<%BANNER%>

Imaging Small Molecules in Tissue by Matrix-Assisted Laser Desorption/Ionization Tandem Mass Spectrometry


PAGE 1

IMAGING SMALL MOLECULES IN TISSU E BY MATRIX-ASSISTED LASER DESORPTION/IONIZATION TA NDEM MASS SPECTROMETRY By TIMOTHY JAMES GARRETT 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 2006

PAGE 2

This document is dedicated to my wife.

PAGE 3

ACKNOWLEDGMENTS Obtaining a graduate degree is not an individual effort, although it shows the efforts of the individual. I thank my parents for providing me with the opportunities to discover the direction my life should take and allowing me to learn fr om my mistakes. I thank my father for teaching me the subtleties of scientific research and guiding me down this path. I have to thank my advisor, Rick, for t eaching me about research and providing an environment in which thoughts and ideas are al lowed to guide the graduate experience. The guidance he has provided in not only science, but also in life, has enabled me to be a valuable member of society. I also have to thank the members of the Yost group for thoughtful discussions of scien ce, research, and life. I thank the city of Gainesville for growing on me and letting me enjoy this community. I must also thank God for providing me w ith guidance and friendships that have given me insight into my ow n life and my research. I must thank Mari, Viatcheslav, Huy, and George at Thermo for allowing me to visit and conduct research at their facility. Working through the difficult aspects of collaborating across country and dealing with the frustrations of instrument developent, their dedication to our efforts has made this re search possible. I must also thank Thermo Electron corporation for their hospitality during each of my research visits. iii

PAGE 4

Finally, I have to thank my wife for helping me through the difficult times of research and giving me the strength to con tinue in this venture of our lives. Her guidance, love, and support have made me who I am today. iv

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................iii LIST OF TABLES ............................................................................................................vii LIST OF FIGURES .........................................................................................................viii ABSTRACT ....................................................................................................................xvii CHAPTER 1 INTRODUCTION........................................................................................................1 Quadrupole Ion Trap Mass Spectrometry .....................................................................2 From 3-D to 2-D Ion Traps ...................................................................................2 3D and 2D Quadrupole Ion Trap (QIT) Theory and Operation............................6 2D/Linear Ion Trap Description ..........................................................................11 Tissue Imaging by Mass Spectrometry .......................................................................15 Background ..........................................................................................................15 Fundamentals .......................................................................................................17 Matrix-Assisted Laser Desorption/Ionization.............................................................24 Background ..........................................................................................................24 Theory of MALDI ...............................................................................................26 Phospholipids and Sphingolipids ................................................................................34 The Structure of Phospholipids ...........................................................................34 Analysis of Lipids by Mass Spectrometry ...........................................................40 Overview of dissertation .............................................................................................51 2 ANALYSIS OF INTACT TISSUE BY INTERMEDIATE-PRESSURE MALDI ON A LINEAR ION TRAP MASS SPECTROMETER............................................52 Introduction .................................................................................................................52 Experimental ...............................................................................................................54 Results and Discussion ...............................................................................................55 Conclusions .................................................................................................................61 3 CHARACTERIZATION OF PROTONATED PHOSPHOLIPIDS AS FRAGILE IONS IN ION TRAP TANDEM MASS SPECTROMETRY....................................63 v

PAGE 6

Introduction .................................................................................................................63 Experimental ...............................................................................................................66 Results and Discussion ...............................................................................................70 Fragility in the trap ..............................................................................................71 Fragility in ion transport ......................................................................................78 Tube lens offset ...................................................................................................82 Initial Studies of Ion Fragili ty in the 2D/Linear Ion Trap ..........................................87 IP-MALDI ...........................................................................................................87 Mass analysis fragility .........................................................................................88 Conclusions .................................................................................................................92 4 IMAGING PHOSPHOLIPIDS IN BRAIN TISSUE BY INTERMEDIATEPRESSURE MALDI ON A LINEAR ION TRAP MASS SPECTROMETER.........95 Introduction .................................................................................................................95 Experimental ...............................................................................................................97 Acquisition of position-specific mass spectra ..................................................101 Results and Discussion .............................................................................................101 Identifying different feat ures of the rat brain. ...................................................119 Quantitation. ......................................................................................................123 Conclusions ...............................................................................................................125 5 CONCLUSION AND FUTURE WORK.................................................................129 LIST OF REFERENCES .................................................................................................135 BIOGRAPHICAL SKETCH ...........................................................................................148 vi

PAGE 7

LIST OF TABLES Table page 1-1. A list of the common fatty acids found in nature. .....................................................38 1-2. The common reactions that occur in chemical ionization when methane, CH4, is used as the reagent gas. ............................................................................................44 2-1. Ions detected from spinal cord in Figure 2A that were chosen for MSn analysis. Only the MS2 major fragment ion m/z is shown in the table, but MS3 was also performed. ................................................................................................................60 3-1. The %CID needed to cause a reduction in the absolute parent ion signal by 50% is shown for each ion studied in the QIT. These results show that the protonated species fragments more readily than the sodiated species, further proof that the protonated ion is fragile. ...........................................................................................77 3-2. A summary of the results from each e xperiment performed in the studies of ion fragility in ion transport. From these resu lts, it was determined that all sodiated species were less susceptible to fragmentation that can occur before mass analysis. ....................................................................................................................87 4-1. A comparison of literature data from a lipid extract of a rat brain to MS data collected directly from a tissue section ..................................................................124 vii

PAGE 8

LIST OF FIGURES Figure page 1-1. A schematic representation of the 3-di mensional (3D) quadrupole ion trap (QIT). The 3D QIT consists of 2 end-cap elect rodes and a ring electrode. The inside dimensions of the 3D QIT are defined by r, the radius of the ring electrode, and z, the distance from the center of the QI T to the apex of the end-cap. For a symmetrical QIT, r is 2 times z. Ions are injected and ejected through holes drilled into each end-cap. ...........................................................................................4 1-2. Diagram of a 2-dimensional (2D) ion trap This 2D ion trap design incorporates three different sections, front and back sections of equal length, and a center section where the ions are trapped. ............................................................................7 1-3. A small portion of the Mathieu stability diagram. Solutions to Equations 1-2 and 1-3 give coordinates in az and qz space that can be mapped onto the above diagram. If the coordinates for an ion fall into the region shown above, then the ion will have a stable trajectory inside the 3D or 2D trap and will be successfully trapped. In normal operation, the DC applied is zero and thus the az is also zero since it is related to the DC voltage a pplied, thus the most important number is the value for qz. The value of qz=0.908 is the edge of the stability diagram. An ion with a qz greater than this value is not trapped. ....................................................9 1-4. A Schematic representation of the 2D ion trap. Ions enter from the left side of the figure. The 2D trap is separated in to 3 sections, two s horter sections of 12 mm in length, and a longer, center, section 37 mm long. Separate DC voltages are applied to all three sections in orde r to trap ions. I ons are ejec ted through slits in the x-rod pairs of the center section. .............................................................13 1-5. Two views of the 2D ion trap looking down the z-axis, along the flight path of the ions. A shows the radial quadrupolar trapping field. The phases of RF are applied to all the electrodes of the ion trap to create the quadrupolar field. B shows the radial dipolar ex citation arrangement. An AC voltage is applied only to the x-rod pairs. The AC voltage is used for isolation and collision-induced dissociation experiments as well as in the ej ection of ions from the trap. Ions are ejected out of slits cut into the x-rod pairs. ..............................................................14 1-6. The process of creating an image by ma ss spectrometry. The two most important stages in tissue analysis by mass spectrometry are 1) cutti ng the tissue to an appropriate thickness with a cryotome, such as 10 m for brain tissue, and 2) viii

PAGE 9

applying the MALDI matrix in a manner th at reduces the possi bility for analyte migration (airbrushing is shown above). When cutting the tissue, it is important to ensure that no OCT compound is on the ti ssue sections as it can interfere with mass analysis. The mass spectrometric stage involves rastering the sample plate at a predefined step size w ith respect to the laser. Each mass spectrum is then position specific so all that is required for generating an image is extracting the m/z value of interest. ................................................................................................19 1-7. The types of sampling used in imagi ng MS. Sampling at the laser resolution is the most widely used data collection scheme because it provides the highest resolution. Under-sampling will take le ss time for acquisition, but is a poor choice when trying to identify changes over a very specific boundary area. Over-sampling is typically not used because overlapping data points may cause confusion in the generated images. ..........................................................................20 1-8. Pipetting the matrix onto a tissue su rface causes analyte migration as shown. A shows the tissue on the MALDI sample pl ate, B shows the pipetting of the matrix onto the tissue, and C shows the how pipetting the matrix causes analyte migration in the tissue. Imaging MS ca nnot be performed wh en this type of matrix application is employed. ...............................................................................22 1-9. The deposition of matrix as a fine mist by either airbrushing or electrospraying. A shows the tissue on the MALDI sample plate before matrix application and B shows the application of ma trix using a device the pr oduced very small droplets such as an airbrush or an electrospray needle with the proper voltage. C shows the final tissue section with matrix applied. The process permits the incorporation of the matrix material into the tissue medi um without causing analytes to migrate across the tissue. .......................................................................23 1-10. Two diagrams of the MALDI proce ss. Diagram A shows how the matrix material dilutes the analyte molecules to reduce the amount of intermolecular interactions, while the desorption of neat analytes especially biomolecules will induce fragmentation. The matrix is presen t at a much higher concentration than the analyte to effectively act as an in termediary by absorbing the laser energy and transferring it to the analyte without causing fragmentation. In Diagram B, the plume generated after a laser pulse contains ions, both positive and negative, related to the analyte an d matrix compound as well as many neutral species. ........28 1-11. Common organic acids used as matrices in MALDI and their abbreviations. No single matrix is compatible with all analyt es, and thus a wide variety of matrices have been employed. ................................................................................................29 1-12. The basic structure of glycerophospholipids (GPLs). All GPLs consist of a polar head group and two non-polar tails, fatty acids. The polar head group is the primary means of separating GPLs into different classes. The names of the four major classes are shown next to the structure of the head group. .....................35 ix

PAGE 10

1-13. The basic structure of sphingolipids. This lipid clas s is derived from sphingosine or another similar base. The structure of sphingosine is shown at the bottom. They are similar to GPLs in that they have a no n-polar tail, usually consisting of one fatty acid, and a polar head group. The polar head group is also used to differentiate the different classes of s phingolipid. The head groups for the major sphingolipids are shown above. ..........................................................37 1-14. A typical fatty acid, olei c acid, is shown in two possi ble configurations. For all double bonds occurring in a fatty acid chain, the double bond is in the cis configuration rather than the trans configuration. The st ructures shown are of the basic shorthand form. .........................................................................................39 1-15. The structure of PC (16:0, 18:1) in a normal structure form is shown. This is the most abundant glycerophospholipid present in ma mmalian cellular membranes. ..............................................................................................................41 1-16. Above is the structure of SPM (18:0). This is the most abundant sphingolipid in mammalian nerve tissue. ..........................................................................................42 1-17. A diagram of the fast atom bombardmen t, FAB, process. In FAB, the analyte, A, is dissolved in a matrix, G, typically glycerol, and an ion or atom beam is directed to the sample surface at an a ngle. Secondary ions of glycerol and analyte are ejected from the surface upon bombardment by the ion beam. These secondary ions generated are from the ma trix and the analyte. This technique limits fragmentation of the analyte, generating intact molecular ions. ....................46 1-18. A diagram of the electrospray process. In electrospray ioni zation (ESI), a liquid solution is pumped through a small capillary needle to which a high voltage is applied. As the solution emerges from the needle, a Taylor cone develops and then small droplets begin to evaporate. These droplets then evaporate further, producing charged species in the gas phase.............................................................49 2-1. A schematic diagram of the Finniga n LTQ linear ion trap instrument with vMALDI source modification. The vMALDI source replaces the standard API source, and includes the sample plate, RF-only quadrupole q00 with auxiliary rods, and skimmer. The RF-only quadrupol e q0, octopole, and 3-section linear ion trap are standard LTQ components. The open arrows show the three stages of differential pumping; the pressure in the vMALDI region is typically 0.17 Torr. ..........................................................................................................................56 2-2. On the left is an optical image of the spinal cord coated with DHB; the locations where the laser was fired to produce spectra A and B are indicated with circles. The spinal cord is outlined in black. Sp ectrum A shows the ions detected from a location away from the tissue, but in an area where the DHB matrix was still present. Most of these ions correspond to DHB clusters such as m/z 409.91, 551.73, and 727.36, as indicated by the dagger symbol. These ions appear at a higher mass than expected because the laser power was increased due to the x

PAGE 11

thickness of the tissue sample and thus wh en the laser fired at a position off the tissue, space-charging was evident. No phospholipids were identified in that location. Spectrum B shows the ion signal from the tissue surface, indicating the presence of phospholipid ions. The ion at m/z 760.82 was determined by tandem MS to be the [M+H]+ ion of phosphatidylcholine 34:1 (PC 34:1). The other starred ions from m/z 700-900 were determined to be either PC or sphingomyelin, SPM. ...............................................................................................58 2-3. Spectrum A is an enlargement of the phospholipid region of the spinal cord tissue section mass spectrum from figure 2B. Spectrum B shows MS2 of m/z 760.82. Only one major fragment ion is produced, m/z 184.00, which corresponds to the phosphocholine head group of phosphatidylc holine. The neutral losses of 18 (water) and 154 could correspond to a DHB cluster ion at the same m/z as the phospholipid ion. The inset displays the basic structure of a phosphatidylcholine, showing the fragme ntation pathway that produces m/z 184. .59 2-4. Spectrum A shows MS2 of m/z 782.82 from figure 2A, while spectrum B displays MS3 (782.82 723.36 ). The neutral loss of 59 after MS2 corresponds to loss of trimethlyamine in the structure for PC (inset of spectrum B). Two major fragment ions from MS3 arise from the neutral loss of 124 (ethylphosphate) with the sodium reta ined on the glycerol backbone, or the neutral loss of 146, when the head group re tains the sodium. This fragmentation pattern is associated with cationized PC and SPM. Fragment ions relating to specific fatty acid tails R1 and R2 were not observed...............................................62 3-1. Chemical structures of the phospholipid s used in the studying ion fragility and source fragmentation. The protonated vers ion of the molecule is shown. SPM stands for sphingomyelin, PC stands for phosphatidylcholine, and PE stands for phosphatidylethanolamine. All three molecules are very similar in structure. SPM and PC share the same head group, but differ in the fatty acid tails, while PC and PE share the same fatty acid ta ils, but have slightly different head groups. ......................................................................................................................68 3-2. Spectrum A, above, is a zoom scan of protonated sphingomyelin (SPM) 16:0. The peak width at 10% peak height (PW10%) was determined to be 0.44. The peak width for the sodiated counterpart, measured from the zoom scan in spectrum B, was 0.37. A narrower isolati on width indicates that the sodiated species is less fragile than the protonated species in ion trap mass analysis. ...........72 3-3. The chart above shows the PW10% for the two ions (prot onated and sodiated) of each phospholipid studied. The peak widths were measured from zoom scan data. The zoom scan data was collecte d for 2 minutes and the peak width was measured from the average spectrum of each ion. For all three phospholipid classes studied, the protonated species was determined to be more fragile because the peak width was wider. ..........................................................................73 xi

PAGE 12

3-4. A chart showing the isolation width needed to efficiently isolat e the parent ion of each ion studied (protonated and sodiated) for each phospholipid. A wider isolation width is typically needed for a more fragile ion. The results from this experiment also indicate that the protonated species is more fragile than the sodiated ion. .............................................................................................................74 3-5. Spectrum A is MS2 of the [M+H]+ for SPM (16:0) and spectrum B is MS2 of the [M+H]+ for PC (16:0, 16:0). Both of these spectra show the same fragmentation pathway producing a predominant ion at m/z 184, corresponding to the polar head group. Spectrum C is MS2 of PE (16:0, 16:0) and shows a different fragmentation pathway. The major ion pr oduced results from the neutral loss of the polar head group. ................................................................................................76 3-6. MS2 spectra of the [M+Na]+ ions for each phospholipid studied. A is from SPM (16:0), B is from PC (16:0, 16:0), an d C is from PE (16:0, 16:0). The fragmentation pathways are very similar fo r all these ions. They all result from neutral loss of the polar head group. A neutral loss of 59 for PC and SPM corresponds to the lo ss of choline (-N(CH3)3) and a neutral loss of 43 corresponds to the loss of ethanimine (-C2H5N). .....................................................77 3-7. The effect of the heated capillary te mperature on TID for the different ions of SPM (16:0) is shown in the three spec tra above. Spectrum A was acquired at 250 C, a normal capillary temperature for most analyses, and shows nearly 100% fragmentation for the [M+H]+ ion of SPM (16:0). In contrast, there is less than 20% fragmentation occurring for the [M+Na]+ ion. Spectrum B shows the spectrum when the temperature is lowered to 190 C and spectrum C shows the spectrum at 130 C. It is clear that lowering the temperature of the heated capillary reduces fragmentati on, TID, of both ions. At 130 C, extra peaks from m/z 300-550 are present. These peaks are mo st likely clusters of solvent ions, because desolvation is not as eff ective at such a low temperature. ..........................79 3-8. A graph showing the intensity of th e parent and daughter ions of the [M+H]+ and the [M+Na]+ ions versus changing the heat ed capillary temperature for SPM (16:0). These data were collected au tomatically using Xcalibur software control. The temperature was adjusted at 20 intervals from 50 C to 250 C. From the graph, it is evident that the pr otonated species is more susceptible to TID than the sodiated species. ..................................................................................81 3-9. A graph showing the effect of changi ng the capillary temperature for the [M+H]+ and [M+Na]+ ions of PC (16:0, 16:0). The same susceptibility of the protonated species to TID is evident, but the temper ature at which the pa rent ion signal is lower than the daughter ion signal occurs at a higher temperature than for the [M+H]+ ion of SPM (16:0). ......................................................................................83 3-10. A graph showing the effect of ch anging the capillary temperature of the [M+H]+ and [M+Na]+ ions for PE (16:0, 16:0). The protonated species showed a greater tendency to fragment upon increasing the capillary te mperature, but the xii

PAGE 13

parent ion signal was still more intense th an the fragment ion even at the highest temperature. This is in contrast to the other two phospholipids studied. ................84 3-11. A graph showing the effect of changing the tube lens offset from 0V to 30V on the [M+H]+ and [M+Na]+ ions for SPM (16:0). The capillary temperature was held at 250 C for each voltage level. At the lo west voltage, fragmentation of the parent ion is still around 50%. Adjusting only the tube lens offset is not enough to limit the fragmentation. The sodiated species appears to be unaffected by a change in the tube lens offset. ..................................................................................85 3-12. Mass spectrum of SPM (from chic ken egg yolk) acquired by MALDI at 10-6 Torr on a 3D-quadrupole ion trap. Fragmentation of the [M+H]+ ion is nearly 90%, while fragmentation of the [M+Na]+ ion is around 50%. The matrix was 6-aza-2-thiothymine. The aste risks indicate matrix ions. .......................................89 3-13. Mass spectrum of SPM (chicken egg yolk) acquired by MALDI at intermediate pressure (10-1 Torr) on a 2-D quadrupole ion trap. Fragmentation of the [M+H]+ ion is reduced, but not removed. Fragmentation of the [M+Na]+ ion is also reduced, but still present. The matrix was 6-aza-2-thiothymine. The asterisks indicate matrix ions. .................................................................................................90 3-14. Zoom scans of the two ions fo r PC (16:0, 16:0): A is the [M+H]+ ion and B is the [M+Na]+ ion. The spectra were acquired by intermediate-p ressure MALDI on a linear ion trap, averaging 75 spectra. As was seen in the ESI studies on a 2D ion trap, the peak width at 10% peak height for the [M+H]+ ion is wider than the [M+Na]+ ion. This wider peak widt h indicates that the [M+H]+ ion is fragile in mass analysis. .......................................................................................................91 4-1. Picture of the airbrush setup used to coat brain tissue samples on glass microscope slides (conductive and n on-conductive). The trigger was held constant by the pressure from a ring stand clamp. For typical coating, the sample was about 6 in. from the airbru sh nozzle. The nozzle can be easily changed, but all experiments were conducted with a nozzle of 0.40 mm. ...............99 4-2. Schematic diagram of the Finnigan LTQ linear ion trap instrument with vMALDI source modification. The vMALDI source replaces the standard API source, and includes the sample plate, RF-onl y quadrupole q00 with auxiliary rods, and skimmer. The RF-only quadrupole q0, oct opole, and 3-section linear ion trap are standard LTQ components. The open arrows show the three stages of differential pumping; the pr essure in the vMALDI re gion is typically 0.17 Torr. .100 4-3. The picture at the top of the figure, A, is an optical image generated from inside the mass spectrometer of a rat brain sec tion coated with DHB matrix. It was acquired with 1 mm x 1 mm square pictures that are stitched together. This creates the lines in the picture. The mass spectrum, B, is the signal from the area on the tissue indicated by the arrow. The spectrum was acquired with 10 xiii

PAGE 14

laser shots. A total of 11,156 spectra we re collected across the tissue section. The open circle is the laser spot size (100 m). .....................................................102 4-4. A is the digital image of the tissue section and B is the mass spectrometric image for the ion at m/z 756. Extracting the intensity from each spectrum in the data file generated this image. This is a raw image, with no further processing done to the data. The spectrum in C is an example of one of the 11,156 spectra that were collected. ........................................................................................................105 4-5. MS image A, top, shows the raw imag e generated by extracting the intensity of m/z 756 with respect to position. MS im age B, bottom, is the normalized MS image for m/z 756. Normalization involve d dividing the intensity of m/z 756 at each pixel by the total ion intensity at that pixel and multiplying by 100000. All MS images further generated used this normalization procedure. .........................106 4-6. The averaged MS2 spectrum of m/z 756.6 (516 spectra were averaged). It is the average of 516 MS2 spectra collected fr om various parts of the tissue section. Not that the intensities of all the ions below m/z 520 have been expanded 50 times. Fragment ions representing th e phosphocholine head group are the most abundant ions, but the less abundant fragment ions ar e used to determine the fatty acid composition of th is phospholipid, identifying the ion as the sodiated adduct of PC (16:0,16:0). The less abundant ions used for identification were m/z 500.3 and m/z 478.4 because they represen t the loss of the fatty acid chain with Na or without, respectively. ...........................................................................108 4-7. The averaged MS2 spectrum of m/z 782.6 (517 spectra were averaged). As in Figure 4-6, the most abundant fragment ions correspond to losses of the phosphocholine head group, while the less abundant fragment ions allow for the correct identification and location of the fatty acid tails. These ions are m/z 526.3 and m/z 500.3 and correspond to the loss of the sn -1 (C1) and sn -2 (C2) fatty acid chains, respectively. The ra tio of these ions allows for proper assignment to the glycerol backbone. Th is ion was correct identified as the sodiated adduct of PC (16:0,18:1). The structure of this compound is shown in the inset with the assignments for the two fragment ions identifying the fatty acid chains. .............................................................................................................109 4-8. MS images are shown of PC ions with 16:0, palmitic acid, at the sn -1 position on the glycerol backbone. A is the MS im age of PC (16:0, 16:0), B is the MS image of PC (16:0, 18:1), C is the MS image of PC (16:0, 20:4), and D is the MS image of PC (16:0, 22:6). The ions were correctly identified from MS2 data; classification is shown above each image. All the ions were the sodiated species, [M+Na]+. The intensity levels were adjusted in images C and D for comparison purposes. As is evident fr om their comparison, the localization of these four PC ions is very different The distribution of m/z 756, PC (16:0, 16:0) is prominently in the gray matter of the brain. .............................................111 xiv

PAGE 15

4-9. MS images showing the distribution of palmitic acid, 16:0, in rat brain tissue. A is an image from MALDI imaging and is the summed image of four intact ions (shown in Figure 4-8) containing palmitic acid at the sn -1 position of the glycerol backbone. B is an image from negative ion SIMS and shows the distribution of only palmitic acid, or an i on that has the m/z of palmitic acid. MALDI enables the analysis of primar ily intact ions, while SIMS provides better spatial resolu tion, but with a high degree of source fragmentation. B is adapted from reference 88. .....................................................................................112 4-10. MS images of PC ions with 18:0 or 18:1 at the sn -1 position of the glycerol backbone. A is the MS image of PC ( 18:0, 18:1), B is the MS image of PC (18:1, 18:1), C is the MS image of PC (18:0, 20:4), and D is the MS image of PC (18:0, 22:6). MS2 data allowed for the correct identification of each ion. ......114 4-11. Three MS images of the SPM ions dete cted from the rat br ain tissue section (A is SPM (18:0), B is SPM (24:0), and C is SPM (24:1)). The structure at the bottom right is SPM where R1 is the variable fatty acid chain. Only m/z 753 was subjected to MS2; the other two ions (835 an d 837) were not chosen for MS2 analysis. The images for these two ions were generated because further research showed that they are primarily present in the white matter of the brain, whereas SPM (18:0) is present in the gray matter. As seen in the images, these three ions show a different distri bution indicative of prior research.121.................115 4-12. MS images of PC (16:0, 16:0), m/ z 756.8 (A and C) and PC (16:0, 20:4), m/z 804.7 (B and D). A and B were generated fr om the first mass an alysis of the rat brain tissue section, whereas C and D were generated from mass analysis that took place after 31 separate experiments ac ross the tissue were performed. It is evident from the pictures that C and D are much more spatially resolved. ...........118 4-13. MS image A was generated from a non-conductive glass microscope slide and shows the distribution of m/z 810.7, PC (18:0, 18:1). MS image B was generated for the same ion from a conductive glass microscope slide. The brain tissues were 10 m thick and are serial sections from the same rat brain. The images are very similar, but the im age from the non-conductive microscope slide (A) appears to be more spatially resolved. ....................................................120 4-14. Spectra from similar spots of the brain tissue. A was acquired from a nonconductive microscope slide and B was acquired from a conductive microscope slide. The TIC for the non-conductive slide is 2x higher than that from the conductive slide. .....................................................................................................121 4-15. A is the MS image of m/z 804.7, PC (16: 0, 20:4) and B is a stereotaxic view of a rat brain section (Bregma .04 mm) show ing the different functional regions of the brain. The MS image shows good correlation to the atlas even with a spatial resolution of only 100 m. ..........................................................................122 xv

PAGE 16

4-16. An average spectrum of 129 different position-specific mass spectra acquired from the rat brain tissue s ection. The complexity of this region of the mass spectrum can be unraveled using MS2 data of each ion. It can therefore be determined that a single compound such as PC (16:0, 18:1) can be represented by four different ions in the mass spectrum, m/z 723.8 ([M+Na-N(CH3)3]+), 782.7 ([M+Na]+, 798.6 ([M+K]+), and 958.4 ([M+Na+(DHB+Na-H)]+). .............126 xvi

PAGE 17

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 IMAGING SMALL MOLECULES IN TISSU E BY MATRIX-ASSISTED LASER DESORPTION/IONIZATION TA NDEM MASS SPECTROMETRY By Timothy James Garrett May 2006 Chair: Richard A. Yost Major Department: Chemistry The use of an intermediate-pressure matr ix-assisted laser desorption/ionization (IPMALDI) source working at 0.17 Torr on a linear ion trap (LIT) was investigated for the analysis of tissue specimens. IP-MALDI, with 2,5-dihydroxybenzoic acid (DHB) as the matrix, was employed for the detection of phospholipids. The results indicate that analyzing tissue specimens at non-traditiona l MALDI vacuum pressures is possible. Coupling MALDI to an LIT permits the use of multiple stages of mass spectrometry, MS n which is critical for the ability to iden tify compounds desorbed directly from tissue specimens. Using MS n ions detected from m/z 600-1000 were characterized as phosphatidlycholines and sphingomyelins. Speci fically using tandem MS, PC ions could be classified as either [M+H] + or [M+Na] + because the fragmentation patterns of protonated and sodiated phosphatidlyc holines follow different pathways. Understanding ionization char acteristics of the ions desorbed from tissue is important in ensuring the stability for ion transport and mass analysis. The analysis of xvii

PAGE 18

the fragility of three phospholip ids by electrospray ionization ion trap mass spectrometry as protonated and sodiated species is discu ssed. The conditions of ion transport and ion trapping that cause head group fragmentation of ions formed by electrospray ionization were evaluated. Ion fragility in ion trapping was evaluated using a slow scan speed, the percent collision-induced di ssociation, and the isolation width needed for effective isolation. Ion fragility in transport was evaluated by ad justing the capillary temperature and the tube lens voltage. Resu lts indicated that the sodiated species was the more stable form of each ion. Finally, mass spectrometric images of phos pholipids in brain tissue were generated by IP-MALDI on a LIT mass spectrometer. El even individual phosphatidylcholines and sphingomyelins were identified by MS 2 with fragment ions that allow for the identification of the fatty acyl chains and their position on the glycerol backbone. The use of non-conductive and conductive glass micros cope slides in connection to ion trap mass analysis was evaluated, and results showed that equivalent data could be obtained on either surface. An artistic airbrush was employed to effectively coat the matrix compound onto tissue sections without the de leterious effects of analyte migration. xviii

PAGE 19

CHAPTER 1 INTRODUCTION Mass spectrometric microprobes with matr ix-assisted laser desorption/ionization (MALDI) offer a unique opportunity to directly probe thin tissue sections for changes in the distribution of a molecular species of interest. Initial designs offered the opportunity to identify changes in the con centration of a desired compound in different regions of a tissue of interest. Investiga ting the distribution of a compoun d within tissue sections can help understand whether an exogenous compoun d administered orally, such as a small drug molecule, can be found within the body, where that compound tends to localize, and possibly how endogenous compounds are affected by the drug. The latter aspect is a new concept to the field of imaging because usi ng a mass spectrometer begets the opportunity to gather more complete information about a tissue section, incl uding unknown analysis. The mapping of endogenous compounds in tissue sections offers an incredibly specific method to identify changes in the profile of the compound in diseased versus normal states or just simply offers a new oppor tunity to re-map entire organs by a more chemically specific process. This area coul d offer supplementary information that may be useful in fully characterizing ce rtain organs, such as the brain. A critical aspect of analyzing compounds fr om tissue is the abili ty to identify the compound with more certainty th an just the expected mass-to -charge (m/z) value. The m/z is defined as the monoisotopic molecular we ight of a molecule plus the weight of one or more protons (or cations, depending on the type of ionization) di vided by the charge. So if the molecular weight of a compound is 400 g/mol and this compound is protonated, 1

PAGE 20

2 then the m/z would be 401. However, if the molecule was able to obtain another proton to have two charges, the ions mass would be 402 g/mol, but the m/z value is divided by 2 and thus would equal 201. The use of tandem mass spectrometry is one way to positively identify ions of the detected m/z value as from the compound of interest. Tandem MS involves isolating the specific m/z value (called the pare nt ion) in the mass analyzer and then subjecting that ion to collision-induced dissociation (CID), thus causing the isolated ion to fragment into sm aller ions, giving a fragmentation spectrum (called a daughter spectrum). Several instruments are capable of performing tandem MS, including a triple quadrupole, time-of-flight/timeof-flight (TOF/TOF), quadrupol e/time-of-flight (Q-TOF), and a quadrupole ion trap (QIT) mass spectro meter. Each instrument has certain advantages in the analysis of tissue sections and all can perform one stage of tandem MS (MS 2 ), but the QIT is unique in its capability to provide for multiple stages of tandem MS (MS n ) as well as the ease of c oupling MALDI to this mass an alyzer. For these reasons, all the research conducted was performed on a QIT-MS and thus the background is a focus on this instrumentation. Quadrupole Ion Trap Mass Spectrometry From 3-D to 2-D Ion Traps Today, the quadrupole ion trap (QIT) is a wi dely used mass analyzer because of the ability to fully characterize a compounds structure with MS n which allows for the unambiguous identification of unknown and know n compounds. The ion trap has a long history before becoming a routine tool for ch emists and biologists. The 3-dimensional (3D) ion trap was first introduced and de scribed as a mass storage device in 1953 by Wolfgang Paul and Helmut Steinwedel. 1 It consisted of two hyperbolic electrodes called

PAGE 21

3 end-caps and a ring electrode centered between them. Figure 1-1 shows a schematic of a 3D ion trap. A few years later, mass-selective detecti on was first employed for the detection of krypton gas. 2 For those studies, the motion of an i on within the ion trap was used to obtain the mass-to-charge (m/z) ratio for that ion. The de tection method was based on the fact that each ion would have a different frequency of motion that correlates to the mass and charge of the ion, similar to mass analysis performed by ion cyclotron resonance. Further development of this inst rumentation led to the mass-selective storage of ions with external detec tion in which ions were ejected through a small hole in one of the end-caps to a detector. 3, 4 These advances were all cruc ial to the development of the ion trap, but the most important advance in making the ion trap useful as a mass spectrometer was the development of mass-selective ejection of ions stored within the ion trap (mass-selective instabil ity mode) in the early 1980s. 5 In this method, all ions are trapped in the quadrupolar field of the ion trap and then seque ntially ejected from the trap in order of increasing m/z by ramping the ra dio-frequency (RF) voltage applied to the ring electrode. This, the addition of He damp ing gas, and stretching the trap led to the production of the first commercial mass spec trometer by Finnigan MAT (now Thermo Electron, San Jose, CA) called the ion trap de tector or ITD. It was a 3D QIT with internal electron ionization and external detection. Until recently, advances in ion trap mass an alysis focused on the injection of ions from external sources and coupling the mass an alyzer to a multitude of ion sources such as matrix-assisted laser desorption ionizati on (MALDI), electrospra y ionization (ESI), atmospheric-pressure chemical ionization (A PCI), and now atmospheric-pressure photo

PAGE 22

4 Entrance End-cap Exit End-cap Ring Electrode r0z0 Entrance End-cap Exit End-cap Ring Electrode r0z0 Figure 1-1. A schematic representation of the 3-dimensional (3D) quadrupole ion trap (QIT). The 3D QIT consists of 2 endcap electrodes and a ring electrode. The inside dimensions of the 3D QIT are defined by r the radius of the ring electrode, and z the distance from the center of the QIT to the apex of the end-cap. For a symmetrical QIT, r is 2 times z Ions are injected and ejected through holes drilled into each end-cap.

PAGE 23

5 ionization (APPI). These multiple sources allowed for a very wide variety of compounds to be studied from the initial elemental studie s to more labile and much larger molecules such as proteins. Virtually any compound that is ionizable can be analyzed with an ion trap, meaning that mass analysis is limite d by the ability of the ionization source to produce ions. Through the years of studying the ion tr ap, the drawbacks have also been documented. Perhaps one of the biggest drawbacks is the concept of space-charging. 6, 7 Due to the small space in which ions are confine d, if the trap is filled with too many ions, then the applied quadrupolar fiel d will not affect all ions in the trap equally because other ions act as shields. This issue tends to cause peak-broadening, mass shifting, and the appearance of ghost peaks. The use of automatic gain control (AGC) reduces this problem by allowing for an initial pre-scan pe riod in which the computer determines how long to fill the trap based on an initial packet of ions. 8 Other disadvantages of the 3D trap are the difficulty of injecting ions because of the small inlet hole and the presence of the quadrupolar field, as well as the loss of half the ions dur ing ejection because ions will exit the trap at both the inlet and the exit holes. In 2002, the linear ion trap, al so called the 2-dimensional (2D) ion trap, was first introduced in commercial instruments. 9, 10 One design was very similar to that of a triple quadrupole mass spectrometer, with the second quadrupole used as the ion trap. In this design, ions are trapped in the second qua drupole and ejected axially (out the back). 9 Operating in this manner does increase the st orage space of ions thus allowing more ions to be injected, but using massselective ejection, ions may still be lost because they will exit out the back of the qua drupole and the front. The second design offered a more

PAGE 24

6 dramatic change in design, in which the trap was divided into 3 parts, a longer center section for trapping ions and two s horter sections (front and back). 10 A diagram of this trap design is shown in Figure 1-2. The significant difference in this design was the machining of two long slits on two opposite r ods of the center section through which ions can be ejected radially from the trap. Placing a dete ctor on both sides immediately doubles the number of detected ions. Both designs offered a substantial increase in the capacity of the ion trap and in the efficiency of ion injection, but the second design also offered an increase in the number of ions detected. 3D and 2D Quadrupole Ion Trap (QIT) Theory and Operation A traditional 3D QIT consists on two identi cal hyperbolic end-cap electrodes with entrance and exit holes in the center of each a nd a hyperbolic ring electrode with radius r o situated between the two end-caps. For an ideal 3D trap, the distance from the entrance end-cap to the exit end-cap is 2z o with z o defined as the distance from the center of the trap to the apex of the end-cap (Figure 1-1). Thus, the theo retical dimensions of the ion trap (also knows as the Paul trap) are shown in Equation 1-1 where r o is the internal radius of the ring electrode. r o= 2z o Equation 1-1 However, due to the existence of entran ce and exit holes in each end-cap which are needed to inject and eject i ons, the quadrupolar field is di sturbed, creating imperfections that affect how ions are tr apped and scanned out, produci ng spectra with improper mass assignments. By moving each end-cap of the trap outward 0.030 of an inch in the z direction (called stretching) the field imperfections are reduced, creating a mass spectrum void of mass shifts. 11

PAGE 25

7 Figure 1-2. Diagram of a 2-dimensional (2 D) ion trap. This 2D ion trap design incorporates three different sections, fr ont and back sections of equal length, and a center section where the ions are trapped.

PAGE 26

8 Ions entering the trap are successfully trapped by the application of a radio frequency (RF) to the ring electrode that crea tes a quadrupolar field in the ion trap due to the shape of the end-caps and ring (hyperbolic ). A direct current (DC) can also be supplied to the end-caps, but t ypically, it is maintained at gr ound. The RF applied to the ring is defined by two parameters, a constant angular frequency ( and a variable amplitude voltage (V). Solutions to the re duced Mathieu equations (Equations 1-2 and 13) where e is the charge (1.602 x 10 -19 C), U is the DC potential, V is the amplitude of the RF, is the angular frequency of the applied RF, and r o and z o are the internal dimensions of the trap, give coordinates in a-q space (a z q z ) that are then mapped onto the Mathieu stability diagram (Figure 1-3). 12 If these coordinates fall into the region where the a z and q z overlap, i.e. stable in both the axial (z) an d radial (r) dimensions, then the ion will be successfully tra pped. In typical operation, th e DC potential,U, is zero and thus all ions have a z equal to zero; thus the location of an ion in a-q space is focused to the q dimension. a z =-2a r =-16eU/m(r o 2 +2z o 2 ) 2 Equation 1-2 q z =-2q r =8eV/m(r o 2 +2z o 2 ) 2 Equation 1-3 Successful trapping and mass analysis of i ons also involves the use of buffer gas, typically helium (He). The addition of a small amount of buffer gas increases the pressure inside the ion trap. As ions enter the trap, they are collisionally cooled, focusing the ions to the center of trap where the qua drupolar field is more uniform, increasing the efficiency of trapping and eventual ejection. Typically, the He buffer gas is maintained at 1 mTorr within in the ion trap. Another adva ntage of collisional cooling is the separation of initial kinetic energies from mass analysis. This means that if a packet of ions of the

PAGE 27

9 Figure 1-3. A small portion of th e Mathieu stability diagram. Solutions to Equations 1-2 and 1-3 give coordinates in a z and q z space that can be mapped onto the above diagram. If the coordinates for an io n fall into the region shown above, then the ion will have a stable trajectory in side the 3D or 2D trap and will be successfully trapped. In normal operation, the DC applied is zero and thus the a z is also zero since it is related to the DC voltage applied, thus the most important number is the value for q z The value of q z =0.908 is the edge of the stability diagram. An ion with a q z greater than this value is not trapped.

PAGE 28

10 same mass has a wide range of initial kinetic energies, as is the case in ions produced by MALDI, they will all have the same kinetic energy once successfully trapped. Perhaps one of the most significant operati ng characteristics of the ion trap is the ability to selectively isolate an ion of interest and then cause it to fragment into smaller pieces called daughter ions. Th is type of analysis is termed tandem MS (MS/MS or MS n ). In some ion traps, isolation of a particular ion of interest is accomplished by using a stored waveform inverse Fourier transform (SWIFT). 13 All ions in the trap exhibit a frequency related to their q value in a-q space. In Figure 1-3, specific points in a-q space correspond to different r and z values. The term is used to describe the secular frequency of an ion in the quadrupolar field and determines the degree to which an ion follows the applied field. The fundamental frequency, z,r of an ion in the quadrupolar field is calculated from the r and z values. Since typically operation involves on the zcomponent, the r value is not used. The z value is approximated from equation 1-4 and then used to calculate the z,0 value using Equation 1-5. z = (q z 2 /2) Equation 1-4 z,0 =0.5( z ) Equation 1-5 In the SWIFT isolation method, a broadband waveform with a notch equal to the frequency of the ion to be is olated is applied to the endcap electrodes. Increasing the voltage of this applied frequency causes ions outside the notch to gain kinetic energy until they are ejected from the trap, leaving only the ion of interest within the trap. Another isolation process involves the use of a 5-500 kHz multi-freque ncy waveform that has sine components at every 0.5 kHz. 14 To isolate an ion of interest, the sine components

PAGE 29

11 corresponding to the frequency of that ion are removed from the multi-frequency waveform, thus creating a not ch similar to SWIFT. Following isolation, the ion of interest can be fragmented by the application of an alternating current voltage (AC) equal to the frequency of th e ion in the trap across the end-caps. With sufficient voltage (typically 1-3V pp ), the ion gains kinetic energy by resonant absorption, which results in more translational motion. With this increased motion, collisions with the He buffer gas presen t in the ion trap become very energetic. These collisions increase the internal ener gy of the ion until sufficient energy is reached to break bonds within the ion. These fragment ions are then held w ithin the trap because of the quadrupolar field except for those fr agments that fall below the low mass cutoff (past the right edge of the stability diagram in Figure 1-3) or those that do not retain a charge. After scanning the ions out, the resu lting spectrum is essentially a mass spectrum of a mass spectrum or MS/MS (MS 2 ). This process is called collisionally activated dissociation (CAD) or collisi on-induced dissociation (CID). 15 Unique to the ion trap is the ability to perform successive CID events This means that not only can an ion of interest can be isolated (the parent ion), fragmented into smaller pieces (the daughter ions), and scanned out resulting in an MS 2 spectrum, but also a daughter ion of the parent can be chosen and further fragmented into its pieces, which can be considered granddaughters of the original parent ion. The spectrum from this event is MS 3 This process can be repeated n nu mber of times, giving a mass spectrum related to each CID event. This is why the term MS n is used in connection with tandem MS on an ion trap. 2D/Linear Ion Trap Description The linear ion trap (LIT), as used in thes e experiments, is composed of 3 sections of linear rods with hyperbolic structure, as shown in Figure 1-2. The front and back

PAGE 30

12 sections are 12 mm long, while the center section is 37 mm long. 10 For the axial trapping of ions, separate DC voltages are applied to all three sectio ns of LIT (Figure 1-4), while RF is applied in two phases to rod pairs for radial trapping (Figure 1-5). For positive ion analysis, the front section has a potential of V, the center section has a potential of 14V, and the back section has a potential of V during ion storage. During mass analysis, the front and back sections have a voltage of +20V, while the center section is maintained at V. A supplementary AC volta ge is also applied in two phases, but only to the x rod pair (Figure 1-5) for isolation and CID. The application of this supplementary AC vo ltage is also used to resonantly eject ions radially, along the x-axis, out of two sm all slits (30 mm long) cut into the x rod pairs. As in the 3D trap, because of the ex istence of these slits, the quadrupolar field is disturbed, and thus the 2D ion tr ap is also stretched (in the xdirection) to help reduce the field imperfections. Ejected ions are th en detected on both sides with conversion dynodes and electron multipliers. This mode func tions very similar to a conventional 3D trap, but has the added benefit of doubling the number of detected ions. Other advantages of the 2D trap in clude increased ion injection e fficiencies, increased trapping efficiencies, and larger storage volume. Alternatively, by controlling the DC applied to the separate sections, ions can be non-mass selectively ejected axially, along the z-axis, at which point they could enter a different mass analyzer such as a time-of-flight, penning trap, or an orbitrap. A signi ficant problem in the production of 2D traps is the machining and mounting of the hyperbolic surfaces. Becau se ions are spread out over 30 mm in the 2D trap versus the 1 mm in the 3D trap, any imperfections in the rods can cause significant decrease in mass resolution. 10 Increased machining costs along with the added

PAGE 31

13 DC 1 DC 3 DC 3 DC 2 DC 2 DC 1 y z y z Figure 1-4. A Schematic representation of the 2D ion trap. Ions enter from the left side of the figure. The 2D trap is separated into 3 sections, two shorter sections of 12 mm in length, and a longer, center, section 37 mm long. Separate DC voltages are applied to all three sections in order to trap ions Ions are ejected through slits in the x-rod pair s of the cente r section.

