGE 130

130 Figure 4-4. Variable hexa -notch SWIFT amplitude ( m/ z 305.8 at q = 0.830). Peak ion intensities of the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2) at different amplitudes (Vp-p) of a hexa-notch SWIFT isolation waveform app lied to the analysis of brain tissue.

PAGE 131

131 Figure 4-5. Mass spectra ( m/z 80 to 2000) of hexa-notch SWIFT at different amplitudes. Hexanotch SWIFT isolation wave form applied to BE, BE-d3, COC, COC-d3, CE, and CEd3 standards that were spiked (1 L; 1 g/mL each) onto blank brain tissue and airbrushed with DHB matrix with a SWIFT amplitude of (a) 0.0 Vp-p (b) 0.4 Vp-p, and (c) 0.8 Vp-p. 0.0 Vp pNL: 1.28E5 0.4 Vp pNL: 6.02E4 0.8 Vp pNL: 2.45E3 200 400 600 800 1000 1200 1400 1600 1800 2000 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 313.15 798.59 577.61 403.34 864.73 518.38 291.04 982.63 1164.65 798.58 782.67 851.67 753.67 864.75 694.58 560.42 318.25 878.75 321.25 307.25 769.67 556.33(a) (b) (c)

PAGE 132

132 Figure 4-6. Mass spectra ( m/z 280 to 330) of hexa-notch SWIFT at different amplitudes. Hexanotch SWIFT isolation wave form applied to BE, BE-d3, COC, COC-d3, CE, and CEd3 standards that were spiked (1 L; 1 g/mL each) onto blank brain tissue and airbrushed with DHB matrix with a SWIFT amplitude of (a) 0.0 Vp-p (b) 0.4 Vp-p, and (c) 0.8 Vp-p. 280 285 290 295 300 305 310 315 320 325 330 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 313.15 314.19 291.04 293.27 308.34 318.30 304.27 302.20 283.35 312.28 329.14 322.31 318.25 304.25 307.25 321.33 293.25 290.25 322.25 308.33 313.17 294.17 323.17 329.25 287.25 301.17 321.25 307.25 322.25 318.33 308.33 0.0 Vp pNL: 1.28E5 0.4 Vp pNL: 7.10E3 0.8 Vp pNL: 2.45E3(a) (b) (c)

PAGE 133

133 Figure 4-7. Pseudopotential well depth ( Dx) of the ion trap. The deepest part of the pseudopotential well is near the q of isolation at qx = 0.83. The sizes of the circles are proportional to the m/z of the ions. Adapted from March, R.E. J. Mass Spectrom. 1997 32 351. 0.91Potential Well Depth, Dx(eV)0.83 10 0 q x 0.30 8RF x xV q D

PAGE 134

134 Table 4-2. Hexa-notch SW IFT properties based on m/z 290.2 at q = 0.830 Analyte [M+H]+ ( m/z ) q value Notch Width (Da) Notch Frequencies (kHz) BE 290.2 0.830 1.5 435.058594 to 439.453125 BE-d3 293.2 0.822 1.5 426.757812 to 430.664062 COC 304.2 0.792 1.5 400.146484 to 403.564453 COC-d3 307.2 0.784 1.5 393.798828 to 396.972656 CE 318.2 0.757 1.5 372.802734 to 375.488281 CE-d3 321.2 0.750 1.5 367.675781 to 370.117187

PAGE 135

135 Figure 4-8. Isolation window width (Da) determined by the preset q of isolation. The q of isolation diverges from q = 0.83 at isolation widths greater than 47 Da to ensure that ions to be isolated are above the low ma ss cutoff (LMCO). The maximum isolation window width allowed by the LTQ software is 100 Da. 0.75 0.76 0.77 0.78 0.79 0.80 0.81 0.82 0.83 0.84 0102030405060708090100110q of IsolationIsolation Window Width (Da) max isolation window width allowed: 100 Da q diverges from 0.83 at isolation widths > 47 Da

PAGE 136

136 Table 4-3. Hexa-notch SW IFT properties based on m/z 290.2 at q = 0.791 Analyte [M+H]+ ( m/z ) q value Notch Width (Da) Notch Frequencies (kHz) BE 290.2 0.791 1.5 399.902344 to 403.320312 BE-d3 293.2 0.783 1.5 393.066406 to 396.484375 COC 304.2 0.755 1.5 371.337891 to 374.023437 COC-d3 307.2 0.747 1.5 365.966797 to 368.652344 CE 318.2 0.721 1.5 347.656250 to 350.097656 CE-d3 321.2 0.715 1.5 343.261719 to 345.458984

PAGE 137

137 Figure 4-9. Hexa-notch SW IFT applied at variable q of isolation. Hexa-notch SWIFT isolation waveform applied to BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 L; 1 g/mL each) onto blank brain tissue and airbrushed with DHB matrix with a SWIFT amplitude of 0.6 Vp-p and the m/z at q of isolation set to (a) m/z 305.8 ( q = 0.830), (b) m/z 290.2 ( q = 0.830), and (c) m/z 290.2 ( q = 0.791); m/z range 280 to 330. 280 285 290 295 300 305 310 315 320 325 330 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 321.30 318.28 307.26 304.28 322.33 293.25 308.32 313.18 290.22 323.09 294.35 329.23 317.15 287.21 298.18 304.27 321.26 318.28 307.26 290.25293.26 308.31 322.26 294.24 313.22 323.23 315.24 328.24 301.21 283.23287.24 318.29 304.26 321.29 307.26 293.25 290.25 308.29 322.28 294.28 313.22 282.39302.24 329.18 317.17 323.38 286.21 m/z 305.8 q = 0.830 0.6 Vp pNL: 4.73E3 m/z 290.2 q = 0.830 0.6 Vp pNL: 5.35E3 m/z 290.2 q = 0.791 0.6 Vp pNL: 4.37E4(a) (b) (c)

PAGE 138

138 Figure 4-10. Variable hexa-notch SWIFT amplitude ( m/z 290.2 at q = 0.791). Peak ion intensities of the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2) at different amplitudes (Vp-p) of a hexa-notch SWIFT isolation waveform based on m/z 290.2 at q = 0.791.

PAGE 139

139 Figure 4-11. Mass spectra ( m/z 280 to 330) of hexa-notch SWIFT at different amplitudes. Hexanotch SWIFT isolation waveform based on m/z 290.2 at q = 0.791 applied to BE, BEd3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 L; 1 g/mL each) onto blank brain tissue and airb rushed with DHB matrix with a SWIFT amplitude of (a) 0.0 Vp-p, (b) 0.6 Vp-p, and (c) 0.9 Vp-p 280 285 290 295 300 305 310 315 320 325 330 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 313.18 318.28 304.25 290.25 321.27 293.23 307.25 282.35 308.32 302.18 314.24 283.28 322.23 329.16 295.15 287.20 318.29 304.26 321.29 307.26 293.25 290.25 308.29 322.28 294.27 313.22 282.39302.23 316.25 329.19 323.38 286.22 318.26 304.26 321.27 307.25 290.25293.25 308.26 322.28 313.15 294.24 282.36301.07 287.21 323.23 317.18 0.0 Vp pNL: 3.02E5 0.6 Vp pNL: 4.47E4 0.9 Vp pNL: 3.11E4(a) (b) (c)

PAGE 140

140 Figure 4-12. Mass spectra ( m/z 80 to 2000) of hexa-notch SWIFT at different amplitudes. Hexanotch SWIFT isolation waveform based on m/z 290.2 at q = 0.791 applied to BE, BEd3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 L; 1 g/mL each) onto blank brain tissue and airb rushed with DHB matrix with a SWIFT amplitude of (a) 0.0 Vp-p, (b) 0.6 Vp-p, and (c) 0.9 Vp-p. 200 400 600 800 1000 1200 1400 1600 1800 2000 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 798.58 273.13 782.61 577.59 848.70 478.41 739.53 313.18 864.70 982.62 258.19 1169.62 1321.91 798.59 782.62 318.29 826.62 577.60 739.55 483.62 864.72 290.25 264.38 982.59 1176.63 1542.24 318.26 290.25 273.14 0.0 Vp pNL: 1.15E5 0.6 Vp pNL: 7.99E4 0.9 Vp pNL: 3.11E4(a) (b) (c)

PAGE 141

141 Figure 4-13. Frequency domain of high mass filte r (HMF). HMF is calculated to excite at frequencies 0 to 338 kHz based on m/z 290.2 at a q of isolation ( q = 0.791). This HMF is designed to eject/excite ions from m/z 325.8 to greater than m/z 2000 (LTQ upper m/z limit; 48 kHz).

PAGE 142

142 Figure 4-14. Variable amplitude of HMF. Peak ion intensities of the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2) CE ( m/z 318.2), and CE-d3 ( m/z 321.2) at different amplitudes (Vp-p) of a high mass filter (HMF) with m/z cutoff at 325.8 applied to the analysis of brain tissue.

PAGE 143

143 Figure 4-15. Mass spectra ( m/z 80 to 2000) of HMF at different amplitudes. HMF SWIFT excitation waveform applied to BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 L; 1 g/mL each) onto blank brain ti ssue and airbrushed with DHB matrix with a SWIFT amplitude of (a) 0.0 Vp-p, (b) 0.4 Vp-p, and (c) 0.5 Vp-p. 200 400 600 800 1000 1200 1400 1600 1800 2000 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 273.12 798.58 577.59 848.69 760.63 478.42 876.70 258.20 296.16 1038.53 1305.901570.12 798.59 273.13 782.62 810.63 739.54 864.70 577.59 483.61 261.17 1028.61 273.13 257.16 0.4 Vp pNL: 3.97E4 0.5 Vp pNL: 5.17E4 0.0 Vp pNL: 1.41E5(a) (b) (c)

PAGE 144

144 Figure 4-16. Mass spectra ( m/z 280 to 330) of HMF at different amplitudes. HMF SWIFT excitation waveform applied to BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 L; 1 g/mL each) onto blank brain ti ssue and airbrushed with DHB matrix with a SWIFT amplitude of (a) 0.0 Vp-p, (b) 0.4 Vp-p, and (c) 0.5 Vp-p. 280 285 290 295 300 305 310 315 320 325 330 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 313.18 318.29 304.27 290.25 321.27 307.26 293.24 282.37 308.30 302.18 314.25 329.14 283.27 295.15 322.25 287.20 318.27 304.24 321.28 307.25 290.23 293.25 313.19 322.27 308.26 296.15 282.33 302.17 315.20 287.20 323.21 318.27 304.25 321.26 307.25 290.25 293.24 313.15 308.26 322.26 296.16 315.19 301.05 282.32 287.19 324.18 0.4 Vp pNL: 3.56E4 0.5 Vp pNL: 4.67E4 0.0 Vp pNL: 2.93E4(a) (b) (c)

PAGE 145

145 Figure 4-17. Frequency domain of two-stage SWIFT isolation. The frequency domain of the two-stage isolation performed by combin ing a HMF SWIFT excitation waveform (red) to eject ions heavier than m/z 325.8 with a hexa-notch SWIFT isolation waveform (blue) to selectively isolate ions at m/z 290.2, 293.2, 304.2, 307.2, 318.2, and 321.2.

PAGE 146

146 Figure 4-18. The time domain of the two-stage SWIFT isolation. The two-stage isolation is composed of a HMF SWIFT excitation waveform (red) occurring from 0 to 4,096 s and the hexa-notch SWIFT isolation wave form (blue) occurring from 4,097 to 8,192 s. A single burst of the tw o-stage isolation pulse (8,192 s) is triggered during the LTQ isolation event (15,500 s).

PAGE 147

147 Figure 4-19. Variable amplitude of two-stage SWIFT isolation. Peak ion intensites of the [M+H]+ ions of BE ( m/z 290.2), BE-d3 ( m/z 293.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2) CE ( m/z 318.2), and CE-d3 ( m/z 321.2) at different amplitudes (Vp-p) of a twostage isolation composed of a HMF SWIF T excitation waveform and a hexa-notch SWIFT isolation waveform applied to the analysis of brain tissue.

