|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 QUANTITATIVE TANDEM MASS SPECTROMETRIC I MAGING WITH MATRIX ASSISTED LASER DESOR PTION/IONIZATION By DAVID ANDREW PIRMAN, JR. A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011
2 2011 David Andrew Pirman, Jr.
3 T o my lat e Mother, may she rest in peace
4 ACKNOWLEDGMENTS I would first like to acknowledge my advisor Dr. Richard A. Yost who has been instrumental in my development as a scientist. Rick has established a research group where independent scientific creativity and thought is encouraged His teaching and guidance has not only bettered my scientifi c understanding of chemistry, but has nurtured a foundation for a lifetime of scientific exploration. His excitement and genuine interest in science has inspired me to attain the same attributes, hopefully leading to a successful career in the field, and for that, I thank you. A special thanks to the past and present Yosties, for creating and establishing an excellent group dynamic allowing for a research environment where help and advice from fellow students is commonplace I would also like to thank my f amily including my parents, specifically my Dad, brother sister, and my late Mom for the ir support throughout my life. Our parents always instilled hard work and perseverance and through their support have encourage us to be successful and happy in wha tever path we may choose in life. I must thank my Grandparents, Fred and Claudia Pirman, for their love and support throughout my life as well. Without them, the last few years would have been nearly impossible A final thank you to my wife, Natasha. Y ou have been an amazing part of my life the past few years. Without your support through some tough times, I may not be at this point. I look forward to the next chapter in our lives together.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ ................... 11 1 INTRODUCTION ................................ ................................ ................................ .... 13 Mass Spectrometric Tissue Imaging ................................ ................................ ....... 13 Historical Perspective ................................ ................................ ....................... 13 Current Technology ................................ ................................ .......................... 14 MALDI MSI Background ................................ ................................ ......................... 16 MALDI Fundamentals ................................ ................................ ............................. 18 Practical Aspe cts ................................ ................................ .............................. 19 Desorption Process ................................ ................................ .......................... 19 Ion Formation Theory ................................ ................................ ....................... 20 Quantitative MALDI ................................ ................................ .................... 21 Mass Spectrome try ................................ ................................ ................................ 22 MALDI LTQ ................................ ................................ ................................ ...... 22 Linear ion trap ................................ ................................ ............................ 23 MS/MS with a LIT ................................ ................................ ....................... 24 Wide isolation MS/MS ................................ ................................ ................ 25 MALDI QTOF ................................ ................................ ................................ ... 26 Mass selection with quadrupole ................................ ................................ 26 Mass analysis with TOF ................................ ................................ ............. 27 MS/MS and wide isolation MS/MS with QTOF ................................ ........... 27 Research Motivation ................................ ................................ ............................... 28 Ketogenic Diet ................................ ................................ ................................ .. 28 Acetyl l Carnitine Role in Treatment ................................ ................................ 29 Overview of Research ................................ ................................ ............................. 30 MALDI MSI ................................ ................................ ................................ ....... 30 Quantitative MALDI MSI Method Development ................................ ................ 32 Piglet Brain Tissue Comparison ................................ ................................ ....... 33 2 QUANTITATIVE IMAGING OF COCAINE FROM BRAIN TISSUE ......................... 48 Introduction To Quantitative MSI ................................ ................................ ............ 48 Experimental ................................ ................................ ................................ ........... 51 Chemicals ................................ ................................ ................................ ......... 51 Tissue Preparation for MSI ................................ ................................ ............... 52 MSI Quantitation ................................ ................................ ............................... 53 Tissue Extraction/Quantitation ................................ ................................ .......... 54