PAGE 32

14 y x y x RF RF RF + RF + RF RF RF + RF + AC + AC GND GND AC + AC GND GND A B Figure 1-5. Two views of the 2D ion trap l ooking down the z-axis, al ong the flight path of the ions. A shows the radial quadr upolar trapping field. The phases of RF are applied to all the elec trodes of the ion trap to create the quadrupolar field. B shows the radial dipolar excitation arra ngement. An AC voltage is applied only to the x-rod pairs. The AC voltage is used for isolation and collisioninduced dissociation experiments as well as in the ejection of ions from the trap. Ions are ejected out of slits cut into the x-rod pairs.

PAGE 33

15 benefits are several reasons w hy the 2D trap is significantly more expensive than the 3D trap. Cost of machining may decrease as the process is refined. Tissue Imaging by Mass Spectrometry Background Prior to the advent of matrix-assisted laser desorption/ionization (MALDI), mass spectrometric analysis of tissue specimens focused primarily on the identification of elements in thin tissue sections, with one of the first experiments involving the detection of heme-bound iron from red blood cells using a laser microprobe mass analyzer, or LAMMA. 16 With the advent of soft ionizati on methods for mass analysis such as MALDI 17 and electrospray ionization (ESI) 18 characterization of traditionally labile and involatile molecules such as proteins and pep tides present in biological tissues has been made possible. The use of softer ionization techniques allows for reduced complexity in analysis of biomolecules with mass spectrome try because intact molecular ions of the particular molecule are produced rather than fragment ions. Using MALDI, mass spectrometric micr oprobes are now offering promising new approaches to map the distribution of sma ll and large molecules directly from tissue sections at biologically significant levels and to help unravel the molecular complexities of cells. 19-22 Microprobe MS offers the unique ability to directly analyze chemical species from tissue samples, either creating images of how a certain m/z value is localized within the tissue section (usually focused on the prot eins and peptides pr esent or absent in normal and diseased tissue 23 ) or for identification of site-specific drug activity as described previously. 24 Caprioli et al. showed the capability of MALDI to create images of spatial localization first using the molecu lar weight of coomassie blue dye to map a

PAGE 34

16 copyright symbol imprinted onto a MALDI target 23 and then showing a map of the location of 3 different pr oteins in mouse brain. 21 Secondary ion mass spectrometry, SIMS, ma y provide similar capabilities to MALDI microprobes, with potentia lly higher spatial resolution due to the smaller size of the primary ion beam compared to the laser beam used in MALDI, albeit with much lower sensitivity because of the monolayer-de pth interrogated by SIMS. Interest in the nervous system using laser microprobes has been related to the analys is of rat brain via SIMS and MALDI, 20 and on neurons from Aplysia californica sea slug, using MALDI and laser desorption/ionizati on on porous silicon (DIOS), pr imarily looking for proteins and peptides. 25, 26 Neurons from Aplysia californica are known to be of the largest diameter in the animal kingdom, up to 1 mm, which allows for intracellular studies using MALDI. Todd et al. have shown the analysis of lipids in brain sections, specifically for phosphatidylcholine, the primary constituent of plasma membranes, and have shown that the disappearance of the m/z 184 fragment ion of phosphatidylcholine in the image created by SIMS following injec tion of lipopolysaccharide, LPS, can be related to brain damage. 19 Injection of LPS into brain sections is known to induce brain damage through demyelination of nerve tissue. Studies in our laboratory have demonstrated the ability to map paclitaxel, a small drug compound, at tr ace (pg/mg) levels in ovarian tumors. 24, 27 Critical to such sensitive and selective analys is was the use of MALD I coupled to an ion trap that provided capability of tandem mass spectrometry (MS n ). Currently, mass analysis for directly probing tissue sections is done with either a time-of-flight (TOF) or a quadrupole ion trap (Q IT). Since tissue samples contain a wide variety of molecular compounds and elements up to and beyond 100,000 Da depending

PAGE 35

17 on the type of tissue analyzed, it is vital to have an instrument capable of efficiently detecting this very wide molecular weight range and capable of real time compound identification. A TOF analyzer is capable of detecting this wide molecular weight range; however, compound identificati on is normally performed offline using an enzymatic cleaving agent. 21 Tandem MS on a TOF using post-source decay is still an inefficient process and not necessarily repeatable, maki ng molecular identification difficult. In contrast, tandem MS with an ion trap is ve ry practical and reproducible. The major draw back of using a QIT in connection to tissue imaging by MALDI is the limited mass range. Current QITs, including the 2D-QIT, are limited in mass range to 4000, even though higher molecular weights, up to m/z 100,000, have been successfully analyzed; this mode of operation is not routinely performed. 28 This means that the primary justification for using QIT technology for imaging MS is for detection of smaller compounds such as drugs, pep tides, or lipids with identi fication by tandem MS and for generating more specific images such as the mapping of a fragment ion produced by MS 2 One could even imagine the mapping of a fragment ion from MS 3 MS 4 or MS n to create a very specific imag e of a desired compound. Fundamentals All organs present in animal or plant sp ecies can be sampled by imaging MS. In animal species, the organs are removed and snap frozen in N 2 (l) and stored in a C freezer until needed. When ready to perform mass analysis, the organ is removed from the freezer and prepared for s ectioning with a cryotome. Sectioning with a cryotome requires the use of optical cutting temperatur e medium (OCT) to affix the organ tissue to a sample stage. Unfortunately, OCT can interfere with mass spectrometric analysis, so

PAGE 36

18 care must be taken to insure that the section of tissue used does not have any OCT on it. The tissue can be cut to any desired thickne ss, but usually the thickness ranges from 10 m to 60 m. Once cut, the tissue section can be placed on a metal, glass, or conductive glass surface for imaging MS. This pro cess is shown graphically in Figure 1-6. Imaging MS can be performed on any mass spectrometer employing a MALDI (or SIMS) source as long as the sample plate, or the laser, can be rastered to move the laser beam across the tissue surface in a predefined pattern. Some researchers have also shown the ability to performing imaging MS studies with a new ionization technique, desorption electrospray ionization (DESI), at atmospheric pressure, but w ith far less resolution than MALDI or SIMS. 29 The laser beam can be moved, or stepped, across the tissue surface at any desired step size. For normal analysis, the step size is equal to the diameter of the laser beam, as shown in Figure 1-7. In MALDI, the laser spot size is typically anywhere from 50-100 m, which is based primarily on the diffr action limit. For a shorter analysis time, under-sampling can be employed in which th e step size is larger than the laser spot diameter. Sampling at twice the laser spot diameter will reduce the analysis time by one half, but will also limit the resolution of the generated image. Over-sampling is the third analysis type, but it is not typically employe d. By stepping at half the laser diameter, 50 m in this example, each new spot overlaps with the previously sampled area thus generating a spectrum that is a combination of an undisturbed area and a sampled area. Instead, it would be preferred to reduce the spot size of the laser beam to achieve a higher resolution rather than try over-sampling. Smaller spot sizes haven been employed on non-commercial instruments typically using a sm all slit to narrow the laser beam with some success. In SIMS, the spot size can be much smaller, usually a few hundred

PAGE 37

19 Figure 1-6. The process of creating an image by mass spectrometry. The two most important stages in tissue analysis by mass spectrometry are 1) cutting the n g e the Section the tissue Place on microscope slide Coat with matrix Raster laser across tissue in MS Send to MS 720 740 760 780 800 820 840 m/z 0 10 20 30 40 50 60 70 80 90 100 782.75 756.85 810.71 798.68 753.89 772.75 832.64 723.87 769.84 826.66 751.87 836.70 820.62 792.01 743.40 721.65 720 740 760 780 800 820 840 m/z 0 10 20 30 40 50 60 70 80 90 100 782.75 756.85 810.71 798.68 753.89 772.75 832.64 723.87 769.84 826.66 751.87 836.70 820.62 792.01 743.40 721.65 Position Specific Mass Spectrum Extract to Generate image tissue to an appropriate thickness with a cryotome, such as 10 m for brai tissue, and 2) applying the MALDI ma trix in a manner that reduces the possibility for analyte migration (airbrushing is shown above). When cuttin the tissue, it is important to ensure that no OCT compound is on the tissu sections as it can interf ere with mass analysis. The mass spectrometric stage involves rastering the sample plate at a predefined step size with respect to laser. Each mass spectrum is then positi on specific so all that is required for generating an image is extracting the m/z value of interest.

PAGE 38

20 Figure 1-7. The types of sampling used in im aging MS. Sampling at the laser resolution is the most widely used data collection scheme because it provides the highest Laser spot size 100 m Step size 100 m Sampling at the laser resolution Step size 200 m Under-sampling Step size50 m Over-sampling Laser spot size 100 m Laser spot size 100 m Laser spot size 100 m Step size 100 m Sampling at the laser resolution Step size 200 m Under-sampling Step size50 m Over-sampling Laser spot size 100 m Laser spot size 100 m resolution. Under-sampling will take le ss time for acquisition, but is a poor choice when trying to identify changes over a very specific boundary area. Over-sampling is typically not used because overlapping data points may cause confusion in the generated images.

PAGE 39

21 nanometers, because of the use of an ion beam, but imaging by SIMS does not allow for the detection of labile molecules and has limited penetration depth into a sample. Unlike SIMS, a requirement to perform MA LDI is the application of a matrix material that absorbs the ener gy of the laser and gently transfers some of that energy to analytes prepared with the matrix. For anal ysis of standards, the matrix, present at a much higher concentration, and the analyte are deposited as solutions on the surface of a metal plate and allowed to dry for co-crystallization to occur. However, this method cannot be employed in tissue imaging MS becau se depositing the matrix as a droplet on to the tissue surface can cause compounds in th e tissue to move, thereby disturbing their original location, 27 as shown in Figure 1-8. Instead, the matrix must be applied in a manner that will not interfere with the actual localization of compounds in the tissue. Several methods have been developed that successfully apply matrix in this manner, including electrospray deposition, 27 nebulization, 30 or airbrushing. The same principl es for ESI are also involved in the deposition of matrix by electros praying. For ESI, it is desi rable to have the analytes evaporate into the gas phase; however, in el ectrospray deposition of matrix this is not desired. The purpose of the elect rospray is to produce a mist of very small droplets of matrix that will wet the tissue surface and mix with molecules present for cocrystallization to occur. A diagram of this process is shown in Figure 1-9. The use of a glass nebulizer or an artistic airbrush acco mplishes the same task, but in a somewhat simpler manner. Both use pressurized gas to aspirate a liquid sample for deposition onto any surface. The airbrush is a new means fo r deposition of matrix onto tissue, and is described in Chapter 4 of this dissertation.

PAGE 40

22 Pipette matrix Analyte migration Tissue Sample plate Analyte A Analyte B Analyte A Analyte B A B C Figure 1-8. Pipetting the matrix onto a tissue surface causes analyte migration as shown. A shows the tissue on the MALDI sample plate, B shows the pipetting of the matrix onto the tissue, and C shows the how pipetting the matrix causes analyte migration in the tissue. Imag ing MS cannot be performed when this type of matrix application is employed.

PAGE 41

23 Matrix incorporation no analyte migration Tissue Sample plate Airbrush or Electrospray matrix Analyte A Analyte B Analyte A Analyte B A B C Figure 1-9. The deposition of matrix as a fine mist, by either airbrushing or electrospraying. A shows the tissue on the MALDI sample plate before matrix application and B shows the application of matrix using a device the produced very small droplets such as an airbrush or an electrospray needle with the proper voltage. C shows the fi nal tissue section with matrix applied. The process permits the incorporation of the matrix material into the tissue medium without causing analytes to migrate across the tissue.

PAGE 42

24 Matrix-Assisted Laser Desorption/Ionization Background The use of lasers coupled to mass spectromete rs to desorb ions fr om a surface is not a new concept. Laser desorption ionization (LDI) has been performed since the 1960s. Early analyses were of atomic ions from me tal surface, primarily because the absorbance of the laser light caused heat build up in the molecules leading, to the breaking of bonds. 31, 32 The practical limit for the mass analysis of intact molecular ions was therefore <1000 amu. Another disadvantage was the pres ence of selective absorption because not all analytes absorbed the laser wavelength used. 32 Thus, analytes that absorbed the wavelength of the laser would be ionized selectively compar ed to those that did not, limiting the dynamic range of the technique. Other techniques for the analysis of surfaces or materials deposited onto surfaces have also been developed. Secondary ion mass spectrometry or SIMS employs an ion gun to bombard a dry sample with a beam of primary ions with energies of 5 keV to 100 MeV. 33 These primary ions strike the analyte on the surface, producing sec ondary ions of the analyte (and anything else near the surface). 34 This technique is in many ways simila r to the action of the cue ball striking another ball, causing movement in billiards. The disadvantages of this technique are the high degree of analyte fragmentation, which limits the analysis to <1000 amu, and the limited depth analysis because only the firs t few nanometers are exposed to the primary ions. An advantage of the technique is the very small diameter of spot sizes, typically around 100 nm or less, that is very useful in th e analysis of tissues at the cellular level as described in the imaging MS section. 35 A similar technique to SIMS is fast atom bombardment (FAB) also known as liquid SIMS (LSIMS). FAB allows for the soft ionization of involatile compounds because th e liquid matrix in which analytes are

PAGE 43

25 dissolved acts as an energy tr ansfer medium, thereby reduci ng the internal energy build up in analyte molecule. This allows the intact molecular ion to be produced rather than a spectrum full of fragment ions of the analyt e. FAB was a widely used technique for many years because it was first ionization me thod to offer the analysis of involatile species without the need for derivatization techniques. In the late 1980s, a new i onization technique emerged th at would soon supplant the use of FAB. The techniqu e was called matrix-assisted laser desorption/ionization (MALDI) and today it is one of the most widely used techniques for the study of involatile species such as proteins, peptides and polymers. The technique began in two different parts of the worl d, but both groups described th e ability to desorb high molecular weight species in the solid phase with the use of a matrix material. In Germany, Franz Hillenkamp and Michael Karas developed a method from observations of the laser desorption of amino acids. 17, 36 Some amino acids absorbed the laser light from a frequency tripled NdYAG laser (355 nm) better than other amino acids. Mixing a highly absorbing amino acid (suc h as tryptophan) with one that absorbed weakly (such as alanine) and subjecting the mixture to laser irradiation (at one tenth of the power needed to desorb just alan ine produced a mass spectrum showing both compounds as intact ions ([M+H] + at m/z 74 for alanine and 189 for typtophan). Further experiments involved mixing a high concentr ation of nicotinic acid (a highly UV absorbing species) and a low concentrati on of bovine albumin (MW 67,000) on a metal probe tip. After drying to allow crystallizati on to occur, the spot was irradiated with a frequency quadrupled Nd-YAG laser (266 nm), producing one of the first mass spectra containing a peak corresponding to a large intact protein ion, [M+H] +

PAGE 44

26 Around the same time, Koichi Tanaka of Shimadzu Corporation in Japan performed experiments in which lysozome (chicken egg) having a molecular weight of 14,306 was mixed with slurry of cobalt powder in glycerol. 37 This solution was allowed to dry before irradiation with a nitrogen laser (337 nm). Th is mixture was al so successful in producing a spectrum containing a signal for the intact mol ecular protein. Both groups postulated that the matrix mate rial, an organic acid in the Hillenkamp and Karas work and a metal in Tanakas wo rk, acted as an intermediary, whereby it absorbed the energy of the laser and transfe rred the energy to the protein causing soft desorption. The technique today is called matrix-assisted laser desorption/ionization (MALDI) and is a combination of these tw o experiments. Typically, the matrix employed today is a small organic acid, using a pulsed nitrogen laser at 337 nm for irradiation. Although most pr actitioners of MALDI follow the procedure presented by Hillenkamp and Karas, in 2002, Tanaka was awarded a share of the Nobel Prize in chemistry for his contribution to the analysis of proteins. Theory of MALDI The theory of MALDI (and laser desorption) can be divided into two categories, the desorption process and the ionization pr ocess, as is true for most desorption techniques. Both categories are currently st udied with a desire to better unde rstand the process of MALDI and thus improve the reproducibility of the technique. It is evident from the research conducted thus far that the most critical aspect for not only the successful desorption of intact molecular ions, but also the ionization of desorbed ions, is proper selection of the matrix compound. In MALDI, the analyte of interest is mixed with an organic acid that absorbs at the wavelength of the laser. The ratio of analyte to matrix is typically 1:1,000 because part

PAGE 45

27 of the aspect in proper soft desorption of the analyte is to completely surround each analyte molecule with many matrix molecules. In this sense, the matrix is used to dilute the analyte molecules and isolate them from each other in the solid state (Figure 1-10). 38, 39 The problem with desorbing biomolecules wit hout matrix is a result of the interaction of biomolecules with other biomolecules. As the size of a molecule increases, the inter molecular forces tend to approach the intra molecular forces in strength. 32 For desorption of a neat analyte sample to occur, the i nter molecular forces have to be broken, releasing molecules into the gas phase. When the intermolecular forces are equal to the intra molecular forces, then the desorption process itself will induce fragmentation. 32 This aspect is part of the reason fo r the limitations of laser desorp tion to ionize only intact ions below a MW of 1000. 40 By isolating biomolecules from each other, the matrix material successfully reduces the inter molecular forces, thus helpi ng to reduce the fragmentation of biomolecules. Typical matrices used in MALDI analysis are show n in Figure 1-11. The desorption process is a fairly well unde rstood process. As long as the energy deposition is large enough, and the time-scale of the laser pulse is short (nanoseconds), a phase transition occurs directly from solid to gas. The key in ener gy deposition is in the amount of energy supplied per pulse, as measured by the fluence (J/cm 2 ), instead of the duration of the laser pulse, as re flected in the irradiance (W/cm 2 ). 41 Research has shown that lengthening the pulse appears to have little or no effect on the mass spectrum generated; however, there is evidence suggest ing the presence of a threshold of energy that must be reached in or der for desorption to occur. 42-45 The threshold of energy is matrix-dependent and suggests that there is a strong connection to the amount of energy

PAGE 46

28 Analyte Matrix Analyte Matrix Laser Mostly fragment ions Mostly intact ions A Laser, N2337 nm + + + -Positive ion Negative ionB Figure 1-10. Two diagrams of the MALDI pr ocess. Diagram A shows how the matrix material dilutes the analyte molecules to reduce the amount of intermolecular interactions, while the desorption of neat analytes especially biomolecules will induce fragmentation. The matrix is pr esent at a much higher concentration than the analyte to effectively act as an intermediary by absorbing the laser energy and transferring it to the anal yte without causing fragmentation. In Diagram B, the plume generated after a laser pulse contains ions, both positive and negative, related to the analyt e and matrix compound as well as many neutral species.

PAGE 47

29 OH O OH OH OH CN O OH OH O OH O O CH3CH3 2,5-dihydryoxybenzoic acid (Gentisic Acid or DHB) -cyano-4-hydroxycinnamic acid (HCCA) 3,5-dimethoxy-4-hyrdoxycinnamic acid (Sinapinic acid or SA)N OH O Nicotinic acid (NA)N NH2 3-aminoquinoline (AQ)OH OH OH OOH NH2NO2 2,4,6-trihydroxy acetophenone (THAP) para-nitroaniline (PNA)N OH O Picolinic acid (PA) Figure 1-11. Common organic aci ds used as matrices in MA LDI and their abbreviations. No single matrix is compatible with a ll analytes, and thus a wide variety of matrices have been employed.

PAGE 48

30 deposited in a single pulse. Once the threshold is met, large chunks of material (clusters) are ablated from the surface, which then undergo the phase transition. 38, 42 The phase transition occurs over a very short time frame, creating a plume of matrix molecules that entrain the analyte mol ecules. This expanding plume is similar to a free jet expansion under vacuum conditions and is an explosive event. An important aspect of the desorption process is the kine tic energy of the ions produced. The average initial kinetic energy of the ions generated ha s been of interest in an attempt to better understand to the MALDI process. One reason fo r these studies is to achieve better mass accuracy in mass analysis, especi ally with a time-of-flight analyzer, since the measured m/z ratio is directly related to the kinetic energy of an ion. An integral finding for peptides and proteins was that the velocity wa s the similar for all ions generated and that it was solely dependent on the matrix us ed and not the wavelength of the laser employed. 46 The average initial velocity of bovi ne insulin can be as high as 520 m/s (1163 mph) when using 2,5-dihydroxybenzoic acid (DHB) as the matrix to as low as 280 m/s (626 mph) with 2-(4-hydroxy-phenylazo )benzoic acid (HABA). Although the average initial velocity for peptides and proteins is very similar, different analyte classes exhibit a different initial velo city under the same experimental conditions. Furthermore, the average initial ion velocity is dependent on the preparation technique of the sample spot to be irradiated. 47 For example, varying the solven t of the matrix 3-hyrdoxypicolinic acid (HPA) changed the average initial velocity of insulin from as high as 620 m/s to a low of 444 m/s. 47 This change was related primarily to poor crystallization of the matrix, causing sample spot heterogeneity.

PAGE 49

31 The ionization process in MALDI is re flected by the mass spectrum produced. This provides a means for studying what may o ccur in the ionization region of the mass spectrometer. A better unders tanding of the mechanisms could assist in developing procedures that will maximize ion yields, provide for improved access to all types of analytes, aid in controlling the charge stat e produced, and control fragmentation induced during laser ablation. Central to the ionization process in MALDI ions is the very abundant production of singly charged species, either positive or ne gative. Multiply charged species are observed, but far less abundantly than in elec trospray ionization, which indicates a different ionization m echanism for the same types of compounds. 41 The production of multiply charged ions is somewhat higher when an infrared laser is used for MALDI, but still the most abundant ion signa l is for the +1 or +2 charge states. 38, 48 The ions generated can be protonated, deprotona ted, cationized, or radical ions and are generated by gas-phase interactions with the matrix rather than from preformed ions in the solid state. 49 The radical ions are usually associated with the matrix species, since radical formation is primaril y a result of photon absorption. 50 It is now generally believed that ioni zation in MALDI occurs via two steps, a primary ionization mechanism and a secondary ionization mechanism. 41, 51 The primary ionization event refers to the production of i ons from the neutral molecules in the solid sample, while the secondary event refers to ions that are generated from those produced in the primary event and are usually the spec ies representing peaks in the mass spectrum. The ions produced in the process are both pos itively and negatively charged and thus ion optics are used to select and gui de ions of the desired polar ity to the detector. Although ions are produced in MALDI, the overall io n-to-neutral ratio for ultra-violet (UV)

PAGE 50

32 MALDI is only about 10 -4 showing that the ionization proces s is inefficient and that the a mass spectrum produced shows the abundance of the minor species in the MALDI process. 52, 53 The primary ionization event is associated wi th the formation of ions from the large excess of matrix molecules. These i ons are the result of photon absorption (h ) causing the neutral matrix molecules (M) to be excited (M ) with a subsequent loss of an electron upon relaxation, leading to a radical cation (M + ), as shown in Equation 1-4. Protonated matrix molecules are generated from interactions of the radical ion with a neutral matrix molecule to form a protonated matrix molecule, [M+H] + and a neutral matrix radical, [M-H] as seen in Equation 1-5. M + h M* M + + e Equation 1-4 M + + M [M+H] + + [M-H] Equation 1-5 Prominent clustering in laser ablation indicates that it is possibl e to excite two matrix neutral molecules, M*M*, which th en pool their energy, creating one radical cation of the matrix and one neutral matrix and an electron, as shown in Equation 1-6. This excited state complex can also ionize an analyte molecule, A (Equation 1-7). Since negative ions are also observed in MALDI, there must also be a mechanism for producing them. A disproportiona tion reaction has been proposed as a possible means to generate negative matrix ions, as shown in Equation 1-8. 41 Other possible reactions have been proposed that could generate matrix a nd analyte ions such as excited state proton transfer, thermal ionization, and desorption of preformed ions. 41, 54 MM + 2h M*M* M + M + + e Equation 1-6 M*M* + A MM + A + + e Equation 1-7

PAGE 51

33 2M + nh (MM)* M + M + Equation 1-8 The secondary ionization event is charact erized by ionization in the expanding MALDI plume through ion-molecule reactions. It is believed that equilibrium has been reached at this time in the MALDI plume, allowing for secondary processes to be described by gas-phase thermodynamics. 54 Ion-molecule interactions occur between matrix ions and molecules and matrix ions w ith analyte molecules. A radical cation of the matrix from the primary ionization event can react with a neutral matrix molecule, producing a protonated matrix molecule, as sh own in Equation 1-9. A process similar to fast atom bombardment (FAB) can also o ccur in which dissociative electron capture occurs, generating a negative matrix ion and a hydrogen radical, H (Equation 1-10). 55 This hydrogen radical is then available for donation to an analyte or matrix molecule. Electrons are produced in some primary ionizat ion event reactions, but it has also been shown that electrons are ejected from the me tal sample plate due to the photoelectric effect. 56, 57 M + + M MH + and [M-H] Equation 1-9 M + e [M-H] + H Equation 1-10 Protonated analyte molecules are generate d from proton transfer reactions with matrix ions, MH + as long as the gas-phase basicity of the analyte is greater than that of the matrix (Equation 1-11). A deprotonated matrix molecule, [M-H] can abstract a proton from an analyte, A, producing a negative analyte ion as long as the gas-phase basicity of the deprotonated matrix is higher, which is typical when using basic matrices (Equation 1-12). MH + + A M + AH + Equation 1-11

PAGE 52

34 [M-H] + A M + [A-H] Equation 1-12 Cationization is also a significant proce ss occurring in the secondary ionization process. Cationization of the analyte is sign ificantly enhanced when an akali metal such as a sodium salt is added to the matrix and is thus present in the solid crystal state. A typical reaction is shown in Equation 1-13 for an analyte molecule with NaCl. A recent finding suggests that this cationi zation event is matrix-dependent and thus care must be taken in choosing the appropriate matrix if cationization is desired. 58 A + NaCl [A+Na] + + Cl Equation 1-13 Phospholipids and Sphingolipids The Structure of Phospholipids There are many classes of lipids present in all living organisms. They are characterized by their ability to be extracted from tissues by organic, non-polar solvents such as chloroform and methanol. They t ypically have a high hydrocarbon content and can have both polar and non-polar regions of the molecular structure. This section focuses on the structure of glycerophospholipid s and sphingolipids, bu t the structures of the other classes of lip ids can be found in most biology textbooks. The basic structure of a glycerophospholipid (GPL) is shown in Figure 1-12. GPLs are derived from the molecule glycerol and consist of two fatty acyl chains esterified to the sn -1 and sn -2 (systematic nomenclature) positions of the gl ycerol backbone that are commonly referred to as non-polar tails. The primary means of differentiating the GPLs is by the molecular composition of the polar head gr oup that is esterified to the sn -3 position of the glycerol backbone. There are four major classes of GPLs: phosphatidylcholines (PCs), phosphatidylethanolamines (PEs), phosphatidyl serines (PSs), and phosphatidylinositols (PIs). The structure of the polar head group for each class is shown in Figure 1-12.

PAGE 53

35 Figure 1-12. The basic structure of glycerophos pholipids (GPLs). All GPLs consist of a polar head group and two non-polar tails, fatty acids. The polar head group is CH2CHCH2O O O C C PO-O O O O Headgroup Fatty acid tail sn-1 Fatty acid tail sn-2Structures of the headgroups for GPLsCH2N+CH3CH3CH3 CH2N+H H H CH2C NH3 +H O-O OH OH OH OH H OH Phosphatidylcholine Phosphatidylethanolamine Phosphatidylserine Phosphatidylinositol Glycerophospholipids (GPLs)CH2CHCH2OH OH OH Glycerol sn-1 sn-2 sn-3 the primary means of separating GPLs in to different classes. The names of the four major classes are shown next to the structure of the head group.

PAGE 54

36 Another class of lipids present in animal cells and tissues is the sphingolipids. These lipids are characterized by the presence of the sphingos ine base (or a related base) as the backbone with a polar head group at the sn -3 position and only one variable fatty acyl chain at the sn -2 position. The head group, shown in Figure 1-13, also distinguishes the three major classes of sphingolipids: ceramides (CER), sphingomyelins (SPM), and cerebrosides (CRB). This class of lipids is primarily found in nerve tissue such as brain, sciatic nerve, and spinal cord. The names a dopted for them reflect this finding, especially sphingomyelins, because they are predominantly found in the myelin sheath, which is the protective covering and insulating me dium of the nervous system. Typical fatty acids present in eukaryotic cells and tissues have between 12 and 24 carbon atoms and can have varying degrees of unsaturation. The common names, number of carbon atoms, and the number of double bonds for many of the common fatty acids are shown in Table 1-1. A saturate d fatty acid does not have any double bonds, while an unsaturated fatty acid can have a nywhere from one to six double bonds present in the carbon chain. All the double bonds pr esent in a fatty acid chain are of the cis configuration instead of the trans configuration, as shown in Figure 1-14 for oleic acid (18:1) in a chemical structure a nd a 3-dimensional representation. The naming style adopted for the research conducted here uses the abbreviated name of the phospholipid followed in parenthesis by two sets of numbers for GPLs and one set for sphingolipids identifying the fatty acyl chain(s). The first set of numbers refers to the sn -1 position and the second set to the sn -2 position. Two numbers separated by a colon identify the fatty acyl chains: the first number refers to the number of carbon atoms in the fatty acyl chain, in cluding the carbonyl carbon, and the number

PAGE 55

37 CHCHCH2CH NH O CH C (CH2)12O OH CH3Headgroup Fatty acid tail sn-2H P O-OCH2N+CH3CH3CH3O O H H H OH OH H OH H OH Ceramide Sphingomyelin Cerebroside SphingolipidsCHCHCH2CH NH2OH CH (CH2)12OH CH3 Sphingosine base Structures of the headgroups for sphingolipids sn-1 sn-2 sn-3 Figure 1-13. The basic structure of sphingolip ids. This lipid class is derived from sphingosine or another similar base. The structure of sphingosine is shown at the bottom. They are similar to GPLs in that they have a non-polar tail, usually consisting of one fatty acid, a nd a polar head group. The polar head group is also used to differentiate the different classes of sphingolipid. The head groups for the major sphingolipids are shown above.

PAGE 56

38 Table 1-1. A list of the comm on fatty acids found in nature. CarbonDouble AtomsBonds Butyric acid 40butanoic acid Caproic Acid 60hexanoic acid Caprylic Acid 80octanoic acid Capric Acid 100decanoic acid Lauric Acid 120dodecanoic acid Myristic Acid 140tetradecanoic acid Palmitic Acid 160hexadecanoic acid Palmitoleic Acid 1619-hexadecenoic acid Stearic Acid 180octadecanoic acid Oleic Acid 1819-octadecenoic acid Vaccenic Acid 18111-octadecenoic acid Linoleic Acid 1829,12-octadecadienoic acid Alpha-Linolenic Acid (ALA) 1839,12,15-octadecatrienoic acid Gamma-Linolenic Acid (GLA)1836,9,12-octadecatrienoic acid Arachidic Acid 200eicosanoic acid Gadoleic Acid 2019-eicosenoic acid Arachidonic Acid (AA) 2045,8,11,14-eicosatetraenoic acid EP A 2055,8,11,14,17-eicosapentaenoic acid Behenic acid 220docosanoic acid Erucic acid 22113-docosenoic acid DH A 2264,7,10,13,16,19-docosahexaenoic acid Lignoceric acid 240tetracosanoic acid Chemical Names and Descriptions of some Common Fatty Acids Common Name Scientific Name

PAGE 57

39 CH3OH O transOH O CH3 cis Oleic acid (18:1) Figure 1-14. A typical fatty acid, oleic acid, is shown in two possibl e configurations. For all double bonds occurring in a fatty acid chain, the double bond is in the cis configuration rather than the trans configuration. The st ructures shown are of the basic shorthand form.

PAGE 58

40 after the colon refers to the number of double bonds present in the carbon chain (degrees of unsaturation). For example, PC (16:0, 18: 1) identifies the lipi d as phosphatidylcholine with two fatty acyl chains. The fatty acyl chain at the sn -1 position is palmitic acid (a 16carbon chain with zero degrees of unsatura tion), while the fatty acyl chain at the sn -2 position is oleic acid (an 18-carbon chain with one degree of uns aturation). This is the most common phospholipid in the cellular membrane of eukaryotic ce lls; its structure is shown in Figure 1-15. Another example is SPM (18:0). This lipid is classified as sphingomyelin with stearic acid, an 18-carbon chain having zero degr ees of unsaturation, at the sn -2 position (Figure 1-16). In general, the fatty acid at s n-1 is saturated, while the fatty acid at sn -2 can be saturated or unsaturated. Analysis of Lipids by Mass Spectrometry Due to the fact that lipids are present in every organism, they have been analyzed by mass spectrometry for many years. Initia l experiments used el ectron ionization (EI) for the analysis of derivatized lipids. In EI, gaseous analyte molecules, M, are ionized within the vacuum chamber usi ng an electron filament. This filament produces energetic electrons (typically at 70 eV ) that collide with the gaseous analyte molecules, causing the ejection of an electron from the analyte mo lecules and producing a positively charged radical, M + of the analyte molecule, as seen in Equation 1-14. M + e M + + 2e Equation 1-14 M + is called the molecular ion because it re presents the mass of the intact analyte molecular ion; however, in EI this ion is often very low in intensity due to the highly energetic process of electr on ionization. Under EI conditions, the analyte is often fragmented extensively, producing a spectrum very populated in the low mass region.

PAGE 59

41 CH3N+CH3O CH3P O O-O O O O O CH3CH3 PC (16:0, 18:1) Basic structure Figure 1-15. The structure of PC (16:0, 18:1) in a normal structure form is shown. This is the most abundant glycerophospholip id present in mammalian cellular membranes.

PAGE 60

42 O O-O P O CH3N+CH3CH3OH CH3NH O CH3 SPM (18:0) Basic structure Figure 1-16. Above is the structure of SPM (18:0). This is the most abundant sphingolipid in mammalian nerve tissue.

PAGE 61

43 One positive aspect of this fragmentation is that every compound will have a characteristic fragmentation pa ttern, allowing for the creation of a searchable database for the identification of co mpounds run by gas chromatography coupled to mass spectrometry (GC/MS) with an EI source. The extensive fragmentation can be a problem in the analysis of lipids, however, especially in the an alysis of unknown lipids. A requirement to perform EI is that the anal yte must have a high enough vapor pressure to be introduced into the ion source as a vapor. Most lipids do not have a sufficiently high vapor pressure to be analyzed directly by EI so they must be derivatized to a more volatile species. There are many techniques to derivatize most lipids, but some lipids, such as phospholipids, do not have a good procedure for derivatiza tion and thus cannot be analyzed by EI. Derivatization of phospho lipids first requires enzymatic cleavage of the phosphate head group, which makes classi fication more difficult because the head group in the distinguishing f eature of the different classes glycerophospholipids. Chemical ionization (CI) reduces the amount of fragmentation during the ionization process and is considered a soft i onization method when compared to EI. Soft ionization refers to the ability of an ionization technique to produce a dominant ion of the intact molecular species with li ttle or no fragmentation. CI is a proton transfer ionization process and can be performed on the same inst rument, but with a tighter ion source than EI. In CI, a reagent gas, typically methane, CH 4 is used to pressurize the source to 1 Torr. The same filament used in EI is tu rned on and the electrons from the filament collide with the reagent gas mo lecules rather than sample molecules because they present in large excess, producing reag ent ions from the methane gas. Table 1-2 shows the primary reactions that occur once a methane molecule, CH 4 is ionized by an electron, e

PAGE 62

44 70eV CH 5 + and C 2 H 5 + are the two primary sources of proton transfer to an analyte molecule because their conjugate bases (CH 4 and C 2 H 4 ) have relatively low proton affinities, 131.6 and 159 kcal/mol, respectively. Table 1-2. The common reactions that occur in chemical ionization when methane, CH 4 is used as the reagent gas. CH 4 + e 70eV CH 4 + + e thermal + e 50eV Ionization CH 4 + CH 3 + + H Fragmentation CH 4 + CH 2 + + H 2 Fragmentation CH 4 + + CH 4 CH 5 + + CH 3 Ion/molecule reaction CH 3 + + CH 4 C 2 H 5 + + H 2 Ion/molecule reaction CH 2 + + CH 4 C 2 H 3 + + H 2 + H Ion/molecule reaction C 2 H 3 + + CH 4 C 3 H 5 + + H 2 Ion/molecule reaction For proton transfer from a r eagent ion to an analyte molecule to occur, the analyte must have a higher proton affinity than the conjugate base of the reagent ion. If this stipulation is met, the analyte will be ionized to a protonated molecule ([M+H] + or [MH] + ). Due to the high pressure of the source, collisional cooling aids in the formation of this ion because it removes excess energy from the ion and thus reduces the amount of fragmentation that occurs. It is important to note that fragmentation of the [M+H] + ion is only minimized, and thus source fragmentation may still occur. Th e fragmentation is more related to the extent of energy transfer fr om a reagent ion to an analyte ion. If there is large difference in the proton affinities of the analyte and the c onjugate base of the reagent ion, there will be excess energy in the [M+H] + ion formed, which can cause fragmentation of the analyte ion thus reducing the signal for [M+H] + By using different reagent gases, the fragmentation can be controlled. Although there is a reduction in fragmentation, this technique still requires the analyte to be volatile and thermally stable for introduction into the ion source, so derivatiz ation of the phospholipid is still required.

PAGE 63

45 Fast atom bombardment (FAB) was devel oped in the early 1980s, and since that time, the analysis of lipids has changed dramatically. 59, 60 FAB was the first softionization technique to offer the widespr ead ability to analy ze non-volatile compounds without the need for time-cons uming derivatization techniques FAB is also called liquid secondary ion mass spectrometry (LSIMS) becau se it was developed from secondary ion mass spectrometry (SIMS). SIMS uses a primary ion beam to bombard a solid surface to eject secondary ions from the surface of the solid that can be mass analyzed, typically by a time-of-flight or magnetic sector instrument The difference between FAB and SIMS is the introduction of a low-vola tility liquid matrix in which the analyte molecules are dissolved or suspended. In FAB, the analyte and the matrix are mixed and then deposited onto a solid surface that is inse rted into the mass spectrometer. The matrix is typically glycerol because it has a very low vapor pr essure and thus does not rapidly evaporate from the solid surface when it is inside th e vacuum chamber. The primary ion beam collides with the matrix/analyte mixture at a 45 angle and ejects ions and neutrals of the analyte and the matrix from the liquid surface as shown in Figure 1-17. FAB is considered a soft ionization technique because the matrix absorbs the energy from the ion beam preventing internal ener gy build up in the analyte molecules, producing ions related to the intact analyte. Anothe r purpose of the liquid matrix is the continual renewal of the ablated area with fresh matrix and analyte. When analyzing a solid sample in SIMS, the area interrogated with the ion beam is dama ged, and the molecular ion signal ceases. The ion beam must be moved to a new area for continued production of molecular ions. The refreshing of the surface in FAB also solves a drawback of SIMS in that the ion beam does not penetrate into the sample very far, typically only a few nano meters, thus limiting

PAGE 64

46 A A A A A Primary ion or atom beam A+A+A+A+A-A-A-A-G+G+G+G+G G G G G GSecondary ions G refers to the matrix A refers to the analyteGSolid support Liquid matrix, G, with dissolved analyte, A Figure 1-17. A diagram of the fast atom bo mbardment, FAB, process. In FAB, the analyte, A, is dissolved in a matrix, G, typically glycerol, and an ion or atom beam is directed to the sample surface at an angle. Secondary ions of glycerol and analyte are ejected from the surface upon bombardment by the ion beam. These secondary ions generated are from the matrix and the analyte. This technique limits fragmentation of the analyte, generating intact molecular ions.

PAGE 65

47 the amount of material ablated. By refres hing the surface, the accumulation of signal can be accomplished which may aid in the anal ysis of lower abundant species. FAB allowed for the direct analys is of phospholipid species without derivatization, thus providing for better characterization of the lipid species in a biological sample. Coupling FAB to a mass analyzer capable of performing tandem MS presented the opportunity to identify the li pids based on their fragmentation patterns under CID. In a sense, this is akin to sp ectra collected by EI, but the fragmentation is better controlled. However, there are a lot of differences between a spectrum collected under EI and one collected by CID of a FAB generated ion. EI cause s a high degree of fragmentation producing many ions related to the structure of the analyte whereas CID imparts less energy to the analyte producing fewe r fragment ions, but still specific to the structure of the analyte. Because CID spectra are quite dependent on CID conditions, and CID is not always performed under the same conditions, there are not libraries of spectra in which a researcher could search to iden tify unknown species. When CID is used, it is better to have a standard of the analyte ion to allow comparison of MS/MS spectra of the standard and the analyte under investigati on collected on the same instrument. A limitation of FAB is the presence of very in tense peaks across the entire low mass range arising from clusters of the matrix, especia lly when glycerol is used. This can make identifying low-mass lipids very difficult, espe cially when they are in low abundance. A significant transformation in mass spectro metry occurred with the introduction of electrospray ionization (ESI). 18 This further advanced the analysis of lipids by mass spectrometry because ions could be generated directly from aqueous/organic solutions. With ESI, there is not a limitation in the size of a molecule, in fact practically every type

PAGE 66

48 of compound from small organic drug molecule s to large protein complexes is amenable to analysis. In ESI, a high voltage from 1000 V to 8000 V is applied to a small metal capillary into which fused silica, usually 100 m inner diameter, is insert ed. The applied voltage creates an electric field at the tip of the capi llary. An aqueous/organic solution is pumped at a constant flow rate thr ough the end of the capillary. As the solution emerges from the capillary and if the electric field is sufficient, a Taylor cone will develop. Small droplets containing both the volatile solvent and samp le ions emerge from the cone. As the solvent evaporates, the charge density in creases until the charge density exceeds the Rayleigh limit and then the ejection of ions occurs. Due to this process, multiply charged ions are routinely observed, especially from proteins and pept ides. The process is shown in Figure 1-18. A prerequisite of ESI is th at the compound must be an ion, or zwitterion, in the solution for ionization to occur. Th is means that non-polar compounds such as steroids cannot be ionized by ESI. This technique permitted the direct coup ling of liquid chromatography (LC) with mass spectrometry due to the ability to ioni ze a wide variety of compounds as intact molecular ions from a liquid phase. 61 By coupling these complementary techniques, more information could be generated from the same analyte solution, a retention time, a mass spectrum, and a MS 2 spectrum. Tandem MS of the intact molecular species has enabled direct structural characterization and identification of phospholipids, and has truly transformed the study of lipids. 62-68 Just as ESI has created a ne w field in the analysis of proteins called proteomics, the impact in lip id analysis has now ge nerated a new field

PAGE 67

49 + Power supply + + + + + ++ + + ++ + + + + + + + + + + + + + + + ++ + + + + + + + + Taylor Cone Inlet Capillary Desolvation region Ion ejection Figure 1-18. A diagram of the electrospray process. In el ectrospray ionization (ESI), a liquid solution is pumped through a sma ll capillary needle to which a high voltage is applied. As the solution em erges from the needle, a Taylor cone develops and then small droplets begin to evaporate. These droplets then evaporate further, producing charge d species in the gas phase.