PAGE 148

148 Figure 4-20. Mass spectra ( m/z 80 to 2000) of two-stage SW IFT isolation at different amplitudes. Two-stage isolation compos ed of a HMF SWIFT excitation waveform and a hexa-notch SWIFT isolation wave form that was applied to BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 L; 1 g/mL each) onto blank brain tissue and airbrushed with DHB matr ix with both SWIFT amplitudes at (a) 0.0 Vp-p, (b) 0.5 Vp-p, and (c) 0.6 Vp-p. 500 1000 1500 2000 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 273.13 798.60 577.62 866.68 483.62 291.09 982.63 1151.70 798.60 782.63 318.27 851.69 753.65 864.71 694.60 483.62 916.72 389.54 1542.24 1086.72 1707.48 307.22 264.33 0.0 Vp pNL: 1.69E5 0.5 Vp pNL: 8.84E3 0.6 Vp pNL: 5.18E3(a) (b) (c)

PAGE 149

149 Figure 4-21. Mass spectra ( m/z 280 to 330) of two-stage SW IFT isolation at different amplitudes. Two-stage isolation compos ed of a HMF SWIFT excitation waveform and a hexa-notch SWIFT isolation wave form that was applied to BE, BE-d3, COC, COC-d3, CE, and CE-d3 standards that were spiked (1 L; 1 g/mL each) onto blank brain tissue and airbrushed with DHB matr ix with both SWIFT amplitudes at (a) 0.0 Vp-p, (b) 0.5 Vp-p, and (c) 0.6 Vp-p. 280 285 290 295 300 305 310 315 320 325 330 m/z 0 50 100 0 50 100 Relative Abundance 0 50 100 282.38 313.19 291.09 283.34 311.19 308.35 314.29 302.19 296.21 321.23 329.17 287.22 318.27 304.25307.23 321.26 290.23 293.22 313.15 308.27 322.22 311.18 294.23 314.26 301.19 282.33 287.23 323.44 329.16 307.22 318.24 293.21 304.22 321.24 290.20 308.24 322.20 313.13 294.21 314.30 301.16 297.16 323.13 285.120.0 Vp pNL: 3.26E4 0.5 Vp pNL: 6.55E3 0.6 Vp pNL: 5.18E3(a) (b) (c)

PAGE 150

150 Figure 4-22. MS/MS product spectra from the app lication of a two-stage isolation. Two-stage SWIFT is composed of a HMF SWIFT exc itation waveform and a hexa-notch SWIFT isolation waveform with both SWIFT amplitudes at 0.6 Vp-p and collision-induced dissociation (CID) set to (a) 0 and (b) 90. 100 150 200 250 300 350 m/z 0 20 40 60 80 100 0 20 40 60 80 100 Relative Abundance 318.26 304.24 321.26 293.24 264.36 323.20 283.11 196.21 199.23 182.18 168.18 318.26 304.23 200.24 150.18 290.19 108.20 245.11 139.17 214.070.6 Vp pCID = 90 NL: 1.43E3 0.6 Vp pCID = 0 NL: 3.68E4(a) (b)

PAGE 151

151 Figure 4-23. Comparison of MS/MS with wide isolation and two-stage SWIFT isolation. MS/MS product spectra from the application of a (a) 40-Da wide isolation with CID = 55 and a (b) two-stage isolation compos ed of a HMF SWIFT excitation waveform and a hexa-notch SWIFT isolation wavefo rm with both SWIFT amplitudes at 0.6 Vp-p and CID = 90. 100 150 200 250 300 350 m/z 0 20 40 60 80 100 0 20 40 60 80 100 Relative Abundance 137.13 147.07 196.22 182.21 318.29 304.26 168.19 290.25 156.99 200.23 104.24 231.05 271.15 323.25 122.21 196.21 199.24 182.19 168.19 318.26 304.24 200.24 150.18 290.22 119.14 245.10 214.07Wide Isolation CID = 55 NL: 9.10E3 SWIFT Isolation CID = 90 NL: 2.17E3(a) (b)

PAGE 152

152 Figure 4-24. Mass spectra comparison of wide isolation and two-stage SWIFT isolation. Isolated ions from the application of a (a ) 40-Da wide isolati on and a (b) two-stage isolation composed of a HMF SWIFT exc itation waveform and a hexa-notch SWIFT isolation waveform with both SWIFT amplitudes at 0.6 Vp-p. 280 285 290 295 300 305 310 315 320 325 330 m/z 0 20 40 60 80 100 0 20 40 60 80 100 Relative Abundance 291.05 318.29 304.29 321.28 307.28 290.30 313.14 293.24 311.09 300.97 296.19 322.25 314.31 323.21 287.22 297.18 318.26 304.24 321.26 307.25 293.24 290.24 308.27 322.24 294.24 313.18 302.20 315.22 297.20 323.20 283.11Wide Isolation NL: 1.63E4 SWIFT Isolation NL: 3.64E4(a) (b)

PAGE 153

153 Figure 4-25. BE calibration curve for BE spiked on intact brain tissue. Five solutions (mixture of BE, COC, and CE at 62, 125, 250, 500, a nd 1000 ng/mL with a mixture of BE-d3, COC-d3, and CE-d3 at 200 ng/mL) were pi petted in triplicate (1 L each) onto blank human brain tissue and airbrushed with DHB matrix. The peak intensities of the product ions of BE and BE-d3 at m/z 168 and m/z 171, respectively, were ratioed from two different MS/MS experiments: MS/M S of the ions isolated by a two-stage SWIFT isolation (CID = 90) and MS/MS of the ions isolated by a 40-Da wide isolation (CID = 55). Th e error bars correspond to the standard error (3 replicates). Mass (pg) SWIFT Iso %RSD Wide Iso %RSD 62 3 4 125 11 4 250 5 11 500 0 2 1000 4 6

PAGE 154

154 Figure 4-26. COC calibration curve for COC spik ed on intact brain tissu e. Five solutions (mixture of BE, COC, and CE at 62, 125, 250, 500, and 1000 ng/mL with a mixture of BE-d3, COC-d3, and CE-d3 at 200 ng/mL) were pipe tted in triplicate (1 L each) onto blank human brain tissue and airbrushed with DHB matrix. The peak intensity of the product ions of COC and COC-d3 at m/z 182 and m/z 185, respectively, were ratioed from two different MS/MS experime nts: MS/MS of the ions isolated by a two-stage SWIFT isolation (CID = 90) and MS/MS of the ions isolated by a 40-Da wide isolation (CID = 55). The error bars correspond to the standard error (3 replicates). Mass (pg) SWIFT Iso %RSD Wide Iso %RSD 62 2 5 125 5 4 250 6 4 500 4 3 1000 1 4

PAGE 155

155 Figure 4-27. Five solutions (mixture of BE, COC, and CE at 62, 125, 250, 500, and 1000 ng/mL with a mixture of BE-d3, COC-d3, and CE-d3 at 200 ng/mL) were pipetted in triplicate (1 L each) onto blank human brain tissue and airbrushed with DHB matrix. The peak intensity of the product ions of CE and CE-d3 at m/z 196 and m/z 199, respectively, were ratioed from two diffe rent MS/MS experiments: MS/MS of the ions isolated by a two-stage SWIFT isola tion (CID = 90) and MS/MS of the ions isolated by a 40-Da wide isolation (CID = 55). The error bars correspond to the standard error (3 replicates). Mass (pg) SWIFT Iso %RSD Wide Iso %RSD 62 2 5 125 5 4 250 6 4 500 4 3 1000 1 4

PAGE 156

156 Table 4-4. Quantification of BE, COC, a nd CE from Unspiked Human Brain Tissue Analyte MS/MS Ion ( m/z ) Two-Stage SWIFT Intact Tissue (ng/g tissue) 40-Da Wide Isolation Intact Tissue (ng/g tissue) 5-Da Wide Isolation Tissue Homogenate (ng/g tissue) BE 168.2 140 170 270 14 COC 182.2 230 60 380 11 CE 196.2 40 30 430 16

PAGE 157

157 Figure 4-28. BE calibration curve for BE standa rds spiked in blank brain tissue homogenate. Mass (ng) % RSD 623 1256 2503 5003 10008

PAGE 158

158 Figure 4-29. COC calibration curve for COC standards spiked into blank brain tissue homogenate. Mass (ng) % RSD 626 12512 2504 5008 10004

PAGE 159

159 Figure 4-30. CE calibration curve for CE standa rds spiked into blank brain tissue homogenate. Mass (ng) % RSD 624 1256 2504 5005 10003

PAGE 160

160 CHAPTER 5 CONCLUSIONS AND FUTURE WORK Conclusions The goal of this research was to develop a quantitative mass spectrometric imaging (MSI) method for determining the regional composition of drugs and their metabolites in postmortem brain tissue. This research focused on the an alysis of cocaine (COC ) and two of its major metabolites, benzoylecgonine (BE) and cocaethylene (CE). COC is the most frequent cause of drug-related deaths in the United States, so it is of particular interest to the field of postmortem toxicology. Conventional quantification of COC in brai n tissue involves hom ogenate preparation, followed by extraction and/or derivatization. Th e extracts are then usua lly analyzed by gas chromatography/mass spectrometry (GC/MS), liquid chromatography/mass spectrometry (LC/MS), GC, or LC. Lengthy extraction pr ocedures are required to remove large concentrations of lipids and ot her endogenous materials present in the brain, which may interfere with the analysis. Multiple sample pretreatment steps also allow opportunity for loss of analyte, and tissue homogenization eliminates spatial in formation, which could provide histologicallyspecific drug distribution. MSI using matrix-assisted laser desorption /ionization (MALDI) mass spectrometry (MS) could provide quantitative inform ation about the distribution of COC and its metabolites in brain tissue more rapidly, with higher spatial resoluti on, and with less sample loss than conventional drug analysis methods that involve tissu e homogenization. Quantitative MALDI-MS is challenging, however, because MALDI exhibits irreproducible signal intensities due to inhomogeneous crystal formation, inconsistent sample preparation, a nd laser shot-to-shot variability. It has been show n though, that normalizing the anal yte ion signal to that of a

PAGE 161

161 structurally similar intern al standard ion (e.g., [M+H]+ ion of COC and COC-d3) can dramatically reduce signal variability ma king quantification by MALDI-MS possible. Mass spectrometry analysis of brain tissue can be very complicated, especially without the benefit of extraction and chromatography met hods to clean up the sample and separate compounds. The presence of isobaric ions in sa mples increases with sample complexity and may interfere with quantificat ion at low analyte concentra tions. Tandem mass spectrometry (MSn) can improve analyte selectivity and produce higher signal-to-noise ratios, resulting in lower detection and quantification limits fo r the analyte. Combining the use of MSn with internal standards is commonly perf ormed by alternating MSn scans of the analyte and the internal standard ions, and then ratioing the resulting product ion signals. This method is effective for use with ionization techniques such as electr ospray and atmospheric pressure chemical ionization; however, due to the shot-to-shot vari ability of MALDI, acquiri ng analyte and internal standard signals in alternating MSn scans may counteract the signa l normalizing effects gained by using an internal standard. Strategies for isolating the analyte and in ternal standard ions during a single MSn scan were investigated in order to improve the precision of MALDI-MSn to allow for quantification of COC and its metabolites in brain tissue. One st rategy was to use a single wide isolation window (e.g., 5 Da) centered at a mass-to-charge ( m/z ) between that of the analyte and internal standard ions. This allows for the simultaneous isolati on and collision-induced di ssociation (CID) of the analyte and internal standard ions. For exampl e, for the analysis of COC, a 5-Da isolation window could be placed at m/z 305.8 to isolate the [M+H]+ ion of COC at m/z 304.2 and the [M+H]+ ion of its trideuterated analog, COC-d3 at m/z 307.2. By applying CID across the isolation window, a MS2 spectrum is produced containing th e product ions of COC and COC-d3