6 Re sults and Discussion ................................ ................................ ........................... 56 Quantitative Detection of COC by MS/MS Imaging ................................ .......... 57 LC and MALDI Quantitation From Tissue Extracts ................................ ........... 62 Conclusions Of Quantitative MSI ................................ ................................ ............ 64 3 USING AN INTERNAL STANDARD TO IDENTIFY TISSUE SPECIFIC ION SIGNAL VARIATION in maldi mass spectrometric imaging ................................ ... 80 Introduction T o Tissue Specific Ion Signal Variation ................................ ............... 80 Materials and Methods ................................ ................................ ............................ 83 Chemicals ................................ ................................ ................................ ......... 83 Tissue Preparation ................................ ................................ ........................... 84 Ion Suppression Studies with Standards and Tissue ................................ ....... 85 Instrumentation ................................ ................................ ................................ 86 Discussion ................................ ................................ ................................ .............. 87 Internal Standard Application ................................ ................................ ........... 87 Io n Suppression with Standards ................................ ................................ ....... 88 Tissue Specific Ion Signal Variability ................................ ................................ 89 Conclusions To Ion Specific Ion Suppression ................................ ......................... 92 4 QUANTITATIVE TANDEM MASS SPECTROMETRIC IMAGING OF ENDOGENOUS ACETYL L CARNITINE FROM PIGLET BRAIN TISSUE USING AN INTERNAL STANDARD ................................ ................................ ..... 101 Introduction to Quantitative MSI for Tissue To Tissue Comparison ...................... 101 Materials and Methods ................................ ................................ .......................... 1 04 Chemicals ................................ ................................ ................................ ....... 104 Analysis of AC Standards ................................ ................................ ............... 104 Tissue Preparation ................................ ................................ ......................... 105 Mass Spectro metry ................................ ................................ ........................ 106 Results and Discussion ................................ ................................ ......................... 108 MS/MS Characterization ................................ ................................ ................. 108 Matrix Application for Solvent Comparisons ................................ ................... 109 Calibration Curve Experiments ................................ ................................ ....... 112 Imaging of Piglet Brain Tissue with Spotted Standards ................................ .. 113 Quantitative Comparison Between Treated and Untreated Tissue ................. 115 Quantitative MSI Of Endogenous AC Conclusions ................................ ............... 116 5 CONCLUSIONS AND FUTURE WORK ................................ ............................... 130 Co nclusions ................................ ................................ ................................ .......... 130 Future Work ................................ ................................ ................................ .......... 133 LIST OF REFERENCES ................................ ................................ ............................. 136 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 142
7 LIST OF TABLES Table page 1 1 Summary of common MSI ionization techniques. ................................ ............... 47 2 1 Summary of calibration plots and cocaine concentration determined by MA LDI MSI analysis with the LTQ ................................ ................................ ...... 77 2 2 Summary of calibration plots and cocaine concentration determined by M ALDI MSI analysis with the QTOF ................................ ................................ ... 78 2 3 Tissue COC concentrations determined by various analytical methods ............. 79 4 1 Matrix dependence on si gnal MALDI ion signal variation ................................ 128 4 2 Signal variation of MS/MS ion signal within c alibration spots ........................... 129
8 LIST OF FIGURES Figure page 1 1 Workflow diagram of MSI. ................................ ................................ .................. 35 1 2 Example of MS I from analysis of piglet brain ................................ ...................... 36 1 3 Illustration of the MALDI process. ................................ ................................ ....... 37 1 4 Photomicrographs showing differences in DHB crystals ................................ .... 38 1 5 Schematic of the Thermo MALDI LTQ XL ................................ .......................... 39 1 6 Schematic of the LIT. ................................ ................................ .......................... 40 1 7 Spectra illustratin g wide isolation MS/MS. ................................ .......................... 41 1 8 Schematic o f Waters Synapt QTOF instrument ................................ .................. 42 1 9 Schematic of quadrupole mass analyzer ................................ ............................ 43 1 10 Schematic of TOF mass analyzer ................................ ................................ ....... 44 1 11 Examples of different car nitine compounds ................................ ........................ 45 1 12 Whole body image of olanzapine distribution in an entire rat ............................. 46 2 1 Structure of cocaine and the two main product ions observed in the MS/MS spectrum ................................ ................................ ................................ ............. 66 2 2 Wide isolation MS/MS spectra of COC and d 3 COC analyzed by the LTQ A) and QTOF (B) instruments ................................ ................................ ................. 67 2 3 Optimization of LS with constant LE using wide isolation MS/MS ...................... 68 2 4 Optimization of LE with 4 LS using wide isolation MS/MS ................................ .. 69 2 5 Experimental setup for quantitative imaging ................................ ....................... 70 2 6 Wide isolation MS/MS imaging experiment for the quantitative analysis of COC from n ucleus accumbens ................................ ................................ ........... 71 2 7 Normalized calibration plot produced from COC spotted beneath a control section of brain tissue ................................ ................................ ......................... 72 2 8 Product ion ratio image generated from wide isolation MS/MS .......................... 73 2 9 Sample LC extracted ion chromatogram ................................ ............................ 74