PAGE 68

50 of analysis called lipidomics. This field is quite young, but it offers the potential to identify the role of lipids in cells, tissue s and organs to bett er understand biological disorders and diseases. 69 Despite the widespread use of both ESI and MALDI in the analysis of proteins and peptides, the adoption of MALDI for the anal ysis of lipids has remained limited. A search of journal articles with the keyw ords MALDI and lip ids produced only 90 publications, while a search w ith the keywords electrospra y and lipids produced 295 publications. Part of the problem with MALD I is the high abundance of matrix ions in the low mass region (under m/z 1000) of a mass spectrum. These ions can interfere in the analysis of lipids because the majority of lip ids as singly charged ions would be detected at a m/z value less than 1000. Since the majority of instruments employing a MALDI source are limited to one stage of mass analys is, distinguishing matrix ions from lipids ions can be difficult. For this reason, MALDI has primarily been a technique for the analysis of higher molecular weight compounds such as proteins, polymers, and DNA. The complexity of lipids in this low-mass region has also been a problem for the use of MALDI in their analysis. Appropriate MALD I preparation can yield very good results for lipids, especially the phospholipids and sphingolipids. 70-74 With the introduction of MALDI onto instruments capable of tandem MS, the analysis of lipids offers structural characterization as well as very good spatial resolution, enabling easier identification of solutions containing many lipids. 75-77 The spatial resolution is important in the analysis of tissue sections by MALDI especially fo r the generation of images related to the distribution of a specific compound in the tissue. Perhaps the analysis of lipid by MALDI will develop

PAGE 69

51 more readily with this capability because of the prominence of lipids in the cellular medium. Overview of dissertation The following chapters describe the an alysis of tissue specimens for small molecules, compounds having a MW less than 1000 amu, by intermediate-pressure (IP) MALDI. The purpose of this research was to develop a new technique that will enable researcher to better characterize compounds present in tissue specimens. Chapter 2 describes the first-ever analysis of tiss ue specimens, spinal cord, using IP-MALDI, operating at a pressure of 170 mTorr for ioni zation. This chapter shows the ability to desorb intact PLs from tissue at this pressure and to correctly identify them, as PCs or SPMs, using tandem MS. Chapter 3 examines the ion fragility of PLs by ESI-QIT-MS and MALDI-QIT-MS. This chapter describes ho w different ions from the same molecule can have different degrees ion fragility causing either source fragmentation or decreased mass resolution in mass analysis by QIT. The formation of cationized species of PLs is proven to show less source fragmentation and better mass resolution. With this understanding, MALDI matrices can be prepar ed and deposited onto tissue specimens to favor the formation of cationized species. Chap ter 4 describes the analysis of brain tissue sections by IP-MALDI for the generation of mass spectrometric images that show the distribution of specific PLs. Chapter 5 offe rs a conclusion to the areas examined and a look to the future of tissue analysis by IP-MALDI as well as a perspective on the use of quadrupole ion traps in conjunction w ith imaging mass spectrometry.

PAGE 70

CHAPTER 2 ANALYSIS OF INTACT TISSUE BY IN TERMEDIATE-PRESSURE MALDI ON A LINEAR ION TRAP MASS SPECTROMETER Introduction Prior to the advent of matrix-assisted la ser desorption/ionization (MALDI), direct mass spectrometric analysis of tissue sections focused primarily on the identification of elements in thin tissue sections, with one of the first experiments involving the detection of heme-bound iron from red blood cells using a laser microprobe mass analyzer, or LAMMA. 16 With the advent of soft ionization methods such as MALDI and electrospray ionization (ESI), characterizat ion of labile and involatile molecules such as proteins, peptides, and lipids present in biological tissu es has become possible. The use of soft ionization techniques allows for reduced comple xity in the analysis of biomolecules with mass spectrometry because intact molecular ions of the particular molecule are produced rather than fragment ions. Mass spectrometric microprobes employing MALDI or secondary ion mass spectrometry (SIMS) are now offering promising new approaches to map the distribution of small and large molecules directly from tissue sections at biologically significant levels 19, 20, 22, 78 and to help unravel the molecular complexities of cells. 79-83 Microprobe MS offers the unique ability to directly an alyze tissue samples for chemical species and to identify changes in the distribution of specific com pounds localized in tissue by generating images of specific ions or by comp aring spectra from different regions of the section. Many of the analyses have focused on the identification of proteins and peptides 52

PAGE 71

53 present or absent in normal and diseased tissue, 23, 84-87 but other applications have focused on the identification of site-specific drug activity 24 as well as lipids. 88-91 Besides tissue samples, artwork has also been analyzed for specific compounds to help determine their authenticity. 92, 93 Studies in our laboratory have demonstrated the ability to map paclitaxel, a small drug compound, at trac e (pg/mg) levels in ovarian tumors. 24, 27 Critical to such sensitive and selective analysis was the unique capability for tandem mass spectrometry (MS n ) that an instrument developed in our laboratory provided. Other studies have shown the need for tandem mass spectrometry for the identification and mapping of small drug molecules such as cocaine 94 and anti-tumor drugs 95 in tissue sections. The analysis of tissue sec tions by MALDI mass spectrometr y has traditionally been performed at low pressure (~10 -6 Torr). This low pressure requires samples to be dried completely, which requires approximately 2 hours, before exposure to vacuum conditions, thus prohibiting the analysis of fr eshly cut tissue. Analyses of phospholipids under traditional vacuum MALDI conditions have shown fragmentation of the phospholipid head group (unpublished result s), making low-level detection difficult; recent analyses of gangliosides, brain glyc olipids, using intermediate-pressure (IP) MALDI have shown decreased fragmentation for these labile biomolecules. 96 Intermediate pressures for IP-MALDI have ranged 97, 98 from 10 -2 Torr to 1 Torr, ten thousand to a million times higher pressure than traditional vacuum MALDI; in all cases, collisional cooling reduces the degree of source fragmentation. Both IP-MALDI, coupled to an FT-ICR, and atmospheric-pr essure MALDI (AP-MA LDI), coupled to a

PAGE 72

54 quadrupole ion trap, have been shown to re duce the amount of source fragmentation for more labile molecules such as gangliosides and phosphopeptides. 99, 100 Experimental The instrument used for all experiment s was a Finnigan LTQ linear ion trap mass spectrometer (San Jose, CA) fitted with a vMALDI source, as shown in Figure 2-1. The source consists of a N 2 laser (337 nm) directed to the s ource by a fiber optic cable; optics inside the source allow for lase r spot diameters from 80 to 120 m at an incident angle of 30 The laser spot size used for da ta collection was adjusted to 120 m. This position can be controlled to +/1 m without removing the plate from the vacuum chamber. If the plate is removed from the chamber and re-inserted, the precision is +/-6 m in the vertical direction and +/7 m in the horizontal. This source is designed to operate around 0.17 Torr. The standard vMALDI soft ware limits sampling to the normal 2 mm diameter sample wells, but tissue samples larger than this can be analyzed with custom software. All chemicals were obtained from Fisher Scie ntific (Fairlawn, NJ). The spinal cord was removed from normal Spague-Dawley rats at the University of California at San Diego and immediatel y stored in N 2 (l). Frozen spinal cord was cut (1 mm thick) with a scalpel and placed onto the vMALDI stainless steel sample plate. For all experiments, tissue specimens were allowed to dry for the typical 2 hours before matrix application. The matrix was electrosprayed from distance of 2 cm onto the surface at a flow rate of 7 L/min for approximately 10 minutes. The matrix used was 2,5 dihydroxybenzoic acid (DHB) at a concentration of 77 mg/mL in 70% methanol/30% water (with 0.1% TFA). Due to the use of a thick sample, the laser was unfocused when it impinged the surface of

PAGE 73

55 the tissue, so the laser power was increased to compensate. Ten laser shots were accumulated per spot to obtain each mass spectrum. Results and Discussion To the authors knowledge, direct analys is of tissue secti ons by MALDI at higher pressures has not previously been reporte d. Analysis of drug compounds in tissue sections using laser desorption/chemical ioni zation (LD/CI) at 1 Torr has been reported by our laboratory. 24, 101 Figure 2-2 shows spectra collected from the analysis of a section of spinal cord, along with an optical image of the area anal yzed (1.8mm x 2.3mm). The entire region was coated with the MALDI matrix. The black line in the optical image on the left indicates the outline of the spinal cord secti on. Spinal cord includes a variety of small and large molecular weight compounds from lipids to proteins, but lipids constitute a larger percentage by weight, and woul d be readily detectable by MALDI mass spectrometry. The phospholipids ions detected as singly charged i ons would range in mass-to-charge (m/z) from around 600 to under 1000. Spectrum A at the top of Figure 2-2 was taken from a location in which the spinal cord was not present but DHB matrix was present. No ions corresponding to phospholipids were detected from this regi on. On the other hand, a spectrum acquired when the laser was fired at a location on th e tissue (spectrum B, bottom of Figure 2-2) indicates ions present from m/z 700-900 (Figur e 2-3A) that were not detected in the spectrum in Figure 2-2A. To ensure that these detected ions corresponded to lipids, tandem MS was performed on eight of them. Table 2-1 lists these ions, along with the major fragment ions observed from MS 2 A neutral loss of 59 or the presence of the fragment ion m/z 184 is indicative of a phosphatidylcholine (PC) or a sphingomyelin

PAGE 74

56 Figure 2-1. A schematic diagram of the Finni gan LTQ linear ion trap instrument with vMALDI source modification. The vMAL DI source replaces the standard ow is Laser beam Sample plate Skimmerq00 q0AUX rods octopoleLens 0 3 section linear ion trap Lens1 MALDI modification API source, and includes the sample plate, RF-only quadrupole q00 with auxiliary rods, and skimmer. The RF-only quadrupole q0, octopole, and 3section linear ion trap are standard LT Q components. The open arrows sh the three stages of differential pumping; the pressure in the vMALDI region typically 0.17 Torr.

PAGE 75

57 sphingomyelin (SPM). The MS 2 spectrum of m/z 760.82 is di splayed in Figure 2-3B. Only one major fragment ion was detected, m/z 184.00 (C 5 H 15 NO 4 P), which is the phosphocholine head group from either a PC or an SPM. Using the nitrogen rule, this even m/z ion cannot be the [M+H] + from SPM due to the presence of an additional nitrogen in the ceramide backbone of SPM. No fragment ions were observed in the MS/MS spectrum that would correspond to the fatty acid tails (R 1 and R 2 ) of PC. MS 3 of m/z 184.00 would not provide any information pertaining to the fatty acid content because the fatty acid tails were lost in MS 2 It is possible to surmise the fatty acid tails based on the relative abundance of fatty acid chai ns expected to be present in spinal cord, but fragmentation information would provide more certainty. MS 2 of m/z 782.82 (Figure 2-4A ) reveals a different fragmentation pattern, a neutral loss of 59 (trimethylamine) to pr oduce m/z 723.36, and then a neutral loss of 124 to produce m/z 599.36 after MS 3 as seen is Figure 2-4B. This fragmentation pattern is characteristic of cationized PC and has been previously observed on triple quadrupole 62, 102 and quadrupole ion trap 103 MS/MS instruments. No ions corresponding to the fatty acid tails were detected from MS 3 Ions related to the fatty acid tails have been detected from the [M+Na] + of PCs using ESI and quadrupole i on trap mass spectrometry, but they were low in abundance. 103 MS 4 of m/z 599.36 was performed to determine if fatty acid information could be obtained, but no fragment ions were observed. In future experiments, to increase less a bundant fragment ions from MS 3 the trap will be filled for a longer period of time or multiple scans will be collected and averaged. Also, increasing the abundance of the cationized lipids by adding sodium to the matrix solution may aid in the identifying lower abundance fragment ions in tandem MS.

PAGE 76

58 Figure 2-2. On the left is an optical image of the spinal cord coated with DHB; the locations where the laser was fired to produce spectra A and B are indicated ions ipid ions. of 0 300 400 500 600 700 800 900 1000 m/z 0 50000 100000 150000 200000 250000 300000 350000Intensity 315.91 551.73 450.09 337.82 727.36 376.00 461.82 587.55 409.91 903.27 513.55 689.18749.45 637.55 925.18 879.45 300 400 500 600 700 800 900 1000 m/z 0 20000 40000 60000 80000 100000 120000 140000 160000 760.82 439.64 315.45 788.73 409.64 480.64 522.73 810.82 615.45 746.91 943.73 851.18 A B 300 400 500 600 700 800 900 1000 m/z 0 50000 100000 150000 200000 250000 300000 350000Intensity 315.91 551.73 450.09 337.82 727.36 376.00 461.82 587.55 409.91 903.27 513.55 689.18749.45 637.55 925.18 879.45 300 400 500 600 700 800 900 1000 m/z 0 20000 40000 60000 80000 100000 120000 140000 160000 760.82 439.64 315.45 788.73 409.64 480.64 522.73 810.82 615.45 746.91 943.73 851.18 A BOptical image of Spinal cord coated with DHB1.8mm by 2.3mm* * * ** DHB Cluster with circles. The spinal cord is outlin ed in black. Spectrum A shows the detected from a location away from the tissue, but in an area where the DHB matrix was still present. Most of these ions correspond to DHB clusters such as m/z 409.91, 551.73, and 727.36, as indicated by the dagger symbol. These ions appear at a higher mass than expected because the laser power was increased due to the thickness of the ti ssue sample and thus when the laser fired at a position off the tissue, space-charging was evident. No phospholipids were identifie d in that location. Spectrum B shows the ion signal from the tissue surface, indicating the presence of phosphol The ion at m/z 760.82 was determined by tandem MS to be the [M+H] + ion phosphatidylcholine 34:1 (PC 34:1). The other starred ions from m/z 700-90 were determined to be either PC or sphingomyelin, SPM.

PAGE 77

59 Figure 2-3. Spectrum A is an enlargement of the phospholipid region of the spinal cord tissue section mass spectrum from figure 2B. Spectrum B shows MS2 of m/z ion at f 760.82. Only one major fragment ion is produced, m/z 184.00, which corresponds to the phosphocholine head group of phosphatidylcholine. The neutral losses of 18 (water) and 154 c ould correspond to a DHB cluster the same m/z as the phospholipid ion. Th e inset displays the basic structure o a phosphatidylcholine, showing the frag mentation pathway that produces m/z 184. 700 720 740 760 780 800 820 840 860 880 900 m/z 0 20 40 60 80 100 760.82 788.73 782.82 810.82 746.91 813.82 798.82 735.00 835.82 851.18 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 184.00 743.09 607.27 568.18 732.00 760.45 700 720 740 760 780 800 820 840 860 880 900 m/z 0 20 40 60 80 100 760.82 788.73 782.82 810.82 746.91 813.82 798.82 735.00 835.82 851.18 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance Full scan MS2of m/z 760.82NL 18 184.00 743.09 607.27 568.18 732.00 760.45 Phosphocholine head group 184Protonated PC 34:1 [M+H]+A BN+O P O OH O O O O CH3CH3CH3O R2R1 NL 154

PAGE 78

60 These results indicated th at m/z 782.82 is the [M+Na] + analog of the [M+H] + ion at m/z 760.82. Even though sodium was not added to the matrix, there was enough sodium present in the tissue section for adduction to o ccur. This PC ion can be classified as either PC (16:0, 18:1) or PC (18:1,16:0), or in general PC (3 4:1) because fatty acid tail information was not obtained under fragmentation. Table 2-1. Ions detected from spinal cord in Figure 2A that were chosen for MS n analysis. Only the MS 2 major fragment ion m/z is shown in the table, but MS 3 was also performed. m/z MS2 major fragment Identification 735.00717.00DHB related 760.82184.00Phospholipid 782.82723.36(NL 59)Phospholipid 788.73184.00Phospholipid 798.82739.55(NL 59)Phospholipid 810.82751.55(NL 59)Phospholipid 835.82776.55(NL 59)Phospholipid 848.85831.09DHB related Other ions studied by MS n were m/z 735.00, 788.82, 798.82, 810.82, 835.82, and 848.85, as listed in Table 2-1. MS 2 of m/z 788.82 was perfor med with the major fragment ion of m/z 184 produced indicating this ion is a pr otonated PC with fatty acids adding to 36:1. MS 2 of m/z 798.82 showed a neutra l loss of 59, producing m/z 739.55, indicating it is cationized. MS 3 of m/z 739.55 produced m/ z 615.00; a neutral loss of 124, confirming that m/z 798.82 is cationized. Using a search engine of phospholipids developed in our lab, this ion was identified as the [M+K] + analog of the [M+H] + ion at m/z 760.82 and the [M+Na] + ion at m/z 782.82 (PC 34:1). MS 2 and MS 3 for m/z 810.82 and 835.82 also identified these ions as cationi zed species. A search through the database

PAGE 79

61 for m/z 810.82 indicates it is the [M+Na] + analog of the [M+H] + ion at m/z 788.82, PC 36:1, while a search for m/z 835.82 indi cates the ion is SPM 24:1, [M+Na] + Finally, the tandem MS spectra for m/z 735.00 and 848.85 both show a neutral loss of 154, which is most likely indicative of a DHB background ion. Conclusions The experiments and data reported here de monstrate that it is possible to analyze tissue specimens at a pressure five orders of magnitude higher th an traditional vacuum MALDI experiments. These results also clearly show the need for MS n for identification of small molecules desorb directly from tissu e sections. The major ions detected from the spinal cord secti on were determined by MS n to be phospholipids, primarily PC and SPM. Due to differences in fragmentation pa tterns, cationized and protonated PCs could be distinguished. Ions desorbed from the spinal cord not corresponding to lipids were also investigated using ta ndem MS. Although the effects of drying tissue specimens before MALDI analysis has not yet been eval uated, preliminary resu lts suggest that IPMALDI may allow for a reduction in amount of drying time necessary to directly analyze tissue specimens and therefore allow for eval uation of the effects of drying such as compound degradation or cell rupture. If the drying time can be reduced, sample throughput can be increased and the analysis of tissue samples within minutes of dissection may be possible.

PAGE 80

62 A B 250 300 350 400 450 500 550 600 650 700 750 800 m/z 0 20 40 60 80 100Relative Abundance 723.36 765.18 602.45 572.18 552.09 510.36 265.27 620.45 739.00 313.27 647.00 377.09437.91 702.91 200 250 300 350 400 450 500 550 600 650 700 750 800 m/z 0 20 40 60 80 100 599.36 577.64 466.91 705.00 661.18 250 300 350 400 450 500 550 600 650 700 750 800 m/z 0 20 40 60 80 100Relative Abundance 723.36 765.18 602.45 572.18 552.09 510.36 265.27 620.45 739.00 313.27 647.00 377.09437.91 702.91 200 250 300 350 400 450 500 550 600 650 700 750 800 m/z 0 20 40 60 80 100 599.36 577.64 466.91 705.00 661.18MS2782.82 MS3782.82 723.36 NL 18 NL 59 NL 124 NL 146 N+O P O O-O O O O CH3CH3CH3O R2R1 Na+ NL 59 NL 124, or 146 if Na+leaves Figure 2-4. Spectrum A shows MS 2 of m/z 782.82 from figure 2A, while spectrum B displays MS 3 (782.82 723.36 ). The neutral loss of 59 after MS 2 corresponds to loss of trimethlyamine in the structure for PC (inset of spectrum B). Two major fragment ions from MS 3 arise from the neutral loss of 124 (ethylphosphate) with the sodium re tained on the glycerol backbone, or the neutral loss of 146, when the head group retains the sodium. This fragmentation pattern is associated w ith cationized PC and SPM. Fragment ions relating to specific fatty acid tails R 1 and R 2 were not observed.

PAGE 81

CHAPTER 3 CHARACTERIZATION OF PROTONATED PHOSPHOLIPIDS AS FRAGILE IONS IN ION TRAP TANDEM MASS SPECTROMETRY Introduction This chapter is focused on the analysis of standard phospholipids and sphingomyelins by ESI and MALDI. Befo re the application of imaging mass spectrometry, it was important to better unde rstand the differences in ionization of phospholipids as protonated and sodiated species particularly the susc eptibility to source fragmentation. If fragmentation of the parent ion is occurring in the source, less of that parent ion will be available for mass analysis, wh ich in turn increases the detection limit. Glycerophospholipids (GPLs) serve a variety of functions from cellular signaling to protein transport and are abunda nt in all living organisms. 104 GPLs are linear in structure and typically consist of a glycerol backbone with a polar head group and two non-polar fatty acid tails. PLs with this basic structure are phosphatid ylcholines (PCs), phosphatidlyethanolamines (PEs), phosphatidylinositols (PIs), and phosphatidylserines (PSs) with the classes determined by the composition of the head group. Because of their biological abundance and importance, phospholipids have been studied by mass spectrometry for many years, from electron ionization coupled to gas chromatography/mass spectrometry (GC/MS) 61 to more recently matrix-assisted laser desorption/ionization (MALDI) 70-72, 105 and electrospray ionization (ESI) 62, 63, 67, 106 Under appropriate experimental conditions, the latter ioni zation techniques produce a predominant ion ([M+H] + or [M+Na] + ) corresponding to the mo lecular weight of the 63

PAGE 82

64 intact molecule whereas GC/MS causes a high degree of fragmentation and requires derivatization to increase volatility, but identification is performed using the immense chemical database available for EI spectra. In ESI, relying on the molecular weight for compound identification is inadequate because of the wide variety of phospholipids present and possible fatty acid combinations ; thus compound identification is typically performed by tandem MS with a triple quadrupole or ion trap mass spectrometer. 62, 67, 102, 107 The initial studies focused on the differe nces in fragmentation between protonated and sodiated phospholipids. For positive ions experiments showed that protonated and sodiated PCs produced very different fragmentation patterns, with sodiated PCs providing a more informativ e fragmentation pattern. 62 Under collision-induced dissociation (CID), protonated PCs produ ced one fragment ion, m/z 184, corresponding to the polar head group, indica ting that the charge is retained on the head group. In contrast, CID of sodiated PCs produced fragments that corresponded to losses of the head group with retention of the charge on th e glycerol backbone. The most abundant fragment in MS/MS showed a neutral loss of 59, which corresponds to the choline group, -N(CH 3 ) 3 62 The difference in fragmentation of protonated and cationized phospholipids has been of considerable intere st, particularly in the st udy of different metals for cationization. Due to the limitation of a single stage of tandem MS ( i.e. MS 2 ) on a triple quadrupole, the need for more in formative fragment ions in MS 2 is desired. The use of lithiated adducts was shown to provide many structurally informative fragment ions for PCs 102 and PEs 65 after a single stage of tandem MS, and many other cations have been

PAGE 83

65 evaluated for other GPLs 107 A structurally significant fr agment ion in the analysis of lipids allows for the correct identification of the fatty acid tails a nd their location on the glycerol backbone. A novel adduction w ith trifluoroacetic acid (TFA)/K + for PCs was determined to provide an abundance of struct urally informative fragment ions as well. 103 The resonating theme of most positive ion studies was that the cationized GPLs, or adducts with other complexes, should be preferred over protonated GPLs if structural identification is desired. This difference in fragmentation behavior under CID is also of concern in ion trap mass spectrometry due to ion fragility. In previous studies from our laboratory, 108 the fragility of an ion within an ion trap was qua ntified and results showed that different ions formed from the same molecule can exhibit varying degrees of fragility. For example, cationized oleandomycin is a stable ion, while its protonated counterpa rt is fragile. A consequence of fragility is an inability to e fficiently isolate a frag ile ion without widening the isolation notch, as well as mass shifts and reduced mass resolution. 108 Although a relationship between differences in fragmenta tion between different ions of the same compound in CID experiments and differences in ion fragility of those ions has not been evaluated, it appears to be an inte resting area for future studies. It has already been shown that cationi zed and protonated species of the same molecule exhibit varying degrees of fragilit y, but the effect on th e desolvation process has not be evaluated. The temperature of the heated capillary, and thus the effective temperature of an ion, has been shown to affect the onset of source fragmentation. 109 In MALDI mass spectrometry, source fragmenta tion of phospholipids was determined to

PAGE 84

66 arise from gas-phase reactions rather than from laser-induced photodissociation. 70 This is an important factor, since gas-phase reacti ons occur in electrospray as well. Fragmentation in ion transport has been studied before. Early studies in electrospray showed the possi bility of performing thermally induced dissociation (TID) of highly charged protein ions. 110 A heated capillary was used for those studies and the results indicated that the higher charged ions (+6, +5, and +4) were more susceptible to TID due to increased coulombic repulsions. Source collision-induced dissociation (SCID) has been used for many years for cont rollable dissociation of complexes and to provide fragment ions that can be further fragmented in a tandem mass spectrometer. The concept of ion fragility in ion trap ma ss analysis is re-examined for a different class of compounds and the effects on fragmentation in ion transport are explored by comparing three different phospholipids as prot onated and sodiated species, namely PC (16:0, 16:0), SPM (16:0), and PE (16:0, 16:0). Experimental For ESI, all experiments were perfor med on the Finnigan LCQ instrument (San Jose, CA), an electrospray ionization quadr upole ion trap (ESI-QIT) mass spectrometer. Analyte solutions were directly infused at a flow rate of 1 L/min using a syringe pump, 4.5 kV applied to the electrospray needle, a nd sheath gas set to 30 arbitrary units. For evaluating differences in ion fragility of protonated and cationized molecules, three phospholipids were chosen: phosphatidylcholine (PC), sphingomyelin (SPM), and phosphatidylethanolamine (PE). PC and SPM were chosen because the protonated and sodiated molecular ions exhi bit different fragmentation pa thways under collision-induced dissociation (CID) and th ey share the same head group. Th e main difference is in their

PAGE 85

67 fatty acid tails because SPM has only one fatty acid chain that can vary, whereas PC has two fatty acid chains that will va ry. PE was chosen because it has a similar head group to PC and SPM and also exhibits a different fragmentation pathway for protonated and cationized species. The only difference in th e head group is the re placement of choline (N(CH 3 ) 3 ) with amine (-NH 3 ), as seen in Figure 3-1. Standards of PE and PC with fixed fatty acid substituents of palmitoyl (16:0) were obtained from Avanti Polar Lipids (Birmingham, AL), while SPM was purchased from Avanti as a chicken egg extract, but with pa lmitoyl (16:0) as the predominant acyl chain (80%). All phospholipids were obtained as powders and prepared to the desired concentrations. PC and SPM were made as stock solutions of 1000 ppm in 50:50 isopropanol:methanol and PE was made as a stock solution of 500 ppm in 75:25 choloroform:methanol. For ESI QIT-MS analys is, the stock solutions were diluted to 10 ppm in methanol. Due to the presence of unwanted source fr agmentation and to uncover the origin of those fragments, solutions were prepared to ensure that either the [M+H] + ion or the [M+Na] + ion was solely present during analysis. For control of protonation, formic acid was added for a final concentration of 0.1% ; for the production of sodiated species, sodium acetate was added to a concentration of 100 M. All solvents were HPLC grade and were obtained from Fisher Scientific (Pittsburgh, PA). Unfortunately, because these lipids tend to prefer sodium, the protonated so lutions contained the sodiated species as well, probably because of sodium ions present in the methanol.

PAGE 86

68 N+O P O OH O O O O CH3CH3CH3CH3O CH3 O O O O CH3O CH3O P O OH NH3 + CH3N+CH3CH3O P O OH O OH NH O CH3CH3 SPM 16:0 PC 16:0, 16:0 PE 16:0, 16:0 Figure 3-1. Chemical structures of the phos pholipids used in the studying ion fragility and source fragmentation. The protonated version of the molecule is shown. SPM stands for sphingomyelin, PC sta nds for phosphatidylcholine, and PE stands for phosphatidylethanolamine. A ll three molecules are very similar in structure. SPM and PC share the same head group, but differ in the fatty acid tails, while PC and PE share the same fatty acid tails, but have slightly different head groups.

PAGE 87

69 Ion fragility in the ion trap was determ ined using three experiments described previously. 111 These experiments were 1) measuring the peak width of the parent ion at 10% peak height (PW 10% ) using a slow scan speed (called zoom scan), 2) finding the isolation width needed for efficiently isola ting the desired ion, and 3) determining the amount of energy required for CID. Typicall y, a fragile ion will need a wider isolation window, will exhibit a wider peak width at a sl ow scan rate, and will require less energy to fragment under CID. These experiments we re set up using Xcalibur software. Data for the zoom scan were averaged over 2 minutes. The isolation width was determined by changing the width from 1 to 4 amu wide in 1 amu increments, and collecting data for 1 minute at each interval. In studying the amount of energy required for CID, all experiments were conducted at a heated capillary temperature of 250 C and a tube lens offset of 30 V while changing the percent CID and collecting data for 1 minute at each level. Values for percent CID were 15% 16%, 17%, 18%, 19%, 20%, 21%, 25%, and 30%. Ion fragility in ion transport, or the susceptibility of ions to source fragmentation, was also studied using two injection parame ters on the LCQ, the temperature of the heated capillary and the voltage applied to th e tube lens (tube lens offset). These two parameters assist in desolva ting and declustering ions during injection. Typical operation involves a high capillary temperature, 250 o C, and a small voltage for the tube lens. The sensitivity of analysis is typically increased with a more positive tube lens offset for positive ions and a more negative value for negative ions; even higher voltages result in ion fragmentation. To determine the effect of temperature on fragmentation of phospholipids, an Xcalibur instrument control file was created that re called different tune

PAGE 88

70 files allowing the capillary temperature to be changed at 10-degree intervals from 50 C to 250 C. Prior to setting up the temperature profile scan, it was determined that a temperature change took 8 minutes before equilibrating; ther efore, the instrument file collected data for 10 minutes at each temperatur e level, but only the last two minutes of each temperature change were used for co mparisons between protonated and sodiated species. The tube lens was maintained at 0 V for these experiments. The final parameter used to understand the suscepti bility of phospholipid ions to source fragmentation was the tube lens voltage. For this experime nt, the heated capillary was held at 250 C and the tube lens voltage was changed; spectra were collected for two minutes at 4 voltage levels (0, 10, 20, and 30 V) and saved as different files. Results and Discussion The three phospholipids studied can form many molecular ions under ESI conditions. For example, the PC (16:0, 16:0), a synthetic phospholipid, if ionized without control of solvent counter ions would be detected as [M+H] + [M+Na] + and [M+K] + because these are the primary cations presen t in most solvents. Determining which molecular ion is the most stable for ion trap mass analysis is of importance to ensure good mass accuracy and lower detection limits (if the more fragile molecular ion is fragmented more easily in ion transport there w ill ultimately be less parent ion present for mass analysis). Up until now, a relationship be tween fragility in the ion trap and source fragmentation has not yet been explored. While operating the LCQ QIT-MS at lower capillary temperatures, a reduction in frag mentation of the head group from [M+H] + of SPM 16:0 was noticed. This reduction in fr agmentation was evalua ted further in the present study.

PAGE 89

71 Fragility in the trap The LCQ zoom scan feature allows for a higher resolution spectrum over a narrower mass range, 10 amu, by using a 20x slower scan rate. A fragile ion will have a broader PW 10% than a more stable ion. Examples of the zoom scan experimental data are shown in Figure 3-2 for the analysis of SPM 16:0 as [M+H] + (A) and [M+Na] + (B). Protonated SPM has a wider PW 10% than the sodiated SPM ion for the monoisotopic peak. A zoom scan was collected for each phospholipid stud ied and the results were summarized in the bar graph of Figure 3-3. All protonated phos pholipids exhibited wider peaks widths than their sodiated counterpart, indicating for this e xperiment that the sodiated species is more stable. PE showed the largest differe nce in peak width between the [M+H] + and [M+Na] + and was the only ion in which the peak was so broad that is otopic resolution was not obtained in a zoom scan. Pr evious studies determined that a PW 10% less than or equal to 0.30 amu was indi cative of a stable ion. 108 Using this value, both the sodiated and protonated species of SPM and PC could be considered fragile ions, while only the sodiated species of PE would be considered stable. It should thus be stated that the sodiated species of SPM and PC exhibit less fragility than the protonated species. For secondary confirmation that the sodiat ed species is more stable than the protonated species, data were collected for determining the isolation width needed to efficiently isolate the parent i on before CID. In this experiment, an efficient isolation width was defined as the width needed to reac h 50% of the intensity at an isolation width of 4.0. As can be seen from the graph in Figure 3-4, all the sodiated species permitted narrower isolation widths than the protonated species. However, the difference for PE is not as great as compared to PC and SPM, po ssibly showing both protonated and sodiated

PAGE 90

72 m/zRelative Abundance 701.5 702.0 702.5 703.0 703.5 704.0 704.5 705.0 705.5 706.0 706.5 0 20 40 60 80 100 703.5 704.3 705.2 723.5 724.0 724.5 725.0 725.5 726.0 726.5 727.0 727.5 728.0 728.5 0 20 40 60 80 100 725.4 726.3 727.3SPM [M+H]+SPM [M+Na]+ 0.44 0.36 m/zRelative Abundance 701.5 702.0 702.5 703.0 703.5 704.0 704.5 705.0 705.5 706.0 706.5 0 20 40 60 80 100 703.5 704.3 705.2 723.5 724.0 724.5 725.0 725.5 726.0 726.5 727.0 727.5 728.0 728.5 0 20 40 60 80 100 725.4 726.3 727.3SPM [M+H]+SPM [M+Na]+ 0.44 0.36m/zRelative Abundance 701.5 702.0 702.5 703.0 703.5 704.0 704.5 705.0 705.5 706.0 706.5 0 20 40 60 80 100 703.5 704.3 705.2 723.5 724.0 724.5 725.0 725.5 726.0 726.5 727.0 727.5 728.0 728.5 0 20 40 60 80 100 725.4 726.3 727.3SPM [M+H]+SPM [M+Na]+ 0.44 0.36A B Figure 3-2. Spectrum A, above, is a zoom scan of protonated sphingomyelin (SPM) 16:0. The peak width at 10% peak height (PW 10% ) was determined to be 0.44. The peak width for the sodiated counterpart, measured from the zoom scan in spectrum B, was 0.37. A narrower isolati on width indicates that the sodiated species is less fragile than the protonated species in ion trap mass analysis.

PAGE 91

73 igure 3-3. The chart above shows the PW10% for the two ions (protonated and sodiated) of each phospholipid studied. The peak widths were measured from zoom 0.44 0.36 0.70 0.29 0.47 0.37 0 0.2 0.4 0.6 0.8Peak width SPM PE PCPhospholipid [M+H] [M+Na]+ + F scan data. The zoom scan data was collected for 2 minutes and the peak width was measured from the average spectrum of each ion. For all three phospholipid classes studied, the protona ted species was determined to be more fragile because the p eak width was wider.

PAGE 92

74 Figure 3-4. A chart showing the isolation width needed to efficiently isolate the parent ion of each ion studied (protonated a nd sodiated) for each phospholipid. A 3.2 1.8 2.8 2.5 2.4 1.6 0 0.5 1 1.5 2 2.5 3 3.5Isolation width SPM PE PCPhospholipid [M+H] [M+Na]+ + wider isolation width is typically needed for a more fragile ion. The results from this experiment also indicate that the protonated species is more fragile than the sodiated ion.

PAGE 93

75 PE are fragile ions. This result is cont rary to the findings from the zoom scan experiment. Both PC and SPM share the same head group, phosphocholine, whereas in PE this head group is amine, -NH 3 rather than trimethylamine, -N(CH 3 ). This small structural difference may cause a dramatic change in intera nd intra-molecular bonding for PE ions in the gas phase. An interesting result was that even the more stable sodiated species required an isolation width of greater than 1 amu wi de implying that this ion is still unstable as compared to previous result s indicating that the stable ion, such as caffeine [M+H] + could be isolated with a 1 amu wide window. 108 Finally, the %CID, given as a percen t of the maximum 5 Vp-p resonance excitation waveform, required to reduce the pa rent ion signal by 50%, was determined for each ion; the results are shown in Table 3-1. These results indicate that the protonated lipids are more fragile than the sodiated li pids because less energy is required for 50% fragmentation. For the benefit of the reader CID spectra of the pr otonated and sodiated GPLs studied are shown in Figures 3-5 and 3-6, respectively. The major fragment ions for MS 2 of the [M+Na] + ions result from partial losse s of the head groups, -N(CH 3 ) 3 (NL of 59) for PC and SPM and C 2 H 5 N (NL of 43) for PE. On the other hand, CID of [M+H] + for PC and SPM produced one dominant fragment ion, m/z 184, the polar head group phosphocholine. This indicates that the head group retains the charge when the molecule is protonated. While CID of the [M+H] + for PE produced an ion corresponding to a neutral loss of 141, loss of the phosphoethanolamine head group, and the charge retained by the glycerol backbone. PE is onl y differentiated from PC by the replacement of three methyl groups with three hydrogens (Figure 3-1). Apparently, this change is enough to alter the fragmentation pathway under CID.

PAGE 94

76 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 184.07 703.53 685.87 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 184.22 734.73 719.63 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 551.53 692.40 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 184.07 703.53 685.87 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 184.22 734.73 719.63 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 551.53 692.40MS2SPM (16:0) MS2PC (16:0, 16:0) MS2PE (16:0, 16:0) [M+H]+ [M+H]+ [M+H]+ A B C Figure 3-5. Spectrum A is MS 2 of the [M+H] + for SPM (16:0) and spectrum B is MS 2 of the [M+H] + for PC (16:0, 16:0). Both of these spectra show the same fragmentation pathway producing a predominant ion at m/z 184, corresponding to the polar head group. Spectrum C is MS 2 of PE (16:0, 16:0) and shows a different frag mentation pathway. The major ion produced results from the neutral loss of the polar head group.

PAGE 95

77 Figure 3-6. MS 2 spectra of the [M+Na] + ions for each phospholipid studied. A is from SPM (16:0), B is from PC (16:0, 16:0), and C is from PE (16:0, 16:0). The fragmentation pathways are very similar for all these ions. They all result from neutral loss of the polar head group. A neutral loss of 59 for PC and SPM corresponds to the loss of choline (-N(CH 3 ) 3 ) and a neutral loss of 43 corresponds to the loss of ethanimine (-C 2 H 5 N). 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 666.27 725.20 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 697.27 756.40 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 671.07 714.27 164.00 458.13573.47 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 666.27 725.20 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 697.27 756.40 200 300 400 500 600 700 800 900 1000 m/z 0 20 40 60 80 100Relative Abundance 671.07 714.27 164.00 458.13573.47MS2SPM (16:0) MS2PC (16:0, 16:0) MS2PE (16:0, 16:0) A B C [M+Na]+ [M+Na]+ [M+Na]+ Table 3-1. The %CID needed to cause a redu ction in the absolute parent ion signal by 50% is shown for each ion studied in th e QIT. These results show that the protonated species fragments more readil y than the sodiated species, further proof that the protonated ion is fragile. Phospholipid [M+H]+[M+Na]+SPM 17.522.0 PE 17.022.0 PC 19.022.0

PAGE 96

78 Fragility in ion transport Fragmentation in ion transport could o ccur in the source region at atmospheric pressure, or in the ion optics re gions where the pressure is lowe r. If an ion is thermally unstable, the heat the ion absorbs could also cause fragmentation, called thermally induced dissociation (TID). TID has been st udied previously for protein ions, but the temperature required for di ssociation was from 250 o C to 400 o C. 110 The temperatures used in this experiment are much lower than th at, but still show the effect of TID. The effect of changing the heated capillary temperature from 50 o C to 250 o C is shown in Figures 3-7 and 3-8 for SPM 16:0. Figure 3-7 shows three different spectra collected at three different temperatures: 250 o C, 190 o C, and 130 o C for SPM (16:0). Because there was some sodium present in the solvent, both the protonated and sodiatied species were detected. With both ions present in the spectra, it is easy to follow the effect of TID on each ion as along, as the ions related to TID can be id entified. Since TID experiments have been shown to cause similar fragmentation pathways to MS 2 comparison of the MS 2 spectra in Figures 3-5 and 3-6 with these spectra permits the identification of the fragment ions associated with TID. 110 The fragment ion from the protonated species, m/z 184, is detected at all three temperatures, but shows a decrease in intensity at 130 o C, whereas the fragment ion from the sodiated species, m/z 666, is only detected at the highest temperature used, 250 o C. Running at 130 C reduces fragmentation for [M+H] + but also increases the S/N and the signal for solvent cluster ions, as would be expected because desolvation is not as effective at lower capillary temperatures. But at 250 o C, the intact molecular ion for the protonated species is almost

PAGE 97

79 150 250 350 450 550 650 750 850 950 0 20 40 60 80 100 184.1 725.5, [M+Na]+125.0 666.6 150 250 350 450 550 650 750 850 950 0 20 40 60 80 100Relative Abundance 725.3 703.3 184.1 365.1 148.9 337.3 413.1 150 250 350 450 550 650 750 850 950 m/z 0 20 40 60 80 100 703.3 725.3 365.1 184.1 337.1 148.9 413.3250C 190C 130CFragment of [M+H]+ 703.5, [M+H]+ Fragment of [M+Na]+ A B C 150 250 350 450 550 650 750 850 950 0 20 40 60 80 100 184.1 725.5, [M+Na]+125.0 666.6 150 250 350 450 550 650 750 850 950 0 20 40 60 80 100Relative Abundance 725.3 703.3 184.1 365.1 148.9 337.3 413.1 150 250 350 450 550 650 750 850 950 m/z 0 20 40 60 80 100 703.3 725.3 365.1 184.1 337.1 148.9 413.3250C 190C 130CFragment of [M+H]+ 703.5, [M+H]+ Fragment of [M+Na]+ A B C Figure 3-7. The effect of the heated capillary temperature on TID for the different ions of SPM (16:0) is shown in the three spec tra above. Spectrum A was acquired at 250 C, a normal capillary temperature for most analyses, and shows nearly 100% fragmentation for the [M+H] + ion of SPM (16:0). In contrast, there is less than 20% fragmentati on occurring for the [M+Na] + ion. Spectrum B shows the spectrum when the temperature is lowered to 190 C and spectrum C shows the spectrum at 130 C. It is clear that lo wering the temperature of the heated capillary reduces fragme ntation, TID, of both ions. At 130 C, extra peaks from m/z 300-550 are pres ent. These peaks are most likely clusters of solvent ions, because desolvat ion is not as effective at such a low temperature.

PAGE 98

80 completely fragmented. Operation of this instrument at low temperatures is not recommended because of this problem. Creating a way to dry or desolvate the i ons before interaction with the heated capillary will allow for proper operation at re duced capillary temperatures. Experiments were performed with such drying and analys is was successfully be performed at low capillary temperatures, but th e purpose of this study was to determine which ionized species would be better for analysis under normal ESI operating conditions. Figure 3-8 shows the complete temperatur e study, graphed, for both protonated and sodiated SPM (16:0). Data for this experiment were collected with a tube lens offset of 30V to avoid confusion from SCID. This comp lete investigation of the heated capillary temperature showed that protonated SPM was greatly affected by the temperature of the heated capillary as is evident from the increas ed fragmentation at a higher temperature. Fragmentation of SPM (16:0) occurs as a resu lt of ion transfer into the vacuum chamber, but can be minimized by operating at lower capillary temperatures. When running at 250 C, a normal operating temperature for most ESI analyses, the fragmentation for [M+H] + was 90%. From this experiment the fragmentation for the [M+H] + ion would be minimized, but no eliminated, by running w ith a capillary temperature under 200 C. On the other hand, the fragmentation of the [M+Na] + ion is minimal and reduced to almost zero at a capillary temperature of 160 C. The signal for the fragment ion at the highest capillary temperature, 250 C, was less than 20% of the parent ion. The apparent increase in fragmentation for this ion at lower temperatures is related more to the poor desolvation aspects of runni ng at such low temperatures. With poor desolvation,

PAGE 99

81 0.0 0.2 0.4 0.6 0.8 1.0 050100150200250300Capillary Temperature (C)Normalized Intensity [M+H]+[M+Na]+Parent ions Daughter ions Figure 3-8. A graph showing the intensity of the parent and daughter ions of the [M+H] + and the [M+Na] + ions versus changing the he ated capillary temperature for SPM (16:0). These data were collected automatically using Xcalibur software control. The temperature was adjusted at 20 intervals from 50 C to 250 C. From the graph, it is evident that the pr otonated species is more susceptible to TID than the sodiated species.

PAGE 100

82 increased signal-to-noise is more evident and this is more likely the cause of apparent fragmentation. This experiment clearly shows a very di fferent fragmentation behavior for the sodiated species and the protonated species of the same phospholipid. If a correlation exists between fragility inside the trap and fragility outside the i on trap, the protonated species would be considered fragile, or in other terms more labile. This fragility makes the protonated species more susceptible to TID and thus care must be taken when analyzing lipid solutions for pr otonated species. However, it would be more beneficial to induce cationization instead of protonation because of this extensive fragmentation. The same temperature experiment was pe rformed on PC (16:0, 16:0) and PE (16:0, 16:0). The effect of temperature on fragment ation for PC (16:0, 16:0) is shown in Figure 3-9. The data for this experiment also show that the protonated species is more susceptible to TID than the sodiated species but the temperature crossover, where the fragment ion equals the in tensity of the parent ion, was higher for the [M+H] + ion of PC (225 C ) than for the [M+H] + ion of SPM (190 C). In contrast to the high degree of TID obs erved for both SPM ( 16:0) and PC (16:0, 16:0), the TID for PE (16:0, 16:0) was not as intense. The graph for PE is shown in Figure 3-10. Although the protonated species di d show increased TID with respect to the sodiated species, the parent in signal was still dominant in the spectra. Tube lens offset To evaluate the effect of the tube le ns offset, the capillary temperature was maintained at 250 o C while adjusting the tube lens voltage. Data for [M+H] + and

PAGE 101

83 0.0 0.2 0.4 0.6 0.8 1.0 050100150200250300Capillary temperature (C)Normalized IntensityParent ions [M+H]+[M+Na]+Daughter ions Figure 3-9. A graph showing the effect of changing the capillary temperature for the [M+H] + and [M+Na] + ions of PC (16:0, 16:0). The same susceptibility of the protonated species to TID is evident, but the temperature at which the parent ion signal is lower than th e daughter ion signal occurs at a higher temperature than for the [M+H] + ion of SPM (16:0).