PAGE 162

162 at m/z 182.2 and m/z 185.2, respectively. This method was shown to provide improved precision (~ 10 to 20 times reduction in %RSD) for quant itative analysis of COC in postmortem brain tissue compared with using two alternating MS2 scans that isolate the analyte and internal standard ions separately. It was also shown that wide isolation can be used for multiple stages of mass analysis (e.g., MS3) as long as the product ion derived from the deuterated internal standard maintains the deuterated tag allowing it to be dist inguished from the product ion of the analyte. For MS3 analysis of COC, the product ion at m/z 150 for COC was ratioed with the product ion at m/z 153 for COC-d3. Multi-notch SWIFT isolation was investigated as an alternative isolation strategy to wide isolation for isolating the analyte and internal standa rd ions during a single MS2 scan for improved MALDI-MS2 precision. SWIFT isolation has hi gher mass selectiv ity than wide isolation and is able to reduce background ions that may complicate or interfere with MS2 analysis (e.g., isobaric product ions ). Also, analysis times and s ubsequently laser shots can be reduced as more frequency notch es are added to the SWIFT is olation waveform. This can become very important when quantitatively imaging several analytes from a large tissue sample. It was determined that multi-notch SWIFT isolation can provide improved precision when compared to using two alternating MS2 scans that isolate the analyt e and internal standard ions separately. However, it was determined that multi-notch SWIFT isolation was not as precise as the wide isolation method. This might be due to frequency shifts of th e analyte and internal standard ions from space-charge effects caused by high m/z background ions from the brain tissue (e.g., lipid region at m/z 700 to 900). A two-stage SWIFT isolation method was devel oped that utilizes a high mass filter (HMF) SWIFT excitation waveform to remove high m/z background ions (e.g., m/ z 325 to 2000) during

PAGE 163

163 the first stage of isolation. This has been shown to reduce the fre quency shifts of the analyte and internal standard ions by preventing them fr om moving outside the notches of the multi-notch SWIFT isolation waveform. This prevents the i ons desired for isolation from being ejected by the SWIFT isolation waveform and resulted in an increased signal for the isolated ions compared to the application of a multi-notch SWIFT isolat ion waveform without the HMF. A hexa-notch SWIFT isolation waveform was used during the second stage of the two-stage SWIFT isolation to mass selectively isolate the [M+H]+ ions of BE ( m/z 168.2), BE-d3 ( m/z 171.2), COC ( m/z 304.2), COC-d3 ( m/z 307.2), CE ( m/z 318.2), and CE-d3 ( m/z 321.2). The hexa-notch SWIFT isolation waveform was able to effectively remove background ions around the analyte and internal standard ions that ma y have interfered with MS/MS analysis. The two-stage SWIFT isolation overall showed similar precision to that of wide isolation when performing MS/MS on the analyte and internal standards spiked on blank brain tissue. Ho wever, the higher mass selectivity of the two-stage SWIFT isolation afford s a significant loss in absolute signal intensity of the analyte and internal sta ndard ions when it is applied. Two-stage SWIFT isolation was compared to wide isolation of intact tissue and SPE extracted homogenized tissue for the MALDI-MS/MS quantification of BE, COC, and CE from unspiked human brain tissue, whose toxicological analysis indicated the presence of cocaine in blood. The two-stage SWIFT isolation method show ed lower analyte signal per gram of tissue than the wide isolation method for BE, COC, a nd CE. The quantificatio n results for the twostage SWIFT isolation and the wide isolation of intact tissue was not comparable to that for the wide isolation analysis of tissu e homogenate (Table 4-4). Howe ver, the intact tissue methods required considerably less sample preparation and smaller sample sizes than the tissue homogenate method. There was no analysis tim e saved or fewer laser shots fired when

PAGE 164

164 comparing the two-stage SWIFT isolati on and wide isolation since the [M+H]+ ions of BE, BEd3, COC, COC-d3, CE, and CE-d3 were all isolated simultaneously during the same MS/MS scan. In conclusion, a wide isolation method may stil l be a better choice over SWIFT isolation for improving MALDI-MS/MS precision for quantificat ion and reducing analysis time, despite the inclusion of unwanted backgr ound ions during isolation. Future Work The linear ion trap has a lim ited ion storage capacity (~107 ions) before coulombic interactions between stored ions degrade th e mass resolution and reduce sensitivity (spacecharge effects). Sensitivity is reduced when space-charge, created by unwanted matrix ions, limits the total number of analyte ions which ma y be trapped. Julian and Cooks first applied SWIFT to the quadrupole ion trap du ring injection to resonantly ej ect these matrix ions and to selectively accumulate and store analyte ions to increase sensitivity and avoid interference.106 For this research, the applica tion of SWIFT during different pe riods of the LTQ scan function was explored to include at the beginning of scan, injection peri od, isolation, activation, and scan out. All of these periods of th e LTQ scan function are included in the programmable trigger provided by the LTQ software; however, applicat ion of SWIFT during the injection period was unsuccessful and requires further investigation to exploit the sensitivity gains promised by selective accumulation and storage of analyte ions. Quantitative imaging of cocaine and its me tabolites from brain tissue of a habitual cocaine user showed no localizat ion of the analytes in the se ction of the nucleus accumbens analyzed. A controlled animal study involving lowe r doses of cocaine may show localization of cocaine and its metabolites within specific region s of the brain, and provide more information about the mechanisms of this drug.

PAGE 165

165 APPENDIX A BETA CALCULATION Beta ( ) is defined precisely by a continued fr action (aqb_conf) expression in terms of a and q Since a = 0, this expression simplifies to Equation A-1 (same as Equation 3-2): (A-1) A LabView subprogram, or subVI (VI = virtual in strument), was written to calculate aqb_conf based on q and inputs. Figure A-1 shows the block diag ram of the aqb_conf sub VI, which is a graphical representation of Equation A-1. Figure A-1. LabView block diag ram of subVI aqb_conf, which is used to calculate the continued fraction given q and inputs. 2 2 2 2 2 2 2 2 2 2 2 2) 12 ( ) 10 ( ) 8 ( ) 6 ( ) 4 ( ) 2 ( q q q q q q conf aqb 2 2 2 2 2 2 2 2 2 2 2 2 2) 12 ( ) 10 ( ) 8 ( ) 6 ( ) 4 ( ) 2 ( q q q q q q

PAGE 166

166 A LabView program called Trap Calc ulator was written to calculate through an iterative process given a specific q input, such as q = 0.83 for isolation. Trap Calculator makes a first guess at the value for based on the Dehmelt approximation given in Equation A-2, same as Equation 3-4: 22q a (A-2) This approximation is assigned the variable x1. Ninety percent of the approximation is assigned the variable x0. Both x0 and x1 are applied as inputs for the aqb_conf sub VI with the desired q as the q inputs resultin g in the function outputs of fx0 and fx1. The variable x2 is used to store the next iterative approximation of in which x2 = x1 – ((x0 – x1) / (fx0 / fx1 1)). Then x0 is replaced with the value of x1 and x1 is replaced w ith the value of x2. This process is repeated until the absolute value of the differen ce between x0 and x1 is greater than 1x10-7, and then the variable beta is assigned the value of x1. Fi gure A-2 shows the block diagram for the Trap Calculator LabView program, which is a graphi cal representation of the iterative process described to calculate Figure A-2. Trap Calculator LabV iew program used to calculate through an iterative process.

PAGE 167

167 This process of calculating is shown graphically in Figure A-3, where was calculated after 5 iterations with q = 0.830. The value for was plotted versus th e number of iterations. Figure A-3. Iterativ e calculation of using the LabView program Trap Calculator. 0.586898628 0.749031089 0.735862704 0.736161786 0.736161658 0.7361616580.58 0.60 0.62 0.64 0.66 0.68 0.70 0.72 0.74 0.76 012345BetaNumber of Iterations

PAGE 168

168 APPENDIX B C++ SWIFT PROGRAM Figure B-1. Program introductory comments. Briefly describes purpose, references, modifications, and operation of program. /*====================================================================================================== Name:REICH HexSWIFT.cpp Author:C++ version by Richard Reich (12 August 2009) using Microsoft Visual C++ 2008 Express Edition Purpose:This program calculates a hextuplet notch stored waveform inverse Fourier transform (SWIFT) pulse and downloads the data points to an arbitrary waveform generator (DS345) to be applied to the X rods of the center section of a linear ion trap (LTQ) for isolation of 3 analytes and their corresponding internal standard ions in the linear ion trap for increased precision during MALDI MS^nquantitation. References:This C++ program is adapted from the original C program written by P.T. Palmer (4 Sep 92). Palmer's program was based on the paper: Chen, L.; Wang, T.C.L.; Ricca, T.L.; Marshall, A.G. "Phase Modulated Stored Waveform Inverse Fourier Transform Excitation for Trapped Ion Mass Spectrometry"; Anal. Chem. 1987, 59, 449 454. The fast Fourier transform (FFT) algorithm and required data format are described in Vetterling, W.T.; Teukolsky, S.A.; Press, W.H.; Flannery, B.P. "Numerical Recipes in C: the Art of Scientific Computing", 2nd ed., New York: Cambridge University Press, 1992, pp. 399 413. Modifications:The following modifications to Peter Palmer's original C SWIFT program were made when writing program in C++ to remove compiler warnings and errors: Removed "huge" from float t[N]; f[N/2+1]; r[N/2+1]; p[N/2+1]; x[N/2+1]; and y[N/2+1], because term caused several compiling errors. (fixed errors C2146, C4430, C2086, C2371) Replaced "gets()" function with "gets_s()" function as recommended by compiler to avoid overrunning buffer. "_s" suffix stands for secure. (fixed warning C4996) Replaced "sscanf()" function with "sscanf_s()" function as recommended by compiler to avoid overrunning buffer. (fixed warning C4996) Replaced "strcpy()" function with "strcpy_s()" function as recommended by compiler to avoid overruningbuffer. (fixed warning C4996) Included "cstdlib" source file to identify "exit". (fixed error C3861) Removed "ieeeio.h" source file and "download" function for downloading time domain waveform over an IEEE 488 to the DS345 waveform generator through a GPIB. Will use RS232 instead. Included "system("PAUSE");" at end of program before function definitions to prevent black window from disappearing before user can read it. Requires user to hit any key to close window. Operation:1. Specify the following parameters: N:number of points in time domain SWIFT waveform freq:sampling rate of function generator (kHz) lfreq:lower limit of frequency pulse (kHz) rfreq:upper limit of frequency pulse (kHz) mag:normalization value for time domain waveform 2. Setup output file (asciiformat) for time and frequency domain waveforms for plotting and inspection. 3. Generate desired frequency domain waveform. Compute magnitude and phase spectra. Use a quadratic function for phase modulation. Convert waveform from polar (magnitude and phase) to a complex array of real and imaginary numbers for Fourier transform. 4. Perform inverse Fourier transform to convert SWIFT waveform from frequency domain to the time domain. 5. Reflect the time domain waveform about its time midpoint to reduce problems with large initial signal transients. 6. Generate the apodizationor smoothing function. This is a quarter wave sinusoid matched to one fourth of the time domain period, followed by unit weighting for the next half period, followed by a quarter wave sinusoid for final one fourth of the period. This function is designed to force the time domain signal smoothly to zero at the beginning and end of the time domain period. 7. Multiply the time domain waveform by the apodizationfunction and normalize.