9 2 10 Calibration curves of the tissue extracts used to quantify COC in the nucleus accumbens ................................ ................................ ................................ ......... 75 2 11 Calibration plots of the product ion signal ratio with different amounts of DHB applied to the tissue extracts analyzed by wide isolation MS/MS on the LTQ .... 76 3 1 Structure of AC and the three main product ions observed in the MS/MS spectrum ................................ ................................ ................................ ............. 93 3 2 MS/MS images comparing the application of the internal standard d 3 AC ........... 94 3 3 MS images of AC mixed with d 3 AC (each ~3 mm across) applied atop a spot of OC T ................................ ................................ ................................ ................ 95 3 4 Multiple tissue analysis of AC ion signal variability using wide isolation MS/MS ................................ ................................ ................................ ................ 96 3 5 Scaled MS/MS intensities of m/z 145 and m/z 148 summed over the raster area ................................ ................................ ................................ .................... 97 3 6 MS image of piglet brain tissue for the detection of endogenous AC ................. 98 3 7 Calibration curve of spotted AC standard plotte d from the intensity of m/z 204 corresponding to the MS image in 3 6 (A). ................................ ......................... 99 3 8 Calibration curve of spotted AC s tandard plotted from the ratio of m/z 204/207 corresponding to the MS image shown in 3 6B. ................................ 100 4 1 Microscope images (20 ) of formed DHB crystals applied with a Meinhard nebulizer onto a glass slide using two different solvents ................................ .. 118 4 2 Extracted wide isolation MS/MS images ................................ .......................... 119 4 3 Line scan intensities normalized to the mean for each ion and ratio ................ 120 4 4 Signal variat ion over subsequent MALDI MSI analysis. ................................ ... 121 4 5 Experimental setup for quantitative analysis of AC standards spotted onto a glass slide ................................ ................................ ................................ ......... 122 4 6 Calibration plots from standard of AC spotted onto a glass microscope slide .. 123 4 7 Wide isolation MS/MS images quantifying AC in piglet brain tissue ................. 124 4 8 Quantitative image of AC in the frontal lobe of a piglet brain slice .................... 125 4 9 Full MS images of AC distribution in piglet brain tissue obtained from the QTOF ................................ ................................ ................................ ............... 126
10 4 10 Standards addition plots for the quantitative determination of AC in piglet brain thin tissue sections ................................ ................................ .................. 127
11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy QUANTITATIVE TANDEM MASS SPECTROMETRIC IMAGING WITH MATRIX ASSISTED LASER DESORPTION / IONIZATION By David Andrew Pirman, Jr. May 2011 Chair: Richard A. Yost Major: Chemistry Mass spectrometric imaging (MSI) allows for d irect analys i s of tissues by matrix assisted laser desorption/ionization (MALDI) tandem mass spectrometry (MS n or MS/MS ) provides spatial distribution of an analyt However, MSI is considered primarily a qual itative technique as a result of the poor quantitative ability of MALDI combined with the challenge of direct analysis from complex tissue samples without cleanup or chromatog raphy Variations in the makeup of tissues results in extra ction and ionization variations making tissue to tissue comparisons difficult. The use of an internal standard to MSI applications has been suggested to control for MALDI signal variation and m ay be used to identify tissue specific ion signal variation. Applying the internal standard in a uniform manner not only leads to improvements in ion signal reproducibility, but can also help overcome variable ion signal as a result of tissue specific ext raction/ionization differences. A wide isolation MS /MS MSI approach was used to increase specificity and sensitivity of the target analytes. Wide isolation MS /MS allows for tandem MS analysis on both the analyte and a deuterated internal standard withi n the same MS scan. By
12 monitoring the ratio of the analyte to internal st andard ion signals, successful quantitative MSI experiments were performed for the analysis of cocaine in a human brain sample. These quantitative results compared well to results f rom high pe rformance liquid chromatography MS/MS experiments performed after tissue e xtraction and homogenization. Furthermore, applying an internal standard coupled with wide isolation MS/MS was successfully used to identify tissue specific ion signal variation. By analyzing five different tissues, variations in the analyte signal applied at a constant concentration were observed between the tissues. However, analyzing the analyte to internal standard ratio resulted in comparable ion signal ratios between the five tissues. Since MSI is a direct analysis technique, differences in the physical or chemical makeup of tissues directl y affects the resulting ion signal; without the use of an internal standard these are difficult to identify and account for. After development of a quantitative method, MSI was employed for quantitative determination of endogenous acetyl l carnitine (AC) concentration in a piglet brain sample. Piglets were f ed either a normal or ketogenic diet and any resulting changes in acetyl l carnitine were monitored by MSI. The results indicate a non significant change between the two samples, but were limited by error in the quantitative estimations. However, the MSI methods developed were used to make quantitative comparisons between two different samples quantitatively which is a significant advancement of the MSI technique.