PAGE 102

84 0.0 0.2 0.4 0.6 0.8 1.0 050100150200250300Capillary Temperature (C)Normalized Intensity [M+H]+[M+Na]+Parent ions Daughter ions Figure 3-10. A graph showing the effect of changing the capillary temperature of the [M+H]+ and [M+Na]+ ions for PE ( 16:0, 16:0). The protonated species showed a greater tendency to frag ment upon increasing the capillary temperature, but the parent ion signal wa s still more intense than the fragment ion even at the highest temperature. This is in contrast to the other two phospholipids studied.

PAGE 103

85 Figure 3-11. A graph showing the effect of ch anging the tube lens offset from 0V to 30V on the [M+H]+ and [M+Na]+ ions for SPM (16:0). The capillary temperature species 0.0 0.2 0.4 0.6 0.8 1.0 05101520253035Tube lens offset (V)Normalized intensity [M+H]+[M+Na]+Parent ion Daughter ion Daughter ion Parent ion was held at 250 C for each voltage level. At the lowest voltage, fragmentation of the parent ion is st ill around 50%. Adjusting only the tube lens offset is not enough to limit the fragmentation. The sodiated appears to be unaffected by a change in the tube lens offset.

PAGE 104

86 [M+Na] + of SPM 16:0 are shown in Figure 3-11. For [M+H] + as the tube lens voltage was increased from 0 V to 30 V, the fragme nt ion, m/z 184, was the major ion detected. However, fragmentation was only reduced by 50% when running at 0 V, showing that by adjusting only the tube lens offset w ithout lowering the cap illary temperature, fragmentation cannot be completely eliminated for the [M+H] + ion of SPM (16:0). Again, fragmentation was seen to only a small degree for sodiated SPM. Data were also collected for PE and PC, and are summarized in Table 3-2. Crossover points were defined as the point at which the intensity of the parent ion was equal to the intensity of its fragment ion. As far as fragility during ion transport to the i on trap, protonated SPM appears to be the most fragile ion, requiring the lowest tube lens offset and the lowest capillary temperature to reduce fragmentation by 50%. Apparently, PE is not appreciably fragmented by conditions outside the trap. This may be re lated to the location of the charge on protonated PE. Unlike protonated SPM and PC, protonated and sodiated PE lose the entire head group as a neutral fragme nt, with the charge re tained on the glycerol backbone. The source fragment ions identified for each ionized species are consistent with the most abundant CID fragments of the same molecular ion. In matrix-assisted laser desorption/ionization (MALD I) of phospholipids, Al-Saad et al. 77 described these fragments as prompt and considered them distinct from post-source decay (PSD), but used PSD has a way to identify the source of t hose fragments. For PC and PE, ions were detected as [M+H] + and [M+Na] + The susceptibility of each ion to produce prompt fragments was discussed as we ll, and the protonated precursor ions tended to fragment more readily possibly due to the differences between ionic binding and proton binding.

PAGE 105

87 Even though that was a MALDI study, the same effect is to be evident in these ESI studies, although the term prompt may not be appropriate in ESI because the fragmentation appears to be pa rtially dependent on the temper ature of ion transfer and the energy needed to remove cluste rs (set by the tube lens offs et) which are parameters that can be adjusted to remove the fragmentat ion. Removal of the prompt fragments in MALDI was not discussed, only an u nderstanding of their source. Table 3-2. A summary of the results from each experiment performed in the studies of ion fragility in ion transport. From th ese results, it was determined that all sodiated species were less susceptible to fragmentation that can occur before mass analysis. Phospholipid Capillary Temp crossover (C) Max Intensity occurred at Capillary T (C) Tube lens offset crossover (V) % CID for 50% fragmentation SPM [M+H]+ 190 120 1.0 17.5 SPM [M+Na]+ Did not occur 170 Did not occur 22.0 PE [M+H]+ Did not occur 150 18.0 17.0 PE [M+Na]+ Did not occur 200 Did not occur 22.0 PC [M+H]+ 230 150 Did not occur 19.0 PC [M+Na]+ Did not occur 190 Did not occur 22.0 Initial Studies of Ion Fragility in the 2D/Linear Ion Trap IP-MALDI As described in Chapter 2, intermediate-pressure MALDI (IP-MALDI) reduces source fragmentation due to col lisional cooling that occurs at the higher pressure used (10 -1 Torr). Because of the difference in th e tendency to fragment, comparing the ionization of the [M+H] + and the [M+Na] + ions by intermediate pressure is of interest especially for imaging MS studies. Figure 3-12 shows a mass spectrum for SPM (chicken extract) collected by MALDI at a pressure of 10 -6 Torr. Figure 3-13 shows a mass spectrum for the same sample collected at a pressure of 10 -1 Torr. They were both acquired using the matrix 6-aza-thiothymine (ATT). Ions related to the matrix are

PAGE 106

88 indicated with asterisks in each figure. Wh en operating at the lower pressure, extensive fragmentation was evident indicated by a low si gnal intensity for the parent ions, [M+H] + and [M+Na] + The fragment ion at m/z 184.1 is the dominant ion in the spectrum and is produced from in-source fragmentation eith er by laser-induced photodissociation or energetic collisions of the [M+H] + ion. Increasing the pressure 100,000x in the source region to 10 -1 Torr reduced fragmentation to only 50% of the parent ion signal for the [M+H] + ion. The sodiated species was also detected at 10 -6 Torr and showed 50% fragmentation, while at 10 -1 Torr, the fragmentation was less than 10%. Each fragmentation percent was calcu lated by adding the intensity of the fragment and parent ions, then dividing the individual signal for each by the summed signal and multiplying by 100. The fragment ion from the sodiated species appears at m/z 666.82 in Figure 3-13 and m/z 666.5 in Figure 3-12. As in ESI, th e source fragmentation associated with the sodiated species is far less than that of the protonated species. Using different matrices may aid in the reduction of the fragmentation for [M+H] + ions, but the key is understanding that the sodiated species is less susceptible to fragmentation and thus would be a better ion to generate in the analysis of phospholipids. Mass analysis fragility With the introduction of the 2D ion trap, a comparison between ion fragility in the 3D ion trap and the 2D ion trap can be undertaken. The zoom scan experiment was repeated using the linear ion trap with a vM ALDI source. A sample of 10 ppm PC (16:0, 16:0) was mixed with 2,5-dihyrdroxybenzoic acid (40 mg/mL 1:1 MeOH/water). Both the [M+H] + and the [M+Na] + ions are present with this preparation. The laser was allowed to randomly fire across the sample pl ate while collecting a full scan and a zoom scan. A total of 75 scans were collected for each scan type. The spectra in Figure 3-14

PAGE 107

89 Figure 3-12. Mass spectrum of SPM (from chicken egg yolk) acquired by MALDI at 10 -6 Torr on a 3D-quadr upole ion trap. Fragmentation of the [M+H] + ion is nearly 90%, while fragmentation of the [M+Na] + ion is around 50%. The matrix was 6-aza-2-thiothymine. The aste risks indicate matrix ions.

PAGE 108

90 Figure 3-13. Mass spectrum of SPM (chicken egg yolk) acquired by MALDI at intermediate pressure (10 -1 Torr) on a 2-D quadrupole ion trap. Fragmentation of the [M+H] + ion is reduced, but not removed. Fragmentation of the [M+Na] + ion is also reduced, but still present. The matrix was 6-aza-2-th iothymine. The asterisks indicate matrix ions.

PAGE 109

91 733.0 733.5 734.0 734.5 735.0 735.5 736.0 736.5 737.0 737.5 738.0 m/z 0 20 40 60 80 100 734.68 735.56 736.54 737.54 755.0 755.5 756.0 756.5 757.0 757.5 758.0 758.5 759.0 759.5 760.0 m/z 0 20 40 60 80 100Relative Abundance 756.50 757.40 758.42 733.0 733.5 734.0 734.5 735.0 735.5 736.0 736.5 737.0 737.5 738.0 m/z 0 20 40 60 80 100 734.68 735.56 736.54 737.54 755.0 755.5 756.0 756.5 757.0 757.5 758.0 758.5 759.0 759.5 760.0 m/z 0 20 40 60 80 100Relative Abundance 756.50 757.40 758.420.60 0.35 PC (16:0, 16:0) [M+H]+PC (16:0, 16:0) [M+Na]+ Figure 3-14. Zoom scans of the two ions for PC (16:0, 16:0): A is the [M+H] + ion and B is the [M+Na] + ion. The spectra were acquired by intermediate-pressure MALDI on a linear ion trap, averaging 75 spectra. As was seen in the ESI studies on a 2D ion trap, the peak widt h at 10% peak height for the [M+H] + ion is wider than the [M+Na] + ion. This wider peak width indicates that the [M+H] + ion is fragile in mass analysis. *

PAGE 110

92 show the zoom scan for both the protonated (A) and sodiated species (B). As was seen in the 3D trap experiment with ESI, the peak wi dth at 10% of the peak height is wider for the protonated species than the sodiated speci es. This wider widt h indicates that the protonated species is a fragile i on. This was only a very initia l look at ion fragility in the 2D trap, but the results indicate a similar effect as in the 3D trap. Conclusions The fragility of the ions formed fr om model compounds of three phospholipid classes was studied, and results showed that all three classes are fragile within the ion trap as the [M+H] + ion, although PE showed fragility as both the [M+H] + and the [M+Na] + ions. From prior experiments, 108 a fragile ion exhibits a PW 10% of greater than or equal to 0.31 amu and requires a minimum of isolation width of 2.0 to isolate 90% of the parent ion signal. Accord ing to these results, both the so diated and protonated species of PC and SPM exhibited a degree of fragility, but the protonated species was determined to be more fragile because it exhibited a wider PW 10% and a required a wider isolation width. The fact that both ions of the same molecule exhibit a degree of ion fragility when compared to very stable ions may indicate that there is a structural characteristic of PC and SPM. Computer modeling of the stru cture of these two compounds as both the [M+H] + and the [M+Na] + ions and the other compounds previously studied may aid in determining if there is a structural correla tion to ion fragility. This result may also suggest that the presence of certain functional groups, such as phosphate, can affect the fragility of the ion. The fragility in ion trap mass analysis for PE was very interesting because it showed the widest peak width in zoom scan for the protonated species, but did not exhibit a large difference in %CID to fragment ei ther the protonated or sodiated species.

PAGE 111

93 Perhaps, PE does not fully desolvate from so lvent ions until much further inside the vacuum chamber, and thus when mass analysis is performed, solvent clusters still loosely bound to PE fall off during the voltage ramp. Two experiments studying fragility in ion transport were developed and showed that protonated SPM (16:0) and PC (16:0, 16:0) were also fragile, with decreased parent ion signals at high capillary temperatures and high tube lens offsets resulting from source fragmentation. Protonated SPM was nearly und etectable at a capillary temperature of 250 o C and tube lens offset of 30 V. Thus, SPM would need to be analyzed under 190 o C and at lower offsets if looking at protonated so lutions. For sodiated SPMs, this is not the case, and thus it may be more beneficial to use cationization when studying positively charged PLs. However, fragmentation cause d by these two ion transport parameters was minimal for protonated and sodiated PE, even though there is only a small change in the head group between P and SPM or PC. Clearly the gas-phase conformation (or site of protonation) or desolvation conditions are very different for PE than for PC and SPM. Using an intermediate-pressure MALDI ion source, fragmentation of both the sodiated and protonated species of PC was reduced. However, under the same conditions (laser power and pressure), the protonated species showed a higher degree of laserinduced photo-dissociation than the sodiated species. This offers more evidence to induce the formation of the ca tionized species esp ecially in the analysis of phospholipids from tissue sections. Finally, the comparison of ion fragility using a 3D versus a 2D ion trap was evaluated with the PC ions, [M+H] + and [M+Na] + These results were obtained with a MALDI source, but showed that ion fragility is conserved using th e newly described 2D

PAGE 112

94 ion trap. Furthermore, the PW 10% for both ions of PC was al so greater than 0.31 amu as was the case with the 3D ion trap indicati ng that both ions exhi bit a degree of ion fragility.

PAGE 113

CHAPTER 4 IMAGING PHOSPHOLIPIDS IN BRAIN TISSUE BY INTERMEDIATE-PRESSURE MALDI ON A LINEAR ION TRAP MASS SPECTROMETER Introduction Imaging tissue sections by matrix-ass isted laser desorption/ionization mass spectrometry (MALDI-MS) is a growing field in mass spectrometry. This is due in part to the prospects of analyzing for very specific molecular ions 80 directly from sectioned tissue and the possibility of correlating histolog ical techniques with the results obtained from imaging mass spectrometry. 112, 113 Another significant opportunity is in the identification of known and unknown compounds from the tissue surface using tandem MS. 95, 114 The field of imaging mass spectrometry (imaging MS) pertains to the direct analysis of surfaces, primarily tissue secti ons, using a focused ion beam (secondary ion mass spectrometry or SIMS) 35 or a focused laser beam with an highly absorbing matrix (MALDI). 23 In order to create an image, the laser or ion beam must be rastered in a discrete pattern across the tissu e surface. Each spot that the laser interrogates generates a mass spectrum and is considered a pixel with di mensions related to the spot size of the laser or ion beam. 19 In this sense, the MS image is e ssentially a post-acquisition feature in which the signal for a desired ion is extrac ted and displayed as a function of position. Recent advances in SIMS has also shown th e ability to analyze a larger area with a primary ion beam, and then mass analyze th e secondary ions, while preserving their spatial arrangement on a nanometer scale. This is a very new approach, and still suffers 95

PAGE 114

96 from the same disadvantages as traditional SIMS, but it does offer the ability to dramatically reduce the time of analysis. SIMS offers smaller spot sizes, typically su b-micrometer in diameter, but results in a high-degree of fragmentation, and is thus limited to compounds of a molecular weight less than 1000, whereas MALDI produces very little fragmentation, but typically with much lager spot sizes (25-150 m). There has been some limited analysis of submicrometer spot sizes for MALDI analysis with limited success. Although cluster SIMS is offering the ability to image intact ions near m/z 1000, 115, 116 MALDI is still better suited for more labile molecules such as proteins and peptides. 85 Because of this, a majority of the work in imaging MS has b een focused on determining the distribution of proteins and peptides in tissue sections. Smaller molecules such as drugs 24, 95, 117 and phospholipids 88, 90, 115 (no images for MALDI) have been analyzed, showing that a wide variety of compounds can be localized in tis sue sections with MALDI-MS and identified by tandem MS. 114 Until recently, tissue analysis by MALDI-MS has been evaluated at low pressures (~10 -6 Torr). The analysis of tissue sections by intermediate-pressure (IP)-MALDI was evaluated and phospholipids were successfully dete cted from rat spinal cord sections at a source pressure of 0.17 Torr, as was describe d in chapter 2 of this dissertation. IPMALDI instruments have ranged in pressure 98, 118 from 10 -2 Torr to 1 Torr for the analysis of standard samples and have shown a reduction of source fragmentation, presumably due to collisional cooling. 99 Although not yet evaluated, analyzing tissue sections at even higher pressure (near atmo sphere) may also provide a means to look at fresh samples without the need for drying.

PAGE 115

97 Tantamount to acquiring an image by mass spectrometry is the deposition of the matrix onto the tissue surface without dist urbing the original distribution of the compounds present (reducing analyte migration) Several techniques have been used to deposit matrix material, most of which aspira te a solution containing the matrix by either applying an electric current (e lectrospray) or a pneumatic ga s (nebulizer), or depositing very small droplets of matrix onto the surface (nano-spotting). 30 The use of an artistic airbrush to effectively coat th e tissue sample with matrix is presented for the first time. The capability to generate mass spectrometric images of phospholipids from brain tissue sections on an intermediate -pressure MALDI instrument operating at 0.17 Torr is discussed for the first time as well. Experimental Brain tissue from a rat (Spague-Dawley) was sectioned to a thickness of 10 m at 22 C using a microtome (Leica). The sections were placed directly onto 3 different types of microscope slides, plain gl ass, plus (polymer coated), and indium tin oxide (ITO) coated. The first two types of microscope slides are non-conduc tive and are typically used in the preparation of samples for hist ology, and the second slid e is conductive. The resistance of the ITO coated microscope slide was 40 over a distance of 1 cm. The tissue was allowed to warm on the microscope slides for 10 seconds before refreezing and storage at C. For mass spectrometric analysis, the tissue sections were removed from the freezer and placed in a dessicator for 30 minutes before applying the MALDI matrix. The matrix used for all experiments was 2,5-dihyrdoxybenzoic acid (DHB) at a concentration of 40 mg/mL in 70/30 methano l/water and final concentration of 20 mM

PAGE 116

98 NaCH 3 COO. All chemicals were obtained from Fisher Scientific. The matrix was applied to the tissue sections using an artistic airbrush (A ztek A470, Testors, Rockford, IL). Electrospray deposition could not be used because some of the sample surfaces used were non-conductive microscope slides. The ai rbrush has controls for solution and gas flow rate. The general-purpose tip (0.40 mm) wa s used for all matrix coatings, but the tip is easily changed from 0.30 mm to 1.02 mm. The solution and gas flow rates were adjusted using a blank piece of glass for ev aluation of the matrix coating. These two controls allowed for the easy adjustment of wetness and droplet size. A good coating wetted the tissue surface, but did not allow the accumulation of large puddles of solvent. A picture of the airbrush set up used is shown in Figure 4-1. After adjusting the airbrush controls, the tissue was passed underneath the spray and allowed to dry after each pass for about 30 seconds. The distance from the tip of the airbrush to the microscope slide was approximately 6 in. Approximately 20 pa sses were required for a good coating. The process took about 10 minutes for each tissue section. The instrument used for all experime nts was the Finnigan LTQ linear ion trap mass spectrometer fitted with the vMALDI s ource (San Jose, CA). A diagram of this instrument is shown in Figure 4-2. The source consists of a N 2 laser (337 nm) directed to the source by a fiber optic cable; optics outsi de the vacuum chamber allow for laser spot diameters from 80 to 120 m at an incident angle of 30 although adjusting the spot size is not trivial. The laser spot size for these experiments was 100 m. The source is designed to operate around 0.17 Torr. The sta ndard vMALDI software limits sampling to the normal 2 mm diameter sample wells, but tissue samples larger than this can be analyzed with custom software from Thermo.

PAGE 117

99 N itrogen line, 20 p si N ozzle, 0.40 m m Solution cup Fine adjustment knob Trigger Microscope slide Figure 4-1. Picture of the ai rbrush setup used to coat brain tissue samples on glass microscope slides (conductive and non-conductive). The trigger was held constant by the pressure from a ring stand clamp. For typical coating, the sample was about 6 in. from the airbru sh nozzle. The nozzle can be easily changed, but all experiments were co nducted with a nozzle of 0.40 mm.

PAGE 118

100 Laser beam Sample plate Skimmerq00 q0AUX rods octopoleLens 0 3 section linear ion trap Lens1 MALDI modification 30 Figure 4-2. Schematic diagram of the Finni gan LTQ linear ion trap instrument with vMALDI source modification. The vMAL DI source replaces the standard API source, and includes the sample plate, RF-only quadrupole q00 with auxiliary rods, and skimmer. The RF-only quadrupole q0, octopole, and 3section linear ion trap are standard LT Q components. The open arrows show the three stages of differential pumping; the pressure in the vMALDI region is typically 0.17 Torr.

PAGE 119

101 Acquisition of position-specific mass spectra For analyzing tissue samples, the custom software takes a picture of a desired area, 1000 m x 1000 m at a time. The area for analysis within this picture is selected by drawing a circle, square, or free draw around the area of interest. The step size can be set to a desired value. In the case of thes e experiments, the step size was set to 100 m, the same size of the laser spot size, to avoid ove r-sampling. To maintain the same number of laser shots at each point across the tissue sp ecimen, automatic gain control (AGC) was turned off. The number of laser shots and the power of the laser were determined by interrogating one spot of the ti ssue sample. Both the power and the number of shots were adjusted in order to avoid space-charging eff ects that can reduce the mass accuracy of the instrument. Typically 10 laser shots were needed to produce a strong signal from each point. Automated MS 2 data was collected from the tissue surface on a series of ions selected by the investigator. More laser shots are typically required when performing MS/MS and MS 3 experiments especially for com pounds present in low abundance to ensure that enough ions are trapped to pr oduce sufficient signal after CID. For the automated MS 2 experiments performed here, the numbe r of laser shots was not increased because sufficient signal was present for the production of good fragmentation patterns. Results and Discussion In Chapter 2, the direct analysis of ti ssue sections by IP-M ALDI pressure was discussed. A direct consequence of that work is the ability to generate images from tissue sections at an increased pr essure regime as long as mass spectra are collected in a position-specific manner. Figure 4-3 shows an optical image of a brain section (A) and a mass spectrum from one point in that se ction. The brain section was cut to 10 m and

PAGE 120

102 15 mm A Figure 4-3. The picture at the top of the figure, A, is an optical image generated from inside the mass spectrometer of a rat brain section coated with DHB matrix. It was acquired with 1 mm x 1 mm square pi ctures that are stitched together. This creates the lines in the picture. The mass spectrum, B, is the signal from the area on the tissue indicated by the arrow. The spectrum was acquired with 10 laser shots. A total of 11,156 spec tra were collected across the tissue section. The open circle is the laser spot size (100 m). B The mass spectrum detected from this spot on the tissue. This is a position-specific mass spectrum 9 mm 200 300 400 500 600 700 800 900 1000 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 782.8 757.0 810.8 754.0 184.5 724.0 832.7 697.9 856.7 958.5 857.6 353.4 673.0 198.6 301.4 551.6 200 300 400 500 600 700 800 900 1000 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 782.8 757.0 810.8 754.0 184.5 724.0 832.7 697.9 856.7 958.5 857.6 353.4 673.0 198.6 301.4 551.6

PAGE 121

103 then placed on a non-conductive glass micros cope slide. The arrow in the figure indicates the area in which this mass spectrum was acquired from (10 laser shots, 1 scan). There is a lot of signal generated in the region from m/z 700-900. Even though a non-conductive sample surface was used, the mass accuracy and peak shape are very good. Surface charging effects when using non-conductive glass microscope slides have been observed in time-of-flight experiments, 112 but may not be as important when using a mass analyzer that does not rely on the ini tial kinetic energy of the ions generated by MALDI. From previous results, 90, 114 these ions can be identified as arising from phospholipids; searching a data base of phospholipid ions (constructed in the lab) generated a large list of possible identities for each ion submitted to the search engine. The low intensity ion at m/z 184 resu lts from source fragmentation of phosphatidylcholines (PCs) and sphingomyelin s (SPMs) and is th e phosphocholine head group from these two phospholipids. This frag ment ion is only seen from protonated PCs and SPMs. 62 The low intensity is primarily due to the fact that sodium was added to the matrix solution in order to force the produc tion of primarily cationized adduct ions. Producing the sodiated adduct has three advantages: 1) the intensity of m/z 184 is significantly reduced, 2) the sodiated species is a more stable ion for QIT mass analysis (chapter 3), and 3) the fragmentation pattern s for CID of cationized phospholipids show more structural information, as was described in Chapter 3. 62, 64, 102 For mass spectrometric imaging, the diffi culty in producing a good image lies in the ability to apply a matrix coating that wi ll minimize the migration of analytes within the tissue specimen, preserving thei r original location, thus producing data that is specific

PAGE 122

104 to the position of the laser spot across the ti ssue section. A simple artistic airbrush was used to deposit the matrix for all the sample s analyzed. This mechanism provided a rapid and inexpensive way to deposit matrix without disturbing the distribut ion of analytes and is similar to the use of a glass nebulizer. The use of an airb rush or nebulizer also allowed for the coating of non-conductive surfaces, whic h electrospray deposition is not capable of. The mass spectrometric im age is created by extracting the intensity of a specific m/z value and the x and y coordinates at each point at which the laser was fired. By then mapping these coordinates into a contour or 3D plot, an image is created, as shown in figure 4-4. Figure 4-5 shows two mass spectrometric imag es generated from rastering the laser across a brain tissue section with a 100 m step size. The image at the top, A, represents the raw data showing the intens ity of an ion, in this case m/ z 756, with respect to position as was shown in Figure 4-4. The bottom image, B, is normalized to the total ion count. Normalization consisted of divi ding the intensity of the desire d ion at a given spot by the total ion intensity at that spot and multiplying by 1x10 5 This was done for each individual pixel to produce a much clearer imag e, as shown in Figure 4-5. The ion at m/z 756 is identified as the sodiated adduct of PC with fatty acid chains 16:0, 16:0 using tandem MS (MS 2 ) data.

PAGE 123

105 Figure 4-4. A is the digital image of the tissue section and B is the mass spectrometric image for the ion at m/z 756. Extract ing the intensity from each spectrum in the data file generated this im age. This is a raw image, with no further processing done to the data. The spectrum in C is an example of one of the 11,156 spectra that were collected. 200 300 400 500 600 700 800 900 1000 m/z 0 10 20 30 40 50 60 70 80 90 10 0 782.8 810.8 756.9 753.9 184.5 828.7 723.9 836.8 958.4 331.6 857.6 698.0 198.5 369.6 200 300 400 500 600 700 800 900 1000 m/z 0 10 20 30 40 50 60 70 80 90 10 0 782.8 810.8 756.9 753.9 184.5 828.7 723.9 836.8 958.4 331.6 857.6 698.0 198.5 369.6

PAGE 124

106 A B Figure 4-5. MS image A, top, shows the raw image generated by extracting the intensity of m/z 756 with respect to position. MS image B, bottom, is the normalized MS image for m/z 756. Normalization i nvolved dividing the intensity of m/z 756 at each pixel by the tota l ion intensity at that pixel and multiplying by 100000. All MS images further generated used this normalization procedure.

PAGE 125

107 Using a linear ion trap, it was not necessary to generate the lithium adduct because the low abundance fragment ions correspondi ng to the losses of fatty acid chains were readily detected from the sodiated species. CID of the lithiated adduct has been shown to produce more intense fragment ions arising from losses of either fatty acid chain that aid in structural characteri zation, after only a single stage of tandem MS (MS 2 ). 65, 102 The averaged MS 2 spectrum of m/z 756.6 is shown in Figure 4-6. Th e most abundant fragment ion, m/z 697.4, corresponds to a ne utral loss of 59, trim ethylamine, -N(CH 3 ) 3 The ions at m/z 573.4 and 551.4 are neutral losses (NLs) of 183 and 205 and can be identified as the loss of the phosphocholine he ad group with and without the retention of sodium, respectively, from sodiated PC. The ions at m/z 500.3, 478.4, and 441.3 provide structural information allowing for the identi fication of the fatty acid chains of this sodiated PC. From similar results with lith iated adducts, these ions are identified as NLs of palmitic acid (m/z 500.3, [M+Na-C 16 H 31 O 2 ] + ), the sodium salt of palmitate (m/z 478.4, [M-C 16 H 31 O 2 Na] + ), and the combined loss of trimethylamine and palmitic acid (m/z 441.3, [M+Na-N(CH 3 ) 3 -C 16 H 31 O 2 ] + ). 114 Since there are no ions re lating to losses of other fatty acid chains and because of the m/z of this ion, it can be identified as [PC (16:0,16:0)+Na] + The averaged MS 2 spectrum of m/z 782.6 is shown in Figure 4-7. This spectrum shows the same losses as m/z 756.6, NL of 59 at m/z 723.4 and NLs of 183 and 205 at m/z 599.4 and 577.5 respectively, but has some other low abundant losses that are different. The ion at m/z 526.3 is the NL of palmitic acid ([M+Na-C 16 H 31 O 2 ] + ), which is the same NL as with m/z 756, but this fragme nt ion is higher in mass by 26 indicating an

PAGE 126

108 200 300 400 500 600 700 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 697.4 500.3 478.4 696.4 441.3 573.4 390.3 551.4 389.0 427.0 698.4 602.8 x50 200 300 400 500 600 700 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 697.4 500.3 478.4 696.4 441.3 573.4 390.3 551.4 389.0 427.0 698.4 602.8 x50 N+O P O O-O O O O CH3CH3CH3O R2R1 Na+ NL 59 NL 183 w/o Na+NL 205 w/Na+ NL 278 R1CO2Na R1=16:0 NL 225 R1CO2H R1=16:0 NL 205 NL 183 NL 154 NL 59 Figure 4-6. The averaged MS 2 spectrum of m/z 756.6 (516 spectra were averaged). It is the average of 516 MS 2 spectra collected from various parts of the tissue section. Not that the in tensities of all the ions below m/z 520 have been expanded 50 times. Fragment ions representing the phosphocholine head group are the most abundant ions, but the less abundant fragment ions are used to determine the fatty acid com position of this phospholipid, identifying the ion as the sodiated adduct of PC (16:0,16:0). The less abundant ions used for identification were m/z 500.3 and m/z 478.4 because they represent the loss of the fatty acid chain with Na or without, respectively.

PAGE 127

109 igure 4-7. The averaged MS2 spectrum of m/z 782.6 (517 spect ra were averaged). As in Figure 4-6, the most abundant fragme nt ions correspond to losses of the r 300 400 500 600 700 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 723.4 526.3 500.3 599.4 478.4 467.3 264.3 577.5 392.3 x50 300 400 500 600 700 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 100Relative Abundance 723.4 526.3 500.3 599.4 478.4 467.3 264.3 577.5 392.3 x50 N+O P O O-O O O O CH3CH3CH3O CH3CH3 Na+1 3 2m/z 526.3 m/z 500.3 NL 282 R2CO2H R1=18:1 NL 256 R1CO2H R1=16:0 NL 205 NL 183 NL 59 F phosphocholine head group, while the less abundant fragment ions allow for the correct identification and location of the fatty acid tails. These ions are m/z 526.3 and m/z 500.3 and correspond to the loss of the sn -1 (C1) and sn -2 (C2) fatty acid chains, respectively. Th e ratio of these ions allows for prope assignment to the glycerol backbone. Th is ion was correct identified as the sodiated adduct of PC (16:0,18:1). Th e structure of this compound is shown in the inset with the assignments for the two fragment ions identifying the fatty acid chains.

PAGE 128

110 addition of ethylene. The ions at m/z 500.3 and 478.4 correspond to NLs of oleic acid ([M+Na-C 18 H 34 O 2 ] + ) and the sodium salt of oleate ([M-C 18 H 33 O 2 Na] + ), respectively. The combined loss of trimethylamine with palmitic acid (m/z 467.3) or oleic acid (m/z 441.3) are also identified. As in previous studies of lithiated and sodiated PCs, the loss of the sn -1 fatty acid chain occurs more readily than the loss of the sn -2 fatty acyl chain. 62 The intensity of m/z 526.3 (loss of palmitic acid) is higher than the intensity of m/z 500.3 (loss of oleic acid) indicating that m/z 782.6 can be identified as [PC (16:0, 18:1)+Na] + rather than [PC (18:1, 16:0) +Na] + The normalized mass spectrometric image of m/z 782.6 is shown in Figure 4-8 along with three other ions, m/z 756.8 [PC (16:0, 16:0+Na] + 804.6 [PC (16:0, 20:4)+Na] + and 828.6 [PC (16:0, 22:6)+Na] + MS 2 data enabled the correct positional identification of the fatty acid chains fo r each ion (data not shown for m/z 804.6 and 828.6). The intensity scales fo r the bottom two images have been adjusted based on the maximum intensity of the ion in the spectrum so all the images can be viewed together more clearly. The four ions presented in Fi gure 4-8 all have the same fatty acyl chain at the sn -1 position (16:0). When all these images are summed into one image, the image would be similar to results from mappi ng the signal for only palmitate (16:0, C 16 H 31 O 2 ) in negative ion SIMS (figure 4-9). 88 This may not be a true comparison because even though the both data sets were collected fr om rat brain, they were sectioned from a different area of the rat brain, and thus the im ages would not be identical. However, they both show that the phospholipids containing palmitic acid are primarily localized in the gray matter of the brain versus the white matter.

PAGE 129

111 PC (16:0, 16:0), m/z 756P C (16:0, 18:1), m/z 782 PC (16:0, 20:4), m/z 804P C (16:0, 22:6), m/z 828 A B D C Figure 4-8. MS images are shown of PC ions with 16:0, palmitic acid, at the sn -1 position on the glycerol bac kbone. A is the MS image of PC (16:0, 16:0), B is the MS image of PC (16:0, 18:1), C is the MS image of PC (16:0, 20:4), and D is the MS image of PC (16:0, 22:6). The ions were correctly identified from MS 2 data; classification is shown above each image. All the ions were the sodiated species, [M+Na] + The intensity levels were adjusted in images C and D for comparison purposes. As is evident from their comparison, the localization of these four PC ions is very different. The distribution of m/z 756, PC (16:0, 16:0) is prominently in the gray matter of the brain.

PAGE 130

112 igure 4-9. MS images showing the distributi on of palmitic acid, 16:0, in rat brain tissue. A is an image from MALDI imaging and is the summed image of four intact A B F ions (shown in Figure 4-8) containing palmitic acid at the sn -1 position of the glycerol backbone. B is an image from negative ion SIMS and shows the distribution of only palmitic acid, or an i on that has the m/z of palmitic acid. MALDI enables the analysis of primar ily intact ions, while SIMS provides better spatial resolu tion, but with a high degree of source fragmentation. B is adapted from reference 88.

PAGE 131

113 Although the spatial resolution of SIMS can be far better than that of MALDI, the increased information that MALDI, when combined with tandem MS, delivers better chemical resolution, and is therefore very useful in the identi fication of individual compounds for more chemically specific and sign ificant images. To il lustrate this point, when the fatty acid chain at the sn -2 position is changed, the localization of the PC ion is shown to be very different. PC (16:0,18:1) m/z 782, seems to be ubiquitous in the brain tissue with some areas of more locali zation, whereas PC (16:0,20:4), m/z 804, is primarily localized in the striata. Figure 4-10 shows four diffe rent PC ions with the sn -1 fatty acid chain as stearic acid (18:0) or oleic acid (18:1). Again, as the sn -2 fatty acid chain changes, the localization of the ion is shown to change very dramatically. In comparing m/z 756 (PC 16:0, 16:0) in Figure 4-8 with m/z 810 (PC 18:0,18:1) in Figure 4-10, the mass spectrometric images are nearly opposite in nature. This is primarily showing the difference between gray and white matter of the brain. These results are similar to previous studies with lipid extractions showi ng variations in the lipid content in these different types of brain tissue. 119, 120 Additionally, sphingomyelin (SPM) ions we re also detected from the rat brain tissue. Only m/z 753 (SPM 18:0) was subjecte d to CID. Identification was similar to that of PC ions except that SPMs are derived from the s phingosine base and thus only have one changing fatty acid chain, at the sn2 position. The struct ure of SPM is shown at the bottom right of Figure 4-11. This was the only ion chosen for CID, because it was the only SPM ion that was identified in the spectrum shown in Figure 4-3 based on the molecular weight. All SPM ions will have an odd molecular weight as an [M+H] + or

PAGE 132

114 PC (18:0, 18:1), m/z 810PC (18:1, 18:1), m/z 808 PC (18:0, 20:4), m/z 832PC (18:0, 22:6), m/z 856 C D A B Figure 4-10. MS images of PC ions with 18:0 or 18:1 at the sn -1 position of the glycerol backbone. A is the MS image of PC ( 18:0, 18:1), B is the MS image of PC (18:1, 18:1), C is the MS image of PC (18:0, 20:4), and D is the MS image of PC (18:0, 22:6). MS 2 data allowed for the correct identification of each ion.

PAGE 133

115 igure 4-11. Three MS images of the SPM ions detected from the rat brain tissue section (A is SPM (18:0), B is SPM (24:0), and C is SPM (24:1)). The structure at the SPM (18:0), m/z 753 SPM (24:0), m/z 837 SPM (24:1), m/z 835 O O O-P O CH3N+CH3CH3OH CH3NH R1O Na+ B C A F bottom right is SPM where R 1 is the variable fatty acid chain. Only m/z 753 was subjected to MS 2 ; the other two ions (835 an d 837) were not chosen for MS 2 analysis. The images for these two ions were generated because further research showed that they are primarily present in the white matter of the brain, whereas SPM (18:0) is present in the gray matter. As seen in the images, these three ions show a differe nt distribution indicative of prior research. 121

PAGE 134

116 [M+Na] + ion because of the extra nitrogen present in the sphingosine base. Images of the other two ions, m/z 835 and 837, were created after consulting a reference on basic neurochemistry. 121 Previous data have shown that lignoceric acid (24:0) and nervonic acid (24:1) are present in brain tissue as well. These two fatty acids are primarily detected in the white matter of the brai n, while stearic acid (m/z 753.9, SPM 18:0) is present in the gray matter. 121 Generating mass spectrometric images of all these ions shows this bimodal distribution (Figure 4-11). There are two significant reasons for missing these two ions in the initial studies: 1) analyzing 11,156 spectra for an unknow n mass is difficult and time-consuming (imagine generating images representing ev ery single m/z value!), and thus the less abundant ions tend to be overlooked because one s instinct is to look at the obvious, and 2) without prior knowledge of the important io ns present in a tissue specimen, the analyst is left to choose the appropriate ions for MS 2 or MS 3 analysis. Collecting an MS 2 image at every m/z value is impossible because th e tissue sample would be exhausted with investigation of unimportant ions, such as t hose related to matrix clusters, before finding an ion of interest. These two problems are not related to the type of mass analyzer or ionization method (MALDI or SIMS), but rather are significant concer ns for the field of imaging mass spectrometry to address. Targeted analysis thus remains the easier analysis to undertake when imaging tissue sections. Unfortunately, it is not possi ble to know which ions are unimportant, unless they can be positively identified as matrix cluster io ns, so it would be beneficial to be able to generate many images from the same tissue se ction. All the images presented up to this point have been generated from one data file, showing the possibility of creating multiple

PAGE 135

117 images from the same data set. All the MS 2 data were collected from the same tissue section as well, but not in a manner specific for generating images. However, the signal present was sufficient enough to enable to fu rther interrogation of that sample, Figure 412 illustrates this point. The mass spectrometric images shown in A (PC (16:0, 16:0), m/z 756.7) and B (PC (16:0, 20:4), m/z 804.7) were collected from the first analysis of a rat brain tissue section (10 laser shots per sport, 120 m spot size). Although this tissue section is from a different part of the rat brain, they show good correlation to the same images in Figure 4-8. The images shown in C and D were generated from mass spectral data that was collected afte r the tissue had been interrogated 31 times for 31 different experiments (10 to 15 laser shots per spot). The purpose of this experiment was to check if the signal was similar to the first run. It is remarkable that the images are still clear, in fact the images look better in C and D as compared to A and B. This may be due to the removal of excess matrix molecules or to an as yet unknown factor but it is clearly a feature of imaging mass spectrometry by MA LDI that will be explored further. Finally, the use of non-conductive glass mi croscopes for the imaging of tissue sections by time-of-flight mass spectrometry has been shown to decrease mass resolution and mass accuracy, especially with the addition of consecutive laser shots. 112 However, using non-conductive glass microscope slides with IP-MALDI on the linear ion trap, spectra produced from the tissue sections were similar in mass resolution and accuracy to those obtained on metal surfaces. Both c onductive and non-conductive glass microscope slides were used in this study and results did not show a significant change in mass accuracy in the spectra with ten laser shots collected per spot. A comparison of m/z 810, PC (18:0, 18:1), on a non-conductive microscope slide and on a conductive microscope

PAGE 136

118 Figure 4-12. MS images of PC (16:0, 16: 0), m/z 756.8 (A and C) and PC (16:0, 20:4), m/z 804.7 (B and D). A and B were generated from the first mass analysis of A PC (16:0, 16:0) m/z 756 PC (16:0, 20:4) m/z 804 PC (16:0, 16:0) m/z 756 PC (16:0, 20:4) m/z 804 B C D the rat brain tissue section, whereas C and D were generated from mass analysis that took place after 31 separa te experiments across the tissue were performed. It is eviden t from the pictures that C and D are much more spatially resolved.

PAGE 137

119 slide shows similar distributions for this phos pholipid ion in the rat brain (Figure 4-13) and similar spectra (Figure 4-14). Upon closer examination, the image generated from the non-conductive microscope slide (A) appears to be better spatially resolved than that from the conductive ITO microscope slide (B). In the cutting of the tissue and the subsequent placement on to microscope slides, it was noticed that the tissue did not seem to adhere to the conductive slid e as easily as the non-conductive slide. It is not clear at this point that differences in surface adhesion will affect the mass spectral analysis, but it is a factor that should be evaluated. Images were also generated from a brain ti ssue section that was placed on different non-conductive microscope slide that has a poly mer coating to help in retaining the tissue on the glass surface, call ed a plus microscope slide. The images and spectra were all similar to those already pres ented, showing that the tec hnique of imaging MS when coupled to an ion trap mass analyzer can ha ndle virtually any type of sample surface. With the ability to use microscope slides, re searchers already familiar with other methods of tissue analysis using a microscope can now submit similar samples for mass spectrometric imaging analysis. The main cr iterion, at least initi ally, is knowing the chemical species of interest. Identifying different featu res of the rat brain. Consulting a stereotaxic view of the rat brain, 122 an approximate section of the rat brain was located and then compared to the MS images generated from the rat brain. Figure 4-15 shows a comparison of the MS image for m/z 804.7, PC (16:0, 20:4), and a section from a rat brain atlas. 122 The intensity scale for the MS image was adjusted to a maximum level of 4000 instead of 5000, as was shown in Figure 4-8 C. This allowed for easier visualization of th e different regions in

PAGE 138

120 A Figure 4-13. MS image A was generated fr om a non-conductive glass microscope slide and shows the distributi on of m/z 810.7, PC (18:0, 18:1). MS image B was generated for the same ion from a conductive glass microscope slide. The brain tissues were 10 m thick and are serial sections from the same rat brain. The images are very similar, but the image from the non-conductive microscope slide (A) appears to be more spatially resolved. B m/z 810 PC (18:0,18:1) m/z 810 PC (18:0,18:1)

PAGE 139

121 200 300 400 500 600 700 800 900 1000 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 782.8 757.0 810.8 754.0 184.5 724.0 832.7 697.9 856.7 958.5 857.6 353.4 673.0 198.6 301.4 551.6 200 300 400 500 600 700 800 900 1000 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 782.8 757.0 810.8 754.0 184.5 724.0 832.7 697.9 856.7 958.5 857.6 353.4 673.0 198.6 301.4 551.6 TIC=2.56E6 A, non-conductive slide 200 300 400 500 600 700 800 900 1000 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 782.7 756.8 783.8 810.7 753.9 184.4 828.6 331.2 723.8 850.8 958.4 697.7 932.4 369.7 198.3 273.4 672.8 567.3 200 300 400 500 600 700 800 900 1000 m/z 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 782.7 756.8 783.8 810.7 753.9 184.4 828.6 331.2 723.8 850.8 958.4 697.7 932.4 369.7 198.3 273.4 672.8 567.3 B, conductive slide TIC=1.30E6 Figure 4-14. Spectra from similar spots of the brain tissue. A was acquired from a nonconductive microscope slide and B was acquired from a conductive microscope slide. The TIC for the non-conductive slide is 2x higher than that from the conductive slide.