PAGE 169

169 Figure B-2. Definition of constant s and declaration of variables. 8. Write waveform data to output file. This data includes: frequency vsmagnitude of specified waveform frequency vsphase of specified waveform time vsamplitude of computed waveform (reflected, apodized, and normalized) 9. Download the time domain waveform over an RS232 serial port to the function generator. 10. Write waveform data to output file. ======================================================================================================*/ /* insert source files from standard C library using the preprocessor #include command */ #include /*(standard input/output header); allows functions from C standard library to be used */ #include /*(standard string header); allows use of string handling and various memory handling functions */ #include /*(standard math header); allows basic mathematical operations */ #include /* added in C++ version to identify function "exit". (fixed error C3861) */ /* define symbolic constants (conventionally name is capitalized) using the preprocessor #define command */ #define SWAP(a,b) tempr= (a); (a) = (b); (b) = tempr/* function swaps values of two variables (a & b) */ #define PI3.141592654 #define N4096/* max # points in SWIFT waveform */ #define MAX2047/* max amplitude for SWIFT waveform */ #define LF1398.681641/* default value for lower limit of first frequency band (kHZ) */ #define RF1401.855469/* default value for upper limit of first frequency band (kHz) */ #define LF2 405.029297/* default value for lower limit of second frequency band (kHz) */ #define RF2 408.203125/* default value for upper limit of second frequency band (kHz) */ #define LF3431.152344/* default value for lower limit of third frequency band (kHZ) */ #define RF3435.058594/* default value for upper limit of third frequency band (kHz) */ #define LF4 439.208984/* default value for lower limit of fourth frequency band (kHz) */ #define RF4 443.603516/* default value for upper limit of fourth frequency band (kHz) */ #define LF5 475.341797/* default value for lower limit of fifth frequency band (kHz) */ #define RF5 481.445312/* default value for upper limit of fifth frequency band (kHz) */ #define LF6 488.037109/* default value for lower limit of sixth frequency band (kHz) */ #define RF6 494.873047/* default value for upper limit of sixth frequency band (kHz) */ #define SF1000/* default sampling frequency (kHz) for DS345 arbitrary waveform generator (AWG) */ #define V1.0/* default value for voltage of output waveform */ /* declaration of variables (property of variable and name); int=integer (whole #), float=floating point (# with fraction) */ unsigned intfreq;/* DS345 sampling frequency (kHz); unsigned means no negative values */ float vpp;/* peak to peak voltage of SWIFT waveform */ float t[N];/* time vector for plotting time domain axis in microseconds */ float f[N/2 + 1];/* frequency vector for plotting freq domain axis in kHz */ float r[N/2 + 1];/* magnitude spectrum of SWIFT waveform (polar coordinate) */ float p[N/2 + 1];/* phase spectrum of SWIFT waveform (polar coordinate) */ float x[N/2 + 1];/* real portion of SWIFT waveform (cartesiancoordinate) */ float y[N/2 + 1];/* imaginary portion of SWIFT waveform (cartesiancoordinate) */ float data[2*N];/* SWIFT waveform of N complex points (real, imag) */ float apod[N];/* apodizationor smoothing function */ intwave[N+1];/* scaled, intformat of SWIFT with checksum at end */ void fft(float data[], intn, intisign);/* void means that function doesn't return a value */ void main() {

PAGE 170

170 Figure B-3. Variab le definitions. /******************************************************************************************************** VARIABLE DEFINITIONS ********************************************************************************************************/ intsflag;/* flag to indicate SWIFT waveform type */ floattempr;/* temporary variable for swap function */ charinbuf[20];/* input buffer array for character strings */ floatl1freq;/* lower limit of first frequency pulse (kHz) */ floatr1freq;/* upper limit of first frequency pulse (kHz) */ floatl2freq;/* lower limit of second frequency pulse (kHz) */ floatr2freq;/* upper limit of second frequency pulse (kHz) */ floatl3freq;/* lower limit of third frequency pulse (kHz) */ floatr3freq;/* upper limit of third frequency pulse (kHz) */ floatl4freq;/* lower limit of fourth frequency pulse (kHz) */ floatr4freq;/* upper limit of fourth frequency pulse (kHz) */ floatl5freq;/* lower limit of fifth frequency pulse (kHz) */ floatr5freq;/* upper limit of fifth frequency pulse (kHz) */ floatl6freq;/* lower limit of sixth frequency pulse (kHz) */ floatr6freq;/* upper limit of sixth frequency pulse (kHz) */ interrflag;/* flag to denote input errors */ floatfreqmax;/* Nyquistfreq or maximum possible fast Fourier transform (FFT) freq (kHz) */ floatdeltat;/* time interval in microseconds */ floatdeltaf;/* frequency interval in kHz */ intl1index;/* lower index of first frequency band */ intr1index;/* upper index of first frequency band */ intl2index;/* lower index of second frequency band */ intr2index;/* upper index of second frequency band */ intl3index;/* lower index of third frequency band */ intr3index;/* upper index of third frequency band */ intl4index;/* lower index of fourth frequency band */ intr4index;/* upper index of fourth frequency band */ intl5index;/* lower index of fifth frequency band */ intr5index;/* upper index of fifth frequency band */ intl6index;/* lower index of sixth frequency band */ intr6index;/* upper index of sixth frequency band */ floata;/* coefficient of x term for quadratic phase modulation */ floatb;/* coefficient of x^2 term for quadratic phase modulation */ inti, j;/* array indices */ charfileflag;/* flag to indicate output desired */ charascfile[20];/* filename for ASCII file output of SWIFT waveform */ charbinfile[20];/* filename for binary file output of SWIFT waveform */ FILE*ascfp;/* file pointer for ASCII file output */ FILE*binfp;/* file pointer for binary file output */ intdlflag;/* flag to denote downloading of SWIFT waveform */ floatmax;/* maximum amplitude (volts) in SWIFT waveform */ floatnorm1, norm2;/* normalization factors computed from time domain maximum */ intchecksum;/* check sum value for downloading SWIFT to AWG (DS345) */

PAGE 171

171 Figure B-4. Initialize variab les for notches 1 through 4. /******************************************************************************************************** INITIALIZE VARIABLES ********************************************************************************************************/ printf("This program can generate SWIFT waveforms for excitation or isolation.\n"); printf("Would you like to generate a SWIFT waveform for excitation (default = Y)? "); gets_s(inbuf, 20); if ((strstr(inbuf,"N") != NULL) || (strstr(inbuf,"n") != NULL)) sflag= 1; /* set flag to 1 if input is N or n */ else sflag= 0;/* if input is other than N or n, set flag to 0 */ printf("Lower limit of first frequency notch in kHz (default = %i)? ", LF1); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) l1freq = LF1; /* if no input from user (i.e, string length is equal to 0), then default LF1 value is used */ else sscanf_s(inbuf, "%f", &l1freq, 20); /* read input from user, format it, and store in variable l1freq */ printf("Upper limit of first frequency notch in kHz (default = %i)? ", RF1); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) r1freq = RF1; /* if no input from user (i.e, string length is equal to 0), then default RF1 value is used */ else sscanf_s(inbuf, "%f", &r1freq, 20); /* read input from user, format it, and store in variable r1freq */ printf("Lower limit of second frequency notch in kHz (default = %i)? ", LF2); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) l2freq = LF2; /* if no input from user (i.e, string length is equal to 0), then default LF2 value is used */ else sscanf_s(inbuf, "%f", &l2freq, 20); /* read input from user, format it, and store in variable l2freq */ printf("Upper limit of second frequency notch in kHz (default = %i)? ", RF2); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) r2freq = RF2; /* if no input from user (i.e, string length is equal to 0), then default RF2 value is used */ else sscanf_s(inbuf, "%f", &r2freq, 20); /* read input from user, format it, and store in variable r2freq */ printf("Lower limit of third frequency notch in kHz (default = %i)? ", LF3); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) l3freq = LF3; /* if no input from user (i.e, string length is equal to 0), then default LF3 value is used */ else sscanf_s(inbuf, "%f", &l3freq, 20); /* read input from user, format it, and store in variable l3freq */ printf("Upper limit of third frequency notch in kHz (default = %i)? ", RF3); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) r3freq = RF3; /* if no input from user (i.e, string length is equal to 0), then default RF3 value is used */ else sscanf_s(inbuf, "%f", &r3freq, 20); /* read input from user, format it, and store in variable r3freq */ printf("Lower limit of fourth frequency notch in kHz (default = %i)? ", LF4); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) l4freq = LF4; /* if no input from user (i.e, string length is equal to 0), then default LF4 value is used */ else sscanf_s(inbuf, "%f", &l4freq, 20); /* read input from user, format it, and store in variable l4freq */

PAGE 172

172 Figure B-5. Initialize vari ables for notches 5 and 6. printf("Upper limit of fourth frequency notch in kHz (default = %i)? ", RF4); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) r4freq = RF4; /* if no input from user (i.e, string length is equal to 0), then default RF4 value is used */ else sscanf_s(inbuf, "%f", &r4freq, 20); /* read input from user, format it, and store in variable r4freq */ printf("Lower limit of fifth frequency notch in kHz (default = %i)? ", LF5); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) l5freq = LF5; /* if no input from user (i.e, string length is equal to 0), then default LF5 value is used */ else sscanf_s(inbuf, "%f", &l5freq, 20); /* read input from user, format it, and store in variable l5freq */ printf("Upper limit of fifth frequency notch in kHz (default = %i)? ", RF5); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) r5freq = RF5; /* if no input from user (i.e, string length is equal to 0), then default RF5 value is used */ else sscanf_s(inbuf, "%f", &r5freq, 20); /* read input from user, format it, and store in variable r5freq */ printf("Lower limit of sixth frequency notch in kHz (default = %i)? ", LF6); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) l6freq = LF6; /* if no input from user (i.e, string length is equal to 0), then default LF6 value is used */ else sscanf_s(inbuf, "%f", &l6freq, 20); /* read input from user, format it, and store in variable l6freq */ printf("Upper limit of sixth frequency notch in kHz (default = %i)? ", RF6); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ if (i== 0) r6freq = RF6; /* if no input from user (i.e, string length is equal to 0), then default RF6 value is used */ else sscanf_s(inbuf, "%f", &r6freq, 20); /* read input from user, format it, and store in variable r6freq */

PAGE 173

173 Figure B-6. Test initi alized variables. if (r1freq < l1freq || r2freq < l2freq || r3freq < l3freq || r4freq < l4freq || r5freq < l5freq || r6freq < l6freq) /* checks tomake sure that the upper limit is not less than the lower limit of freq notch */ { printf("Your upper limit of your frequency notch cannot be less than your lower limit.\n"); printf("This program will now terminate,\n"); exit( 1); /* causes normal program termination */ } do/* Do while loop for acquiring appropriate sampling frequency from user */ { errflag = 0; /* presets errflag to zero */ printf("Sampling frequency for DS345 in kHz (default = 1000)? ", SF); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf);/* counts length of input string from user */ if (i== 0) freq = SF; /* if no input from user (i.e, string length is equal to 0), then default SF value is used */else { sscanf_s(inbuf, "%u", &freq, 20); /* read input from user, format it, &store in variable freq */ if (freq & 1)/* assign to freq the address of 1 */ { printf("Number must be a multiple of 2 try again\n"); errflag= 1; } else if (freq > 40000) { printf("Number must be less than or equal to 40,000 try again\n"); errflag= 1; } } } while (errflag== 1); printf("Peak to peak voltage of output waveform (default = %2.1f)? ", V); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ i= strlen(inbuf);/* counts length of input string from user */ if (i== 0) vpp= V;/* if no input from user (i.e, string length is equal to 0), then default V value is used */ else sscanf_s(inbuf, "%f", &vpp, 20); /* read input from user, format it, and store in variable vpp*/ freqmax= freq/2;/* in kHz */ deltat= 1000/freq; /* in microseconds */ deltaf= 2 freqmax/N; /* in kHz */ l1index = floor(l1freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 1st freq band */ r1index = ceil(r1freq/deltaf); /* ceil rounds up to nearest integer; sets uppperindex of 1st freq band */ l2index = floor(l2freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 2nd freq band */ r2index = ceil(r2freq/deltaf); /* ceil rounds up to nearest integer; sets upper index of 2nd freq band */ l3index = floor(l3freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 3rd freq band */ r3index = ceil(r3freq/deltaf); /* ceil rounds up to nearest integer; sets uppperindex of 3rd freq band */ l4index = floor(l4freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 4th freq band */ r4index = ceil(r4freq/deltaf); /* ceil rounds up to nearest integer; sets upper index of 4th freq band */ l5index = floor(l5freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 5th freq band */ r5index = ceil(r5freq/deltaf); /* ceil rounds up to nearest integer; sets uppperindex of 5th freq band */ l6index = floor(l6freq/deltaf); /* floor rounds down to nearest integer; sets lower index of 6th freq band */ r6index = ceil(r6freq/deltaf); /* ceil rounds up to nearest integer; sets upper index of 6th freq band */