13 CHAPTER 1 INTRODUCTION Mass Spectrometric Tissue Imaging Historical Perspective In most scientific research, the ability to generate visual images of the experiments can result in new interpretations leading to more conclusive results. From the development of X Ray technology 1 to magnetic resonance imaging (MRI) 2 improvements in i nstrument technology have generated new insight s resulting in more accurate determinations about a subject. T he technology is c ontinuously improving imaging da ta from generating highe r resolution MRI scans for tissue differentiation 3, 4 to molecular differentiation with techniques such as optical fluorescent labeling imaging 5 and atomic level imaging with tunneling electron microscopy (TEM). 6 As imaging techniques improve, so does area th ey are employed to investigate. T hrough advancements in biochemistry, chemistry, and medicine, the molecular understanding of biological systems is becoming increasingly important in areas such as disease research. For example, identifying new protein targets or metabolic pathways betwee n normal and diseased state tissues may lead to new therapeutic techniques. Most current imaging technique s; however interrogate sample s with little c hemical speci ficity unless targeted t hrough a label ( e g fluorescent labeling ) This lack of chemical specifi ci ty is a significant limitation of these imaging methods specifically for tissue differentiation at the molecular level New developments in analyte specific imaging techniques for biological samples include monitoring compound specific resonan ces with nuclear magnetic resonance
14 (NMR), Fourier transform infrared spectroscopy (FTIR) and anti stokes Raman scattering 7 T hese techniques are still being developed but one major limitation in them is the extremel y complex spectra generated from direct tissue analysis. Recently, the development of a new imaging technique based on mass spectrometry (MS) has potential to become an important analytical tool to answering biological questions. Mass spectrometric imagi ng (MSI) differs from classical imaging techniques cited above in the fact that it is actually a micro sampling technique performed in a two dimensional array and provides analyte distribution maps in a given sam ple. MSI offers an untargeted imaging method with improved analyte specificity, particularly, when employing either high resolution MS (MS) or tandem MS (MS/MS or MS n ) The coupling of MSI with matrix assisted laser desorption/ionization (MALDI) a surface des orption technique, has resulted in its significant gr owth as an ana lytical method for direct tissue analysis 8 Current Technology Current MSI technology was developed from the idea of using ion microscopy 9 to determine the elemental spatial distribution on a surface. The original ion microscope idea was further developed with the advancement of the ion microscope to secondary ion mass spectrometry (SIMS) and its use for ima ging of biological samples. These ionization techniques; however caused significant fragmentation of the target compounds limiting their application to elemental analysis They were also considered surface analysis techniques (nm sampling depths) resulting in poor sensitivity due to the small sample size of analysis The imaging idea was applied with a new instrument, the laser microprobe mass analyzer (LAMMA), which used laser/desorption ionization (LDI) with an ultra violet (UV) laser to genera te ions from backside illumination, which
15 resulted in complete ablation of the tissue sample 10, 11 However, LDI still induced significant fragmentation of the molecular species present in the tissue sample, resulti ng in mostly elemental analysi s, with improved sensitivity An improvement from these techniques was the coupling of the UV ionization source to an ion trap mass spectrometer that was used to generate quantitative MS/MS information d irectly from liver and brain samples. 12 The MS/MS capabilities allowed for absolute analyte identification during the surface analysis. 12 E ventually the use of softer ionization technique s such as MALDI was employed for MSI allowing for analysis of larger biomolecules 8, 13, 14 In addition, advancements ion gun sources and the development of matrix a ssisted SIMS has allowed for analysis of small intact biomolecules including lipids and peptides. 15, 16 A more recent ionization technique developed by Graham Cooks at Purdue University desor ption electrospray ionization (DESI), has been utilized for MSI. DESI allows for direct analysis of tissue sample with no matrix and at atmospheric pressure, thus simplifying the MSI process. 17, 18 A brief comparison of common ionization techniques for MSI is shown in Table 1 1. Each type of ionization method can be coupled with many types of mass analyzers. Currently, MALDI remains the most common method of ionization for imaging and is t he technique use d throughout this dissertation. Overall, MSI offers the best ability to perform untar geted, molecular specific imaging directly from a biological sample. This unique capability, coupled with the numerous types of available instrumentation has recently generated significant interest amongst the scientific community.