PAGE 140

122 White matter, central hemisphere Figure 4-15. A is the MS image of m/z 804.7, PC (16:0, 20:4) and B is a stereotaxic view of a rat brain section (Bregma .04 mm) showing the different functional regions of the brain. The MS image shows good correlation to the atlas even with a spatial re solution of only 100 m. B A Interpeduncularnucleus, IRP Central gray substance, CG Hippocampus, HPC Dentate gyrus, FD Mesencephalicnucleus of trigeminal nerve Cortex Medial geniculate body, MGV and MGD Substantianigra, SNR Cerebral p eduncle, PC

PAGE 141

123 the rat brain. The rat brain atlas is genera ted from viewing a hematoxylin and eosin (H & E) strained tissue section, through a microsc ope. Using just the signal from this one phospholipid, some of the same areas of the ra t brain can be identified. Even with a spatial resolution of only 100 m, this MS image can be used to identify the cortex, white matter of the central hemisphere the central gray substance, the medial geniculate body, and the dentate gyrus. This result is c onfirmation that the matrix deposition method successfully minimizes analyte migration, while providing a mean s to desorb intact ions from tissue. This also shows the ability to use chemical-specific images to identify the regions of a rat brain as a possibl e comparison to current techniques. Quantitation. Quantitation by MALDI is rarely attempted due to the inherent variations in shotto-shot laser fluence, 54, 95 in matrix deposition, 123 in crystal formation, 54 analyte or matrix suppression, 124, 125 and in the distribution of analyte. 126 Although MALDI has been used in some cases for quantitation, it has never showed the reproducibility from sample to sample that has enabled liquid chromatogr aphy with ESI-MS to be so effectively employed for quantitation. An extensive study wa s recently performed for the analysis of a exogenous drug molecule from tissue specimens. 127 This study focused on the deposition of different concentrations of the drug into the tissue specimen to evaluate laser fluence effects, sample heterogeneity, ma trix deposition, and ioni zation efficiencies. The results showed the ability to detect varying concentrations of a drug compound, but did not investigate the analysis of varyi ng drug doses to a sample. A correlation to endogenous compounds was not evaluated either. Nevertheless, an in itial investigation into the prospects for quantif ying the chemical species present in a tissue section by

PAGE 142

124 imaging MS is worthwhile. This study did not begin with a goal of quantifying compounds present in tissue, so this comparis on is only a glimpse into the beginning of a new area for the field of imaging MS to evaluate. Table 4-1 compares published 121 data from an extraction of phospholipids, primarily PCs, from a rat brain as mol% with the data acquired in this study from the rat brai n tissue section by MS. For the MS data, all the spectra collected across the section (Figure 4-4) were averaged into one spectrum. The intensity for each ion in the table was ta ken from this spectrum. For generating the %int, the intensities of all the ions in the table were summed and each individual ion intensity was divided by this summed PC in tensity. As the table shows, the MS data show remarkable correlation to the lipid extract data even without adding an internal standard or other means of standardization. This agreement to th e literature values suggests that the ionization e fficiencies of all the PCs ar e similar, which is not too surprising given their structural similarities. It may be more difficult to quantify multiple species when the ionization efficiencies are not similar such as with a polar and a nonpolar compound. The literature data is from th e lipid extraction of an entire rat brain, while the MS data is only from one 10 m section of a rat brain so deviations may also be a result of comparing only a thin section to the whole brain. Table 4-1. A comparison of literature data from a lipid extract of a rat brain to MS data collected directly from a tissue section sn -1 sn -2 Literature (mol%)MS data (%Int) 16:022:6 3.3 5.9% 16:020:4 4.4 7.3% 18:120:3 Trace 18:020:4 3.8 6.4% 18:022:6 2.5 2.8% 14:016:0 3.1 1.0% 18:022:5 0.4 1.4% 18:118:1 3.4 3.9% 16:018:1 36.2 36.7% 16:016:0 19.2 22.7% 18:018:1 14.1 11.8%

PAGE 143

125 Finally, an average spectrum of 129 indivi dual spectra collected from the center area of the tissue section is s hown in Figure 4-16. The primary purpose of this figure is to show the complexity of analyzing GPLs using MALDI-MS. For PC (16:0, 18:1), there are four ions in the spectrum representing this compound. The most abundant ion corresponding to this co mpound is m/z 782.7, the [M+Na] + ion as expected because sodium acetate was added to the matrix soluti on to force this adduction. The next most abundant ion corresponding to PC (16:0, 18:1) is m/z 798.6. Tandem MS data of this ion determines it is the potassiated adduct, [M+K] + of PC (16:0, 18:1). The ion at m/z 723.8 is identified as a source fragment of the [M+Na] + ion, m/z 782.7, arising from loss of N(CH 3 ) 3 because it shows a neutral loss of 124 after MS 2 which corresponds to the MS 3 data of m/z 782.7, as was shown in Figur e 2-4 of Chapter 2. The fourth ion corresponding to PC (16:0, 18:1) is at m/z 958.4, which is identified as the [M+Na+(DHB+Na-H)] + the sodiated adduct clustered with a DHB matrix molecule. Again, MS 2 of this ion allowed for proper iden tification. The complexity of this spectrum is thus not just in the number of different compounds iden tified, but also in determining which ions are actually differe nt compounds and which ions are a different ion from the same compound. Using MS 2 data for each ion in the spectrum, this complexity can be unraveled, showing that most compounds identified have more than one ion in the spectrum. Conclusions The research reported in this chapter has shown the ability to generate mass spectrometric images of phospholipids in rat brai n at intermediate vacuum pressure on a

PAGE 144

126 m/z 650 700 750 800 850 900 950 1000 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 782.7 756.8 753.9 798.6 810.7 832.6 723.8 697.8 751.8 958.4 932.4 856.7 896.5 672.8 986.5 772.7 769.7 694.9 PC 16:0, 18:1 PC 16:0, 16:0 SPM 18:0 PC 16:0, 20:4 PC 18:0, 18:1 PC 16:0, 22:6 PC 18:0, 20:4 PC 18:0, 22:6 804.7 828.6 848.6 m/z 650 700 750 800 850 900 950 1000 0 10 20 30 40 50 60 70 80 90 100Relative Abundance 782.7 756.8 753.9 798.6 810.7 832.6 723.8 697.8 751.8 958.4 932.4 856.7 896.5 672.8 986.5 772.7 769.7 694.9 PC 16:0, 18:1 PC 16:0, 16:0 SPM 18:0 PC 16:0, 20:4 PC 18:0, 18:1 PC 16:0, 22:6 PC 18:0, 20:4 PC 18:0, 22:6 804.7 828.6 848.6 Figure 4-16. An average spectrum of 129 different position-specific mass spectra acquired from the rat brain tissue secti on. The complexity of this region of the mass spectrum can be unraveled using MS 2 data of each ion. It can therefore be determined that a singl e compound such as PC (16:0, 18:1) can be represented by four different ions in the mass spectrum, m/z 723.8 ([M+Na-N(CH 3 ) 3 ] + ), 782.7 ([M+Na] + 798.6 ([M+K] + ), and 958.4 ([M+Na+(DHB+Na-H)] + ).

PAGE 145

127 linear ion trap with a spatial resolution of 100 m. The images generated showed a varied distribution of PC and SPM ions in th e rat brain. Specific parts of the rat brain were identified based on a comparison of th e image of m/z 804.7, PC (16:0, 20:4) to a stereotaxic atlas of the rat brain. The locati ons of the fatty acid chains were correctly identified using tandem MS of the sodiated po sitive ion directly from the tissue section. The results also showed the ability to desorb ions from a tissue s ection attached to nonconductive or conductive microscope slides, wi thout any significant differences in mass accuracy or intensity. This suggests that when analyzing phospholipids from tissue by IP-MALDI on an ion trap, the section can be placed on nearly any surface. The effect of the different mounting surfaces on the ioni zation of other compounds has not been evaluated, but it is presumed that similar resu lts would be obtained. Finally, the use of an artistic airbrush for the application of the matrix to brain tissue sections was shown to effectively limit analyte migration and thus prepared tissue sections to produce ion specific images in an inexpensive and rapid manner. The mass spectrometric images of the phos pholipids in rat brain presented showed a remarkable variation in distribution. Th is varied distribution allowed from the identification of sub-structur es of the rat brain. The use on tandem MS provided a means to properly identify the ions detected and thus properly locate th e compound in the tissue section. An interesting finding was that th ere are sometimes up to 5 different ions detected and characterized by tandem mass spectrometry that represent the same compound. These different ions include protona ted, sodiated, and potassiated species as well as ions that are adducted with DHB matr ix molecules. This finding further shows

PAGE 146

128 the need to tandem mass spectrometry in th e analysis of small molecules from tissue sections. Most of the images presented in this chap ter were generated from one mass spectral acquisition showing the wealth of information hidden behind the data. The continued analysis of individual brai n sections was mentioned br iefly, but offers a unique opportunity to continually probe one tissue section for furthe r information. The example shown described that even after approximately 3,000,000 laser shots were fired at 10,000 (300 shots per spot) spots across the tissue, th ere was enough signal left to generate more MS images. In fact, the image generated afte r those shots were of better quality than the initial images. Finally, the potential for quantifyi ng endogenous compounds in tissue by IPMALDI was evaluated. Results showed th at phospholipids were detected with remarkable accuracy when compared to prev iously published results. This quantitation effort was very minimal, and thus future efforts will require a much more tedious approach, using some form of standardi zation. The compounds quantified most likely showed the same ionization potential; ther efore, compound having different ionization potentials will have to be evaluated for accuracy in quantitation. This may be of particular concern for drug discovery, especi ally if the active form of the drug has a different ionization potential than the dosed version of the drug.

PAGE 147

CHAPTER 5 CONCLUSION AND FUTURE WORK The goals of this research were to deve lop an instrument capable of analyzing tissue samples by intermediate-pressure matr ix-assisted laser desorption/ionization (IPMALDI) on an ion trap mass spectrometer and to make the technique of imaging mass spectrometry available to a broader segment of the scientific community. The advantages of IP-MALDI are reduced source fragmenta tion and the possibilities of looking at samples that are hydrated, although that last part may require operation at even higher pressures (atmospheric pressure). The research centered on the study of a particular class of compounds, phospholipids, because of thei r abundance and relevance to nerve tissue specimens. The use of IP-MALDI in the pressure re gion of 0.17 Torr has been presented as a new means for the imaging of thin tissu e sections by MALDI-MS. The chemical specificity that tandem mass spectrometry prov ides was shown to be instrumental in the identification of glcerophospho lipids (GPL) from both spinal cord and brain tissue and the identification of matrix ions. The initial research on spinal cord showed that very thick samples could be successfully analyzed by MALDI-MS with the proper adjustment of laser power due to cha nges in the focal point. Since ionization is key in mass analysis, a significant aspect of analyzing GPLs by ion trap mass spectrometry was the characteriza tion of the most stable form of GPLs as ions. In electrospray ionizati on (ESI), this research concluded that the cationized species, namely the [M+Na] + ion, was the most stable form for the phosphatidylcholines (PCs), 129

PAGE 148

130 phosphatidlyethanolamines (PEs), and sphingomyelins (SPMs) in both ion trap mass analysis and in ion transport. Studies of i on fragility in mass analysis showed that the protonated species of PC (16:0, 16:0), SPM ( 16:0), and PE (16:0, 16: 0) exhibited a wider peak width under slow scanning conditions, requ ired a wider isolation range for effective isolation, and fragmented more easily under collision-induced dissociation. A significant new finding in th is research was relating ion fragility inside the ion trap to ion fragility of get ting ions from an atmospheric pressure ionization source into vacuum and into the ion trap (ion transport). For PC (16:0, 16:0) and SPM (16:0), the protonated species tended to fragment very readily with increased temperature of the heated capillary, while the sodi ated species showed very little fragmentation. On the other hand, the [M+H] + ion of PE (16:0, 16:0) only showed a slightly higher tendency to fragment in ion transport. This may be re lated to the fragmentation mechanisms of PE versus SPM and PC. Under collisions-indu ced dissociation conditions, PC and SPM, as [M+H] + ions, fragment to an ion relating to their head group, phosphocholine, while the [M+H] + of PE fragments to an ion relating to the loss of the entir e head group. The different mechanism may mean that there is a different gas-phase structure of these PLs or that the difference in the head group stru cture is enough to change where the charge is located in the molecule. With protonated PC and SPM, the charge is retained on the head group, while for protonated PE the charge is retained on the backbone. Both head groups contain a phosphate group, so it would seem that the charge would be on the phosphate oxygen, particularly if it is a prot on. Further studies on the ga s-phase structure of these compounds will allow for clarif ication of this phenomenon. A possible route for this

PAGE 149

131 future study is using high-field asymmetric waveform ion mobility (FAIMS) with mass spectrometry to aid in identifying if the structure is signifi cantly different when adduction occurs or when protonation occurs or between PE and PC ions with the same fatty acid chain lengths. A comparison between ion fragility of ions stored in 2D and 3D ion traps showed that the aspects of ion fragility in mass analys is are the same in both arrangements. This shows that the phenomenon is not related to the design of the ion trap, but more related to the ions that are contained within and how those ions in teract with each other, the quadrupolar field, and the He damping gas. Further studies comparing these two different ion traps should include a re-e xamination of the same compounds already studied on the 2D trap such as caffeine, oleandomycin, and many of the explosive compounds. Ion transport issues on the 2D ion trap were not examined because the comparison of mass analysis used a different ionization means, namely MALDI. Future studies must also include ESI comparisons to MALDI. It was interesting to see that spectra generated by both MALDI and ESI showed the same type of source fragmentation tendencies for the protonated and sodiated species. The protonated species showed around 50% fragmentation under IP-MALDI conditions, while the sodiated species showed less than 10% for both SPM and PC. At lower pressure (higher vacuum) conditions (10 -6 Torr), the protonated species was nearly 100%, fragmented showing that collisional-cooling at high source pressure assisted in keeping this ion intact. In ESI, with the highest cap illary temperature and the highest tube lens offset, the protonated species showed greater than 90% fragmentation, while the sodiated species again showed less than 10% fragmentation. It was thus concluded that all

PAGE 150

132 analyses of PLs should be performed with a cationized species such as the [M+Na] + ion, the [M+Li] + ion, or the [M+K] + ion. A future study would be to investigate if these different cationized species exhibit varying de grees of ion fragility in mass analysis and ion transport and thus determine the best adduction ion for their analysis by mass spectrometry. Using position-specific mass spectra collect ed from brain tissue, chemical-specific images were generated that showed the distribution of individua l phospholipids in the brain tissue. These images represented the first MS images generated by IP-MALDI. These first images of the distribution of phospholipids in brain tissue offer a new possibility to characterize the brain in a very chemically si gnificant manner. An entire brain could be sectioned and mapped by MS, crea ting an image to show the levels of PLs in a normal brain. Such an approach could be invaluable in correlating the effects of aging to the deterioration of specific co mpounds or substructures of the brain. Tandem mass spectrometry provided the ability to positively identify the phospholipids as PCs and SPMs with the correc t fatty acid chain assignments using CID of the more stable ion, [M+Na] + These chemical-specific images provided a new view of the rat brain showing the va ried distribution of individual PC and SPM species. The PLs could be differentiated between white a nd gray matter of the brain. The different substructures of the brain were identified using the distribution of PC (16:0, 20:4), but other PLs could be used for this purpose as well. The use of an artistic airbrush was shown to be an effective technique to evenly coat the tissue surface with a matrix while inhibiting analyte migration. This provided a

PAGE 151

133 rapid and cheap method for coating multiple tissue specimens with the same matrix composition, but still requires a trai ned eye to identify a good coating. The intensities of the specific PC ions detected from a brain tissue slice were compared to previous data showing the concentr ation of fatty acids in the entire brain of a rat. This preliminary comparison showed that the levels detected from brain tissue by mass spectrometry were similar in concentra tion to those previous studies, and thus provided the initial view of using imagi ng MS to determine the concentration of compounds detected directly from the tissue sample. Quantitation studies in imaging MS will require the development of a more reproducible coating mechanism. The use of inkjet technology, cu rrently underway in our group, will provide a very reproducible co ating process and allow for the coating of multiple tissue samples with very little hum an error and effort. The use of mixed matrices in tissue analysis should be examined to determine if a wider variety of ions could be released from the surface of the tissu e, since it is known that one matrix cannot ionize all compound classes. Ot her experiments with the i nkjet printer would be the deposition of a standard compound with the ma trix at varied concentrations that will allow for absolute quantita tion of compounds. This woul d require examining how a standard is incorporated into the tissue medi um and the choice of solvents for uniform incorporation. The use of an inkjet printer will hopefully aid in the widespread adoption of the imaging mass spectrometry technique by producing a more robust and repeatable matrix deposition process. The identification of other phospholipids in brain tissue should involve studies in negative ion mode and the evalua tion of different matrices for negative ion MALDI. The

PAGE 152

134 correlation of phospholipid distribution to brai n function is an inte resting aspect for future studies and will involve significant collaboration with other fields of science, especially medicine. The successful us e of non-conductive microscope slides will hopefully increase the use of this imaging mass spectrometry techni que in conjunction with other well-known optical techniques for analyzing tissue specimens because of the similarity in sample preparation.

PAGE 153

LIST OF REFERENCES 1. Paul, W.; Steinwedel, H.: Patent 2,939,952 Germany, 1960. 2. Fischer, E. Z. Phys. 1959, 156, 1-26. 3. Dawson, P. H.; Whetten, N. R. Ion storag e in 3-dimensional ro tationally symmetric quadupole fields. II. A sensitive mass spectrometer. J. Vac. Sci. Technol. 1968, 5, 11-18. 4. Dawson, P. H.; Whetten, N. R. Ion storag e in 3-dimensional ro tationally symmetric quadrupole fields. I. Theoretical treatment. J. Vac. Sci. Technol. 1968, 5, 1-10. 5. Stafford, G.C; Kelley, P. E.; Syka, J. E. P.; Reynolds, W. E.; Todd, J. F. J. Recent improvements in and analytical applica tions of advanced ion trap technology. Int. J. Mass Spectrom. Ion Processes 1984, 60, 85-98. 6. Eades, D. M.; Johnson, J. V.; Yost, R. A. Resonance in a quadrupole ion trap. J. Amer. Soc. Mass Spectrom. 1993, 4, 917-929. 7. Guan, S.; Marshall, A. G. Space charge in an ion trap. J. Amer. Soc. Mass Spectrom. 1994, 5, 757-764. 8. Stafford, G.C; Taylor, D. M.; Bradshaw, S. C.; Syka, J. E. P.; Uhrich, M., 36 th ASMS conference on mass spectrometry and allied topics, San Francisco, CA 1987. 9. Hager, J. W. A new linear ion trap mass spectrometer. Rapid Commun. Mass Spectrom. 2002, 16, 512-526. 10. Schwartz, J. C.; Senko, M. W.; Syka, J. E. P. A two-dimensional quadrupole ion trap mass spectrometer. J. Amer. Soc. Mass Spectrom. 2002, 13, 659-669. 135

PAGE 154

136 11. Stafford, G.C. Ion trap mass spect rometry: A personal perspective. J. Amer. Soc. Mass Spectrom. 2002, 13, 589-596. 12. March, R. E. An introduction to qua drupole ion trap mass spectrometry. J. Mass Spectrom. 1997, 32, 351-369. 13. Soni, M. H.; Cooks, R. G. Selective inje ction and isolation of ions in quadrupole ion-trap mass-spectrometry using notched wave-forms created using the inverse fourier-transform. Anal. Chem. 1994, 66, 2488-2496. 14. Louris, J. N.; Taylor, D. M.: Patent 5,324,939 US, 1993. 15. Louris, J. N.; Cooks, R. G.; Syka, J. E. P.; Kelley, P. E.; Stafford, G. C.; Todd, J. F. J. Instrumentation, applications, and energy deposition in quadrupole ion-trap tandem mass-spectrometry. Anal. Chem. 1987, 59, 1677-1685. 16. Kaufmann, R.; Hillenkamp, R.; Nitsche, R.; Schurmann, M.; Unsold, E. Biomedical Application of Laser Microprobe Analysis. J. Micro. Biol. Cell. 1975, 22, 389-398. 17. Karas, M.; Bachmann, D.; Bahr, U.; Hille nkamp, F. Matrix-assisted ultravioletlaser desorption of nonvolatile compounds. Int. J. Mass Spectrom. Ion Processes 1987, 78, 53-68. 18. Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Electrospray ionization for mass spectrometry of large biomolecules. Science 1989, 246, 64-71. 19. Todd, P. J.; Schaaf, T. G.; Chaurand, P.; Caprioli, R. M. Orga nic ion imaging of biological tissue with secondary ion mass spectrometry and matrix-assisted laser desorption/ionization. J. Mass Spectrom. 2001, 36, 355-369. 20. Pacholski, M. L.; Winograd, N. Imaging with Mass Spectrometry. Chem. Rev. 1999, 99, 2977-3005. 21. Stoeckli, M.; Chaurand, P.; Hallahn, D. E.; Caprioli, R. M. Imaging mass spectrometry: A new technology for the analysis of protein expression in mammalian tissues. Nature Med. 2001 7, 493-496.

PAGE 155

137 22. Chaurand, P.; Dague, B. B.; Pearsall, R. S.; Threadgill, D. W.; Caprioli, R. M. Profiling proteins from azoxymethane-induc ed colon tumors at the molecular level by matrix-assisted laser deso rption/ionization mass spectrometry. Proteomics 2001, 1, 1320-1326. 23. Caprioli, R. M.; Farmer, T. B.; Gile, J. Molecular imaging of biological samples: localization of peptides and proteins using MALDI-TOF MS. Anal. Chem. 1997, 69, 4751-4760. 24. Troendle, F. J.; Reddick, C. D.; Yost, R. A. Detection of pharmaceutical compounds in tissue by matrix-assisted laser desorption/io nization and laser desorption/chemical ionization tandem ma ss spectrometry with a quadrupole ion trap. J. Amer. Soc. Mass Spectrom. 1999, 10, 1315-1321. 25. Garden, R. W.; Moroz, L. L.; Moroz, T. P.; Shippy, S. A.; Sweedler, J. V. Excess salt removal with matrix rinsing: Direct peptide profiling of neurons from marine invertabrates using matrix-assisted lase r desorption/ionizatio n time-of-flight mass spectrometry. J. Mass Spectrom. 1996, 31, 1126-1130. 26. Kruse, R. A.; Rubakhin, S. S.; Romanova, E. V.; Bohn, P. W.; Sweedler, J. V. Direct analysis of aplysia tissues and cells with laser desorption/ionization timeof-flight mass spectrometry on porous silicon. J. Mass Spectrom. 2001, 36, 13171322. 27. Troendle, F. J. A Quadrupole ion trap laser microprobe for the mapping of pharmaceutical compounds in intact tissue. Ph.D.Dissertation, University of Florida 2000 28. Kaiser, R. E.; Cooks, R. G.; Stafford, G. C.; Syka, J. E. P.; Hemberger, P. H. Operation of a quadrupole ion trap ma ss spectrometer to achieve high mass charge ratios. Int. J. Mass Spectrom. Ion Processes 1991, 106, 79-115. 29. Cooks, R. G.; Ouyang, Z.; Takats, Z.; Wi seman, J. M. Ambient mass spectrometry. Science 2006 311, 1566-1570. 30. Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. Direct tissue analysis using matrixassisted laser desorption/ionization mass sp ectrometry: practical aspects of sample preparation. J. Mass Spectrom. 2003, 38, 699-708. 31. Honig, R. E.; Woolston, J. R. Appl. Phys. Lett. 1963, 2, 138-139.

PAGE 156

138 32. Spengler, B. The basics of matrix-assiste d laser desorption/ionization time-of-flight mass spectrometry and post-source decay analysis, in Proteome Research: Mass Spectrometry ; Peter, J., Ed.; Springer: Berlin, 2001, pp 33-53. 33. Harrison, A. G.; Cotter, R. J. Methods of ionization, in Mass Spectrometry ; McCloskey, J. A., Ed.; Academic Pr ess: San Diego, 1990; Vol. 193, pp 3-37. 34. Vickerman, J. C.; Brown, A.; Reed, N. M. Secondary ion mass spectrometry: Principles and applications ; Oxford University Press: New York, 1990. 35. Winograd, N. Prospects for imaging TOF-SIMS: from fundamentals to biotechnology. Appl. Surface Sci. 2003, 203-204, 13-19. 36. Karas, M.; Hillenkamp, F. Laser desorption ionisation of proteins with molecular masses in excess of 10,000 daltons. Anal. Chem. 1988, 60, 2301-2303. 37. Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T.; Matsuo, T. Protein and polymer analyses up to m/z 100,000 by laser ionization time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom. 1988, 8, 151-153. 38. Karas, M.; Gluckmann, M.; Schafer, J. Ionization in matrix-assisted laser desorption/ionization: singly charged mol ecular ions are the lucky survivors. J. Mass Spectrom. 2000, 35, 1-12. 39. Beavis, R. C.; Chait, B. T. Matrix-a ssisted laser desorption/ionization mass spectrometry of proteins, in Methods in Enzymology 1996; Vol. 270. 40. Hillenkamp, F.; Karas, M., Eds. Mass Spectrometry of peptides and proteins by matrix-assisted ultraviolet laser desorption/ionization ; Academic Press, 1990. 41. Zenobi, R.; Knochenmuss, R. Ion fo rmation in MALDI mass spectrometry. Mass Spectrom. Rev. 1998, 17, 337-366. 42. Zhigilei, L. V.; Kodali, P. B. S.; Garrison, B. J. On the threshold behavior in laser ablation of organic solids. Chem. Phys. Lett. 1997 276, 269-273.

PAGE 157

139 43. Chevrier, M. R.; Cotter, R. J. A matrix-a ssisted laser desorption time-of-flight mass spectrometer based on a 600 ps, 1.2 mL nitrogen laser. Rapid Commun. Mass Spectrom. 1991, 5, 611-617. 44. Dreisewerd, K.; Schurenberg, M.; Karas, M. ; Hillenkamp, F. Influence of the laser intensity and spot size on the desorption of molecules and ions in matrix-assisted laser-desorption ionization w ith a uniform beam profile. Int. J. Mass Spectrom. Ion Processes 1995, 141, 127-148. 45. Westmacott, G.; Ens, W.; Hillenkamp, F. ; Dreisewerd, K.; Schurenberg, M. The influence of laser fluence on ion yield in matrix-assited laser desorption/ionization mass spectrometry. Int. J. Mass Spectrom. 2002, 221, 67-81. 46. Juhasz, P.; Vestal, M. L.; Martin, S. A. On the initial velocity of ions generated by matrix-assisted laser desorption ionizati on and its effect on the calibration of delayed extraction timeof-flight mass spectra. J. Amer. Soc. Mass Spectrom. 1997, 8, 209-217. 47. Gluckmann, M.; Karas, M. The initial i on velocity and its de pendence on matrix, analyte and preparation method in ultraviolet matrix-assisted laser desorption/ionization. J. Mass Spectrom. 1999, 34, 467-477. 48. Niu, S.; Zhang, W.; Chait, B. T. Direct comparison of infrared and ultraviolet matrix-assisted laser desorption/ioni zation mass spectrometry of proteins. J. Amer. Soc. Mass Spectrom. 1998, 9, 1-7. 49. Wang, B. H.; Dreisewerd, K.; Bahr, U.; Karas, M.; Hillenkamp, F. Gas-phase cationization and protonati on of neutrals generated by matrix-assisted laser desorption. J. Amer. Soc. Mass Spectrom. 1993, 4, 393-398. 50. Ehring, H.; Karas, M.; Hillenkamp, F. Role of photoionization and photochemistry in ionization processes of organic molecules and relevance for matrix-assisted laser desorption ionizat ion mass spectrometry. Org. Mass Spectrom. 1992, 27, 472-480. 51. Knochenmuss, R. A quantitative model of ultraviolet matrix-assisted laser desorption/ionization. J. Mass Spectrom. 2002, 37, 867-877.

PAGE 158

140 52. Puretzky, A. A.; Geohegan, D. B. Gas-phase diagnostics and LIF-imaging of 3hydroxypicolinic acid maldi-matrix plumes. Chem. Phys. Lett. 1998, 286, 425432. 53. Mowry, C. D.; Johnston, M. V. Internal energy of neutral molecules ejected by matrix-assisted laser-desorption. J. Phys. Chem. 1994, 98, 1904-1909. 54. Knochenmuss, R.; Zenobi, R. MALDI ionizatio n: The role of in-plume processes. Chem. Rev. 2003, 103, 441-452. 55. Sunner, J. Ionization in liquid sec ondary ion mass spectrometry (LSIMS). Org. Mass Spectrom. 1993, 28, 805-823. 56. Frankevich, V.; Knochenmuss, R.; Zenobi, R. The origin of electrons in MALDI and their use for sympathetic coo ling of negative ions in FTICR. Int. J. Mass Spectrom. 2002, 220, 11-19. 57. Frankevich, V. E.; Zhang, J.; Friess, S. D.; Dashtiev, M.; Zenobi, R. Role of electrons in laser desorption/ionization mass spectrometry. Anal. Chem. 2003, 75, 6063-6067. 58. Zhang, J.; Zenobi, R. Matrix-depen dent cationization in MALDI mass spectrometry. J. Mass Spectrom. 2004, 39, 808-816. 59. Barber, M.; Bordoli, R. S.; Sedgwick, R. D. ; Tyler, A. N. Fast atom bombardment of solids (FAB): A new ion source for mass spectrometry. J. Chem. Soc. Chem. Commun. 1981, 325-327. 60. Barber, M.; Bordoli, R. S.; Elliot, G. J. ; Sedgwick, R. D.; Tyler, A. N. FAB-MS. Anal. Chem. 1982, 54, 645A-657A. 61. Murphy, R. C. Mass Spectrometry of Lipids ; Plenum Press: New York, 1993. 62. Han, X.; Gross, R. W. Structural de termination of picomole amounts of phospholipids via electrospray ioni zation tandem mass spectrometry. J. Amer. Soc. Mass Spectrom. 1995, 6, 1201-1210.

PAGE 159

141 63. Han, X.; Gross, R. W. Electrospray ioni zation mass spectroscopic analysis of human erythrocyte plasma membrane phospholipids. Proc. Nat. Acad. Sci.USA 1994, 91, 10635-10639. 64. Hsu, F.-F.; Turk, J. Structural dete rmination of sphingomyelin by tandem mass spectrometry with electrospray ionization. J. Amer. Soc. Mass Spectrom. 2000, 11, 437-449. 65. Hsu, F.-F.; Turk, J. Characterization of phosphatidylethanolamine as a lithiated adduct by triple quadrupole tandem mass sp ectometry with electrospay ionization. J. Mass Spectrom. 2000, 35, 596-606. 66. Hsu, F.-F.; Turk, J. Studies on phosphatidylserine by tandem quadrupole and multiple stage quadrupole ion-trap mass spectrometry with electrospray ionization: structural characteriza tion and the fragmentation processes. J. Amer. Soc. Mass Spectrom. 2005, 16, 1510-1522. 67. Pulfer, M.; Murphy, R. C. Electrospra y mass spectrometry of phospholipids. Mass Spectrom. Rev. 2003, 22, 332-364. 68. Murphy, R. C.; Fiedler, J.; Hevko, J. Analysis of nonvolatile lipids by mass spectrometry. Chem. Rev. 2001, 101, 479-526. 69. Wenk, M. R. The emerging field of lipidomics. Nature Rev. Drug Discov. 2005, 4, 594-610. 70. Harvey, D. J. Matrix-assisted laser desorption/ionization mass spectrometry of phospholipids. J. Mass Spectrom. 1995, 30, 1333-1346. 71. Harvey, D. J. Matrix-assisted laser desorption/ionization mass spectrometry of sphingoand glycoshingo-lipids. J. Mass Spectrom. 1995, 30, 1311-1324. 72. Schiller, J.; Arnhold, J.; Benard, S.; Mu ller, M.; Reichl, S.; Arnold, K. Lipid analysis by matrix-assisted laser desorption/ionization mass spectrometry: A methodological approach. Anal. Biochem. 1999, 267, 46-56. 73. Estrada, R.; Yappert, M. C. Alternative approaches for the detection of various phospholipid classes by matrix -assisted laser desorption /ionization time-of-flight mass spectrometry. J. Mass Spectrom. 2004, 39, 412-422.

PAGE 160

142 74. Petkovic, M.; Schiller, J.; Muller, M.; Benard, S.; Reichl, S.; Arnold, K.; Arnhold, J. Detection of individual phospholipids in lipid mixtures by matrix-assisted laser desorption/ionization time-of-flight ma ss spectrometry: Phosphatidylcholine prevents the detection of further species. Anal. Biochem. 2001, 289, 202-216. 75. Woods, A. S.; Ugarov, M.; Egan, T.; Koome n, J.; Gillig, K. J.; Fuhrer, K.; Gonin, M.; Schultz, J. A. Lipid/peptide/nucle otide separation with MALDI-Ion mobilityTOF MS. Anal. Chem. 2004, 76, 2187-2195. 76. Hunnam, V.; Harvey, D. J. Ionization and fragmentation of neutral and acidic glycosphingolipids with a Q-TOF mass spectrometer fitted with a MALDI ion source. J. Amer. Soc. Mass Spectrom. 2001, 12, 1220-1225. 77. Al-Saad, K. A.; Zabrouskov, V.; Siems, W. F.; Knowles, N. R.; Hannan, R. M.; Hill, J., Herbert H. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry of lipids: ionization an d prompt fragmentation patterns. Rapid Commun. Mass Spectrom. 2003, 17, 87-96. 78. Stoeckli, M.; Chaurand, P.; Hallahn, D. E.; Caprioli, R. M. Imaging mass spectrometry: A new technology for the analysis of protein expression in mammalian tissues. Nature Med. 2001 7, 493-496. 79. Jurchen, J. C.; Rubakhin, S. S.; Sweedler, J. V. MALDI-MS imaging of features smaller than the size of the laser beam. J. Amer. Soc. Mass Spectrom. 2005, 16, 1654-1659. 80. Monroe, E. B.; Jurchen, J. C.; Lee, J.; R ubakhin, S. S.; Sweedler, J. V. Vitamin E imaging and localization in the neuronal membrane. J. Amer. Chem. Soc. 2005, 127, 12152-12153. 81. McDonnell, L. A.; Piersma, S. R.; Altelaar A. F. M.; Mize, T. H.; Luxembourg, S. L.; Verhaert, P.; van Minnen, J.; Heeren, R. M. A. Subcellular imaging mass spectrometry of brain tissue. J. Mass Spectrom. 2005, 40, 160-168. 82. Altelaar, A. F. M.; van Minnen, J.; Jimenez, C. R.; Heeren, R. M. A.; Piersma, S. R. Direct molecular imaging of Lymnaea stagnalis nervous tissue at subcellular spatial resolution by mass spectrometry. Anal. Chem. 2005, 77, 735-741.

PAGE 161

143 83. Sjovall, P.; Lausmaa, J.; Nygren, H.; Carlsson, L.; Malmberg, P. Imaging of membrane lipids in single cells by impr ing-imaging time-of-flight secondary ion mass spectrometry. Anal. Chem. 2003, 75, 3429-3434. 84. Chaurand, P.; Schwartz, S. A.; Caprioli, R. M. Profiling and imaging proteins in tissue sections by IMS. Anal. Chem. 2004, 76, 86A-93A. 85. Chaurand, P.; Caprioli, R. M. Direct profili ng and imaging of peptides and proteins from mammalian cells and tissu e sections by mass spectrometry. Electrophoresis 2002, 23, 3125-3135. 86. Chaurand, P.; Schwartz, S. A.; Caprioli, R. M. Assessing protein patterns in disease using imaging mass spectrometry. J. Proteome Res. 2004, 3, 245-252. 87. Laurent, C.; Levinson, D. F.; Schwartz, S. A.; Harrington, P. B.; Markey, S. P.; Caprioli, R. M.; Levitt, P. Direct profiling of the cerebellum by matrix-assisted laser desorption/ionizatio n time-of-flight mass spectrometry: A methodological study in postnatal and adult mouse. J. Neurosci. Res. 2005, 81, 613-621. 88. Sjovall, P.; Lausmaa, J.; Johansson, B. Ma ss spectrometric imaging of lipids in brain tissue. Anal. Chem. 2004, 76, 4271-4278. 89. Pacholski, M. L.; D.M.Cannon, J.; Ewing, A. G.; Winograd, N. Imaging of exposed headgroups and tailgroups of phospholipid membranes by mass spectrometry. J. Amer. Chem. Soc. 1999, 121, 4716-4717. 90. Jackson, S. N.; Wang, H.-Y. J.; Woods, A. S. Direct profiling of lipid distribution in brain tissue using MALDI-TOFMS. Anal. Chem. 2005, 77, 4523-4527. 91. Todd, P. J.; McMahon, J. M.; McCandlish, C. A. Secondary ion images of the developing rat brain. J. Amer. Soc. Mass Spectrom. 2004, 15, 1116-1122. 92. Keune, K.; Boon, J. J. Imaging secondary ion mass spectrometry of a paint cross section taken from an early netherland ish painting by Rogeir van der Weyden. Anal. Chem. 2004, 76, 1374-1385.

PAGE 162

144 93. van den Brink, O. F.; Boon, J. J.; O'Connor, P. B.; Duursma, M. C.; Heeren, R. M. A. Matrix-assisted laser desorpti on/ionization Fourie r transform mass spectrometric analysis of oxygenated tr iglycerides and phosphatidylcholines in egg tempera paint dosimeters used for environmental monitoring of museum display conditions. J. Mass Spectrom. 2001, 36, 479-492. 94. Wang, H.-Y. J.; Jackson, S. N.; McEuen, J.; Woods, A. Localization and analyses of small drug molecules in rat brain tissue sections. Anal. Chem. 2005, 77, 66826686. 95. Reyzer, M. L.; Hsieh, Y.; Ng, K.; Korfmacher, W. A.; Caprioli, R. M. Direct analysis of drug candidates in tissue by ma trix-assisted laser desorption/ionization mass spectrometry. J. Mass Spectrom. 2003, 38, 1081-1092. 96. O'Connor, P. B.; Mirgorodskaya, E.; Costello, C. E. High pressure matrix-assisted laser desorption/ionization Fourier transf orm mass spectrometry for minimization of ganglioside fragmentation. J. Amer. Soc. Mass Spectrom. 2002, 13, 402-407. 97. Baldwin, M. A.; Medzihradszky, K. F.; Lock, C. M.; Fisher, B.; Settineri, T. A.; Burlingame, A. L. Matrix-assisted la ser desorption/ionization coupled with quadrupole/orthogonal acceleration time-offlight mass spectrometry for protein discovery, identification, and structural analysis. Anal. Chem. 2001, 73, 17071720. 98. Loboda, A. V.; Krutchinsky, A. N.; Bromir ski, M.; Ens, W.; Standing, K. G. A tandem quadrupole/time-of-flight mass spectr ometer with a matrix-assisted laser desorption/ionization source: design and performance. Rapid Commun. Mass Spectrom. 2000, 14, 1047-1057. 99. O'Connor, P. B.; Costello, C. E. A high pressure matrix-assisted laser desorption/ionization Fourie r transform mass spectromet ry ion source for thermal stabilization of labile molecules. Rapid Commun. Mass Spectrom. 2001, 15, 18621868. 100. Doroshenko, V. M.; Laiko, V. V.; Taranenko, N. I.; Berkout, V. D.; Lee, H. S. Recent developments in atmospheri c pressure MALDI mass spectrometry. Int. J. Mass Spectrom. 2002, 221, 39-58. 101. Perchalski, R. J. Characteristics and app lication of a laser io nization/evaporation source for tandem mass spectrometry. Ph.D.Dissertation, University of Florida 1985.

PAGE 163

145 102. Hsu, F.-F.; Bohrer, A.; Turk, J. Formation of lithiated adducts of glycerophosphocholine lipids f acilitates their identification by electrospray ionization tandem mass spectrometry. J. Amer. Soc. Mass Spectrom. 1997, 9, 516526. 103. Ho, Y.-P.; Huang, P.-C. A novel structural analysis of glycerophosphocholines as TFA/K+ adducts by electrospray ionizat ion ion trap tandem mass spectrometry. Rapid Commun. Mass Spectrom. 2002, 16, 1582-1589. 104. Alberts, B.; Bray, D.; Lewis, J.; Raff, M.; Roberts, K.; Watson, J. D. Membrane structure, in Molecular Biology of the Cell 3rd ed.; Garland: New York, 1994, pp 477-506. 105. Guittard, J.; Hronowski, X. L.; Costello, C. E. Direct matrix-assisted laser desorption/ionization mass spectrometric analysis of glycoshingolipids on thin layer chromatographic plates and transfer membranes. Rapid Commun. Mass Spectrom. 1999, 13, 1838-1849. 106. Larsen, A.; Uran, S.; Jacobsen, P. B.; Skotland, T. Collision-induced dissociation of glycero phospholipids using electrospray ion-trap mass spectrometry. Rapid Commun. Mass Spectrom. 2001, 15, 2393-2398. 107. Ho, Y.-P.; Huang, P.-C.; Deng, K.-H. Me tal ion complexes in the structural analysis of phospholipids by electrospray ionization tandem mass spectrometry. Rapid Commun. Mass Spectrom. 2003, 17, 114-121. 108. McClellan, J. E.; Murphy, I., James P.; Mul holland, J. J.; Yost, R. A. Effects of fragile ions on mass resolution and on is olation for tandem mass spectrometry in the quadrupole ion trap mass spectrometer. Anal. Chem. 2002, 74, 402-412. 109. Gabelica, V.; De Pauw, E.; Karas, M. Infl uence of the capillary temperature and the source pressure on the internal energy distribution of electrosprayed ions. Int. J. Mass Spectrom. 2004, 231, 189-195. 110. Rockwood, A. L.; Busman, M.; Udseth, H. R.; Smith, R. D. Thermally induced dissociation of ions from el ectrospray mass spectrometry. Rapid Commun. Mass Spectrom. 1991, 5, 582-585.

PAGE 164

146 111. McClellan, J. E.; James P. Murphy, I.; Mulholland, J. J.; Yost, R. A. Effects of fragile ions on mass resolution and on is olation for tandem mass spectrometry in the quadrupole ion trap mass spectcrometer. Anal. Chem. 2002, 74, 402-412. 112. Chaurand, P.; Schwartz, S. A.; Billheimer, D. ; Xu, B. J.; Crecelius, A.; Caprioli, R. M. Integrating histology and imaging mass spectrometry. Anal. Chem. 2004, 76, 1145-1155. 113. Maddalo, G.; Petrucci, F.; Iezzi, M.; Pannell ini, T.; Del Boccio, P.; Ciavardelli, D.; Biroccio, A.; Forli, F.; Di Ilio, C.; Ballone E.; Urbani, A.; Federici, G. Analytical assessment of MALDI-TOF imaging mass spectrometry on this histological samples. An insight into proteome investigation. Clin. Chim. Acta 2005, 357 210218. 114. Jackson, S. N.; Wang, H.-Y. J.; Woods, A. S. In situ structural characterization of phosphatidylcholines in brai n tissue using MALDI-MS/MS. J. Amer. Soc. Mass Spectrom. 2005, 16, 2052-2056. 115. Touboul, D.; Kollmer, F.; Niehuis, E.; Laprevote, O.; Brunelle, A. Improvement of biological time-of-flight-secondary i on mass spectrometry imaging with a bismuth cluster ion source. J. Amer. Soc. Mass Spectrom. 2005 16, 1608-1618. 116. Winograd, N. The magic of cluster SIMS. Anal. Chem. 2005 77, 143A-149A. 117. Wang, H.-Y. J.; Jackson, S. N.; McEue n, J.; Woods, A. S. Localization and analysis of small drug molecule s in rat brain tissue sections. Anal. Chem. 2005, 77, 6682-6686. 118. Baldwin, M. A.; Medzihradszky, K. F.; Lock, C. M.; Fisher, B.; Settineri, T. A.; Burlingame, A. L. Matrix-assisted la ser desorption/ionization coupled with quadrupole/orthogonal acceleration time-offlight mass spectrometry for protein discovery, identification, and structural analysis. Anal. Chem. 2001, 73, 17071720. 119. Xiao, Y.; Huang, Y.; Chen, Z.-Y. Dist ribution, depletion and recovery of docosahexanoic acid are region-specific in rat brain. Brit. J. Nutr. 2005, 94, 544550. 120. Martinez, M.; Mougan, I. Fatty acid co mposition of human brain phospholipids during normal development. J. Neurochem. 1998, 71, 2528-2533.