PAGE 174

174 Figure B-7. Setup output files. /******************************************************************************************************** SETUP OUTPUT FILES ********************************************************************************************************/ fileflag= 0; /* fileflagstores the type of output files desired (1 = ASCII, 2 = binary) */ printf("Write SWIFT waveform data to an ASCII file (default = N)? "); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ if ( (strstr(inbuf, "Y") != NULL) || (strstr(inbuf, "y") != NULL) ) { fileflag= 1; do { printf("Filename for the ASCII data file (default = SWIFT.DAT)? "); gets_s(inbuf, 20);/* reads the input from the user &stores it in the character string "inbuf" */ i= strlen(inbuf); /* counts length of input string from user */ sscanf_s(inbuf, "%s", ascfile, 20);if (i== 0)/* if no input from user, then ascfileset to SWIFT.DAT */ strcpy_s(ascfile, "SWIFT.DAT"); if (i> 16) printf("Filename must be less than or equal to 16 characters try again\n"); } while (i> 16); if ( (strstr(ascfile, ".DAT") == NULL) && (strstr(ascfile, ".dat") == NULL) ) strcat(ascfile, ".DAT"); if ( (ascfp= fopen(ascfile, "wt")) == NULL){ printf("Error could not open %s\n", ascfile); fileflag= 0; } } printf("Write SWIFT waveform data to a binary file (default = N)? "); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ if ( (strstr(inbuf, "Y") != NULL) || (strstr(inbuf, "y") != NULL) ) { fileflag= fileflag+ 2; do { printf("Filename for binary data file (default = SWIFT.ARB)? "); gets_s(inbuf, 20);/* reads the input from the user &stores it in the character string "inbuf" */ i= strlen(inbuf);/* counts length of input string from user */ sscanf_s(inbuf, "%s", binfile, 20); if (i== 0)/* if no input from user, then binfileset to SWIFT.ARB */ strcpy(binfile, "SWIFT.ARB"); if (i> 16) printf("Filename must be less than or equal to 16 characters please try again\n"); } while (i> 16);

PAGE 175

175 Figure B-8. Build components 1, 2, 3, 4, 5, 7, 9 and 11 of excitation waveform. if ( (strstr(binfile, ".ARB") == NULL) && (strstr(binfile, ".arb") == NULL) ) strcat(binfile, ".ARB"); if ( (binfp= fopen(binfile, "wb")) == NULL) { printf("Error could not open %s\n", binfile); fileflag= fileflag 2; } } dlflag= 0; printf("Download SWIFT waveform to the DS345 waveform generator (default = N)? "); gets_s(inbuf, 20);/* reads the input from the user and stores it in the character string "inbuf" */ if ( (strstr(inbuf, "Y") != NULL) || (strstr(inbuf, "y") != NULL) ) dlflag= 1; /******************************************************************************************************** GENERATE DESIRED FREQUENCY DOMAIN WAVEFORM ********************************************************************************************************/ printf("Generating hexa notch SWIFT waveform ...\n"); for (i= 0; i< N; i++) /* increment iup to N (number of points in time domain SWIFT waveform) */ t[i] = i* deltat; /* initialize time axis array in microseconds */ if (sflag== 0) /* if SWIFT waveform type is excitation (sflagequal to 0) */ { a = PI/2; /* component "a" of quadratic phase modulation equation */ /*************** Building component 1, 3, 5, 7, 9, and 11 of excitation SWIFT waveform******************************/ for (i = 0; i <= N/2; i++) { f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 0; /* zero fill r (magnitude, polar coordinate) array */ p[i] = 0; /* zero fill p (phase, polar coordinate) array */ x[i] = 0; /* zero fill x (real, cartesiancoordinate) array */ y[i] = 0; /* zero fill y (imaginary, cartesiancoordinate) array */ } /*************** Building component 2 of excitation SWIFT waveform**************************************/ b = PI/(r1index l1index);/* component "b" of quadratic phase modulation equation */ for (i= l1index; i<= r1index; i++) /* increment iup to right side of 1st freq notch */ {j = i l1index; /* increment j used in quadratic function */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 4 of excitation SWIFT waveform**************************************/ b = PI/(r2index l2index);/* component "b" of quadratic phase modulation equation */ for (i= l2index; i<= r2index; i++)/* increment iup to right side of 2nd freq notch */ { j = i l2index; /* increment j (shifted from i) used in quadratic function */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ }

PAGE 176

176 Figure B-9. Build components 6, 8, 10 and 12 of excitation waveform. /*************** Building component 6 of excitation SWIFT waveform**************************************/ b = PI/(r3index l3index);/* component "b" of quadratic phase modulation equation */ for (i= l3index; i<= r3index; i++)/* increment iup to right side of 3rd freq notch */ { j = i l3index; /* increment j (shifted from i) used in quadratic function */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 8 of excitation SWIFT waveform**************************************/ b = PI/(r4index l4index);/* component "b" of quadratic phase modulation equation */ for (i= l4index; i<= r4index; i++) /* increment iup to right side of 4th freq notch */ { j = i l4index; /* increment j (shifted from i) used in quadratic function */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 10 of excitation SWIFT waveform**************************************/ b = PI/(r5index l5index);/* component "b" of quadratic phase modulation equation */ for (i= l5index; i<= r5index; i++) /* increment iup to right side of 5th freq notch */ { j = i l5index; /* increment j (shifted from i) used in quadratic function */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 12 of excitation SWIFT waveform**************************************/ b = PI/(r6index l6index);/* component "b" of quadratic phase modulation equation */ for (i= l6index; i<= r6index; i++) /* increment iup to right side of 6th freq notch */ { j = i l6index; /* increment j (shifted from i) used in quadratic function */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } } else/* for isolation SWIFT waveform */ {

PAGE 177

177 Figure B-10. Build components 1 through 5 of isolation waveform. /*************** Building component 1 of isolation SWIFT waveform**************************************/ a = PI/2; /* component "a" of quadratic phase modulation equation */ b = PI/l1index; /* component "b" of quadratic phase modulation equation */ for (i= 0; i< l1index; i++) /* increment iup to left side of 1st freq notch */ { f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 1; /* scale magnitude array to unity scale later */ p[i] = (a i) + (b/2 i* i); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 2 of isolation SWIFT waveform**************************************/ for (i= l1index; i<= r1index; i++)/* increment ifrom left side to right side of 1st freq notch */ { f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 0; /* set magnitude to zero */ p[i] = 0; /* set phase to zero */ x[i] = 0; /* set real portion of complex number to zero */ y[i] = 0; /* set imaginary portion of complex number to zero */ } /*************** Building component 3 of isolation SWIFT waveform**************************************/ b = PI/(l2index r1index); /* component "b" of quadratic phase modulation */ for (i= r1index + 1; i<= l2index; i++) /* increment ifrom right of 1st freq notch to left of 2nd freq notch */ { j = i r1index; /* increment j (shifted from i) used in quadratic function */ f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 4 of isolation SWIFT waveform**************************************/ for (i= l2index + 1; i<= r2index; i++)/* increment ifrom left side to right side of 2nd freq notch */ { f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 0; /* set magnitude to zero */ p[i] = 0; /* set phase to zero */ x[i] = 0; /* set real portion of complex number to zero */ y[i] = 0; /* set imaginary portion of complex number to zero */ } /*************** Building component 5 of isolation SWIFT waveform**************************************/ b = PI/(l3index r2index); /* component "b" of quadratic phase modulation */ for (i= r2index + 1; i<= l3index; i++) /* increment ifrom right of 2nd freq notch to left of 3rd freq notch */ { j = i r2index; /* increment j (shifted from i) used in quadratic function */ f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ }

PAGE 178

178 Figure B-11. Build components 6 through 10 of isolation waveform. /*************** Building component 6 of isolation SWIFT waveform**************************************/ for (i= l3index + 1; i<= r3index; i++)/* increment ifrom left side to right side of 3rd freq notch */ { f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 0; /* set magnitude to zero */ p[i] = 0; /* set phase to zero */ x[i] = 0; /* set real portion of complex number to zero */ y[i] = 0; /* set imaginary portion of complex number to zero */ } /*************** Building component 7 of isolation SWIFT waveform**************************************/ b = PI/(l4index r3index); /* component "b" of quadratic phase modulation */ for (i= r3index + 1; i<= l4index; i++) /* increment ifrom right of 3rd freq notch to left of 4th freq notch */ { j = i r3index; /* increment j (shifted from i) used in quadratic function */ f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 8 of isolation SWIFT waveform**************************************/ for (i= l4index + 1; i<= r4index; i++)/* increment ifrom left side to right side of 4th freq notch */ { f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 0; /* set magnitude to zero */ p[i] = 0; /* set phase to zero */ x[i] = 0; /* set real portion of complex number to zero */ y[i] = 0; /* set imaginary portion of complex number to zero */ } /*************** Building component 9 of isolation SWIFT waveform**************************************/ b = PI/(l5index r4index); /* component "b" of quadratic phase modulation */ for (i= r4index + 1; i<= l5index; i++) /* increment ifrom right of 4th freq notch to left of 5th freq notch */ { j = i r4index; /* increment j (shifted from i) used in quadratic function */ f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 10 of isolation SWIFT waveform**************************************/ for (i= l5index + 1; i<= r5index; i++)/* increment ifrom left side to right side of 5th freq notch */ { f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 0; /* set magnitude to zero */ p[i] = 0; /* set phase to zero */ x[i] = 0; /* set real portion of complex number to zero */ y[i] = 0; /* set imaginary portion of complex number to zero */ }

PAGE 179

179 Figure B-12. Build components 11 through 13 of isolation waveform. /*************** Building component 11 of isolation SWIFT waveform**************************************/ b = PI/(l6index r5index); /* component "b" of quadratic phase modulation */ for (i= r5index + 1; i<= l6index; i++) /* increment ifrom right of 5th freq notch to left of 6th freq notch */ { j = i r5index; /* increment j (shifted from i) used in quadratic function */ f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 1; /* set magnitude array to unity scale later */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } /*************** Building component 12 of isolation SWIFT waveform**************************************/ for (i= l6index + 1; i<= r6index; i++)/* increment ifrom left side to right side of 6th freq notch */ { f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 0; /* set magnitude to zero */ p[i] = 0; /* set phase to zero */ x[i] = 0; /* set real portion of complex number to zero */ y[i] = 0; /* set imaginary portion of complex number to zero */ } /*************** Building component 13 of isolation SWIFT waveform**************************************/ b = PI/(N/2 r6index); /* component "b" of quadratic phase modulation */ for (i= r6index + 1; i<= N/2; i++)/* increment ifrom right side of 6th freq notch to N/2 */ { j = i r6index; /* increment j (shifted from i) used in quadratic function */ f[i] = deltaf* i; /* initialize frequency axis array in kHz */ r[i] = 1; /* set magnitude array to 1 */ p[i] = (a j) + (b/2 j j); /* set phase to quadratic function */ x[i] = cos(p[i]); /* set real portion of complex number */ y[i] = sin(p[i]); /* set imaginary portion of complex number */ } }

PAGE 180

180 Figure B-13. Create data array for frequency data. /******************************************************************************************************** Create data array, which will be filled with frequency data. The contents of this array are shown below for f = 100 and N = 8: freal data ptsimaginary data pts 0data[0] = x[0]data[1] = y[0] 12.5data[2] = x[1]data[3] = y[1] 25data[4] = x[2]data[5] = y[2] 37.5data[6] = x[3]data[7] = y[3] 50data[8] = x[4]data[9] = y[4] 37.5data[10] = x[3]data[11] = y[3] 25data[12] = x[2]data[13] = y[2] 12.5data[14] = x[1]data[15] = y[1] From this example, its obvious (hopefully) why the f, r, p, x, and y arrays were dimensioned by N+1 (since we need values from f[0] = 0 to f[N/2 + 1] = freqmax).However, we need only calculate values for the positive frequencies, since we can compute the values for the negative frequencies, which are simply the complex conjugate of the positive portion (complex conjugate of (x + yi) = (x yi)). ********************************************************************************************************/ i= 0; for (j = 0; j <= N/2; j++) { data[i++] = x[j]; data[i++] = y[j]; } for (j = N/2 1; j > 0; j ) { data[i++] = x[j]; data[i++] = y[j]; }