16 MALDI MSI Background any surface so long as ions can be generated and detected via MS. The most common application of the technique is for the analysis of biological samples, and is the primary focus of this research. A workflow illustrating the steps in generating MS images via MALDI MS n is described in Figure 1 1 For quantitative MSI, the addition of calibration standards and internal standards is necessary. A brief summary is described here, and f urther details will be discussed as they are used in each chapter. After the addition of the standard, e xcised tissue is first prepared by thin layer sectioning (10 20 m thick) performed on a cryo microtome instrument at temperatures between negative 15 and 25 C T he use of a tissue mounting medium to assist in sectioning is avoided as these cutting mediums may con taminate the tissue surface, interfering with MS analysis. In this research, tissue samples were mounted to the cryostat stage using de ionized water which froze upon the addition of frozen tissue. T he thin layer tissue section is the n placed on the MAL DI target surface which may be a standard glass or stainless steel microscope slide or a MALDI target plate For MALDI MSI, a matrix must be applied to the surface of the sample. This is a significant step in order to produce quality MS images, as the matrix serves numerous purposes. First, the MALD I matrix serves as a primary intermediary between the lase r source and the target analyte ( ionization theory is further discussed in the MALDI subsection). Second, the matrix must be applied uniformly to th e tissue section to ensure variations in ion signal are not a result of matrix crystallization differences. The matrix may be applied in numerous ways, including manual pneumatic spray coating
17 with an artistic airbrush or nebulizer 19, 20 micro spotted in a confined area with automated instrumentation such as the LabCyte 21, 22 or Shimadzu CHiP 23, 24 or by automated nebulization with instruments such as a Bruker ImagePrep which creates fine droplets off the surface of a piezo electric mate rial or even a modified inkjet printer. 76 Other methods w hich require no solvent, such as sublimation 25 and sifting fine matrix crystals 26 28 o nto the tissue have been shown as alterna tives to spray coating methods The absence of solvent reduces lateral diffusion of analytes within the tissue; however, i t also reduces the extraction of analytes from the tissue. Regardless of the technique chosen, the matrix must be applied uniformly producing crystal sizes smaller than the spot diameter of the MALDI laser. Once the tissue is coated with matrix, it is loaded into the vacuum chamber and moved step wise with respect to the ionization source. This patterned movement is referred to as r astering. At each raster step, which has specific x and y axis value s, a mass spectrum is recorded via MS or MS n The spatial resolution of the resulting MS images is dependent on the focusing of the MALDI laser as well as the raster step size between c ollected mass spectra. For MALDI and DESI MSI, typically resolution can be between 20 200 m Imaging software is then used to generate specific ion intensity images versus position, as exemplified in Figure 1 2. The figure shows two examples of MS imag es generated by plotting the intensity of the [M+Na] + ions of two lipids versus (x,y) position. From the resulting MS images, it i s conclu ded that the sphingomyelin ( m/z 751) is more abundant in the white matter of the brain, whereas the distributi on of a pho s phatidyl choline ( m/z 734) is more abundant in the grey matter of the brain. These types of images can be generated for any ion in the MS or MS n spectra.