PAGE 165

147 121. Agranoff, B. W.; Benjamins, J. A.; Hajra, A. A. Lipids, in Basic Neurochemistry. Molecular, Cellular and Medical Aspects, 6th ed.; Siegel, G. J., Agranoff, B. W., Fisher, S. K., Albers, R. W., Uhler, M. D., Eds.; Lippincott Raven: Philadelphia, 1998, pp 47-67. 122. Paxinos, G.; Watson, C. The rat brain in stereotaxic coordinates 4th ed.; Academic Press: San Diego, 1998. 123. Hensel, R. R.; King, R. C.; Owens, K. G. Electrospray sample preparation for improved quantitation in matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom. 1997, 11, 1785-1793. 124. Knochenmuss, R. A quantitative model of ultra-violet matrix-assisted laser desorption/ionization includi ng analyte ion generation. Anal. Chem. 2003 75, 2199-2207. 125. Knochenmuss, R.; Dubois, F.; Dale, M. J.; Zenobi, R. The matrix suppression effect and ionization mechanisms in ma trix-assisted laser desorption/ionization. Rapid Commun. Mass Spectrom. 1996, 10, 871-877. 126. McCombie, G.; Knochenmuss, R. Small-molecule MALDI using the matrix suppression effect to reduce or elimin ate matrix background interferences. Anal. Chem. 2004, 76, 4990-4997. 127. Hsieh, Y.; Casale, R.; Fukuda, E.; Chen, J. ; Knemeyer, I.; Wingate, J.; Morrison, R.; Korfmacher, W. Matrix-assisted laser desorption/ioni zation imaging mass spectrometry for direct measurement of clozapine in rat brain tissue. Rapid Commun. Mass Spectrom. 2006, 20, 965-972.

PAGE 166

BIOGRAPHICAL SKETCH Timothy James Garrett was born on Novemb er 8, 1976, as the second of twins and the fourth child to parents David and Sharon in Newark, OH. Growing up with very well educated parents induced an educationa l experience around ever y corner. Although Timothy did not enjoy chemistry in the be ginning, having a father with a Ph.D. in physical organic chemistry instilled the scientif ic thought at an early age. Eventually a love for chemistry grew from those seeds. Timothy was first introduced to mass spect rometry while studying at the University of Georgia and working in the lab of Professor I. Jonathan Amster. He gained valuable insight into the emerging field of proteomics while conducting research on the proteins expressed in bacteria samples and comple ted an honors thesis discussing this work entitled Improved methods for on-probe cl eanup of unpurified protein samples for MALDI-TOF mass spectrometry. During these studies he met his future wife, Jennifer, during an interview for a volunteer or ganization, alternat ive spring break. Timothy graduated from the University of Georgia in May of the year 1999 Summa cum Laude with highest honors and honors in chemistry. After 2 years working in the corporate envi ronment, Timothy decided to return to school and work towards a graduate degree in analytical chemistry at the University of Florida under the direction of Richard A. Yost. He married Jennifer on July 7, 2001, and returned to school in August of that year. 148


Permanent Link: http://ufdc.ufl.edu/UFE0013807/00001

Material Information

Title: Imaging Small Molecules in Tissue by Matrix-Assisted Laser Desorption/Ionization Tandem Mass Spectrometry
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013807:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013807/00001

Material Information

Title: Imaging Small Molecules in Tissue by Matrix-Assisted Laser Desorption/Ionization Tandem Mass Spectrometry
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013807:00001


This item has the following downloads:


Full Text












IMAGING SMALL MOLECULES IN TISSUE BY MATRIX-ASSISTED LASER
DESORPTION/IONIZATION TANDEM MASS SPECTROMETRY















By

TIMOTHY JAMES GARRETT


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


2006

































This document is dedicated to my wife.















ACKNOWLEDGMENTS

Obtaining a graduate degree is not an individual effort, although it shows the

efforts of the individual. I thank my parents for providing me with the opportunities to

discover the direction my life should take and allowing me to learn from my mistakes. I

thank my father for teaching me the subtleties of scientific research and guiding me down

this path.

I have to thank my advisor, Rick, for teaching me about research and providing an

environment in which thoughts and ideas are allowed to guide the graduate experience.

The guidance he has provided in not only science, but also in life, has enabled me to be a

valuable member of society. I also have to thank the members of the Yost group for

thoughtful discussions of science, research, and life.

I thank the city of Gainesville for growing on me and letting me enjoy this

community.

I must also thank God for providing me with guidance and friendships that have

given me insight into my own life and my research.

I must thank Mari, Viatcheslav, Huy, and George at Thermo for allowing me to

visit and conduct research at their facility. Working through the difficult aspects of

collaborating across country and dealing with the frustrations of instrument development,

their dedication to our efforts has made this research possible. I must also thank Thermo

Electron corporation for their hospitality during each of my research visits.









Finally, I have to thank my wife for helping me through the difficult times of

research and giving me the strength to continue in this venture of our lives. Her

guidance, love, and support have made me who I am today.
















TABLE OF CONTENTS



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

LIST OF TABLES .............. ................. ........... ............... ............ vii

LIST OF FIGURES ..................................................... .......... ................ viii

ABSTRACT ........................................................... xvii

CHAPTER

1 IN T R O D U C T IO N ............................................................................. .....................

Quadrupole Ion Trap Mass Spectrometry ............... .............................................2
From 3-D to 2-D Ion Traps ........................ .. ... ........ ...................2
3D and 2D Quadrupole Ion Trap (QIT) Theory and Operation ........................6
2D/Linear Ion Trap Description ....... .. ......... ...............11
Tissue Imaging by Mass Spectrometry .............................................. ...............15
B a ck g ro u n d .....................................................................................1 5
F undam entals .......................... ....................................................................... 17
Matrix-Assisted Laser Desorption/Ionization........................................ 24
B background ................................................................... ............ 24
Theory of MALDI .................. ........................ .. ........ ................. 26
Phospholipids and Sphingolipids...................................................... 34
The Structure of Phospholipids ........................................ ....................... 34
Analysis of Lipids by Mass Spectrometry....................................................40
O overview of dissertation ............................................................................. ... ........5 1

2 ANALYSIS OF INTACT TISSUE BY INTERMEDIATE-PRES SURE MALDI
ON A LINEAR ION TRAP MASS SPECTROMETER ......................................52

In tro d u ctio n ............. .. ........... .......................................................................... 5 2
Experimental ................................ ............................... 54
R results and D discussion ........................................ ................... .. ...... 55
C o n c lu sio n s ........................................................................................................... 6 1

3 CHARACTERIZATION OF PROTONATED PHOSPHOLIPIDS AS FRAGILE
IONS IN ION TRAP TANDEM MASS SPECTROMETRY ...................................63









In tro d u ctio n .......................................................................................6 3
E x p e rim e n ta l ....................................................................................................6 6
R results and D discussion ........................ ................ ................... .. ...... 70
Fragility in the trap ......................... ........... ........ ...............71
F ragility in ion tran sport......................................................................... ... ... 78
Tube lens offset ....................................... ................. 82
Initial Studies of Ion Fragility in the 2D/Linear Ion Trap ................................. 87
IP -M A L D I ..................................................................................................... 8 7
M ass analysis fragility ........ ... ........ ................. ............... .............. 88
C o n c lu sio n s ........................................................................................................... 9 2

4 IMAGING PHOSPHOLIPIDS IN BRAIN TISSUE BY INTERMEDIATE-
PRESSURE MALDI ON A LINEAR ION TRAP MASS SPECTROMETER .........95

Intro du action .................................................................................................... 9 5
Experimental .................. ...... .........................97
Acquisition of position-specific mass spectra ............................... .. .............101
R results and D discussion .............. ..................................................... .............. 101
Identifying different features of the rat brain. ................................................. 119
Quantitation ................................................................123
C o n c lu sio n s ......................................................................................................... 12 5

5 CONCLUSION AND FUTURE WORK ................... ....... ..... ............... 129

L IST O F R E F E R E N C E S ......... .................................... .............................................. 135

BIOGRAPH ICAL SKETCH .............. ......................... ................... ............... 148















LIST OF TABLES


Table page

1-1. A list of the common fatty acids found in nature. ....................................................38

1-2. The common reactions that occur in chemical ionization when methane, CH4, is
used as the reagent gas. ............................... .... .......... ................ ............. 44

2-1. Ions detected from spinal cord in Figure 2A that were chosen for MSn analysis.
Only the MS2 major fragment ion m/z is shown in the table, but MS3 was also
perform ed. ...........................................................................60

3-1. The %CID needed to cause a reduction in the absolute parent ion signal by 50%
is shown for each ion studied in the QIT. These results show that the protonated
species fragments more readily than the sodiated species, further proof that the
protonated ion is fragile ............................................................... .......... .... 77

3-2. A summary of the results from each experiment performed in the studies of ion
fragility in ion transport. From these results, it was determined that all sodiated
species were less susceptible to fragmentation that can occur before mass
analy sis. .............................................................................87

4-1. A comparison of literature data from a lipid extract of a rat brain to MS data
collected directly from a tissue section ...................................... ............... 124















LIST OF FIGURES


Figure page

1-1. A schematic representation of the 3-dimensional (3D) quadrupole ion trap (QIT).
The 3D QIT consists of 2 end-cap electrodes and a ring electrode. The inside
dimensions of the 3D QIT are defined by ro, the radius of the ring electrode, and
zo, the distance from the center of the QIT to the apex of the end-cap. For a
symmetrical QIT, ro is '/2 times zo. Ions are injected and ejected through holes
drilled into each end-cap. ........................ ................ .............. ................ .4

1-2. Diagram of a 2-dimensional (2D) ion trap. This 2D ion trap design incorporates
three different sections, front and back sections of equal length, and a center
section where the ions are trapped. ........................................ ........................ 7

1-3. A small portion of the Mathieu stability diagram. Solutions to Equations 1-2 and
1-3 give coordinates in az and qz space that can be mapped onto the above
diagram. If the coordinates for an ion fall into the region shown above, then the
ion will have a stable trajectory inside the 3D or 2D trap and will be successfully
trapped. In normal operation, the DC applied is zero and thus the az is also zero
since it is related to the DC voltage applied, thus the most important number is
the value for qz. The value of qz=0.908 is the edge of the stability diagram. An
ion with a qz greater than this value is not trapped........................ .. ............... 9

1-4. A Schematic representation of the 2D ion trap. Ions enter from the left side of
the figure. The 2D trap is separated into 3 sections, two shorter sections of 12
mm in length, and a longer, center, section 37 mm long. Separate DC voltages
are applied to all three sections in order to trap ions. Ions are ejected through
slits in the x-rod pairs of the center section ........... ............................................ 13

1-5. Two views of the 2D ion trap looking down the z-axis, along the flight path of
the ions. A shows the radial quadrupolar trapping field. The phases ofRF are
applied to all the electrodes of the ion trap to create the quadrupolar field. B
shows the radial dipolar excitation arrangement. An AC voltage is applied only
to the x-rod pairs. The AC voltage is used for isolation and collision-induced
dissociation experiments as well as in the ejection of ions from the trap. Ions are
ejected out of slits cut into the x-rod pairs. ........... ............................................ 14

1-6. The process of creating an image by mass spectrometry. The two most important
stages in tissue analysis by mass spectrometry are 1) cutting the tissue to an
appropriate thickness with a cryotome, such as 10 |tm for brain tissue, and 2)









applying the MALDI matrix in a manner that reduces the possibility for analyte
migration airbrushingg is shown above). When cutting the tissue, it is important
to ensure that no OCT compound is on the tissue sections as it can interfere with
mass analysis. The mass spectrometric stage involves rastering the sample plate
at a predefined step size with respect to the laser. Each mass spectrum is then
position specific so all that is required for generating an image is extracting the
m /z value of interest. ............. ................................................ .. .... ......... ... 19

1-7. The types of sampling used in imaging MS. Sampling at the laser resolution is
the most widely used data collection scheme because it provides the highest
resolution. Under-sampling will take less time for acquisition, but is a poor
choice when trying to identify changes over a very specific boundary area.
Over-sampling is typically not used because overlapping data points may cause
confusion in the generated images. ............................................... ............... 20

1-8. Pipetting the matrix onto a tissue surface causes analyte migration as shown. A
shows the tissue on the MALDI sample plate, B shows the pipetting of the
matrix onto the tissue, and C shows the how pipetting the matrix causes analyte
migration in the tissue. Imaging MS cannot be performed when this type of
m atrix application is em played. ........................................ ......................... 22

1-9. The deposition of matrix as a fine mist, by either airbrushing or electrospraying.
A shows the tissue on the MALDI sample plate before matrix application and B
shows the application of matrix using a device the produced very small droplets
such as an airbrush or an electrospray needle with the proper voltage. C shows
the final tissue section with matrix applied. The process permits the
incorporation of the matrix material into the tissue medium without causing
analytes to m igrate across the tissue. ............................................ ............... 23

1-10. Two diagrams of the MALDI process. Diagram A shows how the matrix
material dilutes the analyte molecules to reduce the amount of intermolecular
interactions, while the desorption of neat analytes especially biomolecules will
induce fragmentation. The matrix is present at a much higher concentration than
the analyte to effectively act as an intermediary by absorbing the laser energy
and transferring it to the analyte without causing fragmentation. In Diagram B,
the plume generated after a laser pulse contains ions, both positive and negative,
related to the analyte and matrix compound as well as many neutral species.........28

1-11. Common organic acids used as matrices in MALDI and their abbreviations. No
single matrix is compatible with all analytes, and thus a wide variety of matrices
have been employed. ............................................... ...............29

1-12. The basic structure of glycerophospholipids (GPLs). All GPLs consist of a
polar head group and two non-polar tails, fatty acids. The polar head group is
the primary means of separating GPLs into different classes. The names of the
four major classes are shown next to the structure of the head group......................35









1-13. The basic structure of sphingolipids. This lipid class is derived from
sphingosine or another similar base. The structure of sphingosine is shown at
the bottom. They are similar to GPLs in that they have a non-polar tail, usually
consisting of one fatty acid, and a polar head group. The polar head group is
also used to differentiate the different classes of sphingolipid. The head groups
for the major sphingolipids are shown above ........... .... ............... ................... 37

1-14. A typical fatty acid, oleic acid, is shown in two possible configurations. For all
double bonds occurring in a fatty acid chain, the double bond is in the cis
configuration rather than the trans configuration. The structures shown are of
the basic shorthand form ............................................... .............................. 39

1-15. The structure of PC (16:0, 18:1) in a normal structure form is shown. This is
the most abundant glycerophospholipid present in mammalian cellular
m em branes. .......................................... ............................ 41

1-16. Above is the structure of SPM (18:0). This is the most abundant sphingolipid in
m am m alian nerve tissue. ..... ........................... .........................................42

1-17. A diagram of the fast atom bombardment, FAB, process. In FAB, the analyte,
A, is dissolved in a matrix, G, typically glycerol, and an ion or atom beam is
directed to the sample surface at an angle. Secondary ions of glycerol and
analyte are ejected from the surface upon bombardment by the ion beam. These
secondary ions generated are from the matrix and the analyte. This technique
limits fragmentation of the analyte, generating intact molecular ions ...................46

1-18. A diagram of the electrospray process. In electrospray ionization (ESI), a liquid
solution is pumped through a small capillary needle to which a high voltage is
applied. As the solution emerges from the needle, a Taylor cone develops and
then small droplets begin to evaporate. These droplets then evaporate further,
producing charged species in the gas phase. ..................................... ...............49

2-1. A schematic diagram of the Finnigan LTQ linear ion trap instrument with
vMALDI source modification. The vMALDI source replaces the standard API
source, and includes the sample plate, RF-only quadrupole qOO with auxiliary
rods, and skimmer. The RF-only quadrupole qO, octopole, and 3-section linear
ion trap are standard LTQ components. The open arrows show the three stages
of differential pumping; the pressure in the vMALDI region is typically 0.17
Torr .......... ..... .... ............ .................. 56

2-2. On the left is an optical image of the spinal cord coated with DHB; the locations
where the laser was fired to produce spectra A and B are indicated with circles.
The spinal cord is outlined in black. Spectrum A shows the ions detected from a
location away from the tissue, but in an area where the DHB matrix was still
present. Most of these ions correspond to DHB clusters such as m/z 409.91,
551.73, and 727.36, as indicated by the dagger symbol. These ions appear at a
higher mass than expected because the laser power was increased due to the









thickness of the tissue sample and thus when the laser fired at a position off the
tissue, space-charging was evident. No phospholipids were identified in that
location. Spectrum B shows the ion signal from the tissue surface, indicating
the presence of phospholipid ions. The ion at m/z 760.82 was determined by
tandem MS to be the [M+H] ion of phosphatidylcholine 34:1 (PC 34:1). The
other starred ions from m/z 700-900 were determined to be either PC or
sphingom yelin, SPM .............................. ..... ... .... ..... ............ 58

2-3. Spectrum A is an enlargement of the phospholipid region of the spinal cord tissue
section mass spectrum from figure 2B. Spectrum B shows MS2 of m/z 760.82.
Only one major fragment ion is produced, m/z 184.00, which corresponds to the
phosphocholine head group of phosphatidylcholine. The neutral losses of 18
(water) and 154 could correspond to a DHB cluster ion at the same m/z as the
phospholipid ion. The inset displays the basic structure of a
phosphatidylcholine, showing the fragmentation pathway that produces m/z 184..59

2-4. Spectrum A shows MS2 of m/z 782.82 from figure 2A, while spectrum B
displays MS3 (782.82-723.36-...). The neutral loss of 59 after MS2
corresponds to loss of trimethlyamine in the structure for PC (inset of spectrum
B). Two major fragment ions from MS3 arise from the neutral loss of 124
(ethylphosphate) with the sodium retained on the glycerol backbone, or the
neutral loss of 146, when the head group retains the sodium. This fragmentation
pattern is associated with cationized PC and SPM. Fragment ions relating to
specific fatty acid tails R1 and R2 were not observed .............................................62

3-1. Chemical structures of the phospholipids used in the studying ion fragility and
source fragmentation. The protonated version of the molecule is shown. SPM
stands for sphingomyelin, PC stands for phosphatidylcholine, and PE stands for
phosphatidylethanolamine. All three molecules are very similar in structure.
SPM and PC share the same head group, but differ in the fatty acid tails, while
PC and PE share the same fatty acid tails, but have slightly different head
group s. ............................................................................... 68

3-2. Spectrum A, above, is a zoom scan of protonated sphingomyelin (SPM) 16:0.
The peak width at 10% peak height (PW10%) was determined to be 0.44. The
peak width for the sodiated counterpart, measured from the zoom scan in
spectrum B, was 0.37. A narrower isolation width indicates that the sodiated
species is less fragile than the protonated species in ion trap mass analysis............72

3-3. The chart above shows the PWio% for the two ions (protonated and sodiated) of
each phospholipid studied. The peak widths were measured from zoom scan
data. The zoom scan data was collected for 2 minutes and the peak width was
measured from the average spectrum of each ion. For all three phospholipid
classes studied, the protonated species was determined to be more fragile
because the peak width was wider. ............................................... ............... 73









3-4. A chart showing the isolation width needed to efficiently isolate the parent ion of
each ion studied (protonated and sodiated) for each phospholipid. A wider
isolation width is typically needed for a more fragile ion. The results from this
experiment also indicate that the protonated species is more fragile than the
sodiated ion ....................................................... ................. 74

3-5. Spectrum A is MS2 of the [M+H] for SPM (16:0) and spectrum B is MS2 of the
[M+H] for PC (16:0, 16:0). Both of these spectra show the same fragmentation
pathway producing a predominant ion at m/z 184, corresponding to the polar
head group. Spectrum C is MS2 of PE (16:0, 16:0) and shows a different
fragmentation pathway. The major ion produced results from the neutral loss of
the polar head group ................................................ ... .... .... .... ........... 76

3-6. MS2 spectra of the [M+Na] ions for each phospholipid studied. A is from SPM
(16:0), B is from PC (16:0, 16:0), and C is from PE (16:0, 16:0). The
fragmentation pathways are very similar for all these ions. They all result from
neutral loss of the polar head group. A neutral loss of 59 for PC and SPM
corresponds to the loss of choline (-N(CH3)3) and a neutral loss of 43
corresponds to the loss of ethanimine (-C2H N). ............. ....................... ......... 77

3-7. The effect of the heated capillary temperature on TID for the different ions of
SPM (16:0) is shown in the three spectra above. Spectrum A was acquired at
2500C, a normal capillary temperature for most analyses, and shows nearly
100% fragmentation for the [M+H]+ ion of SPM (16:0). In contrast, there is less
than 20% fragmentation occurring for the [M+Na] ion. Spectrum B shows the
spectrum when the temperature is lowered to 1900C and spectrum C shows the
spectrum at 1300C. It is clear that lowering the temperature of the heated
capillary reduces fragmentation, TID, of both ions. At 1300C, extra peaks from
m/z 300-550 are present. These peaks are most likely clusters of solvent ions,
because desolvation is not as effective at such a low temperature...........................79

3-8. A graph showing the intensity of the parent and daughter ions of the [M+H]+ and
the [M+Na] ions versus changing the heated capillary temperature for SPM
(16:0). These data were collected automatically using Xcalibur software
control. The temperature was adjusted at 200 intervals from 500C to 2500C.
From the graph, it is evident that the protonated species is more susceptible to
TID than the sodiated species........................................................ ............... 81

3-9. A graph showing the effect of changing the capillary temperature for the [M+H]
and [M+Na]+ ions of PC (16:0, 16:0). The same susceptibility of the protonated
species to TID is evident, but the temperature at which the parent ion signal is
lower than the daughter ion signal occurs at a higher temperature than for the
[M + H ]+ ion of SP M (16 :0) ............................................................................ .... 83

3-10. A graph showing the effect of changing the capillary temperature of the
[M+H]+ and [M+Na]+ ions for PE (16:0, 16:0). The protonated species showed
a greater tendency to fragment upon increasing the capillary temperature, but the









parent ion signal was still more intense than the fragment ion even at the highest
temperature. This is in contrast to the other two phospholipids studied.................84

3-11. A graph showing the effect of changing the tube lens offset from 0V to 30V on
the [M+H] and [M+Na] ions for SPM (16:0). The capillary temperature was
held at 2500C for each voltage level. At the lowest voltage, fragmentation of the
parent ion is still around 50%. Adjusting only the tube lens offset is not enough
to limit the fragmentation. The sodiated species appears to be unaffected by a
change in the tube lens offset. ............................................................................85

3-12. Mass spectrum of SPM (from chicken egg yolk) acquired by MALDI at 10-6
Torr on a 3D-quadrupole ion trap. Fragmentation of the [M+H]+ ion is nearly
90%, while fragmentation of the [M+Na] ion is around 50%. The matrix was
6-aza-2-thiothymine. The asterisks indicate matrix ions. .................................... 89

3-13. Mass spectrum of SPM (chicken egg yolk) acquired by MALDI at intermediate
pressure (10-1 Torr) on a 2-D quadrupole ion trap. Fragmentation of the [M+H]
ion is reduced, but not removed. Fragmentation of the [M+Na] ion is also
reduced, but still present. The matrix was 6-aza-2-thiothymine. The asterisks
indicate m atrix ions. ........................ .... .................. .. ........ .. ............... 90

3-14. Zoom scans of the two ions for PC (16:0, 16:0): A is the [M+H] ion and B is
the [M+Na] ion. The spectra were acquired by intermediate-pressure MALDI
on a linear ion trap, averaging 75 spectra. As was seen in the ESI studies on a
2D ion trap, the peak width at 10% peak height for the [M+H]+ ion is wider than
the [M+Na] ion. This wider peak width indicates that the [M+H]+ ion is fragile
in m ass analy sis. ................................................................... ... 9 1

4-1. Picture of the airbrush setup used to coat brain tissue samples on glass
microscope slides (conductive and non-conductive). The trigger was held
constant by the pressure from a ring stand clamp. For typical coating, the
sample was about 6 in. from the airbrush nozzle. The nozzle can be easily
changed, but all experiments were conducted with a nozzle of 0.40 mm ...............99

4-2. Schematic diagram of the Finnigan LTQ linear ion trap instrument with vMALDI
source modification. The vMALDI source replaces the standard API source,
and includes the sample plate, RF-only quadrupole qOO with auxiliary rods, and
skimmer. The RF-only quadrupole qO, octopole, and 3-section linear ion trap
are standard LTQ components. The open arrows show the three stages of
differential pumping; the pressure in the vMALDI region is typically 0.17 Torr..100

4-3. The picture at the top of the figure, A, is an optical image generated from inside
the mass spectrometer of a rat brain section coated with DHB matrix. It was
acquired with 1 mm x 1 mm square pictures that are stitched together. This
creates the lines in the picture. The mass spectrum, B, is the signal from the
area on the tissue indicated by the arrow. The spectrum was acquired with 10









laser shots. A total of 11,156 spectra were collected across the tissue section.
The open circle is the laser spot size (100 [tm). ................................................. 102

4-4. A is the digital image of the tissue section and B is the mass spectrometric image
for the ion at m/z 756. Extracting the intensity from each spectrum in the data
file generated this image. This is a raw image, with no further processing done
to the data. The spectrum in C is an example of one of the 11,156 spectra that
w ere c o lle cte d ............................... ................................... .............. 10 5

4-5. MS image A, top, shows the raw image generated by extracting the intensity of
m/z 756 with respect to position. MS image B, bottom, is the normalized MS
image for m/z 756. Normalization involved dividing the intensity of m/z 756 at
each pixel by the total ion intensity at that pixel and multiplying by 100000. All
MS images further generated used this normalization procedure ........................106

4-6. The averaged MS2 spectrum of m/z 756.6 (516 spectra were averaged). It is the
average of 516 MS spectra collected from various parts of the tissue section.
Not that the intensities of all the ions below m/z 520 have been expanded 50
times. Fragment ions representing the phosphocholine head group are the most
abundant ions, but the less abundant fragment ions are used to determine the
fatty acid composition of this phospholipid, identifying the ion as the sodiated
adduct of PC (16:0,16:0). The less abundant ions used for identification were
m/z 500.3 and m/z 478.4 because they represent the loss of the fatty acid chain
with N a or without, respectively. ........................................ ....................... 108

4-7. The averaged MS2 spectrum of m/z 782.6 (517 spectra were averaged). As in
Figure 4-6, the most abundant fragment ions correspond to losses of the
phosphocholine head group, while the less abundant fragment ions allow for the
correct identification and location of the fatty acid tails. These ions are m/z
526.3 and m/z 500.3 and correspond to the loss of the sn-i (Cl) and sn-2 (C2)
fatty acid chains, respectively. The ratio of these ions allows for proper
assignment to the glycerol backbone. This ion was correct identified as the
sodiated adduct of PC (16:0,18:1). The structure of this compound is shown in
the inset with the assignments for the two fragment ions identifying the fatty
acid chains. .......................................... ........................... 109

4-8. MS images are shown of PC ions with 16:0, palmitic acid, at the sn-i position on
the glycerol backbone. A is the MS image of PC (16:0, 16:0), B is the MS
image of PC (16:0, 18:1), C is the MS image of PC (16:0, 20:4), and D is the
MS image of PC (16:0, 22:6). The ions were correctly identified from MS2 data;
classification is shown above each image. All the ions were the sodiated
species, [M+Na] The intensity levels were adjusted in images C and D for
comparison purposes. As is evident from their comparison, the localization of
these four PC ions is very different. The distribution of m/z 756, PC (16:0,
16:0) is prominently in the gray matter of the brain. ...... .. .... ......... ...........111









4-9. MS images showing the distribution of palmitic acid, 16:0, in rat brain tissue. A
is an image from MALDI imaging and is the summed image of four intact ions
(shown in Figure 4-8) containing palmitic acid at the sn-1 position of the
glycerol backbone. B is an image from negative ion SIMS and shows the
distribution of only palmitic acid, or an ion that has the m/z of palmitic acid.
MALDI enables the analysis of primarily intact ions, while SIMS provides
better spatial resolution, but with a high degree of source fragmentation. B is
adapted from reference 88 ......... ................................................. ............... 112

4-10. MS images of PC ions with 18:0 or 18:1 at the sn-i position of the glycerol
backbone. A is the MS image of PC (18:0, 18:1), B is the MS image of PC
(18:1, 18:1), C is the MS image of PC (18:0, 20:4), and D is the MS image of
PC (18:0, 22:6). MS2 data allowed for the correct identification of each ion.......114

4-11. Three MS images of the SPM ions detected from the rat brain tissue section (A
is SPM (18:0), B is SPM (24:0), and C is SPM (24:1)). The structure at the
bottom right is SPM where R1 is the variable fatty acid chain. Only m/z 753
was subjected to MS2; the other two ions (835 and 837) were not chosen for
MS2 analysis. The images for these two ions were generated because further
research showed that they are primarily present in the white matter of the brain,
whereas SPM (18:0) is present in the gray matter. As seen in the images, these
three ions show a different distribution indicative of prior research.121.................. 115

4-12. MS images of PC (16:0, 16:0), m/z 756.8 (A and C) and PC (16:0, 20:4), m/z
804.7 (B and D). A and B were generated from the first mass analysis of the rat
brain tissue section, whereas C and D were generated from mass analysis that
took place after 31 separate experiments across the tissue were performed. It is
evident from the pictures that C and D are much more spatially resolved. ...........118

4-13. MS image A was generated from a non-conductive glass microscope slide and
shows the distribution of m/z 810.7, PC (18:0, 18:1). MS image B was
generated for the same ion from a conductive glass microscope slide. The brain
tissues were 10 |tm thick and are serial sections from the same rat brain. The
images are very similar, but the image from the non-conductive microscope
slide (A) appears to be more spatially resolved. ............. ................................... 120

4-14. Spectra from similar spots of the brain tissue. A was acquired from a non-
conductive microscope slide and B was acquired from a conductive microscope
slide. The TIC for the non-conductive slide is 2x higher than that from the
co n d u ctiv e slid e .................................................................................. 12 1

4-15. A is the MS image of m/z 804.7, PC (16:0, 20:4) and B is a stereotaxic view of
a rat brain section (Bregma -6.04 mm) showing the different functional regions
of the brain. The MS image shows good correlation to the atlas even with a
spatial resolution of only 100 |tm ...................................................... ..... ..... 122









4-16. An average spectrum of 129 different position-specific mass spectra acquired
from the rat brain tissue section. The complexity of this region of the mass
spectrum can be unraveled using MS2 data of each ion. It can therefore be
determined that a single compound such as PC (16:0, 18:1) can be represented
by four different ions in the mass spectrum, m/z 723.8 ([M+Na-N(CH3)3]+),
782.7 ([M+Na] 798.6 ([M+K] ), and 958.4 ([M+Na+(DHB+Na-H)]+)..............126















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

IMAGING SMALL MOLECULES IN TISSUE BY MATRIX-ASSISTED LASER
DESORPTION/IONIZATION TANDEM MASS SPECTROMETRY


By

Timothy James Garrett

May 2006

Chair: Richard A. Yost
Major Department: Chemistry

The use of an intermediate-pressure matrix-assisted laser desorption/ionization (IP-

MALDI) source working at 0.17 Torr on a linear ion trap (LIT) was investigated for the

analysis of tissue specimens. IP-MALDI, with 2,5-dihydroxybenzoic acid (DHB) as the

matrix, was employed for the detection of phospholipids. The results indicate that

analyzing tissue specimens at non-traditional MALDI vacuum pressures is possible.

Coupling MALDI to an LIT permits the use of multiple stages of mass spectrometry,

MSn, which is critical for the ability to identify compounds desorbed directly from tissue

specimens. Using MSn, ions detected from m/z 600-1000 were characterized as

phosphatidlycholines and sphingomyelins. Specifically using tandem MS, PC ions could

be classified as either [M+H] or [M+Na] because the fragmentation patterns of

protonated and sodiated phosphatidlycholines follow different pathways.

Understanding ionization characteristics of the ions desorbed from tissue is

important in ensuring the stability for ion transport and mass analysis. The analysis of









the fragility of three phospholipids by electrospray ionization ion trap mass spectrometry

as protonated and sodiated species is discussed. The conditions of ion transport and ion

trapping that cause head group fragmentation of ions formed by electrospray ionization

were evaluated. Ion fragility in ion trapping was evaluated using a slow scan speed, the

percent collision-induced dissociation, and the isolation width needed for effective

isolation. Ion fragility in transport was evaluated by adjusting the capillary temperature

and the tube lens voltage. Results indicated that the sodiated species was the more stable

form of each ion.

Finally, mass spectrometric images of phospholipids in brain tissue were generated

by IP-MALDI on a LIT mass spectrometer. Eleven individual phosphatidylcholines and

sphingomyelins were identified by MS2 with fragment ions that allow for the

identification of the fatty acyl chains and their position on the glycerol backbone. The

use of non-conductive and conductive glass microscope slides in connection to ion trap

mass analysis was evaluated, and results showed that equivalent data could be obtained

on either surface. An artistic airbrush was employed to effectively coat the matrix

compound onto tissue sections without the deleterious effects of analyte migration.


xviii














CHAPTER 1
INTRODUCTION

Mass spectrometric microprobes with matrix-assisted laser desorption/ionization

(MALDI) offer a unique opportunity to directly probe thin tissue sections for changes in

the distribution of a molecular species of interest. Initial designs offered the opportunity

to identify changes in the concentration of a desired compound in different regions of a

tissue of interest. Investigating the distribution of a compound within tissue sections can

help understand whether an exogenous compound administered orally, such as a small

drug molecule, can be found within the body, where that compound tends to localize, and

possibly how endogenous compounds are affected by the drug. The latter aspect is a new

concept to the field of imaging because using a mass spectrometer begets the opportunity

to gather more complete information about a tissue section, including unknown analysis.

The mapping of endogenous compounds in tissue sections offers an incredibly specific

method to identify changes in the profile of the compound in diseased versus normal

states or just simply offers a new opportunity to re-map entire organs by a more

chemically specific process. This area could offer supplementary information that may

be useful in fully characterizing certain organs, such as the brain.

A critical aspect of analyzing compounds from tissue is the ability to identify the

compound with more certainty than just the expected mass-to-charge (m/z) value. The

m/z is defined as the monoisotopic molecular weight of a molecule plus the weight of one

or more protons (or cations, depending on the type of ionization) divided by the charge.

So if the molecular weight of a compound is 400 g/mol and this compound is protonated,









then the m/z would be 401. However, if the molecule was able to obtain another proton

to have two charges, the ion's mass would be 402 g/mol, but the m/z value is divided by

2 and thus would equal 201. The use of tandem mass spectrometry is one way to

positively identify ions of the detected m/z value as from the compound of interest.

Tandem MS involves isolating the specific m/z value (called the parent ion) in the mass

analyzer and then subjecting that ion to collision-induced dissociation (CID), thus

causing the isolated ion to fragment into smaller ions, giving a fragmentation spectrum

(called a daughter spectrum).

Several instruments are capable of performing tandem MS, including a triple

quadrupole, time-of-flight/time-of-flight (TOF/TOF), quadrupole/time-of-flight (Q-TOF),

and a quadrupole ion trap (QIT) mass spectrometer. Each instrument has certain

advantages in the analysis of tissue sections and all can perform one stage of tandem MS

(MS2), but the QIT is unique in its capability to provide for multiple stages of tandem MS

(MSn) as well as the ease of coupling MALDI to this mass analyzer. For these reasons,

all the research conducted was performed on a QIT-MS and thus the background is a

focus on this instrumentation.

Quadrupole Ion Trap Mass Spectrometry

From 3-D to 2-D Ion Traps

Today, the quadrupole ion trap (QIT) is a widely used mass analyzer because of the

ability to fully characterize a compounds structure with MSn, which allows for the

unambiguous identification of unknown and known compounds. The ion trap has a long

history before becoming a routine tool for chemists and biologists. The 3-dimensional

(3D) ion trap was first introduced and described as a mass storage device in 1953 by

Wolfgang Paul and Helmut Steinwedel.1 It consisted of two hyperbolic electrodes called









end-caps and a ring electrode centered between them. Figure 1-1 shows a schematic of a

3D ion trap.

A few years later, mass-selective detection was first employed for the detection of

krypton gas.2 For those studies, the motion of an ion within the ion trap was used to

obtain the mass-to-charge (m/z) ratio for that ion. The detection method was based on

the fact that each ion would have a different frequency of motion that correlates to the

mass and charge of the ion, similar to mass analysis performed by ion cyclotron

resonance. Further development of this instrumentation led to the mass-selective storage

of ions with external detection in which ions were ejected through a small hole in one of

the end-caps to a detector.3' 4 These advances were all crucial to the development of the

ion trap, but the most important advance in making the ion trap useful as a mass

spectrometer was the development of mass-selective ejection of ions stored within the ion

trap (mass-selective instability mode) in the early 1980s.5 In this method, all ions are

trapped in the quadrupolar field of the ion trap and then sequentially ejected from the trap

in order of increasing m/z by ramping the radio-frequency (RF) voltage applied to the

ring electrode. This, the addition of He damping gas, and stretching the trap led to the

production of the first commercial mass spectrometer by Finnigan MAT (now Thermo

Electron, San Jose, CA) called the ion trap detector or ITD. It was a 3D QIT with

internal electron ionization and external detection.

Until recently, advances in ion trap mass analysis focused on the injection of ions

from external sources and coupling the mass analyzer to a multitude of ion sources such

as matrix-assisted laser desorption ionization (MALDI), electrospray ionization (ESI),

atmospheric-pressure chemical ionization (APCI), and now atmospheric-pressure photo-



















Entrance __ o 9 Exit
End-cap zo End-cap












Figure 1-1. A schematic representation of the 3-dimensional (3D) quadrupole ion trap
(QIT). The 3D QIT consists of 2 end-cap electrodes and a ring electrode. The
inside dimensions of the 3D QIT are defined by ro, the radius of the ring
electrode, and zo, the distance from the center of the QIT to the apex of the
end-cap. For a symmetrical QIT, ro is ~/2 times zo. Ions are injected and
ejected through holes drilled into each end-cap.









ionization (APPI). These multiple sources allowed for a very wide variety of compounds

to be studied from the initial elemental studies to more labile and much larger molecules

such as proteins. Virtually any compound that is ionizable can be analyzed with an ion

trap, meaning that mass analysis is limited by the ability of the ionization source to

produce ions.

Through the years of studying the ion trap, the drawbacks have also been

documented. Perhaps one of the biggest drawbacks is the concept of space-charging.6' 7

Due to the small space in which ions are confined, if the trap is filled with too many ions,

then the applied quadrupolar field will not affect all ions in the trap equally because other

ions act as shields. This issue tends to cause peak-broadening, mass shifting, and the

appearance of ghost peaks. The use of automatic gain control (AGC) reduces this

problem by allowing for an initial pre-scan period in which the computer determines how

long to fill the trap based on an initial packet of ions.8 Other disadvantages of the 3D trap

are the difficulty of injecting ions because of the small inlet hole and the presence of the

quadrupolar field, as well as the loss of half the ions during ejection because ions will

exit the trap at both the inlet and the exit holes.

In 2002, the linear ion trap, also called the 2-dimensional (2D) ion trap, was first

introduced in commercial instruments.9' 10 One design was very similar to that of a triple

quadrupole mass spectrometer, with the second quadrupole used as the ion trap. In this

design, ions are trapped in the second quadrupole and ejected axially (out the back).9

Operating in this manner does increase the storage space of ions thus allowing more ions

to be injected, but using mass-selective ejection, ions may still be lost because they will

exit out the back of the quadrupole and the front. The second design offered a more









dramatic change in design, in which the trap was divided into 3 parts, a longer center

section for trapping ions and two shorter sections (front and back).10 A diagram of this

trap design is shown in Figure 1-2. The significant difference in this design was the

machining of two long slits on two opposite rods of the center section through which ions

can be ejected radially from the trap. Placing a detector on both sides immediately

doubles the number of detected ions. Both designs offered a substantial increase in the

capacity of the ion trap and in the efficiency of ion injection, but the second design also

offered an increase in the number of ions detected.

3D and 2D Quadrupole Ion Trap (QIT) Theory and Operation

A traditional 3D QIT consists on two identical hyperbolic end-cap electrodes with

entrance and exit holes in the center of each and a hyperbolic ring electrode with radius ro

situated between the two end-caps. For an ideal 3D trap, the distance from the entrance

end-cap to the exit end-cap is /2zo, with Zo defined as the distance from the center of the

trap to the apex of the end-cap (Figure 1-1). Thus, the theoretical dimensions of the ion

trap (also knows as the Paul trap) are shown in Equation 1-1 where ro is the internal

radius of the ring electrode.

ro,=2zo Equation 1-1

However, due to the existence of entrance and exit holes in each end-cap which are

needed to inject and eject ions, the quadrupolar field is disturbed, creating imperfections

that affect how ions are trapped and scanned out, producing spectra with improper mass

assignments. By moving each end-cap of the trap outward 0.030 of an inch in the z

direction (called stretching), the field imperfections are reduced, creating a mass

spectrum void of mass shifts.11











30mm long exit aperture


Bark
Section
12mm


Front
Section
12mm


Figure 1-2. Diagram of a 2-dimensional (2D) ion trap. This 2D ion trap design
incorporates three different sections, front and back sections of equal length,
and a center section where the ions are trapped.









Ions entering the trap are successfully trapped by the application of a radio

frequency (RF) to the ring electrode that creates a quadrupolar field in the ion trap due to

the shape of the end-caps and ring (hyperbolic). A direct current (DC) can also be

supplied to the end-caps, but typically, it is maintained at ground. The RF applied to the

ring is defined by two parameters, a constant angular frequency (Q) and a variable

amplitude voltage (V). Solutions to the reduced Mathieu equations (Equations 1-2 and 1-

3) where e is the charge (1.602 x 10-19 C), U is the DC potential, V is the amplitude of the

RF, Q is the angular frequency of the applied RF, and ro and Zo are the internal

dimensions of the trap, give coordinates in a-q space (az, qz) that are then mapped onto

the Mathieu stability diagram (Figure 1-3).12 If these coordinates fall into the region

where the az and qz overlap, i.e. stable in both the axial (z) and radial (r) dimensions, then

the ion will be successfully trapped. In typical operation, the DC potential,U, is zero and

thus all ions have az equal to zero; thus the location of an ion in a-q space is focused to

the q dimension.

az=-2ar=- 6eU/m(ro2+2Zo2)Q2 Equation 1-2

qz=-2qr=8eV/m(ro2+2Zo2)Q2 Equation 1-3

Successful trapping and mass analysis of ions also involves the use of buffer gas,

typically helium (He). The addition of a small amount of buffer gas increases the

pressure inside the ion trap. As ions enter the trap, they are collisionally cooled, focusing

the ions to the center of trap where the quadrupolar field is more uniform, increasing the

efficiency of trapping and eventual ejection. Typically, the He buffer gas is maintained at

1 mTorr within in the ion trap. Another advantage of collisional cooling is the separation

of initial kinetic energies from mass analysis. This means that if a packet of ions of the











0.2


0.0 --- -----------'-|--- -



0.5

-0.2-

0N6



-0.4-
0.8
0.9
1.0

-0.6-





-0.8 I

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4



Figure 1-3. A small portion of the Mathieu stability diagram. Solutions to Equations 1-2
and 1-3 give coordinates in az and qz space that can be mapped onto the above
diagram. If the coordinates for an ion fall into the region shown above, then
the ion will have a stable trajectory inside the 3D or 2D trap and will be
successfully trapped. In normal operation, the DC applied is zero and thus the
az is also zero since it is related to the DC voltage applied, thus the most
important number is the value for qz. The value of qz=0.908 is the edge of the
stability diagram. An ion with a qz greater than this value is not trapped.









same mass has a wide range of initial kinetic energies, as is the case in ions produced by

MALDI, they will all have the same kinetic energy once successfully trapped.

Perhaps one of the most significant operating characteristics of the ion trap is the

ability to selectively isolate an ion of interest and then cause it to fragment into smaller

pieces called daughter ions. This type of analysis is termed tandem MS (MS/MS or

MSn). In some ion traps, isolation of a particular ion of interest is accomplished by using

a stored waveform inverse Fourier transform (SWIFT).13 All ions in the trap exhibit a

frequency related to their q value in a-q space. In Figure 1-3, specific points in a-q space

correspond to different 3r and 3P values. The 3 term is used to describe the secular

frequency of an ion in the quadrupolar field and determines the degree to which an ion

follows the applied field. The fundamental frequency, COz,r, of an ion in the quadrupolar

field is calculated from the Pr and 3P values. Since typically operation involves on the z-

component, the 3r value is not used. The Pz value is approximated from equation 1-4 and

then used to calculate the cz,o value using Equation 1-5.

Bz=~(qz2/2) Equation 1-4

C0z,o=0.5(pzQ) Equation 1-5

In the SWIFT isolation method, a broadband waveform with a notch equal to the

frequency of the ion to be isolated is applied to the end-cap electrodes. Increasing the

voltage of this applied frequency causes ions outside the notch to gain kinetic energy

until they are ejected from the trap, leaving only the ion of interest within the trap.