PAGE 181

181 Figure B-14. Inverse Fourie r transform, midpoint time re flection, and apodization. /******************************************************************************************************** PERFORM INVERSE FOURIER TRANSFORM TO COMPUTE TIME DOMAIN WAVEFORM ********************************************************************************************************/ fft(data 1,N, 1); for (i= 0; i< 2*N; i++)/* normalize results */ data[i] = data[i]/N; /******************************************************************************************************** REFLECT TIME DOMAIN SIGNAL ABOUT MIDPOINT ********************************************************************************************************/ /*enclose the SWAP statement in {} since it compiles to multiple statements */ for (i = 0; i < N; i++) { SWAP(data[i], data[N+i]); } /******************************************************************************************************** GENERATE APODIZATION FUNCTION AND SMOOTH TIME DOMAIN WAVEFORM ********************************************************************************************************/ for (i = 0; i < N; i++) { if (i< .25 N) /* set 1st quarter = 1st quarter of sine wave */ apod[i] = sin(2*PI*i/N); else if (i< .75 N) /* set next half = unit */ apod[i] = 1; else/* set last quarter = last quarter of sine wave */ apod[i] = apod[N 1 i]; /* compute from reflection of 1st quarter */ } for (i = j = 0; i < 2*N; i+=2, j++) { data[i] = data[i] apod[j]; data[i+1] = data[i+1] apod[j]; }

PAGE 182

182 Figure B-15. Normalize waveform a nd download to function generator. /******************************************************************************************************** NORMALIZE WAVEFORM AND DOWNLOAD TO FUNCTION GENERATOR ********************************************************************************************************/ max = 0; for (i = 0; i < 2*N; i++) if (fabs(data[i]) > max) max = fabs(data[i]); norm1 = MAX/max; /* calc scaling factor for wave (intdata) */ norm2 = vpp/max; /* calc scaling factor for data (real data) */ checksum = 0; for (i= 0; i< N; i++)/* norm waveform to give max amp */ { wave[i] = (int) (data[2*i] norm1); data[2*i] = data[2*i] norm2; checksum += wave[i]; /* and keep running checksum */ } wave[N] = checksum;/* tack checksum onto end of array */ if (dlflag== 1) { printf("Still trying to figure out how to download using RS232 instead of GPIB any suggestions? ...\n"); }

PAGE 183

183 Figure B-16. Write waveform data to output files. /******************************************************************************************************** WRITE WAVEFORM DATA TO OUTPUT FILES ********************************************************************************************************/ if (fileflag& 1) { printf("Outputting SWIFT data to ASCII file ...\n"); fprintf(ascfp, "\\DATASET 1: frequency vsmagnitude of SWIFT waveform\n"); fprintf(ascfp, "%i%i\n", N/2 + 1, 2); /* print number of rows and columns */ for (i = 0; i <= N/2; i++) fprintf(ascfp, "%f %f\n", f[i], r[i]); fprintf(ascfp, "\\DATASET 2: frequency vsphase of SWIFT waveform\n"); fprintf(ascfp, "%i %i\n", N/2 + 1, 2); for (i = 0; i <= N/2; i++) fprintf(ascfp, "%f %f\n", f[i], p[i]); fprintf(ascfp, "\\DATASET 3: time domain SWIFT waveform (apodizedand normalized)\n"); fprintf(ascfp, "%i %i\n", N, 2); for (i = 0; i < N; i++) fprintf(ascfp, "%f %f\n", t[i], data[2*i]); for (i= 0; i< N; i++)/* normalize waveform to give original amplitude */ data[2*i] = data[2*i]/norm2; fft(data 1,N,1);/* perform forward Fourier transform */ for (i= 0; i<= N/2; i++) /* compute magnitude */ r[i] = sqrt( pow(data[2*i],2) + pow(data[2*i+1],2) ); fprintf(ascfp, "\\DATASET 4: frequency domain SWIFT waveform (zero filled, magmode)\n"); fprintf(ascfp, "%i%i\n", N/2 + 1, 2); /* print number rows and columns */ for (i = 0; i <= N/2; i++) fprintf(ascfp, "%f %f\n", f[i], r[i]); fclose(ascfp); } if (fileflag& 2) { printf("Outputting SWIFT data to binary file ...\n"); fwrite(wave, sizeof(int), N+1, binfp); fclose(binfp); } system("PAUSE"); }

PAGE 184

184 Figure B-17. Fast Four ier transfor m function. /*======================================================================================================= Function:FFT Purpose:This function performs a discrete fast Fourier transform and is adapted fro the algorithm in "Numberical Recipes in C", pp. 411 412. If isignis 1, it replaces data with a Fourier transform. If isignis 1, it replaces data by nntimes its inverse Fourier transform. Data is a complex array of length nninput as a real array data[1..2*nn]. nnmust be an integer power of 2 (this is not checked for). =======================================================================================================*/ void fft(float data[], int nn, int isign) { intn,mmax,m,j,istep,i; double wtemp,wr,wpr,wpi,wi,theta; floattempr,tempi; n=nn<< 1; j=1; for (i=1; i i) { SWAP(data[j],data[i]); SWAP(data[j+1],data[i+1]); } m=n >> 1; while (m >= 2 && j > m) { j = m; m >>= 1; } j += m; } mmax=2; while (n > mmax)/* perform Danielson Lanczostransform */ { istep=2*mmax; theta=6.28318530717959/(isign*mmax); wtemp=sin(0.5*theta); wpr= 2.0*wtemp*wtemp; wpi=sin(theta); wr=1.0; wi=0.0; for (m=1; m<=mmax; m+=2) { for (i=m; i<=n; i+=istep) { j=i+mmax; tempr=wr*data[j] wi*data[j+1]; tempi=wr*data[j+1]+wi*data[j]; data[j]=data[i] tempr; data[j+1]=data[i+1] tempi; data[i] += tempr; data[i+1] += tempi; } wr=(wtemp=wr)*wpr wi*wpi+wr; wi=wi*wpr+wtemp*wpi+wi; } mmax=istep; } }

PAGE 185

185 APPENDIX C LTQ MODIFICATIONS The SWIFT waveform is calculated from the C++ program in Appendix B and stored in an arbitrary waveform generator (AWG) (Stanf ord Research Systems Model DS345, Sunnyvale, CA, USA). The SWIFT output of the AWG is then connected to the LTQ analog printed circuit board (PCB). The LTQ analog PCB contains a pair of AD734 multiplier/divider microchips shown in Figure C-1. Chip U46 multiplies the LTQ isolation waveform signal (X1 – X2) by the isolation waveform gain (Y1 – Y2). This product is then divided by the denominator interface (U1 – U2) and then the resu lts of chip U64 (Z1 – Z2) are subtracted from this quotient. Chip U64 multiplies the resonant ejection/excitation (Res Ej/Ex) waveform (X1 – X2) by the Res Ej/Ex waveform gain (Y1 – Y2). This product is then divided by the denominator interface (U1 – U2). The resulting quotient is subtracted by (Z1 – Z2), which is usually set to ground. Z2 corresponds to pin 10 of chip U64. The pin was lifted from the PCB and was wired to the center contact of the BNC (Bayonet Neill-Concelman) cable connect ed to the output of the AWG (Figure C-2). The grounding sheath of the AWG BNC cable was wired to a ground pin on the analog PCB. When SWIFT is not being applied to the LTQ, a BNC grounding cap is placed on the BNC connector wired to pin 10 of chip U64. The LTQ software has a programmable trigger th at allows an external waveform to be triggered during a designated location of the sc an function. The location of the programmable trigger in the LTQ software is in the Diagnostics menu underneath the Tools list. By clicking on Triggers, the window shown in Figure C-3 appears, which allo ws the user to input two arguments (ARG1 and ARG2), which defines th e ion trap control la nguage (ITCL) trigger function, trig(ARG1, ARG2). ARG1 is the trigger location type (e .g., -1 = all triggers off, 0 = beginning of scan, 1 = injection period, 2 = is olation, 3 = activation, a nd 4 = scan out), and

PAGE 186

186 ARG2 is the nth position of that type. If no ARG1 or ARG2 is given, the trigger is set at the start of the analytical scan. If no ARG2 is given, it is set to 0 (the first position of that location type). The available values for ARG2 will depend on the mode and the trigger location type (Figure C3). If ARG2 = -1, then triggers are on for all pe riods of that trigger loca tion type. Figure C-4 shows a diagram of the mass spectrometer scan func tion with the different locations labeled with the corresponding trigger functions values. The programmable trigger was accessed by connec ting a wire to pin 14 of the J1 connector of the LTQ digital PCB (Figure C-5). The othe r end of the wire was connected to the center contact of the BNC cable leading to the trigge r input on the back of the AWG. The grounding sheath of the AWG BNC cable was wire d to a ground pin on the digital PCB. The locations of the programmable trigger we re tested by using a two-channel digitizing oscilloscope with screen capture capability (Model TDS 540, Tektroni x Inc., Beaverton, OR, USA). The trigger pulse from the LTQ dig ital PCB was connected to channel 1 of the oscilloscope, and the waveform output after amp lification was connected to channel 2 of the oscilloscope through an RCA (R adio Corporation of America) connector on the LTQ analog PCB. Figure C-6 shows the oscilloscope image of channels 1 and 2 with the trigger set to the beginning of the scan [trig(0,0)] with automatic ga in control (AGC) on. No te that there are two different triggers shown for channel 1, one duri ng the beginning of the pr escan and the other one during the beginning of the analytical scan. Th e end of the prescan is indicated by the large AGC signal from channel 2, where the radiofrequency (RF) voltage is ramped up to scan out the trapped ions from the ion trap. Channel 2 also shows two signals, one for the prescan and one for the analytical scan, that correspond to the LTQ isolation (ISO) waveform.

PAGE 187

187 Figure C-7 shows the oscilloscope image of ch annels 1 and 2 with th e trigger set to the injection period [trig(1,0)] a nd AGC on. Channel 1 shows no visible trigger pulses, which indicate that the programmable trigger is not sending a measur able trigger pulse during the injection period highlighted in Figure C-4. Cha nnel 2 however does indicate the presence of the prescan and analytical scan each containing an isolation waveform signal and a scan out signal. Figure C-8 shows the oscilloscope image of ch annels 1 and 2 with th e trigger set to the isolation period [trig(2,0)] and AGC on. Channel 1 shows visible tr igger pulses that line up with the isolation waveform signals of the prescan and analytical scan shown from channel 2. Before the SWIFT isolation waveform can be triggered duri ng the isolation period, it is important to turn off the LTQ isolation waveform to avoid interfer ence. Under the Diagnostics Tools menu in the LTQ software, Toggles can be selected (Figure C-9). Then the Isolation waveform can be highlighted and the off or on radi o buttons selected before the Set button is pressed. It is important to note that if the Define Scan wi ndow is opened in the LTQ software, the toggles return to their default factory setti ngs (i.e., LTQ isolation waveform on). Figure C-10 shows the mass spectra of cocaine (COC) with the isolation waveform toggled on (Figure C-10a) and off (Figure C-10b). COC standard was spotted (1 L, 1 ng/L) onto a MALDI plate and airbrushed w ith 2,5-dihydroxybenzoic acid as the matrix. The [M+H]+ ion of cocaine at m/z 304 was isolated with a 1-Da isolation window centered at m/z 304. With the LTQ isolation waveform on (Figure C-10a), the background ions are ejected leaving a visible [COC+H]+ ion at m/z 304; however, with the LTQ isolati on waveform turned off (Figure C-10b) the background ions overwhelm the m/z 304 ion and it is not visi ble. The low mass cutoff (LMCO) is visible with the LTQ isolation wa veform turned off (Figure C-10b), which is calculated to be m/z 277 based on m/z 304 being isolated at a q of 0.83, [LMCO =

PAGE 188

188 (0.83)(304)/(0.91) = 277]. Figure C-11 shows the oscilloscope images of channel 1 and 2 with the trigger at isolation, AGC on, and the LTQ is olation waveform toggled on (Figure C-11a) and off (Figure C-11b). Channel 2 of Figure C-11b clearly shows that the isolation waveforms disappear from the prescan and the analytical scan when the LTQ isolation waveform is toggled off. Figure C-12 shows the oscilloscope image of channels 1 and 2 with the trigger set at activation [trig(3,0)] and AGC on. Channel 1 shows that activati on occurs between the isolation event and the scan out event of both the prescan and the analytical scan. Figure C-13 shows the oscilloscope image of channels 1 and 2 with th e trigger set at scan out [trig(4,0)] and AGC on. Channel 1 shows only a single tr igger pulse during the scan out of the analytical scan.