18 Instrument parameters may also be varied, such as the applied laser energy (LE) and the number of laser shots (LS) performed at each raster step is determined on random areas of the tissue. Varying the LE and LS to achieve adequate ion signal is important for MS analysis. Depending on the type of mass analyzer, these numbers may vary greatly. For instance, analysis with a time of flight (TOF) instrument may require a large number of laser shots (~200 LS). In comparison, the linear ion trap (LIT) typically requires only 4 LS to achieve adequate ion signal. However, with the LIT, space charging is a concern when analyzing a large number of ions (discussed further in the ion trap section). The LE can also be varied to improve ion signal; however, too high a LE may result in increased fragmentation. MALDI F undamentals The coupling of MALDI with M SI offered a significant improvement over previous ionization methods. 8 MALDI produces mainly singly charged ions e.g., [ M+H] + across a broad mass range (50 200,000 Daltons). MALDI is also considered a sensitive technique with detection limits in the lower attomole levels. 29 The MALDI technique was invented and named by Frans Hillenkamp and Michael Karas 30, 31 The technique developed by Hillenkamp and Karas used small organic acids as the MALDI matrix, which ha ve proven to be the matrices of choice for most current MALDI applications The technique was later shown to ionize large biomolecules including prote ins by Koichi Tanaka 32 ue employed cobalt resin suspended in glycerol and was the first to show ionization of large intact bio molecules for which he shared one quarter of the Nob el P rize in Chemistry in 2002.
19 Practical A spects First, the MALDI matrix is applied to the sample, which can be small molecules, lipids, peptides, proteins, or intact tissue and the matrix/analyte mixture is allowed to dry producing a matrix crystal layer wit h analyte incorporated within the crystal lattice An illustration of the MALDI process is shown in Figure 1 3. This incorporation of the analyte may not actu ally be inside a crystal, but may be surrounding the matrix crystals. The important step is des orption of both the matrix and analyte during laser irradiation. If working with standards, the matrix to analyte ratio (M/A) is important in producing adequate ion signal. A high M/A results in mostly matrix ions in the mass spectrum. Conversely, a lo w M/A also results in inadequate signal as the concentration of matrix is too low to produce analyte ions. There exists a range of M/As in which the ion signal observed is mainly from the analyte; however, this M/A range varies with matrix type and analyt e. 34 For imaging applications, the M/A ratio is difficult to determine as the matrix is generally applied atop the tissue samples until a thin layer o f crystals is observe d and the amount of matrix is dif ficult to quantify. Equally difficult to quantify is the amount of analytes possible from a given area of tissue. The tissue and matrix layer may vary in thickness at each raster step which resulted in variations in the amount of sample analyzed. Also the incorporation of matrix crystals into the tissue may vary based on the chemical and physical composition of the tissue. Desorption P rocess The first step in ion production is the desorption process resulting from irradiation from the laser. 35 As the laser irradiates the sample and a certain amount of energy applied per area is reached, desorption occurs. This energy level for desorption is matrix dependant and can vary greatly with matrices and samples. Desorption results
20 in sublimation of the matrix along with the analyte which can be considered the first role of the MALDI matrix in generating ions 35 The MALDI matrix is usually chosen to have a very large mol ar absorptivity at the irradiation wavelength which allows it to absorb most of the energy imparted by the laser. This results in a lower initial kinetic energy of the target analyte s during the ablation process, thus limiting analyte fragmentation from t he ion source and producing mainly [M+H] + analyte ions. The plume is considered to form on the nanosecond time scale and may last milliseconds above the sample surface. This lasting plume has a high molecular density resultin g in a high molecular colli sion rate and leading to the formation of ions through subsequent ion molecule reactions or declustering of preformed ions 36, 37 Ion F ormation T heory The generation of ions from th e MALDI process is still not completely understood; however, two models of ion generation currently exist. One theory of ion formation suggests that, as, i n the resulting plume after desorption primary and secondary ions are formed The primary ions are generally matrix ions resulting from photo ionization effects from the laser irradiation. 35 Secondary ions are formed in the MALDI plume through gas phase ion molecule collisions in which the primary ion tr ansfers a charge in most cases a proton to the target analyte creating an [M+H] + ion. These secondary reactions occur between either a matrix ion and matrix neutral, a matrix ion and analyte neutral, or a hydrogen ion and analyte neutral. 35 Aside from ions created from the addition of a proton, the addition of cations is also observed and can lead to more stable analyte ions and more informative fragmentation pathways in some MS/MS applications. 20