Another isolation process involves the use of a 5-500 kHz multi-frequency waveform that

has sine components at every 0.5 kHz.14 To isolate an ion of interest, the sine components









corresponding to the frequency of that ion are removed from the multi-frequency

waveform, thus creating a notch similar to SWIFT.

Following isolation, the ion of interest can be fragmented by the application of an

alternating current voltage (AC) equal to the frequency of the ion in the trap across the

end-caps. With sufficient voltage (typically 1-3Vpp), the ion gains kinetic energy by

resonant absorption, which results in more translational motion. With this increased

motion, collisions with the He buffer gas present in the ion trap become very energetic.

These collisions increase the internal energy of the ion until sufficient energy is reached

to break bonds within the ion. These fragment ions are then held within the trap because

of the quadrupolar field except for those fragments that fall below the low mass cutoff

(past the right edge of the stability diagram in Figure 1-3) or those that do not retain a

charge. After scanning the ions out, the resulting spectrum is essentially a mass spectrum

of a mass spectrum or MS/MS (MS2). This process is called collisionally activated

dissociation (CAD) or collision-induced dissociation (CID).15 Unique to the ion trap is

the ability to perform successive CID events. This means that not only can an ion of

interest can be isolated (the parent ion), fragmented into smaller pieces (the daughter

ions), and scanned out resulting in an MS2 spectrum, but also a daughter ion of the parent

can be chosen and further fragmented into its pieces, which can be considered

granddaughters of the original parent ion. The spectrum from this event is MS3. This

process can be repeated n number of times, giving a mass spectrum related to each CID

event. This is why the term MSn is used in connection with tandem MS on an ion trap.

2D/Linear Ion Trap Description

The linear ion trap (LIT), as used in these experiments, is composed of 3 sections

of linear rods with hyperbolic structure, as shown in Figure 1-2. The front and back









sections are 12 mm long, while the center section is 37 mm long.10 For the axial trapping

of ions, separate DC voltages are applied to all three sections of LIT (Figure 1-4), while

RF is applied in two phases to rod pairs for radial trapping (Figure 1-5). For positive ion

analysis, the front section has a potential of -9V, the center section has a potential of-

14V, and the back section has a potential of-12V during ion storage. During mass

analysis, the front and back sections have a voltage of +20V, while the center section is

maintained at -14V. A supplementary AC voltage is also applied in two phases, but only

to the x rod pair (Figure 1-5) for isolation and CID.

The application of this supplementary AC voltage is also used to resonantly eject

ions radially, along the x-axis, out of two small slits (30 mm long) cut into the x rod

pairs. As in the 3D trap, because of the existence of these slits, the quadrupolar field is

disturbed, and thus the 2D ion trap is also stretched (in the x-direction) to help reduce the

field imperfections. Ejected ions are then detected on both sides with conversion

dynodes and electron multipliers. This mode functions very similar to a conventional 3D

trap, but has the added benefit of doubling the number of detected ions. Other

advantages of the 2D trap include increased ion injection efficiencies, increased trapping

efficiencies, and larger storage volume. Alternatively, by controlling the DC applied to

the separate sections, ions can be non-mass selectively ejected axially, along the z-axis, at

which point they could enter a different mass analyzer such as a time-of-flight, penning

trap, or an orbitrap. A significant problem in the production of 2D traps is the machining

and mounting of the hyperbolic surfaces. Because ions are spread out over 30 mm in the

2D trap versus the 1 mm in the 3D trap, any imperfections in the rods can cause

significant decrease in mass resolution.10 Increased machining costs along with the added













y









Figure 1-4. A Schematic representation of the 2D ion trap. Ions enter from the left side
of the figure. The 2D trap is separated into 3 sections, two shorter sections of
12 mm in length, and a longer, center, section 37 mm long. Separate DC
voltages are applied to all three sections in order to trap ions. Ions are ejected
through slits in the x-rod pairs of the center section.








RF+





RF RF -






RF+

GND
B




AC+ AC-






GND
Figure 1-5. Two views of the 2D ion trap looking down the z-axis, along the flight path
of the ions. A shows the radial quadrupolar trapping field. The phases of RF
are applied to all the electrodes of the ion trap to create the quadrupolar field.
B shows the radial dipolar excitation arrangement. An AC voltage is applied
only to the x-rod pairs. The AC voltage is used for isolation and collision-
induced dissociation experiments as well as in the ejection of ions from the
trap. Ions are ejected out of slits cut into the x-rod pairs.









benefits are several reasons why the 2D trap is significantly more expensive than the 3D

trap. Cost of machining may decrease as the process is refined.

Tissue Imaging by Mass Spectrometry

Background

Prior to the advent of matrix-assisted laser desorption/ionization (MALDI), mass

spectrometric analysis of tissue specimens focused primarily on the identification of

elements in thin tissue sections, with one of the first experiments involving the detection

of heme-bound iron from red blood cells using a laser microprobe mass analyzer, or

LAMMA.16 With the advent of soft ionization methods for mass analysis such as

MALDI17 and electrospray ionization (ESI)18, characterization of traditionally labile and

involatile molecules such as proteins and peptides present in biological tissues has been

made possible. The use of softer ionization techniques allows for reduced complexity in

analysis of biomolecules with mass spectrometry because intact molecular ions of the

particular molecule are produced rather than fragment ions.

Using MALDI, mass spectrometric microprobes are now offering promising new

approaches to map the distribution of small and large molecules directly from tissue

sections at biologically significant levels and to help unravel the molecular complexities

of cells.19-22 Microprobe MS offers the unique ability to directly analyze chemical species

from tissue samples, either creating images of how a certain m/z value is localized within

the tissue section (usually focused on the proteins and peptides present or absent in

normal and diseased tissue23) or for identification of site-specific drug activity as

described previously.24 Caprioli et al. showed the capability of MALDI to create images

of spatial localization first using the molecular weight of coomassie blue dye to map a









copyright symbol imprinted onto a MALDI target23 and then showing a map of the

location of 3 different proteins in mouse brain.21

Secondary ion mass spectrometry, SIMS, may provide similar capabilities to

MALDI microprobes, with potentially higher spatial resolution due to the smaller size of

the primary ion beam compared to the laser beam used in MALDI, albeit with much

lower sensitivity because of the monolayer-depth interrogated by SIMS. Interest in the

nervous system using laser microprobes has been related to the analysis of rat brain via

SIMS and MALDI,20 and on neurons from Aplysia californica, sea slug, using MALDI

and laser desorption/ionization on porous silicon (DIOS), primarily looking for proteins

and peptides.25'26 Neurons from Aplysia californica are known to be of the largest

diameter in the animal kingdom, up to 1 mm, which allows for intracellular studies using

MALDI. Todd et al. have shown the analysis of lipids in brain sections, specifically for

phosphatidylcholine, the primary constituent of plasma membranes, and have shown that

the disappearance of the m/z 184 fragment ion of phosphatidylcholine in the image

created by SIMS following injection of lipopolysaccharide, LPS, can be related to brain

damage.19 Injection of LPS into brain sections is known to induce brain damage through

demyelination of nerve tissue. Studies in our laboratory have demonstrated the ability to

map paclitaxel, a small drug compound, at trace (pg/mg) levels in ovarian tumors.24'27

Critical to such sensitive and selective analysis was the use of MALDI coupled to an ion

trap that provided capability of tandem mass spectrometry (MSn).

Currently, mass analysis for directly probing tissue sections is done with either a

time-of-flight (TOF) or a quadrupole ion trap (QIT). Since tissue samples contain a wide

variety of molecular compounds and elements up to and beyond 100,000 Da depending









on the type of tissue analyzed, it is vital to have an instrument capable of efficiently

detecting this very wide molecular weight range and capable of real time compound

identification. A TOF analyzer is capable of detecting this wide molecular weight range;

however, compound identification is normally performed offline using an enzymatic

cleaving agent.21 Tandem MS on a TOF using post-source decay is still an inefficient

process and not necessarily repeatable, making molecular identification difficult. In

contrast, tandem MS with an ion trap is very practical and reproducible. The major draw

back of using a QIT in connection to tissue imaging by MALDI is the limited mass range.

Current QITs, including the 2D-QIT, are limited in mass range to 4000, even

though higher molecular weights, up to m/z 100,000, have been successfully analyzed;

this mode of operation is not routinely performed.28 This means that the primary

justification for using QIT technology for imaging MS is for detection of smaller

compounds such as drugs, peptides, or lipids with identification by tandem MS and for

generating more specific images such as the mapping of a fragment ion produced by

MS2. One could even imagine the mapping of a fragment ion from MS3, MS4, or MSn to

create a very specific image of a desired compound.

Fundamentals

All organs present in animal or plant species can be sampled by imaging MS. In

animal species, the organs are removed and snap frozen in N2 (1) and stored in a -80C

freezer until needed. When ready to perform mass analysis, the organ is removed from

the freezer and prepared for sectioning with a cryotome. Sectioning with a cryotome

requires the use of optical cutting temperature medium (OCT) to affix the organ tissue to

a sample stage. Unfortunately, OCT can interfere with mass spectrometric analysis, so









care must be taken to insure that the section of tissue used does not have any OCT on it.

The tissue can be cut to any desired thickness, but usually the thickness ranges from 10

|tm to 60 |tm. Once cut, the tissue section can be placed on a metal, glass, or conductive

glass surface for imaging MS. This process is shown graphically in Figure 1-6.

Imaging MS can be performed on any mass spectrometer employing a MALDI (or

SIMS) source as long as the sample plate, or the laser, can be rastered to move the laser

beam across the tissue surface in a predefined pattern. Some researchers have also shown

the ability to performing imaging MS studies with a new ionization technique, desorption

electrospray ionization (DESI), at atmospheric pressure, but with far less resolution than

MALDI or SIMS.29 The laser beam can be moved, or stepped, across the tissue surface

at any desired step size. For normal analysis, the step size is equal to the diameter of the

laser beam, as shown in Figure 1-7. In MALDI, the laser spot size is typically anywhere

from 50-100 |tm, which is based primarily on the diffraction limit. For a shorter analysis

time, under-sampling can be employed in which the step size is larger than the laser spot

diameter. Sampling at twice the laser spot diameter will reduce the analysis time by one

half, but will also limit the resolution of the generated image. Over-sampling is the third

analysis type, but it is not typically employed. By stepping at half the laser diameter, 50

|tm in this example, each new spot overlaps with the previously sampled area thus

generating a spectrum that is a combination of an undisturbed area and a sampled area.

Instead, it would be preferred to reduce the spot size of the laser beam to achieve a higher

resolution rather than try over-sampling. Smaller spot sizes haven been employed on

non-commercial instruments typically using a small slit to narrow the laser beam with

some success. In SIMS, the spot size can be much smaller, usually a few hundred
















- C a


Section the tissue


Go


Place on microscope slide


100 7827 Extract to
90 Generate i
80
70
60 Position S
50o 756 Mass Spe,
810 71
753 79
30
20 723 7 275 8324
,9 7 8 82b l 8
10
0 2 743 4
,21 74kil0 76 78 I 11, 1i,,,
720 740 760 780 800 820 840
rn/z


mage

specific
ctrum


Send to MS


Coat with matrix Raster laser across tissue in MS


Figure 1-6. The process of creating an image by mass spectrometry. The two most
important stages in tissue analysis by mass spectrometry are 1) cutting the
tissue to an appropriate thickness with a cryotome, such as 10 Ltm for brain
tissue, and 2) applying the MALDI matrix in a manner that reduces the
possibility for analyte migration airbrushingg is shown above). When cutting
the tissue, it is important to ensure that no OCT compound is on the tissue
sections as it can interfere with mass analysis. The mass spectrometric stage
involves rastering the sample plate at a predefined step size with respect to the
laser. Each mass spectrum is then position specific so all that is required for
generating an image is extracting the m/z value of interest.


I


/7\


B












f
Laser spot size
100 ipm


f
Laser spot size
100 pm


Step size 100 im






Step size 200 im


Sampling at the
laser resolution




I
Under-sampling


Over-sampling


f
Laser spot size
100 pm


U
Step size 50 tim


Figure 1-7. The types of sampling used in imaging MS. Sampling at the laser resolution
is the most widely used data collection scheme because it provides the highest
resolution. Under-sampling will take less time for acquisition, but is a poor
choice when trying to identify changes over a very specific boundary area.
Over-sampling is typically not used because overlapping data points may
cause confusion in the generated images.









nanometers, because of the use of an ion beam, but imaging by SIMS does not allow for

the detection of labile molecules and has limited penetration depth into a sample.

Unlike SIMS, a requirement to perform MALDI is the application of a matrix

material that absorbs the energy of the laser and gently transfers some of that energy to

analytes prepared with the matrix. For analysis of standards, the matrix, present at a

much higher concentration, and the analyte are deposited as solutions on the surface of a

metal plate and allowed to dry for co-crystallization to occur. However, this method

cannot be employed in tissue imaging MS because depositing the matrix as a droplet on

to the tissue surface can cause compounds in the tissue to move, thereby disturbing their

original location,27 as shown in Figure 1-8.

Instead, the matrix must be applied in a manner that will not interfere with the

actual localization of compounds in the tissue. Several methods have been developed

that successfully apply matrix in this manner, including electrospray deposition,27

nebulization,30 or airbrushing. The same principles for ESI are also involved in the

deposition of matrix by electrospraying. For ESI, it is desirable to have the analytes

evaporate into the gas phase; however, in electrospray deposition of matrix this is not

desired. The purpose of the electrospray is to produce a mist of very small droplets of

matrix that will wet the tissue surface and mix with molecules present for co-

crystallization to occur. A diagram of this process is shown in Figure 1-9. The use of a

glass nebulizer or an artistic airbrush accomplishes the same task, but in a somewhat

simpler manner. Both use pressurized gas to aspirate a liquid sample for deposition onto

any surface. The airbrush is a new means for deposition of matrix onto tissue, and is

described in Chapter 4 of this dissertation.



















Pipette matrix

B


Tissue

SSample plate

SAnalyte A


SAnalyte B


Ippppff X X 21


Analyte migration


Figure 1-8. Pipetting the matrix onto a tissue surface causes analyte migration as shown.
A shows the tissue on the MALDI sample plate, B shows the pipetting of the
matrix onto the tissue, and C shows the how pipetting the matrix causes
analyte migration in the tissue. Imaging MS cannot be performed when this
type of matrix application is employed.


C ~~ ~ r ~ 7~











S4 -Tissue

Sample plate

SAnalyte A


Airbrush or
Electrospray matrix


000
00 O
0 0 0
0 0 0
0 00 0
0 0 0 0 0 0
O O000 0 0
0 0 0 0 00 0


I.... S b&


Analyte B


Matrix incorporation
no analyte migration


Figure 1-9. The deposition of matrix as a fine mist, by either airbrushing or
electrospraying. A shows the tissue on the MALDI sample plate before
matrix application and B shows the application of matrix using a device the
produced very small droplets such as an airbrush or an electrospray needle
with the proper voltage. C shows the final tissue section with matrix applied.
The process permits the incorporation of the matrix material into the tissue
medium without causing analytes to migrate across the tissue.


A r
p


0 00 0 0 0 0 0 Ol









Matrix-Assisted Laser Desorption/Ionization

Background

The use of lasers coupled to mass spectrometers to desorb ions from a surface is not

a new concept. Laser desorption ionization (LDI) has been performed since the 1960s.

Early analyses were of atomic ions from metal surface, primarily because the absorbance

of the laser light caused heat build up in the molecules leading, to the breaking of

bonds.31 32 The practical limit for the mass analysis of intact molecular ions was therefore

<1000 amu. Another disadvantage was the presence of selective absorption because not

all analytes absorbed the laser wavelength used.32 Thus, analytes that absorbed the

wavelength of the laser would be ionized selectively compared to those that did not,

limiting the dynamic range of the technique. Other techniques for the analysis of

surfaces or materials deposited onto surfaces have also been developed. Secondary ion

mass spectrometry or SIMS employs an ion gun to bombard a dry sample with a beam of

primary ions with energies of 5 keV to 100 MeV.33 These primary ions strike the analyte

on the surface, producing secondary ions of the analyte (and anything else near the

surface).34 This technique is in many ways similar to the action of the cue ball striking

another ball, causing movement in billiards. The disadvantages of this technique are the

high degree of analyte fragmentation, which limits the analysis to <1000 amu, and the

limited depth analysis because only the first few nanometers are exposed to the primary

ions. An advantage of the technique is the very small diameter of spot sizes, typically

around 100 nm or less, that is very useful in the analysis of tissues at the cellular level as

described in the imaging MS section.35 A similar technique to SIMS is fast atom

bombardment (FAB) also known as liquid SIMS (LSIMS). FAB allows for the soft

ionization of involatile compounds because the liquid matrix in which analytes are









dissolved acts as an energy transfer medium, thereby reducing the internal energy build

up in analyte molecule. This allows the intact molecular ion to be produced rather than a

spectrum full of fragment ions of the analyte. FAB was a widely used technique for

many years because it was first ionization method to offer the analysis of involatile

species without the need for derivatization techniques.

In the late 1980s, a new ionization technique emerged that would soon supplant the

use of FAB. The technique was called matrix-assisted laser desorption/ionization

(MALDI) and today it is one of the most widely used techniques for the study of

involatile species such as proteins, peptides, and polymers. The technique began in two

different parts of the world, but both groups described the ability to desorb high

molecular weight species in the solid phase with the use of a matrix material.

In Germany, Franz Hillenkamp and Michael Karas developed a method from

observations of the laser desorption of amino acids.17'36 Some amino acids absorbed the

laser light from a frequency tripled Nd-YAG laser (355 nm) better than other amino

acids. Mixing a highly absorbing amino acid (such as tryptophan) with one that absorbed

weakly (such as alanine) and subjecting the mixture to laser irradiation (at one tenth of

the power needed to desorb just alanine produced a mass spectrum showing both

compounds as intact ions ([M+H]+ at m/z 74 for alanine and 189 for typtophan). Further

experiments involved mixing a high concentration of nicotinic acid (a highly UV

absorbing species) and a low concentration of bovine albumin (MW 67,000) on a metal

probe tip. After drying to allow crystallization to occur, the spot was irradiated with a

frequency quadrupled Nd-YAG laser (266 nm), producing one of the first mass spectra

containing a peak corresponding to a large intact protein ion, [M+H]+.









Around the same time, Koichi Tanaka of Shimadzu Corporation in Japan

performed experiments in which lysozome (chicken egg) having a molecular weight of

14,306 was mixed with slurry of cobalt powder in glycerol.37 This solution was allowed

to dry before irradiation with a nitrogen laser (337 nm). This mixture was also successful

in producing a spectrum containing a signal for the intact molecular protein.

Both groups postulated that the matrix material, an organic acid in the Hillenkamp

and Karas work and a metal in Tanaka's work, acted as an intermediary, whereby it

absorbed the energy of the laser and transferred the energy to the protein causing soft

desorption. The technique today is called matrix-assisted laser desorption/ionization

(MALDI) and is a combination of these two experiments. Typically, the matrix

employed today is a small organic acid, using a pulsed nitrogen laser at 337 nm for

irradiation. Although most practitioners of MALDI follow the procedure presented by

Hillenkamp and Karas, in 2002, Tanaka was awarded a share of the Nobel Prize in

chemistry for his contribution to the analysis of proteins.

Theory of MALDI

The theory of MALDI (and laser desorption) can be divided into two categories,

the desorption process and the ionization process, as is true for most desorption

techniques. Both categories are currently studied with a desire to better understand the

process of MALDI and thus improve the reproducibility of the technique. It is evident

from the research conducted thus far that the most critical aspect for not only the

successful desorption of intact molecular ions, but also the ionization of desorbed ions, is

proper selection of the matrix compound.

In MALDI, the analyte of interest is mixed with an organic acid that absorbs at the

wavelength of the laser. The ratio of analyte to matrix is typically 1:1,000 because part









of the aspect in proper soft desorption of the analyte is to completely surround each

analyte molecule with many matrix molecules. In this sense, the matrix is used to dilute

the analyte molecules and isolate them from each other in the solid state (Figure 1-10).38,

39 The problem with desorbing biomolecules without matrix is a result of the interaction

of biomolecules with other biomolecules. As the size of a molecule increases, the

intermolecular forces tend to approach the intramolecular forces in strength.32 For

desorption of a neat analyte sample to occur, the intermolecular forces have to be broken,

releasing molecules into the gas phase. When the intermolecular forces are equal to the

intramolecular forces, then the desorption process itself will induce fragmentation.32 This

aspect is part of the reason for the limitations of laser desorption to ionize only intact ions

below a MW of 1000.40 By isolating biomolecules from each other, the matrix material

successfully reduces the intermolecular forces, thus helping to reduce the fragmentation

ofbiomolecules. Typical matrices used in MALDI analysis are shown in Figure 1-11.

The desorption process is a fairly well understood process. As long as the energy

deposition is large enough, and the time-scale of the laser pulse is short (nanoseconds), a

phase transition occurs directly from solid to gas. The key in energy deposition is in the

amount of energy supplied per pulse, as measured by the fluence (J/cm2), instead of the

duration of the laser pulse, as reflected in the irradiance (W/cm2).41 Research has shown

that lengthening the pulse appears to have little or no effect on the mass spectrum

generated; however, there is evidence suggesting the presence of a threshold of energy

that must be reached in order for desorption to occur.42-45 The threshold of energy is

matrix-dependent and suggests that there is a strong connection to the amount of energy









Mostly fragment ions


Laser


SAnalyte
* Matrix


Laser, N2 337 nm


.4


S~
* -


Figure 1-10. Two diagrams of the MALDI process. Diagram A shows how the matrix
material dilutes the analyte molecules to reduce the amount of intermolecular
interactions, while the desorption of neat analytes especially biomolecules will
induce fragmentation. The matrix is present at a much higher concentration
than the analyte to effectively act as an intermediary by absorbing the laser
energy and transferring it to the analyte without causing fragmentation. In
Diagram B, the plume generated after a laser pulse contains ions, both positive
and negative, related to the analyte and matrix compound as well as many
neutral species.


Positive ion
Negative ion


Mostly intact ions









O O
HO
HO OH 1 OH
CN
OH HO

2,5-dihydryoxybenzoic acid a-cyano-4-hydroxycinnamic acid
(Gentisic Acid or DHB) (HCCA)


CH3 0 O

0 OH OH

HO N

H3c/O
3,5-dimethoxy-4-hyrdoxycinnamic acid Nicotinic acid (NA)
(Sinapinic acid or SA)


0

OH



3-aminoquinoline (AQ) Picolinic acid (PA)



HO 0 NH2

HO, y OH




OH NO2


2,4,6-trihydroxy acetophenone para-nitroaniline (PNA)

(THAP)


Figure 1-11. Common organic acids used as matrices in MALDI and their abbreviations.
No single matrix is compatible with all analytes, and thus a wide variety of
matrices have been employed.









deposited in a single pulse. Once the threshold is met, large chunks of material (clusters)

are ablated from the surface, which then undergo the phase transition.38 42

The phase transition occurs over a very short time frame, creating a plume of

matrix molecules that entrain the analyte molecules. This expanding plume is similar to a

free jet expansion under vacuum conditions and is an explosive event. An important

aspect of the desorption process is the kinetic energy of the ions produced. The average

initial kinetic energy of the ions generated has been of interest in an attempt to better

understand to the MALDI process. One reason for these studies is to achieve better mass

accuracy in mass analysis, especially with a time-of-flight analyzer, since the measured

m/z ratio is directly related to the kinetic energy of an ion. An integral finding for

peptides and proteins was that the velocity was the similar for all ions generated and that

it was solely dependent on the matrix used and not the wavelength of the laser

employed.46 The average initial velocity of bovine insulin can be as high as 520 m/s

(1163 mph) when using 2,5-dihydroxybenzoic acid (DHB) as the matrix to as low as 280

m/s (626 mph) with 2-(4-hydroxy-phenylazo)benzoic acid (HABA). Although the

average initial velocity for peptides and proteins is very similar, different analyte classes

exhibit a different initial velocity under the same experimental conditions. Furthermore,

the average initial ion velocity is dependent on the preparation technique of the sample

spot to be irradiated.47 For example, varying the solvent of the matrix 3-hyrdoxypicolinic

acid (HPA) changed the average initial velocity of insulin from as high as 620 m/s to a

low of 444 m/s.47 This change was related primarily to poor crystallization of the matrix,

causing sample spot heterogeneity.









The ionization process in MALDI is reflected by the mass spectrum produced.

This provides a means for studying what may occur in the ionization region of the mass

spectrometer. A better understanding of the mechanisms could assist in developing

procedures that will maximize ion yields, provide for improved access to all types of

analytes, aid in controlling the charge state produced, and control fragmentation induced

during laser ablation. Central to the ionization process in MALDI ions is the very

abundant production of singly charged species, either positive or negative. Multiply

charged species are observed, but far less abundantly than in electrospray ionization,

which indicates a different ionization mechanism for the same types of compounds.41 The

production of multiply charged ions is somewhat higher when an infrared laser is used for

MALDI, but still the most abundant ion signal is for the +1 or +2 charge states.38' 48 The

ions generated can be protonated, deprotonated, cationized, or radical ions and are

generated by gas-phase interactions with the matrix rather than from preformed ions in

the solid state.49 The radical ions are usually associated with the matrix species, since

radical formation is primarily a result of photon absorption.50

It is now generally believed that ionization in MALDI occurs via two steps, a

primary ionization mechanism and a secondary ionization mechanism.41' 51 The primary

ionization event refers to the production of ions from the neutral molecules in the solid

sample, while the secondary event refers to ions that are generated from those produced

in the primary event and are usually the species representing peaks in the mass spectrum.

The ions produced in the process are both positively and negatively charged and thus ion

optics are used to select and guide ions of the desired polarity to the detector. Although

ions are produced in MALDI, the overall ion-to-neutral ratio for ultra-violet (UV)-









MALDI is only about 10-4, showing that the ionization process is inefficient and that the a

mass spectrum produced shows the abundance of the minor species in the MALDI

process.52 53

The primary ionization event is associated with the formation of ions from the large

excess of matrix molecules. These ions are the result of photon absorption (hv) causing

the neutral matrix molecules (M) to be excited (M*) with a subsequent loss of an electron

upon relaxation, leading to a radical cation (M+ ), as shown in Equation 1-4. Protonated

matrix molecules are generated from interactions of the radical ion with a neutral matrix

molecule to form a protonated matrix molecule, [M+H] and a neutral matrix radical,

[M-H]', as seen in Equation 1-5.

M + hv- M* M+' + e- Equation 1-4

M + M [M+H] + [M-H]' Equation 1-5

Prominent clustering in laser ablation indicates that it is possible to excite two

matrix neutral molecules, M*M*, which then pool their energy, creating one radical

cation of the matrix and one neutral matrix and an electron, as shown in Equation 1-6.

This excited state complex can also ionize an analyte molecule, A (Equation 1-7). Since

negative ions are also observed in MALDI, there must also be a mechanism for

producing them. A disproportionation reaction has been proposed as a possible means to

generate negative matrix ions, as shown in Equation 1-8.41 Other possible reactions have

been proposed that could generate matrix and analyte ions such as excited state proton

transfer, thermal ionization, and desorption of preformed ions.41'54

MM + 2hv -M*M* M + M+ + e- Equation 1-6

M*M* + A MM + A+ + e- Equation 1-7









2M + nhv (MM)* M- + M' Equation 1-8

The secondary ionization event is characterized by ionization in the expanding

MALDI plume through ion-molecule reactions. It is believed that equilibrium has been

reached at this time in the MALDI plume, allowing for secondary processes to be

described by gas-phase thermodynamics.54 Ion-molecule interactions occur between

matrix ions and molecules and matrix ions with analyte molecules. A radical cation of

the matrix from the primary ionization event can react with a neutral matrix molecule,

producing a protonated matrix molecule, as shown in Equation 1-9. A process similar to

fast atom bombardment (FAB) can also occur in which dissociative electron capture

occurs, generating a negative matrix ion and a hydrogen radical, H' (Equation 1-10).55

This hydrogen radical is then available for donation to an analyte or matrix molecule.

Electrons are produced in some primary ionization event reactions, but it has also been

shown that electrons are ejected from the metal sample plate due to the photoelectric

effect.56' 57

M + + M MH+ and [M-H]' Equation 1-9

M + e- [M-H]- + H' Equation 1-10

Protonated analyte molecules are generated from proton transfer reactions with

matrix ions, MH as long as the gas-phase basicity of the analyte is greater than that of

the matrix (Equation 1-11). A deprotonated matrix molecule, [M-H]-, can abstract a

proton from an analyte, A, producing a negative analyte ion as long as the gas-phase

basicity of the deprotonated matrix is higher, which is typical when using basic matrices

(Equation 1-12).


MH+ + A M + AH


Equation I -I









[M-H]- + A M + [A-H]- Equation 1-12

Cationization is also a significant process occurring in the secondary ionization

process. Cationization of the analyte is significantly enhanced when an akali metal such

as a sodium salt is added to the matrix and is thus present in the solid crystal state. A

typical reaction is shown in Equation 1-13 for an analyte molecule with NaC1. A recent

finding suggests that this cationization event is matrix-dependent and thus care must be

taken in choosing the appropriate matrix if cationization is desired.58

A + NaCl [A+Na]+ + C1- Equation 1-13

Phospholipids and Sphingolipids

The Structure of Phospholipids

There are many classes of lipids present in all living organisms. They are

characterized by their ability to be extracted from tissues by organic, non-polar solvents

such as chloroform and methanol. They typically have a high hydrocarbon content and

can have both polar and non-polar regions of the molecular structure. This section

focuses on the structure of glycerophospholipids and sphingolipids, but the structures of

the other classes of lipids can be found in most biology textbooks. The basic structure of

a glycerophospholipid (GPL) is shown in Figure 1-12. GPLs are derived from the

molecule glycerol and consist of two fatty acyl chains esterified to the sn-1 and sn-2

(systematic nomenclature) positions of the glycerol backbone that are commonly referred

to as non-polar tails. The primary means of differentiating the GPLs is by the molecular

composition of the polar head group that is esterified to the sn-3 position of the glycerol

backbone. There are four major classes of GPLs: phosphatidylcholines (PCs),

phosphatidylethanolamines (PEs), phosphatidylserines (PSs), and phosphatidylinositols

(PIs). The structure of the polar head group for each class is shown in Figure 1-12.













Glycerophospholipids (GPLs)

oHeadgroup


O=P-O
O O
0
sn-I sn-2 I
H2C-CH-CH2
I I sn-3
0 0





~-g o
I I
O= C C-O







LL -L
"U "U


Structures of the headgroups for GPLs


--CH2 CH3
I
N-CH3
CH3

-CH2 H
I+
N-H
H

-CH2 H
C. +
O NH3
0


Phosphatidylcholine




Phosphatidylethanolamine




Phosphatidylserine





Phosphatidylinositol


Glycerol OH

H2C-CH-CH2
I I
OH OH


Figure 1-12. The basic structure of glycerophospholipids (GPLs). All GPLs consist of a
polar head group and two non-polar tails, fatty acids. The polar head group is
the primary means of separating GPLs into different classes. The names of
the four major classes are shown next to the structure of the head group.









Another class of lipids present in animal cells and tissues is the sphingolipids.

These lipids are characterized by the presence of the sphingosine base (or a related base)

as the backbone with a polar head group at the sn-3 position and only one variable fatty

acyl chain at the sn-2 position. The head group, shown in Figure 1-13, also distinguishes

the three major classes of sphingolipids: ceramides (CER), sphingomyelins (SPM), and

cerebrosides (CRB). This class of lipids is primarily found in nerve tissue such as brain,

sciatic nerve, and spinal cord. The names adopted for them reflect this finding, especially

sphingomyelins, because they are predominantly found in the myelin sheath, which is the

protective covering and insulating medium of the nervous system.

Typical fatty acids present in eukaryotic cells and tissues have between 12 and 24

carbon atoms and can have varying degrees ofunsaturation. The common names,

number of carbon atoms, and the number of double bonds for many of the common fatty

acids are shown in Table 1-1. A saturated fatty acid does not have any double bonds,

while an unsaturated fatty acid can have anywhere from one to six double bonds present

in the carbon chain. All the double bonds present in a fatty acid chain are of the cis

configuration instead of the trans configuration, as shown in Figure 1-14 for oleic acid

(18:1) in a chemical structure and a 3-dimensional representation.

The naming style adopted for the research conducted here uses the abbreviated

name of the phospholipid followed in parenthesis by two sets of numbers for GPLs and

one set for sphingolipids identifying the fatty acyl chain(s). The first set of numbers

refers to the sn-1 position and the second set to the sn-2 position. Two numbers

separated by a colon identify the fatty acyl chains: the first number refers to the number

of carbon atoms in the fatty acyl chain, including the carbonyl carbon, and the number










Sphingolipids


sn-1 /
0
HO sn-2
I
CH- CH-CH2
I I sn-3
CH NH
I I
CH C=O
(CH2)12 -
CH3 |


TO
>.,


LL
U-


HeadarouD


Structures of the headgroups for sphingolipids


-H


Ceramide


0

-P-O-0CH2 H3
o- \ N-CH3 Sphingomyelin
CH3


Cerebroside


OH
HO
I I
CH-CH-CH2
CH NH2
I I

CH
I Sphingosine base
(CH2)12
CH3


Figure 1-13. The basic structure of sphingolipids. This lipid class is derived from
sphingosine or another similar base. The structure of sphingosine is shown at
the bottom. They are similar to GPLs in that they have a non-polar tail,
usually consisting of one fatty acid, and a polar head group. The polar head
group is also used to differentiate the different classes of sphingolipid. The
head groups for the major sphingolipids are shown above.


A.-- d















Table 1-1. A list of the common fatty acids found in nature.


Chemical Names and Descriptions of some Common Fatty Acids
Carbon Double
Common Name Atoms Bonds Scientific Name
Butyric acid 4 0 butanoic acid
Caproic Acid 6 0 hexanoic acid
Caprylic Acid 8 0 octanoic acid
Capric Acid 10 0 decanoic acid
Lauric Acid 12 0 dodecanoic acid
Myristic Acid 14 0 tetradecanoic acid
Palmitic Acid 16 0 hexadecanoic acid
Palmitoleic Acid 16 1 9-hexadecenoic acid
Stearic Acid 18 0 octadecanoic acid
Oleic Acid 18 1 9-octadecenoic acid
Vaccenic Acid 18 1 11-octadecenoic acid
Linoleic Acid 18 2 9,12-octadecadienoic acid
Alpha-Linolenic Acid (ALA) 18 3 9,12,15-octadecatrienoic acid
Gamma-Linolenic Acid (GLA) 18 3 6,9,12-octadecatrienoic acid
Arachidic Acid 20 0 eicosanoic acid
Gadoleic Acid 20 1 9-eicosenoic acid
Arachidonic Acid (AA) 20 4 5,8,11,14-eicosatetraenoic acid
EPA 20 5 5,8,11,14,17-eicosapentaenoic acid
Behenic acid 22 0 docosanoic acid
Erucic acid 22 1 13-docosenoic acid
DHA 22 6 4,7,10,13,16,19-docosahexaenoic acid
Lignoceric acid 24 0 tetracosanoic acid









O 0
HO HO











trans c





Oleic acid (18:1)
CH3




CH3


Figure 1-14. A typical fatty acid, oleic acid, is shown in two possible configurations. For
all double bonds occurring in a fatty acid chain, the double bond is in the cis
configuration rather than the trans configuration. The structures shown are of
the basic shorthand form.









after the colon refers to the number of double bonds present in the carbon chain (degrees

ofunsaturation). For example, PC (16:0, 18:1) identifies the lipid as phosphatidylcholine

with two fatty acyl chains. The fatty acyl chain at the sn-1 position is palmitic acid (a 16-

carbon chain with zero degrees of unsaturation), while the fatty acyl chain at the sn-2

position is oleic acid (an 18-carbon chain with one degree of unsaturation). This is the

most common phospholipid in the cellular membrane of eukaryotic cells; its structure is

shown in Figure 1-15. Another example is SPM (18:0). This lipid is classified as

sphingomyelin with stearic acid, an 18-carbon chain having zero degrees of unsaturation,

at the sn-2 position (Figure 1-16). In general, the fatty acid at sn-1 is saturated, while the

fatty acid at sn-2 can be saturated or unsaturated.

Analysis of Lipids by Mass Spectrometry

Due to the fact that lipids are present in every organism, they have been analyzed

by mass spectrometry for many years. Initial experiments used electron ionization (EI)

for the analysis of derivatized lipids. In EI, gaseous analyte molecules, M, are ionized

within the vacuum chamber using an electron filament. This filament produces energetic

electrons (typically at 70 eV) that collide with the gaseous analyte molecules, causing the

ejection of an electron from the analyte molecules and producing a positively charged

radical, M+, of the analyte molecule, as seen in Equation 1-14.

M + e- M+' + 2e- Equation 1-14

M+' is called the molecular ion because it represents the mass of the intact analyte

molecular ion; however, in El this ion is often very low in intensity due to the highly

energetic process of electron ionization. Under El conditions, the analyte is often

fragmented extensively, producing a spectrum very populated in the low mass region.











PC (16:0, 18:1)
Basic structure


CH3


Figure 1-15. The structure of PC (16:0, 18:1) in a normal structure form is shown. This
is the most abundant glycerophospholipid present in mammalian cellular
membranes.









CH3
H\ 3C
xN
OH3


SPM (18:0)
Basic structure


Figure 1-16. Above is the structure of SPM (18:0). This is the most abundant
sphingolipid in mammalian nerve tissue.









One positive aspect of this fragmentation is that every compound will have a

characteristic fragmentation pattern, allowing for the creation of a searchable database for

the identification of compounds run by gas chromatography coupled to mass

spectrometry (GC/MS) with an El source. The extensive fragmentation can be a problem

in the analysis of lipids, however, especially in the analysis of unknown lipids. A

requirement to perform El is that the analyte must have a high enough vapor pressure to

be introduced into the ion source as a vapor. Most lipids do not have a sufficiently high

vapor pressure to be analyzed directly by EI, so they must be derivatized to a more

volatile species. There are many techniques to derivatize most lipids, but some lipids,

such as phospholipids, do not have a good procedure for derivatization and thus cannot

be analyzed by EI. Derivatization of phospholipids first requires enzymatic cleavage of

the phosphate head group, which makes classification more difficult because the head

group in the distinguishing feature of the different classes glycerophospholipids.

Chemical ionization (CI) reduces the amount of fragmentation during the

ionization process and is considered a soft ionization method when compared to EI. Soft

ionization refers to the ability of an ionization technique to produce a dominant ion of the

intact molecular species with little or no fragmentation. CI is a proton transfer ionization

process and can be performed on the same instrument, but with a tighter ion source than

EI. In CI, a reagent gas, typically methane, CH4, is used to pressurize the source to 1

Torr. The same filament used in El is turned on and the electrons from the filament

collide with the reagent gas molecules rather than sample molecules because they present

in large excess, producing reagent ions from the methane gas. Table 1-2 shows the

primary reactions that occur once a methane molecule, CH4, is ionized by an electron, e









70ev. CH5 and C2H5 are the two primary sources of proton transfer to an analyte

molecule because their conjugate bases (CH4 and C2H4) have relatively low proton

affinities, 131.6 and 159 kcal/mol, respectively.



Table 1-2. The common reactions that occur in chemical ionization when methane, CH4,
is used as the reagent gas.
CH4 + e-70eV CH4+ + e-thermal + e50eV Ionization
CH4+' CH3 + H' Fragmentation
CH4+' CH2+ + H2 Fragmentation
CH4+* + CH4 CH5+ + CH3* Ion/molecule reaction
CH3+ + CH4 C2H5+ + H2 Ion/molecule reaction
CH2+' + CH4 C2H3+ + H2 + H Ion/molecule reaction
C2H3 + CH4 C3H5+ + H2 Ion/molecule reaction


For proton transfer from a reagent ion to an analyte molecule to occur, the analyte

must have a higher proton affinity than the conjugate base of the reagent ion. If this

stipulation is met, the analyte will be ionized to a protonated molecule ([M+H] or

[MH]+). Due to the high pressure of the source, collisional cooling aids in the formation

of this ion because it removes excess energy from the ion and thus reduces the amount of

fragmentation that occurs. It is important to note that fragmentation of the [M+H] ion is

only minimized, and thus source fragmentation may still occur. The fragmentation is

more related to the extent of energy transfer from a reagent ion to an analyte ion. If there

is large difference in the proton affinities of the analyte and the conjugate base of the

reagent ion, there will be excess energy in the [M+H] ion formed, which can cause

fragmentation of the analyte ion thus reducing the signal for [M+H]+. By using different

reagent gases, the fragmentation can be controlled. Although there is a reduction in

fragmentation, this technique still requires the analyte to be volatile and thermally stable

for introduction into the ion source, so derivatization of the phospholipid is still required.









Fast atom bombardment (FAB) was developed in the early 1980s, and since that

time, the analysis of lipids has changed dramatically.59'60 FAB was the first soft-

ionization technique to offer the widespread ability to analyze non-volatile compounds

without the need for time-consuming derivatization techniques. FAB is also called liquid

secondary ion mass spectrometry (LSIMS) because it was developed from secondary ion

mass spectrometry (SIMS). SIMS uses a primary ion beam to bombard a solid surface to

eject secondary ions from the surface of the solid that can be mass analyzed, typically by

a time-of-flight or magnetic sector instrument. The difference between FAB and SIMS is

the introduction of a low-volatility liquid matrix in which the analyte molecules are

dissolved or suspended. In FAB, the analyte and the matrix are mixed and then deposited

onto a solid surface that is inserted into the mass spectrometer. The matrix is typically

glycerol because it has a very low vapor pressure and thus does not rapidly evaporate

from the solid surface when it is inside the vacuum chamber. The primary ion beam

collides with the matrix/analyte mixture at a 450 angle and ejects ions and neutrals of the

analyte and the matrix from the liquid surface as shown in Figure 1-17. FAB is

considered a soft ionization technique because the matrix absorbs the energy from the ion

beam preventing internal energy build up in the analyte molecules, producing ions related

to the intact analyte. Another purpose of the liquid matrix is the continual renewal of the

ablated area with fresh matrix and analyte. When analyzing a solid sample in SIMS, the

area interrogated with the ion beam is damaged, and the molecular ion signal ceases. The

ion beam must be moved to a new area for continued production of molecular ions. The

refreshing of the surface in FAB also solves a drawback of SIMS in that the ion beam

does not penetrate into the sample very far, typically only a few nanometers, thus limiting









Primary ion
* or atom beam
0* .
*


Liquid matrix, G, with
dissolved analyte, A


Secondary ions
G refers to the matrix
A refers to the analyte
G+ A- G+
G_ A+ G

A+ A+
At- +


A A
A A A
---->^A A A


Solid support


Figure 1-17. A diagram of the fast atom bombardment, FAB, process. In FAB, the
analyte, A, is dissolved in a matrix, G, typically glycerol, and an ion or atom
beam is directed to the sample surface at an angle. Secondary ions of glycerol
and analyte are ejected from the surface upon bombardment by the ion beam.
These secondary ions generated are from the matrix and the analyte. This
technique limits fragmentation of the analyte, generating intact molecular
ions.


I _


G+









the amount of material ablated. By refreshing the surface, the accumulation of signal can

be accomplished which may aid in the analysis of lower abundant species.

FAB allowed for the direct analysis of phospholipid species without

derivatization, thus providing for better characterization of the lipid species in a

biological sample. Coupling FAB to a mass analyzer capable of performing tandem MS

presented the opportunity to identify the lipids based on their fragmentation patterns

under CID. In a sense, this is akin to spectra collected by EI, but the fragmentation is

better controlled. However, there are a lot of differences between a spectrum collected

under El and one collected by CID of a FAB generated ion. El causes a high degree of

fragmentation producing many ions related to the structure of the analyte whereas CID

imparts less energy to the analyte producing fewer fragment ions, but still specific to the

structure of the analyte. Because CID spectra are quite dependent on CID conditions, and

CID is not always performed under the same conditions, there are not libraries of spectra

in which a researcher could search to identify unknown species. When CID is used, it is

better to have a standard of the analyte ion to allow comparison of MS/MS spectra of the

standard and the analyte under investigation collected on the same instrument. A

limitation of FAB is the presence of very intense peaks across the entire low mass range

arising from clusters of the matrix, especially when glycerol is used. This can make

identifying low-mass lipids very difficult, especially when they are in low abundance.