PAGE 189

189 Figure C-1. Adding SWIFT waveform to AD734 chip (U64) on LTQ Analog PCB. AD734 Multiplier/Divider Chips Isolation Waveform Isolation Waveform Gain Waveform Output Waveform Output Res Ej/Ex Waveform Res Ej/Ex Waveform Gain 1K WW 2.49K SWIFT Use BNC grounding cap to ground Pin 10 when SWIFT is not applied.

PAGE 190

190 Figure C-2. Analog PCB modifica tion to apply SWIFT waveform. Add SWIFT to Pin 10 of Chip U64 on Analog PCB. Ground

PAGE 191

191 Figure C-3. LTQ progr ammable trigger.

PAGE 192

192 Figure C-4. Programmabl e trigger locations. Time (ms) Axial Modulation Collision Induced Dissociation (CID) Waveform Broadband Waveform Pre Isolation Waveform Radio Frequency (RF) Voltage Mass Analysis Scan Out –Trig(4,0) Activation Trig(3,0) Isolation –Trig(2,0) Beginning of Scan –Trig(0,0) Injection Period –Trig(1,0)

PAGE 193

193 Figure C-5. Digital PCB modifications to access program mable trigger. Programmable trigger on Pin 14 of J1 on the Digital PCB.

PAGE 194

194 Figure C-6. Trigger at beginning of scan. Trigger at beginning of scan; trig (0,0); AGC = on. prescananalytical scan 1sttrigger 2ndtriggerTrigger at Beginning of ScanTrigger RCA AGC ISO ISO Scan Out

PAGE 195

195 Figure C-7. Trigge r at injection. Trigger at injection period; trig (1,0); AGC = on. no visible trigger pulses on scopeTrigger at InjectionTrigger RCA ISO AGC ISO Scan Out

PAGE 196

196 Figure C-8. Trigge r at isolation. Trigger at isolation; trig (2,0); AGC = on.Trigger at IsolationTrigger RCA 1sttrigger 2ndtrigger ISO ISO Scan Out AGC

PAGE 197

197 Figure C-9. Toggling LTQ isolat ion waveform on and off.

PAGE 198

198 Figure C-10. Mass spectra of cocaine with isol ation waveform toggled: (a) on and (b) off. 100 150 200 250 300 350 m/z 0 20 40 60 80 100 0 20 40 60 80 100 Relative Abundance 304.17 321.99 286.25339.65 272.25 317.36 335.40 289.45 282.46 190.40 172.43 146.39 212.23 232.52 123.37 107.25 Isolation Waveform Off Isolation Waveform On LMCO NL: 5.37E4 NL: 9.83E5 [COC+H]+(a) (b)

PAGE 199

199 Figure C-11. Trigger at is olation with isolation toggl ed: (a) on and (b) off. Isolation on Isolation off Trigger Trigger RCA RCA Scan Out Scan Out ISO ISO AGC AGC Isolation disappears from RCA output. 1sttrigger 1sttrigger 2ndtrigger 2ndtrigger(a) (b)

PAGE 200

200 Figure C-12. Trigge r at activation. Trigger at activation; trig (3,0); AGC = on.Trigger at ActivationTrigger RCA 1sttrigger 2ndtrigger ISO ISO AGC Scan Out

PAGE 201

201 Figure C-13. Trigger at scan out. Trigger at scan out; trig (4,0); AGC = on.Trigger at Scan OutScan Out Trigger Trigger RCA ISOISO AGC

PAGE 202

202 LIST OF REFERENCES (1) Karch, S. B. Karch's Pathology of Drug Abuse 3rd ed.; CRC Press LLC: Boca Raton, 2002. (2) Substance Abuse and Mental Health Servi ces Administration, Office of Applied Studies Drug Abuse Warning Network, 2007: Area Profiles of Drug-Related Mortality. HHS Publication No. SMA 09-4407, DAWN Series D-31. Rockville, MD, 2009. (3) Human Illnesses and Behavioral Healt h. http://www.humanillnesses.com (accessed 30 June 2008). (4) Drummer, O. H.; Gerstamoulos, J. Therap. Drug Monitor. 2002 24 199-209. (5) Stimpfl, T.; Reichel, S. Forensic Sci. Int. 2007 170 179-182. (6) Moriya, F.; Hashimoto, Y. J. Forensic Sci. 1996 41 612-616. (7) Moriya, F.; Hashimoto, Y. J. Forensic Sci. 1996 41 129-133. (8) Spiehler, V. R.; Reed, D. J. Forensic Sci. 1985 30 1003-1011. (9) Giroud, C.; Michaud, K.; Sporkert, F.; Eap, C.; Augsburger, M.; Cardinal, P.; Mangin, P. J. Anal. Toxicol. 2004 28 464-474. (10) In National Institute on Drug Abuse Research Monograph 163 ; Majewska, M. D., Ed., 1996, pp 1-340. (11) Ritz, M. C.; Lamb, R. T.; Goldberg, S. R.; Kuhar, M. J. Science 1987 237 1219-1223. (12) Kalasinsky, K. S.; Bosy, T. Z.; Schmunk, G. A.; Ang, L.; Adams, V.; Gore, S. B.; Smialek, J.; Furukawa, Y.; Guttman, M.; Kish, S. J. J. Forensic Sci. 2000 45 1041-1048. (13) Alburges, M. E.; Crouch, D. J. ; Andrenyak, D. M.; Wamsley, J. K. Neurochem. Int. 1996 28 51-57. (14) James, C. A.; Breda, M.; Baratte, S.; Ca sati, M.; Grassi, S.; Pellegatta, B.; Sarati, S.; Frigerio, E. Chromatographia 2004 59 S149-S156. (15) Kusumoto, M.; Ikeda, K.; Nishiya, Y.; Kawamura, Y. Anal. Biochem. 2001 294 185186. (16) Schiffer, W. K.; Liebling, C. N. B.; Patel, V.; Dewey, S. L. Nucl. Med. Biol. 2007 34 833-847. (17) Gatley, S. J.; Volkow, N. D. Drug Alcohol Depend. 1998 51 97-108.

PAGE 203

203 (18) Sosnovik, D. E.; Weissleder, R. Curr. Opin. Biotechnol. 2007 18 4-10. (19) Wang, G.; Yu, H.; De Man, B. Med. Phys. 2008 35 1051-1064. (20) Gumbleton, M.; Stephens, D. J. Adv. Drug Delivery Rev. 2005 57 5-15. (21) Contag, C. H.; Ross, B. D. J. Magn. Reson. Imaging 2002 16 378-387. (22) Denoyer, A.; Ossant, F.; Arbeille, B.; Fetisso f, F.; Patat, F.; Pour celot, L.; Pisella, P.-J. Ophthalmic Res. 2008 40 298-308. (23) Som, P.; Oster, Z. H.; Wang, G. -J.; Volkow, N. D.; Sacker, D. F. Life Sci. 1994 55 1375-1382. (24) Garidel, P.; Boese, M. Microsc. Res. Tech. 2007 70 336-349. (25) Mahmoudi, M.; Simchi, A.; Imani, M.; Hafeli, U. O. J. Phys. Chem. C 2009 113 81248131. (26) Niu, G.; Chen, X. Drugs R&D 2008 9 351-368. (27) Chaurand, P.; Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. Toxicol. Pathol. 2005 33 92-101. (28) Pacholski, M. L.; Winograd, N. Chem. Rev. 1999 99 2977-3005. (29) Caprioli, R. M.; Farmer, T. B.; Gile, J. Anal. Chem. 1997 69 4751-4760. (30) Troendle, F. J.; Reddick, C. D.; Yost, R. A. J. Am. Soc. Mass Spectrom. 1999 10 13151321. (31) Reyzer, M. L.; Hsieh, Y.; Ng, K.; Korfmacher, W. A.; Caprioli, R. M. J. Mass Spectrom. 2003 38 1081-1092. (32) Bunch, J.; Clench, M. R.; Richards, D. S. Rapid Commun. Mass Spectrom. 2004 18 3051-3060. (33) Wang, H.-Y. J.; Jackson, S. N.; McEuen, J.; Woods, A. S. Anal. Chem. 2005 77 66826686. (34) Cristoni, S.; Brioschi, M.; Rizzi, A.; Sironi, L.; Gelosa, P.; Tremoli, E.; Bernardi, L. R.; Banfi, C. Rapid Commun. Mass Spectrom. 2006 20 3483-3487. (35) Crossman, L.; McHugh, N. A.; Hsieh, Y.; Korfmacher, W. A.; Chen, J. Rapid Commun. Mass Spectrom. 2006 20 284-290.

PAGE 204

204 (36) Hsieh, Y.; Casale, R.; Fukuda, E.; Chen, J.; Knemeyer, I.; Wingate, J.; Morrison, R.; Korfmacher, W. Rapid Commun. Mass Spectrom. 2006 20 965-972. (37) Khatib-Shahidi, S.; Andersson, M.; Herman J. L.; Gillespie, T. A.; Caprioli, R. M. Anal. Chem. 2006 78 6448-6456. (38) Drexler, D. M.; Garrett, T. J.; Cantone, J. L.; Diters, R. W.; Mitroka, J. G.; Prieto Conaway, M. C.; Adams, S. P.; Yost, R. A.; Sanders, M. J. Pharmacol. Toxicol. Methods 2007 55 279-288. (39) Hsieh, Y.; Chen, J.; Korfmacher, W. A. J. Pharmacol. Toxicol. Methods 2007 55 193200. (40) Stoeckli, M.; Staab, D.; Schweitzer, A. Int. J. Mass Spectrom. 2007 260 195-202. (41) Chen, J.; Hsieh, Y.; Knemeyer, I.; Crossman, L.; Korfmacher, W. A. Drug Metab. Lett. 2008 2 1-4. (42) Cornett, D. S.; Frappier, S. L.; Caprioli, R. M. Anal. Chem. 2008 80 5648-5653. (43) Hopfgartner, G.; Varesio, E.; Stoeckli, M. Rapid Commun. Mass Spectrom. 2009 23 733-736. (44) Li, F.; Hsieh, Y.; Kang, L.; Sondey, C.; Lachowicz, J.; Korfmacher, W. A. Bioanalysis 2009 1 299-307. (45) Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Science 2004 306 471-473. (46) Nemes, P.; Vertes, A. Anal. Chem. 2007 79 8098-8106. (47) Cooks, R. G.; Ouyang, Z.; Takats, Z.; Wiseman, J. M. Science 2006 311 1566-1570. (48) Zenobi, R.; Knochenmuss, R. Mass Spectrom. Rev. 1998 17, 337-366. (49) Niu, S.; Zhang, W.; Chait, B. T. J. Am. Soc. Mass Spectrom. 1998 9 1-7. (50) Berkenkamp, S.; Menzel, C. ; Karas, M.; Hillenkamp, F. Rapid Commun. Mass Spectrom. 1997 11 1399-1406. (51) Zhang, W.; Niu, S.; Chait, B. T. J. Am. Soc. Mass Spectrom. 1998 9 879-884. (52) Dreisewerd, K. Chem. Rev. 2003 103 395-425. (53) Schriver, K. E.; Chaurand, P.; Caprioli, R. M. Proceedings of the 51st ASMS Conference on Mass Spectrometry and Allied Topics : Montreal, Canada, 2003.