A significant transformation in mass spectrometry occurred with the introduction of

electrospray ionization (ESI).18 This further advanced the analysis of lipids by mass

spectrometry because ions could be generated directly from aqueous/organic solutions.

With ESI, there is not a limitation in the size of a molecule, in fact practically every type









of compound from small organic drug molecules to large protein complexes is amenable

to analysis.

In ESI, a high voltage from 1000 V to 8000 V is applied to a small metal capillary

into which fused silica, usually 100 |tm inner diameter, is inserted. The applied voltage

creates an electric field at the tip of the capillary. An aqueous/organic solution is pumped

at a constant flow rate through the end of the capillary. As the solution emerges from the

capillary and if the electric field is sufficient, a Taylor cone will develop. Small droplets

containing both the volatile solvent and sample ions emerge from the cone. As the

solvent evaporates, the charge density increases until the charge density exceeds the

Rayleigh limit and then the ejection of ions occurs. Due to this process, multiply charged

ions are routinely observed, especially from proteins and peptides. The process is shown

in Figure 1-18. A prerequisite of ESI is that the compound must be an ion, or zwitterion,

in the solution for ionization to occur. This means that non-polar compounds such as

steroids cannot be ionized by ESI.

This technique permitted the direct coupling of liquid chromatography (LC) with

mass spectrometry due to the ability to ionize a wide variety of compounds as intact

molecular ions from a liquid phase.61 By coupling these complementary techniques, more

information could be generated from the same analyte solution, a retention time, a mass

spectrum, and a MS2 spectrum. Tandem MS of the intact molecular species has enabled

direct structural characterization and identification of phospholipids, and has truly

transformed the study of lipids.62-68 Just as ESI has created a new field in the analysis of

proteins called proteomics, the impact in lipid analysis has now generated a new field











Taylor Cc

Capillary


ne





+ +

Desolvation Ion ejection
region


Power supply


Figure 1-18. A diagram of the electrospray process. In electrospray ionization (ESI), a
liquid solution is pumped through a small capillary needle to which a high
voltage is applied. As the solution emerges from the needle, a Taylor cone
develops and then small droplets begin to evaporate. These droplets then
evaporate further, producing charged species in the gas phase.









of analysis called lipidomics. This field is quite young, but it offers the potential to

identify the role of lipids in cells, tissues and organs to better understand biological

disorders and diseases.69

Despite the widespread use of both ESI and MALDI in the analysis of proteins and

peptides, the adoption of MALDI for the analysis of lipids has remained limited. A

search of journal articles with the keywords "MALDI" and "lipids" produced only 90

publications, while a search with the keywords "electrospray" and "lipids" produced 295

publications. Part of the problem with MALDI is the high abundance of matrix ions in

the low mass region (under m/z 1000) of a mass spectrum. These ions can interfere in the

analysis of lipids because the majority of lipids as singly charged ions would be detected

at a m/z value less than 1000. Since the majority of instruments employing a MALDI

source are limited to one stage of mass analysis, distinguishing matrix ions from lipids

ions can be difficult.

For this reason, MALDI has primarily been a technique for the analysis of higher

molecular weight compounds such as proteins, polymers, and DNA. The complexity of

lipids in this low-mass region has also been a problem for the use of MALDI in their

analysis. Appropriate MALDI preparation can yield very good results for lipids,

especially the phospholipids and sphingolipids.7074 With the introduction of MALDI onto

instruments capable of tandem MS, the analysis of lipids offers structural characterization

as well as very good spatial resolution, enabling easier identification of solutions

containing many lipids.7577 The spatial resolution is important in the analysis of tissue

sections by MALDI especially for the generation of images related to the distribution of a

specific compound in the tissue. Perhaps the analysis of lipid by MALDI will develop









more readily with this capability because of the prominence of lipids in the cellular

medium.

Overview of dissertation

The following chapters describe the analysis of tissue specimens for small

molecules, compounds having a MW less than 1000 amu, by intermediate-pressure (IP)

MALDI. The purpose of this research was to develop a new technique that will enable

researcher to better characterize compounds present in tissue specimens. Chapter 2

describes the first-ever analysis of tissue specimens, spinal cord, using IP-MALDI,

operating at a pressure of 170 mTorr for ionization. This chapter shows the ability to

desorb intact PLs from tissue at this pressure and to correctly identify them, as PCs or

SPMs, using tandem MS. Chapter 3 examines the ion fragility of PLs by ESI-QIT-MS

and MALDI-QIT-MS. This chapter describes how different ions from the same molecule

can have different degrees ion fragility causing either source fragmentation or decreased

mass resolution in mass analysis by QIT. The formation of cationized species of PLs is

proven to show less source fragmentation and better mass resolution. With this

understanding, MALDI matrices can be prepared and deposited onto tissue specimens to

favor the formation of cationized species. Chapter 4 describes the analysis of brain tissue

sections by IP-MALDI for the generation of mass spectrometric images that show the

distribution of specific PLs. Chapter 5 offers a conclusion to the areas examined and a

look to the future of tissue analysis by IP-MALDI as well as a perspective on the use of

quadrupole ion traps in conjunction with imaging mass spectrometry.














CHAPTER 2
ANALYSIS OF INTACT TISSUE BY INTERMEDIATE-PRESSURE MALDI ON A
LINEAR ION TRAP MASS SPECTROMETER

Introduction

Prior to the advent of matrix-assisted laser desorption/ionization (MALDI), direct

mass spectrometric analysis of tissue sections focused primarily on the identification of

elements in thin tissue sections, with one of the first experiments involving the detection

of heme-bound iron from red blood cells using a laser microprobe mass analyzer, or

LAMMA.16 With the advent of soft ionization methods such as MALDI and electrospray

ionization (ESI), characterization of labile and involatile molecules such as proteins,

peptides, and lipids present in biological tissues has become possible. The use of soft

ionization techniques allows for reduced complexity in the analysis of biomolecules with

mass spectrometry because intact molecular ions of the particular molecule are produced

rather than fragment ions.

Mass spectrometric microprobes employing MALDI or secondary ion mass

spectrometry (SIMS) are now offering promising new approaches to map the distribution

of small and large molecules directly from tissue sections at biologically significant

levels19, 20, 22, 78 and to help unravel the molecular complexities of cells.79-83 Microprobe

MS offers the unique ability to directly analyze tissue samples for chemical species and

to identify changes in the distribution of specific compounds localized in tissue by

generating images of specific ions or by comparing spectra from different regions of the

section. Many of the analyses have focused on the identification of proteins and peptides









present or absent in normal and diseased tissue,23' 84-87 but other applications have focused

on the identification of site-specific drug activity24 as well as lipids.88-91 Besides tissue

samples, artwork has also been analyzed for specific compounds to help determine their

authenticity.92 93 Studies in our laboratory have demonstrated the ability to map

paclitaxel, a small drug compound, at trace (pg/mg) levels in ovarian tumors.24' 27 Critical

to such sensitive and selective analysis was the unique capability for tandem mass

spectrometry (MSn) that an instrument developed in our laboratory provided. Other

studies have shown the need for tandem mass spectrometry for the identification and

mapping of small drug molecules such as cocaine94 and anti-tumor drugs95 in tissue

sections.

The analysis of tissue sections by MALDI mass spectrometry has traditionally been

performed at low pressure (-10-6 Torr). This low pressure requires samples to be dried

completely, which requires approximately 2 hours, before exposure to vacuum

conditions, thus prohibiting the analysis of freshly cut tissue. Analyses of phospholipids

under traditional vacuum MALDI conditions have shown fragmentation of the

phospholipid head group (unpublished results), making low-level detection difficult;

recent analyses of gangliosides, brain glycolipids, using intermediate-pressure (IP)

MALDI have shown decreased fragmentation for these labile biomolecules.96

Intermediate pressures for IP-MALDI have ranged97'98 from 10-2 Torr to 1 Torr, ten

thousand to a million times higher pressure than traditional vacuum MALDI; in all cases,

collisional cooling reduces the degree of source fragmentation. Both IP-MALDI,

coupled to an FT-ICR, and atmospheric-pressure MALDI (AP-MALDI), coupled to a









quadrupole ion trap, have been shown to reduce the amount of source fragmentation for

more labile molecules such as gangliosides and phosphopeptides.99' 100

Experimental

The instrument used for all experiments was a Finnigan LTQ linear ion trap mass

spectrometer (San Jose, CA) fitted with a vMALDI source, as shown in Figure 2-1. The

source consists of a N2 laser (337 nm) directed to the source by a fiber optic cable; optics

inside the source allow for laser spot diameters from 80 to 120 |tm at an incident angle of

300. The laser spot size used for data collection was adjusted to 120 |tm. This position

can be controlled to +/- 1 |tm without removing the plate from the vacuum chamber. If

the plate is removed from the chamber and re-inserted, the precision is +/-6 |tm in the

vertical direction and +/- 7 |tm in the horizontal. This source is designed to operate

around 0.17 Torr. The standard vMALDI software limits sampling to the normal 2 mm

diameter sample wells, but tissue samples larger than this can be analyzed with custom

software.

All chemicals were obtained from Fisher Scientific (Fairlawn, NJ). The spinal cord

was removed from normal Spague-Dawley rats at the University of California at San

Diego and immediately stored in N2 (1). Frozen spinal cord was cut (1 mm thick) with a

scalpel and placed onto the vMALDI stainless steel sample plate. For all experiments,

tissue specimens were allowed to dry for the typical 2 hours before matrix application.

The matrix was electrosprayed from distance of 2 cm onto the surface at a flow rate of 7

[tL/min for approximately 10 minutes. The matrix used was 2,5 dihydroxybenzoic acid

(DHB) at a concentration of 77 mg/mL in 70% methanol/30% water (with 0.1% TFA).

Due to the use of a thick sample, the laser was unfocused when it impinged the surface of









the tissue, so the laser power was increased to compensate. Ten laser shots were

accumulated per spot to obtain each mass spectrum.

Results and Discussion

To the author's knowledge, direct analysis of tissue sections by MALDI at higher

pressures has not previously been reported. Analysis of drug compounds in tissue

sections using laser desorption/chemical ionization (LD/CI) at 1 Torr has been reported

by our laboratory.24' 101

Figure 2-2 shows spectra collected from the analysis of a section of spinal cord,

along with an optical image of the area analyzed (1.8mm x 2.3mm). The entire region

was coated with the MALDI matrix. The black line in the optical image on the left

indicates the outline of the spinal cord section. Spinal cord includes a variety of small

and large molecular weight compounds from lipids to proteins, but lipids constitute a

larger percentage by weight, and would be readily detectable by MALDI mass

spectrometry. The phospholipids ions detected as singly charged ions would range in

mass-to-charge (m/z) from around 600 to under 1000.

Spectrum A at the top of Figure 2-2 was taken from a location in which the spinal

cord was not present but DHB matrix was present. No ions corresponding to

phospholipids were detected from this region. On the other hand, a spectrum acquired

when the laser was fired at a location on the tissue (spectrum B, bottom of Figure 2-2)

indicates ions present from m/z 700-900 (Figure 2-3A) that were not detected in the

spectrum in Figure 2-2A. To ensure that these detected ions corresponded to lipids,

tandem MS was performed on eight of them. Table 2-1 lists these ions, along with the

major fragment ions observed from MS2. A neutral loss of 59 or the presence of the

fragment ion m/z 184 is indicative of a phosphatidylcholine (PC) or a sphingomyelin














SLaser
Sbeam
/ qOO



AUX rods
Sample plate I


K


-i


MALDI
modification


a-

q0


Skimmer


1


Lens 0 Lens] 3 section
linear ion trap
octopole

II I


- I- -


Figure 2-1. A schematic diagram of the Finnigan LTQ linear ion trap instrument with
vMALDI source modification. The vMALDI source replaces the standard
API source, and includes the sample plate, RF-only quadrupole qOO with
auxiliary rods, and skimmer. The RF-only quadrupole qO, octopole, and 3-
section linear ion trap are standard LTQ components. The open arrows show
the three stages of differential pumping; the pressure in the vMALDI region is
typically 0.17 Torr.









sphingomyelin (SPM). The MS2 spectrum of m/z 760.82 is displayed in Figure 2-3B.

Only one major fragment ion was detected, m/z 184.00 (C5Hi5NO4P), which is the

phosphocholine head group from either a PC or an SPM. Using the nitrogen rule, this

even m/z ion cannot be the [M+H] from SPM due to the presence of an additional

nitrogen in the ceramide backbone of SPM. No fragment ions were observed in the

MS/MS spectrum that would correspond to the fatty acid tails (R1 and R2) of PC. MS3 of

m/z 184.00 would not provide any information pertaining to the fatty acid content

because the fatty acid tails were lost in MS2. It is possible to surmise the fatty acid tails

based on the relative abundance of fatty acid chains expected to be present in spinal cord,

but fragmentation information would provide more certainty.

MS2 of m/z 782.82 (Figure 2-4A) reveals a different fragmentation pattern, a

neutral loss of 59 (trimethylamine) to produce m/z 723.36, and then a neutral loss of 124

to produce m/z 599.36 after MS3, as seen is Figure 2-4B. This fragmentation pattern is

characteristic of cationized PC and has been previously observed on triple quadruplee6'
102 and quadrupole ion trap103 MS/MS instruments. No ions corresponding to the fatty

acid tails were detected from MS3. Ions related to the fatty acid tails have been detected

from the [M+Na] of PCs using ESI and quadrupole ion trap mass spectrometry, but they

were low in abundance.103 MS4 of m/z 599.36 was performed to determine if fatty acid

information could be obtained, but no fragment ions were observed. In future

experiments, to increase less abundant fragment ions from MS3, the trap will be filled for

a longer period of time or multiple scans will be collected and averaged. Also, increasing

the abundance of the cationized lipids by adding sodium to the matrix solution may aid in

the identifying lower abundance fragment ions in tandem MS.















Optical image of


300 400 500 600 700 800 900 1000
m/z


Figure 2-2. On the left is an optical image of the spinal cord coated with DHB; the
locations where the laser was fired to produce spectra A and B are indicated
with circles. The spinal cord is outlined in black. Spectrum A shows the ions
detected from a location away from the tissue, but in an area where the DHB
matrix was still present. Most of these ions correspond to DHB clusters such
as m/z 409.91, 551.73, and 727.36, as indicated by the dagger symbol. These
ions appear at a higher mass than expected because the laser power was
increased due to the thickness of the tissue sample and thus when the laser
fired at a position off the tissue, space-charging was evident. No
phospholipids were identified in that location. Spectrum B shows the ion
signal from the tissue surface, indicating the presence of phospholipid ions.
The ion at m/z 760.82 was determined by tandem MS to be the [M+H] ion of
phosphatidylcholine 34:1 (PC 34:1). The other starred ions from m/z 700-900
were determined to be either PC or sphingomyelin, SPM.


315.91













760.82


Full scan


78 73
782.8
810.82
79.82 843.82
Lf 5 wK.8


720 740 760 780 800
m/z


835.82 851 18


820 840 860 880 900


MS2 of m/z 760.82


Phosphocholine head group


Protonated PC 34:1
[M+H]+


200 300 400


NL 154
--NL 18
743.09
568.18 607.27 732.0 .
500 600 700 800
m/z


Figure 2-3. Spectrum A is an enlargement of the phospholipid region of the spinal cord
tissue section mass spectrum from figure 2B. Spectrum B shows MS2 of m/z
760.82. Only one major fragment ion is produced, m/z 184.00, which
corresponds to the phosphocholine head group of phosphatidylcholine. The
neutral losses of 18 (water) and 154 could correspond to a DHB cluster ion at
the same m/z as the phospholipid ion. The inset displays the basic structure of
a phosphatidylcholine, showing the fragmentation pathway that produces m/z
184.


73.0074.91


S20
-: 0 ,-I
ro
S700

B
S 184.00
100


900 1000









These results indicated that m/z 782.82 is the [M+Na] analog of the [M+H] ion at

m/z 760.82. Even though sodium was not added to the matrix, there was enough sodium

present in the tissue section for adduction to occur. This PC ion can be classified as

either PC (16:0, 18:1) or PC (18:1,16:0), or in general PC (34:1) because fatty acid tail

information was not obtained under fragmentation.



Table 2-1. Ions detected from spinal cord in Figure 2A that were chosen for MSn
analysis. Only the MS2 major fragment ion m/z is shown in the table, but MS3
was also performed.
m/z MS2 major fragment Identification
735.00 717.00 DHB related
760.82 184.00 Phospholipid
782.82 723.36(NL 59) Phospholipid
788.73 184.00 Phospholipid
798.82 739.55(NL 59) Phospholipid
810.82 751.55(NL 59) Phospholipid
835.82 776.55(NL 59) Phospholipid
848.85 831.09 DHB related


Other ions studied by MSn were m/z 735.00, 788.82, 798.82, 810.82, 835.82, and

848.85, as listed in Table 2-1. MS2 of m/z 788.82 was performed with the major

fragment ion of m/z 184 produced indicating this ion is a protonated PC with fatty acids

adding to 36:1. MS2 of m/z 798.82 showed a neutral loss of 59, producing m/z 739.55,

indicating it is cationized. MS3 of m/z 739.55 produced m/z 615.00; a neutral loss of

124, confirming that m/z 798.82 is cationized. Using a search engine of phospholipids

developed in our lab, this ion was identified as the [M+K] analog of the [M+H] ion at

m/z 760.82 and the [M+Na]+ ion at m/z 782.82 (PC 34:1). MS2 and MS3 for m/z 810.82

and 835.82 also identified these ions as cationized species. A search through the database









for m/z 810.82 indicates it is the [M+Na] analog of the [M+H] ion at m/z 788.82, PC

36:1, while a search for m/z 835.82 indicates the ion is SPM 24:1, [M+Na] Finally, the

tandem MS spectra for m/z 735.00 and 848.85 both show a neutral loss of 154, which is

most likely indicative of a DHB background ion.

Conclusions

The experiments and data reported here demonstrate that it is possible to analyze

tissue specimens at a pressure five orders of magnitude higher than traditional vacuum

MALDI experiments. These results also clearly show the need for MSn for identification

of small molecules desorb directly from tissue sections. The major ions detected from

the spinal cord section were determined by MSn to be phospholipids, primarily PC and

SPM. Due to differences in fragmentation patterns, cationized and protonated PCs could

be distinguished. Ions desorbed from the spinal cord not corresponding to lipids were

also investigated using tandem MS. Although the effects of drying tissue specimens

before MALDI analysis has not yet been evaluated, preliminary results suggest that IP-

MALDI may allow for a reduction in amount of drying time necessary to directly analyze

tissue specimens and therefore allow for evaluation of the effects of drying such as

compound degradation or cell rupture. If the drying time can be reduced, sample

throughput can be increased and the analysis of tissue samples within minutes of

dissection may be possible.
















MS2 782.82-...


723.36


265.27 31.27

250 300


37 .09 437.91
350 400 450


146 if Na+ leaves


60

40

20
o
0
C
n


100-

80

60

40-

20

0-


577.64


200 250 300 350 400 450 500 550
m/z


572.18 602.45
51 .36 55.09 I 600.45647.00 702.91


500
m/z


4 NNL 59


NL 18

765.18

73 900
-i


550 600 650 700 750


800


- 7 NL 124


NL 146




661.18 70?.00


600 650 700 750 800


Figure 2-4. Spectrum A shows MS2 of m/z 782.82 from figure 2A, while spectrum B
displays MS3 (782.82-723.36-...). The neutral loss of 59 after MS2
corresponds to loss of trimethlyamine in the structure for PC (inset of
spectrum B). Two major fragment ions from MS3 arise from the neutral loss
of 124 (ethylphosphate) with the sodium retained on the glycerol backbone, or
the neutral loss of 146, when the head group retains the sodium. This
fragmentation pattern is associated with cationized PC and SPM. Fragment
ions relating to specific fatty acid tails Ri and R2 were not observed.


466.91


MS3 782.82-723.36-...
599.36














CHAPTER 3
CHARACTERIZATION OF PROTONATED PHOSPHOLIPIDS AS FRAGILE IONS
IN ION TRAP TANDEM MASS SPECTROMETRY

Introduction

This chapter is focused on the analysis of standard phospholipids and

sphingomyelins by ESI and MALDI. Before the application of imaging mass

spectrometry, it was important to better understand the differences in ionization of

phospholipids as protonated and sodiated species, particularly the susceptibility to source

fragmentation. If fragmentation of the parent ion is occurring in the source, less of that

parent ion will be available for mass analysis, which in turn increases the detection limit.

Glycerophospholipids (GPLs) serve a variety of functions from cellular signaling to

protein transport and are abundant in all living organisms.104 GPLs are linear in structure

and typically consist of a glycerol backbone with a polar head group and two non-polar

fatty acid tails. PLs with this basic structure are phosphatidylcholines (PCs),

phosphatidlyethanolamines (PEs), phosphatidylinositols (PIs), and phosphatidylserines

(PSs) with the classes determined by the composition of the head group.

Because of their biological abundance and importance, phospholipids have been

studied by mass spectrometry for many years, from electron ionization coupled to gas

chromatography/mass spectrometry (GC/MS)61 to more recently matrix-assisted laser

desorption/ionization (MALDI)70-72 105 and electrospray ionization (ESI)62, 63,67, 106

Under appropriate experimental conditions, the latter ionization techniques produce a

predominant ion ([M+H]+ or [M+Na] ) corresponding to the molecular weight of the









intact molecule whereas GC/MS causes a high degree of fragmentation and requires

derivatization to increase volatility, but identification is performed using the immense

chemical database available for El spectra. In ESI, relying on the molecular weight for

compound identification is inadequate because of the wide variety of phospholipids

present and possible fatty acid combinations; thus compound identification is typically

performed by tandem MS with a triple quadrupole or ion trap mass spectrometer.62'67, 102,
107

The initial studies focused on the differences in fragmentation between protonated

and sodiated phospholipids. For positive ions, experiments showed that protonated and

sodiated PCs produced very different fragmentation patterns, with sodiated PCs

providing a more informative fragmentation pattern.62 Under collision-induced

dissociation (CID), protonated PCs produced one fragment ion, m/z 184, corresponding

to the polar head group, indicating that the charge is retained on the head group. In

contrast, CID of sodiated PCs produced fragments that corresponded to losses of the head

group with retention of the charge on the glycerol backbone. The most abundant

fragment in MS/MS showed a neutral loss of 59, which corresponds to the choline group,

-N(CH3)3.62

The difference in fragmentation of protonated and cationized phospholipids has

been of considerable interest, particularly in the study of different metals for

cationization. Due to the limitation of a single stage of tandem MS (i.e., MS2) on a triple

quadrupole, the need for more informative fragment ions in MS2 is desired. The use of

lithiated adducts was shown to provide many structurally informative fragment ions for

PCsl02and PEs65 after a single stage of tandem MS, and many other cations have been









evaluated for other GPLs107 A structurally significant fragment ion in the analysis of

lipids allows for the correct identification of the fatty acid tails and their location on the

glycerol backbone. A novel adduction with trifluoroacetic acid (TFA)/K+ for PCs was

determined to provide an abundance of structurally informative fragment ions as well.103

The resonating theme of most positive ion studies was that the cationized GPLs, or

adducts with other complexes, should be preferred over protonated GPLs if structural

identification is desired.

This difference in fragmentation behavior under CID is also of concern in ion trap

mass spectrometry due to ion fragility. In previous studies from our laboratory,108 the

fragility of an ion within an ion trap was quantified and results showed that different ions

formed from the same molecule can exhibit varying degrees of fragility. For example,

cationized oleandomycin is a stable ion, while its protonated counterpart is fragile. A

consequence of fragility is an inability to efficiently isolate a fragile ion without widening

the isolation notch, as well as mass shifts and reduced mass resolution.108 Although a

relationship between differences in fragmentation between different ions of the same

compound in CID experiments and differences in ion fragility of those ions has not been

evaluated, it appears to be an interesting area for future studies.

It has already been shown that cationized and protonated species of the same

molecule exhibit varying degrees of fragility, but the effect on the desolvation process

has not be evaluated. The temperature of the heated capillary, and thus the effective

temperature of an ion, has been shown to affect the onset of source fragmentation.109 In

MALDI mass spectrometry, source fragmentation of phospholipids was determined to









arise from gas-phase reactions rather than from laser-induced photodissociation.70 This is

an important factor, since gas-phase reactions occur in electrospray as well.

Fragmentation in ion transport has been studied before. Early studies in

electrospray showed the possibility of performing thermally induced dissociation (TID)

of highly charged protein ions.110 A heated capillary was used for those studies and the

results indicated that the higher charged ions (+6, +5, and +4) were more susceptible to

TID due to increased coulombic repulsions. Source collision-induced dissociation

(SCID) has been used for many years for controllable dissociation of complexes and to

provide fragment ions that can be further fragmented in a tandem mass spectrometer.

The concept of ion fragility in ion trap mass analysis is re-examined for a different

class of compounds and the effects on fragmentation in ion transport are explored by

comparing three different phospholipids as protonated and sodiated species, namely PC

(16:0, 16:0), SPM (16:0), and PE (16:0, 16:0).

Experimental

For ESI, all experiments were performed on the Finnigan LCQ instrument (San

Jose, CA), an electrospray ionization quadrupole ion trap (ESI-QIT) mass spectrometer.

Analyte solutions were directly infused at a flow rate of 1 L/min using a syringe pump,

4.5 kV applied to the electrospray needle, and sheath gas set to 30 arbitrary units. For

evaluating differences in ion fragility of protonated and cationized molecules, three

phospholipids were chosen: phosphatidylcholine (PC), sphingomyelin (SPM), and

phosphatidylethanolamine (PE). PC and SPM were chosen because the protonated and

sodiated molecular ions exhibit different fragmentation pathways under collision-induced

dissociation (CID) and they share the same head group. The main difference is in their









fatty acid tails because SPM has only one fatty acid chain that can vary, whereas PC has

two fatty acid chains that will vary. PE was chosen because it has a similar head group to

PC and SPM and also exhibits a different fragmentation pathway for protonated and

cationized species. The only difference in the head group is the replacement of choline (-

N(CH3)3) with amine (-NH3), as seen in Figure 3-1.

Standards of PE and PC with fixed fatty acid substituents of palmitoyl (16:0) were

obtained from Avanti Polar Lipids (Birmingham, AL), while SPM was purchased from

Avanti as a chicken egg extract, but with palmitoyl (16:0) as the predominant acyl chain

(80%). All phospholipids were obtained as powders and prepared to the desired

concentrations. PC and SPM were made as stock solutions of 1000 ppm in 50:50

isopropanol:methanol and PE was made as a stock solution of 500 ppm in 75:25

choloroform:methanol. For ESI QIT-MS analysis, the stock solutions were diluted to 10

ppm in methanol.

Due to the presence of unwanted source fragmentation and to uncover the origin of

those fragments, solutions were prepared to ensure that either the [M+H] ion or the

[M+Na] ion was solely present during analysis. For control of protonation, formic acid

was added for a final concentration of 0.1%; for the production of sodiated species,

sodium acetate was added to a concentration of 100 pM. All solvents were HPLC grade

and were obtained from Fisher Scientific (Pittsburgh, PA). Unfortunately, because these

lipids tend to prefer sodium, the protonated solutions contained the sodiated species as

well, probably because of sodium ions present in the methanol.














16:0


H3C CH
/CH3

PC 16:0, 16:0
0

HOF \


NH3

SPE 16:0, 16:0
0

HO~ \


Figure 3-1. Chemical structures of the phospholipids used in the studying ion fragility
and source fragmentation. The protonated version of the molecule is shown.
SPM stands for sphingomyelin, PC stands for phosphatidylcholine, and PE
stands for phosphatidylethanolamine. All three molecules are very similar in
structure. SPM and PC share the same head group, but differ in the fatty acid
tails, while PC and PE share the same fatty acid tails, but have slightly
different head groups.









Ion fragility in the ion trap was determined using three experiments described

previously."1 These experiments were 1) measuring the peak width of the parent ion at

10% peak height (PWioo) using a slow scan speed (called zoom scan), 2) finding the

isolation width needed for efficiently isolating the desired ion, and 3) determining the

amount of energy required for CID. Typically, a fragile ion will need a wider isolation

window, will exhibit a wider peak width at a slow scan rate, and will require less energy

to fragment under CID. These experiments were set up using Xcalibur software. Data

for the zoom scan were averaged over 2 minutes. The isolation width was determined by

changing the width from 1 to 4 amu wide in 1 amu increments, and collecting data for 1

minute at each interval. In studying the amount of energy required for CID, all

experiments were conducted at a heated capillary temperature of 2500C and a tube lens

offset of 30 V while changing the percent CID and collecting data for 1 minute at each

level. Values for percent CID were 15%, 16%, 17%, 18%, 19%, 20%, 21%, 25%, and

30%.

Ion fragility in ion transport, or the susceptibility of ions to source fragmentation,

was also studied using two injection parameters on the LCQ, the temperature of the

heated capillary and the voltage applied to the tube lens (tube lens offset). These two

parameters assist in desolvating and declustering ions during injection. Typical operation

involves a high capillary temperature, 250C, and a small voltage for the tube lens. The

sensitivity of analysis is typically increased with a more positive tube lens offset for

positive ions and a more negative value for negative ions; even higher voltages result in

ion fragmentation. To determine the effect of temperature on fragmentation of

phospholipids, an Xcalibur instrument control file was created that recalled different tune









files allowing the capillary temperature to be changed at 10-degree intervals from 500C to

2500C. Prior to setting up the temperature profile scan, it was determined that a

temperature change took 8 minutes before equilibrating; therefore, the instrument file

collected data for 10 minutes at each temperature level, but only the last two minutes of

each temperature change were used for comparisons between protonated and sodiated

species. The tube lens was maintained at 0 V for these experiments. The final parameter

used to understand the susceptibility of phospholipid ions to source fragmentation was

the tube lens voltage. For this experiment, the heated capillary was held at 2500C and the

tube lens voltage was changed; spectra were collected for two minutes at 4 voltage levels

(0, 10, 20, and 30 V) and saved as different files.

Results and Discussion

The three phospholipids studied can form many molecular ions under ESI

conditions. For example, the PC (16:0, 16:0), a synthetic phospholipid, if ionized without

control of solvent counter ions would be detected as [M+H] [M+Na] and [M+K]

because these are the primary cations present in most solvents. Determining which

molecular ion is the most stable for ion trap mass analysis is of importance to ensure

good mass accuracy and lower detection limits (if the more fragile molecular ion is

fragmented more easily in ion transport there will ultimately be less parent ion present for

mass analysis). Up until now, a relationship between fragility in the ion trap and source

fragmentation has not yet been explored. While operating the LCQ QIT-MS at lower

capillary temperatures, a reduction in fragmentation of the head group from [M+H] of

SPM 16:0 was noticed. This reduction in fragmentation was evaluated further in the

present study.









Fragility in the trap

The LCQ zoom scan feature allows for a higher resolution spectrum over a narrower

mass range, 10 amu, by using a 20x slower scan rate. A fragile ion will have a broader

PW1io than a more stable ion. Examples of the zoom scan experimental data are shown

in Figure 3-2 for the analysis of SPM 16:0 as [M+H] (A) and [M+Na] (B). Protonated

SPM has a wider PW1io than the sodiated SPM ion for the monoisotopic peak. A zoom

scan was collected for each phospholipid studied and the results were summarized in the

bar graph of Figure 3-3. All protonated phospholipids exhibited wider peaks widths than

their sodiated counterpart, indicating for this experiment that the sodiated species is more

stable. PE showed the largest difference in peak width between the [M+H] and

[M+Na] and was the only ion in which the peak was so broad that isotopic resolution

was not obtained in a zoom scan. Previous studies determined that a PW10o less than or

equal to 0.30 amu was indicative of a stable ion.108 Using this value, both the sodiated

and protonated species of SPM and PC could be considered fragile ions, while only the

sodiated species of PE would be considered stable. It should thus be stated that the

sodiated species of SPM and PC exhibit less fragility than the protonated species.

For secondary confirmation that the sodiated species is more stable than the

protonated species, data were collected for determining the isolation width needed to

efficiently isolate the parent ion before CID. In this experiment, an efficient isolation

width was defined as the width needed to reach 50% of the intensity at an isolation width

of 4.0. As can be seen from the graph in Figure 3-4, all the sodiated species permitted

narrower isolation widths than the protonated species. However, the difference for PE is

not as great as compared to PC and SPM, possibly showing both protonated and sodiated










703.5


SPM [M+H]+


704.3


5.44


705.2


100
80
S60-
0 40-
1 20-
C 0-
3
-Q

> 100
80
() 60-
40-
20-
0


725.4
SPM [M+Na]
726.3

.3

723.5 724.0 724.5 725.0 725.5 726.0 726.5
m/z


727.3

727.0 727.5 728.0 728.5


Figure 3-2. Spectrum A, above, is a zoom scan of protonated sphingomyelin (SPM) 16:0.
The peak width at 10% peak height (PWio%) was determined to be 0.44. The
peak width for the sodiated counterpart, measured from the zoom scan in
spectrum B, was 0.37. A narrower isolation width indicates that the sodiated
species is less fragile than the protonated species in ion trap mass analysis.


701.5 702.0 702.5 703.0 703.5 704.0 704.5 705.0 705.5 706.0 706.5












0.8 0.70 [M+H]
U [M+Na]+

0.6
0.47
0.44

0.4 0.36 0.37
0.29
(L

0.2



0-
SPM PE PC
Phospholipid



Figure 3-3. The chart above shows the PWioo for the two ions (protonated and sodiated)
of each phospholipid studied. The peak widths were measured from zoom
scan data. The zoom scan data was collected for 2 minutes and the peak width
was measured from the average spectrum of each ion. For all three
phospholipid classes studied, the protonated species was determined to be
more fragile because the peak width was wider.






74





3.5 3.2 E[M+H]+
3 2.8 2 [M+Na]+
2.5
2.4
2.5

2 1.8 1.6
0
1.5

U) 1

0.5

0
SPM PE PC
Phospholipid



Figure 3-4. A chart showing the isolation width needed to efficiently isolate the parent
ion of each ion studied (protonated and sodiated) for each phospholipid. A
wider isolation width is typically needed for a more fragile ion. The results
from this experiment also indicate that the protonated species is more fragile
than the sodiated ion.









PE are fragile ions. This result is contrary to the findings from the zoom scan

experiment. Both PC and SPM share the same head group, phosphocholine, whereas in

PE this head group is amine, -NH3, rather than trimethylamine, -N(CH3). This small

structural difference may cause a dramatic change in inter- and intra-molecular bonding

for PE ions in the gas phase. An interesting result was that even the more stable sodiated

species required an isolation width of greater than 1 amu wide implying that this ion is

still unstable as compared to previous results indicating that the stable ion, such as

caffeine [M+H]+, could be isolated with a 1 amu wide window.108

Finally, the %CID, given as a percent of the maximum 5 Vp-p resonance

excitation waveform, required to reduce the parent ion signal by 50%, was determined for

each ion; the results are shown in Table 3-1. These results indicate that the protonated

lipids are more fragile than the sodiated lipids because less energy is required for 50%

fragmentation. For the benefit of the reader, CID spectra of the protonated and sodiated

GPLs studied are shown in Figures 3-5 and 3-6, respectively. The major fragment ions

for MS2 of the [M+Na] ions result from partial losses of the head groups, -N(CH3)3 (NL

of 59) for PC and SPM and -C2H5N (NL of 43) for PE. On the other hand, CID of

[M+H] for PC and SPM produced one dominant fragment ion, m/z 184, the polar head

group phosphocholine. This indicates that the head group retains the charge when the

molecule is protonated. While CID of the [M+H]+ for PE produced an ion corresponding

to a neutral loss of 141, loss of the phosphoethanolamine head group, and the charge

retained by the glycerol backbone. PE is only differentiated from PC by the replacement

of three methyl groups with three hydrogens (Figure 3-1). Apparently, this change is

enough to alter the fragmentation pathway under CID.


























"I UULT
2 80
-0=
1 60
< 40-
20-
g 0


MS2 SPM (16:0)


200 300 400 500 600
m/z


MS2 PC (16:0, 16:0)


200 300 400 500 600
m/z


[M+H]


703.53
685.87
700 800 900 1000


[M+H]

I


700 800 900 1000


MS2 PE (16:0, 16:0)


551.53


200 300 400 500 600
m/z


[M+H]+


692.40


700


800 900 1000


Figure 3-5. Spectrum A is MS2 of the [M+H] for SPM (16:0) and spectrum B is MS2 of
the [M+H] for PC (16:0, 16:0). Both of these spectra show the same
fragmentation pathway producing a predominant ion at m/z 184,
corresponding to the polar head group. Spectrum C is MS2 of PE (16:0, 16:0)
and shows a different fragmentation pathway. The major ion produced results
from the neutral loss of the polar head group.


S10G 184.07
o80
60
<
. 40
S20
nl)


g 100-
I 80
S60
., 40
S20
Q 0











'100-
o
80-
C:
5 60
S40
S20




010
o
S 80
S60
8o

< 40-
20
cr 0



S100
I 80
I 60
, 40


MS2 SPM (16:0)






200 300 400 500 600
m/z


MS2 PC (16:0, 16:0)


200 300 400


500 600
m/z


MS2 PE (16:0, 16:0)


666.27
[M+Na]*

725.20



700 800 900


1000


697.27 B

[M+Na]+

756.40


700 800 900 1000


[M+Na]+


S20164.00 458.13 573.47
0 0- I P F IT P I
200 300 400 500 600 700 800 900
m/z
Figure 3-6. MS2 spectra of the [M+Na] ions for each phospholipid studied. A is from
SPM (16:0), B is from PC (16:0, 16:0), and C is from PE (16:0, 16:0). The
fragmentation pathways are very similar for all these ions. They all result
from neutral loss of the polar head group. A neutral loss of 59 for PC and
SPM corresponds to the loss of choline (-N(CH3)3) and a neutral loss of 43
corresponds to the loss of ethanimine (-C2HsN).


Table 3-1. The %CID needed to cause a reduction in the absolute parent ion signal by
50% is shown for each ion studied in the QIT. These results show that the
protonated species fragments more readily than the sodiated species, further
proof that the protonated ion is fragile.

Phospholipid [M+H]* [M+Na]*
SPM 17.5 22.0
PE 17.0 22.0
PC 19.0 22.0









Fragility in ion transport

Fragmentation in ion transport could occur in the source region at atmospheric

pressure, or in the ion optics regions where the pressure is lower. If an ion is thermally

unstable, the heat the ion absorbs could also cause fragmentation, called thermally

induced dissociation (TID). TID has been studied previously for protein ions, but the

temperature required for dissociation was from 250C to 4000C.110 The temperatures used

in this experiment are much lower than that, but still show the effect of TID.

The effect of changing the heated capillary temperature from 50C to 2500C is

shown in Figures 3-7 and 3-8 for SPM 16:0. Figure 3-7 shows three different spectra

collected at three different temperatures: 250C, 190C, and 130C for SPM (16:0).

Because there was some sodium present in the solvent, both the protonated and sodiatied

species were detected. With both ions present in the spectra, it is easy to follow the effect

of TID on each ion as along, as the ions related to TID can be identified. Since TID

experiments have been shown to cause similar fragmentation pathways to MS2,

comparison of the MS2 spectra in Figures 3-5 and 3-6 with these spectra permits the

identification of the fragment ions associated with TID.110 The fragment ion from the

protonated species, m/z 184, is detected at all three temperatures, but shows a decrease in

intensity at 130C, whereas the fragment ion from the sodiated species, m/z 666, is only

detected at the highest temperature used, 250C. Running at 1300C reduces

fragmentation for [M+H] but also increases the S/N and the signal for solvent cluster

ions, as would be expected because desolvation is not as effective at lower capillary

temperatures. But at 2500C, the intact molecular ion for the protonated species is almost












184.1 --Fragment of [M+H]
100L
80
60
40
20 5.0 Frag

150 250 350


100-
80 184.1
60
0 148.9
0
150 250


703.5, [M+H] -
725.5, [I

ment of [M+Na]+ 0 66.6 .

450 550 650 750


337.33651413.1 __
350 450 550 650


M+Na]


2500C


850 950

B

1900C


750 850


703.3 C

80725.3 1300C
60-
40- 148.9 184.1 337.1 65.1.
20 .4.
150 250 350 450 550 650 750 850 950
rr/z



Figure 3-7. The effect of the heated capillary temperature on TID for the different ions of
SPM (16:0) is shown in the three spectra above. Spectrum A was acquired at
250C, a normal capillary temperature for most analyses, and shows nearly
100% fragmentation for the [M+H]+ ion of SPM (16:0). In contrast, there is
less than 20% fragmentation occurring for the [M+Na]+ ion. Spectrum B
shows the spectrum when the temperature is lowered to 1900C and spectrum
C shows the spectrum at 1300C. It is clear that lowering the temperature of
the heated capillary reduces fragmentation, TID, of both ions. At 130C,
extra peaks from m/z 300-550 are present. These peaks are most likely
clusters of solvent ions, because desolvation is not as effective at such a low
temperature.









completely fragmented. Operation of this instrument at low temperatures is not

recommended because of this problem.

Creating a way to dry or desolvate the ions before interaction with the heated

capillary will allow for proper operation at reduced capillary temperatures. Experiments

were performed with such drying and analysis was successfully be performed at low

capillary temperatures, but the purpose of this study was to determine which ionized

species would be better for analysis under normal ESI operating conditions.

Figure 3-8 shows the complete temperature study, graphed, for both protonated and

sodiated SPM (16:0). Data for this experiment were collected with a tube lens offset of

30V to avoid confusion from SCID. This complete investigation of the heated capillary

temperature showed that protonated SPM was greatly affected by the temperature of the

heated capillary as is evident from the increased fragmentation at a higher temperature.

Fragmentation of SPM (16:0) occurs as a result of ion transfer into the vacuum chamber,

but can be minimized by operating at lower capillary temperatures. When running at

2500C, a normal operating temperature for most ESI analyses, the fragmentation for

[M+H] was 90%. From this experiment, the fragmentation for the [M+H] ion would be

minimized, but no eliminated, by running with a capillary temperature under 2000C.

On the other hand, the fragmentation of the [M+Na] ion is minimal and reduced to

almost zero at a capillary temperature of 1600C. The signal for the fragment ion at the

highest capillary temperature, 2500C, was less than 20% of the parent ion. The apparent

increase in fragmentation for this ion at lower temperatures is related more to the poor

desolvation aspects of running at such low temperatures. With poor desolvation,









1.0
Parent ions

.0.8 /


c 0.6
o .6 [M+H]+

0N *[M+Na]
= 0.4 -


Z 0.2 -

Daughter ions A
0.0
0 50 100 150 200 250 300
Capillary Temperature (C)


Figure 3-8. A graph showing the intensity of the parent and daughter ions of the [M+H]
and the [M+Na] ions versus changing the heated capillary temperature for
SPM (16:0). These data were collected automatically using Xcalibur software
control. The temperature was adjusted at 200 intervals from 500C to 2500C.
From the graph, it is evident that the protonated species is more susceptible to
TID than the sodiated species.









increased signal-to-noise is more evident and this is more likely the cause of apparent

fragmentation.

This experiment clearly shows a very different fragmentation behavior for the

sodiated species and the protonated species of the same phospholipid. If a correlation

exists between fragility inside the trap and fragility outside the ion trap, the protonated

species would be considered fragile, or in other terms more labile. This fragility makes

the protonated species more susceptible to TID and thus care must be taken when

analyzing lipid solutions for protonated species. However, it would be more beneficial to

induce cationization instead of protonation because of this extensive fragmentation.

The same temperature experiment was performed on PC (16:0, 16:0) and PE (16:0,

16:0). The effect of temperature on fragmentation for PC (16:0, 16:0) is shown in Figure

3-9. The data for this experiment also show that the protonated species is more

susceptible to TID than the sodiated species, but the temperature crossover, where the

fragment ion equals the intensity of the parent ion, was higher for the [M+H] ion of PC

(2250C ) than for the [M+H] ion of SPM (1900C).

In contrast to the high degree of TID observed for both SPM (16:0) and PC (16:0,

16:0), the TID for PE (16:0, 16:0) was not as intense. The graph for PE is shown in

Figure 3-10. Although the protonated species did show increased TID with respect to the

sodiated species, the parent in signal was still dominant in the spectra.

Tube lens offset

To evaluate the effect of the tube lens offset, the capillary temperature was

maintained at 2500C while adjusting the tube lens voltage. Data for [M+H]+ and