PAGE 205

205 (54) Jurchen, J. C.; Rubakhin, S. S.; Sweedler, J. V. J. Am. Soc. Mass Spectrom. 2005 16 1654-1659. (55) Cohen, S. L.; Chait, B. T. Anal. Chem. 1996 68 31-37. (56) Nakanishi, T.; Ohtsu, I.; Fu ruta, M.; Ando, E.; Nishimura, O. J. Proteome Res. 2005 4 743-747. (57) Baluya, D. L.; Garrett, T. J.; Yost, R. A. Anal. Chem. 2007 79 6862-6867. (58) Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. J. Mass Spectrom. 2003 38 699-708. (59) Landgraf, R. R., Analysis of Lipi ds in Nerve Tissue by MALDI Tandem Mass Spectrometric Imaging. Ph.D. Dissertation, Univ ersity of Florida, Gainesville, FL, 2009. (60) Garrett, T. J.; Yost, R. A. Anal. Chem. 2006 78 2465-2469. (61) Ayorinde, F. O.; Hambright, P. ; Porter, T. N.; Keith, Q. L., Jr. Rapid Commun. Mass Spectrom. 1999 13 2474-2479. (62) Cohen, L. H.; Gusev, A. I. Anal. Bioanal. Chem. 2002 373 571-586. (63) Karas, M.; Kruger, R. Chem. Rev. 2003 103 427-439. (64) Chapman, J. R. Practical Organic Mass Spectromet ry: A Guide for Chemical and Biochemical Analysis 2nd ed.; John Wiley and Sons: New York, 1998. (65) Hensel, R. R.; King, R. C.; Owens, K. G. Rapid Commun. Mass Spectrom. 1997 11 1785-1793. (66) Nicola, A. J.; Gusev, A. I.; Proctor, A.; Jackson, E. K.; Hercules, D. M. Rapid Commun. Mass Spectrom. 1995 9 1164-1171. (67) Garrett, T. J.; Prieto-Conaway, M. C.; K ovtoun, V.; Bui, H.; Izgarian, N.; Stafford, G.; Yost, R. A. Int. J. Mass Spectrom. 2007 260 166-176. (68) Sugiura, Y.; Shimma, S.; Setou, M. Anal. Chem. 2006 78 8227-8235. (69) Hankin, J. A.; Barkley, R. M.; Murphy, R. C. J. Am. Soc. Mass Spectrom. 2007 18 1646-1652. (70) Aerni, H.-R.; Cornett, D. S.; Caprioli, R. M. Anal. Chem. 2006 78 827-834. (71) Puolitaival, S. M.; Burnum, K. E. ; Cornett, D. S.; Caprioli, R. M. J. Am. Soc. Mass Spectrom. 2008 19 882-886.

PAGE 206

206 (72) Atkinson, S. J.; Loadman, P. M.; Sutton, C.; Patterson, L. H.; Clench, M. R. Rapid Commun. Mass Spectrom. 2007 21 1271-1276. (73) Trim, P. J.; Henson, C. M.; Avery, J. L.; McEwen, A.; Snel, M. F.; Claude, E.; Marshall, P. S.; West, A.; Princivalle, A. P.; Clench, M. R. Anal. Chem. 2008 80 8628-8634. (74) Luxembourg, S. L.; McDonnell, L. A.; D uursma, M. C.; Guo, X.; Heeren, R. M. A. Anal. Chem. 2003 75 2333-2341. (75) Lemaire, R.; Tabet, J.; Ducoroy, P.; Hendra, J. B.; Salzet, M.; Fournier, I. Anal. Chem. 2006 78 809-819. (76) Onnerfjord, P.; Ekstrom, S.; Bergquist, J. ; Nilsson, J.; Laurell, T.; Marko-Varga, G. Rapid Commun. Mass Spectrom. 1999 13 315-322. (77) Sleno, L.; Volmer, D. A. Rapid Commun. Mass Spectrom. 2006 20 1517-1524. (78) Nicola, A. J.; Gusev, A. I.; Hercules, D. M. Appl. Spectrosc. 1996 50 1479-1482. (79) Ling, Y.-C.; Lin, L.; Chen, Y.-T. Rapid Commun. Mass Spectrom. 1998 12 317-327. (80) Hatsis, P.; Brombacher, S.; Co rr, J.; Kovarik, P.; Volmer, D. A. Rapid Commun. Mass Spectrom. 2003 17 2303-2309. (81) Cui, M.; McCooeye, M. A.; Fraser, C.; Mester, Z. Anal. Chem. 2004 76 7143-7148. (82) Gusev, A. I.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M. Fresenius. J. Anal. Chem. 1996 354 455-463. (83) Kang, M.-J.; Tholey, A.; Heinzle, E. Rapid Commun. Mass Spectrom. 2001 15 13271333. (84) Krutchinsky, A. N.; Chait, B. T. J. Am. Soc. Mass Spectrom. 2002 129-134. (85) FinniganTM vMALDI Source Hardware Manual ; Thermo Electron Corporation: San Jose, CA, 2003. (86) Paul, W.; Steinwedel, H. Apparatus fo r Separating Charged Particles of Different Specific Charges. U.S. Patent 2,939,952, June 7, 1960. (87) Schwartz, J. C.; Senko, M. W.; Syka, J. E. P. J. Am. Soc. Mass Spectrom. 2002 13 659669. (88) Stafford, G. C.; Kelley, P. E.; Syka, J. E. P.; Rey nolds, W. E.; Todd, J. F. J. Int. J. Mass Spectrom. Ion Processes 1984 60 85-98.

PAGE 207

207 (89) March, R. E.; Todd, J. F. J., Eds. Quadrupole Ion Trap Mass Spectrometry 2nd ed.; John Wiley & Sons, Inc.: Hoboken, New Jersey, 2005. (90) Douglas, D. J.; Frank, A. J.; Mao, D. Mass Spectrom. Rev. 2005 24 1-29. (91) Cox, K. A.; Cleven, C. D.; Cooks, R. G. Int. J. Mass Spectrom. Ion Processes 1995 144 47-65. (92) Cole, R., Ed. Electrospray Ionization Mass Spectrometry ; John Wiley & Sons, Inc.: New York, New York, 1997. (93) Stafford, G.; Taylor, D.; Bradshaw, S.; Syka, J., Proceedings of the 35th ASMS Conference on Mass Spectrometry and Allied Topics: Denver, CO, 1987. (94) FinniganTM vMALDI Getting Started ; Thermo Electron Corpor ation: San Jose, CA, 2004. (95) Kishore, M.; Ghosh, P. Int. J. Mass Spectrom. Ion Processes 1979 29 345-350. (96) March, R.; Todd, T., Eds. Practical Aspects of Ion Trap Mass Spectrometry ; CRC Press: New York, NY, 1995; Vol. 1. (97) Bier, M.; Schwartz, J.; Z hou, J.; Taylor, D.; Syka, J.; James, M.; Fies, W.; Stafford, G., Proceedings of the 43rd ASMS Conference on Mass Spectrometry and Allied Topics: Atlanta, GA, 1995. (98) Browne, S. P.; Moore, C. M.; Scheurer, J.; Tebbett, I. R.; Logan, B. K. J. Forensic Sci. 1991 36 1662-1665. (99) Srinivasan, K.; Wang, P.; Eley, A. T.; White, C. A.; Bartlett, M. G. J. Chromatogr. B 2000 745 287-303. (100) Sershen, H.; Reith, M. E. A.; Lajtha, A. Neuropharmacology 1980 19 1145-1148. (101) Cognard, E.; Bouchonnet, S.; Staub, C. J. Pharm. Biomed. Anal. 2006 41 925-934. (102) Ferrer, I.; Furlong, E. T. Environ. Sci. Technol. 2001 35 2583-2588. (103) Reich, R. F.; Cudzilo, K.; Levisky, J. A.; Yost, R. A. J. Am. Soc. Mass Spectrom. 2010 21 564-571. (104) Marshall, A. G.; Wang, T. C. L.; Ricca, T. L. J. Am. Chem. Soc. 1985 107 7893-7897. (105) Guan, S.; Marshall, A. G. Anal. Chem. 1993 65 1288-1294. (106) Julian, R. K., Jr.; Cooks, R. G. Anal. Chem. 1993 65 1827-1833.

PAGE 208

208 (107) Soni, M. H.; Cooks, R. G. Anal. Chem. 1994 66 2488-2496. (108) March, R. E. J. Mass Spectrom. 1997 32 351-369. (109) Embree, P. M.; Kimbel, B. C. Language Algorithms for Digital Signal Processing ; Prentice Hall: Englewood Cliffs, NJ, 1991. (110) Chen, L.; Wang, T. C. L.; Ricca, T. L.; Marshall, A. G. Anal. Chem. 1987 59 449-454. (111) Wang, T. C. L.; Ricca, T. L.; Marshall, A. G. Anal. Chem. 1985 107 7893-7897. (112) Guan, S.; Marshall, A. G. Int. J. Mass Spectrom. Ion Processes 1996 157/158 5-37. (113) Kelley, P. E. Mass Spectrometry Method Using Notch Filter. U.S. Patent 5,134,286, July 28, 1992. (114) Nappi, M.; Frankevich, V. ; Soni, M. H.; Cooks, R. G. Int. J. Mass Spectrom. 1998 177 91-104. (115) Popoulis, A. The Fourier Integral and Its Applications ; McGraw-Hill: New York, 1962. (116) Williams, J. D.; Cox, K. A.; Cooks, R. G. Anal. Chem. 1994 66 725-729.

PAGE 209

209 BIOGRAPHICAL SKETCH Richard Fred Reich, Jr., graduated from the University of Arizona in December, 1995, with a B.S. in chemistry and a B.A. in German, and was commissioned in the United States Air Force. Richard has served 14 years activ e duty in the Air Force as a chemical research officer. At his first assignment, he served as an explosives chemist at the Energe tic Materials Branch, Munitions Directorate, Air Force Research La boratory at Eglin Air Force Base in Fort Walton Beach, Florida, from June 1996 – July 1999. Richard was the principal investigator of melt-cast energetics and Deput y Program Manager for the in-house High Energy Explosives Development Program. He was resp onsible for formulating and characterizing the sensitivity and performance of elev en explosive formulations to qua lify their use in advanced Air Force munitions. Richard was competitively selected by the Air Force to attend the Air Force Institute of Technology (AFIT) at the Universi ty of Florida in August, 1999, to pursue a Master’s degree in analytical chemistry. His graduate thesis i nvolved the trace detection of explosives using atmospheric pressure chemical ionization tandem mass spectrometry (APCI-MS/MS). He graduated from the University of Florida in Febr uary, 2001, with an M.S. in analytical chemistry. Richard was then assigned to the Combusti on Branch, Propulsion Directorate, Air Force Research Laboratory at Wright-P atterson Air Force Base in Da yton, Ohio. From February 2001 to June 2004, he served as the Deputy Branch Chief of the Combustion Branch in which he assisted the Branch Chief in leading 25 scientists and engineers in the de velopment of state-ofthe-art combustor technology for legacy and future Air Force turbine engines. He also served as

PAGE 210

210 the principal chemist responsible for characterization of part iculate matter from research combustors and development of fuel addi tives designed to mitigate soot production. From June, 2004, to July, 2007, Richard was assi gned to the Department of Chemistry at the U.S. Air Force Academy in Colorado Spri ngs, Colorado. As Assistant Professor of chemistry, he taught general chemistry, analytical chemistry, and chemistry of weapons. He was the course director for analyti cal chemistry and chemistry of weapons. He also served as the executive officer of the Department of Chemis try, assisting the Department Head, Colonel Van Valkenburg, in leading a department of 65 faculty and staff. In 2007, Richard was competitively selected to a ttend AFIT at the University of Florida to pursue his PhD in analytical chemistry w ith a completion date of August 2010. Upon graduation, Richard will be assigne d to the Air Force Technical A pplications Center (AFTAC) at Patrick AFB, Florida, as the De puty Chief of the Verification Sc iences Division. There he will be responsible for establishing a new radiochemi stry effluent laborator y. Richard is then scheduled to return to the U.S. Air Force Acad emy to teach chemistry in 2013. Richard plans to retire from the Air Force in 2016 after 20 years of service.