21 l ucky s urvivor m 38 This model retains the idea of the initial desorption and photo ionization event but states that analytes retain thei r charge state from the solution phase and are incorporated into the matrix partially solvated. In the desorption step, the crystal lattice is broken up with varying sizes of molecule clusters and statistically some of these fragments will be a single an alyte ion. The analyte clusters (with many charges) continue to lose neutral solvent or matrix ions and counterions as the plume expands resulting in neutralization of the c lusters charges During the neutralization process as the cluster is reduced in size, only one ch he l ucky l urvivor. 37, 38 Quantitative MALDI The use of MALDI for quantitative purposes has been limited by the relatively poor ion signal reproducibility. The quantitative possibilities of the technique have been explored previously; however, quantitative data has proven difficult to generate using these techniques. 39 43 Significant ion signal variation and non linear increases in analyte signal independent of M/A ratio have contributed to the poor quantitative ability of MALDI MS. Ion signal variation results from laser shot to shot energy variability ; however, in most MSI instruments this effect is minimal because multiple laser shots were used Inhomogenous MALDI matrix crystallization is the largest contributor to varying MALDI ion signal, thus the importance of producing homogenous matrix layers consisting of uniform small crystals atop the imaging sample. Figure 1 4 shows optical images of different matrix crystals formed for MALDI experiments. On a scale similar to the laser diameter, the ion signal is dependent on the crystal quality at a given location. Other factors limiting the quantitative ability of MALDI for MSI will be discussed in Chapter 3.
22 Mass Spectrometry As the ions are produced from the MALDI process, they a re then detected via MS or MS n in terms of the mass to charge ratio ( m/z ) Several types of commercial mass analyzers are available for MSI experimentation. Many of these instruments are based on TOF technology and also include hybrid TOF instruments to provide MS/MS possibilities. There are also instruments equipped with ion trap mass analyzers offering MS n capabilities and even newer instrumentation offering high resolution MS n analysis for MSI. Two instruments were used in the MSI experiments disc ussed here: a Thermo Scientific MALDI LTQ XL (Thermo Scientific; San Jose, CA, USA) and a Waters Synapt MALDI Q Ion Mobility (T wave cell) TOF ( QTOF ) (Waters Corp. Milford, MA, USA). The former instrument is a linear ion trap (LIT) mass analyzer, and th e latter, a quadrupole coupled with a TOF mass analyzer The Thermo instrument will be referred to as the LTQ and the Waters instrument will be referred to as the Q TOF MALDI LTQ The Thermo Scientific MALDI LTQ XL is a commercially available MSI instrument capable of generating MS and MS n data with the LIT as shown in Figure 1 5 The instrument has previously been described in extensive detail 19 It is equipped with an intermediate pressure MALDI source (0.07 Torr) and a 337 nm nitrogen laser which fires at a frequency of 60 Hz and variable laser energy (maximum 250 J per pulse). The laser is focused to spot size of ~100 m at an in cident angle of 32 Upon laser irradiation of the sample, ions are transferred through a series of focusing multipoles, lenses and injected in to the LIT for mass analysis.
23 Linear ion trap The LIT is a two dimensional variation of the classical three d imensional quadrupole ion trap (QIT) developed by Wolfgang Paul and Helmut Steinwedel. 44, 45 The LIT differs from the QIT in that the ions are stored axially in two dimensions opposed t o three dimensional storage capabilities of the QIT. The LIT consists of three sections of hyperbolic quadrupole rods as shown in Figure 1 6 A Discre t e direct current (DC) voltages are applied to the front and back sections of the LIT, which trap the io ns in the center section, and also reduce the possibility of fringe electric field effects during trapping and resonant ejection of the ions. Trapping in the center section is performed by applying an oscillating radio frequency (RF) voltage to the X and Y rod pairs as shown in Figure 1 6 B Alternating current (AC) is applied across the X rods for isolation, activation and ejection of ions also shown in Figure 1 6 C The ions are ejected through 30 mm long slits along the X rods, strike two conversion dynodes and are ultimately detected by two electron multipliers one on each side of the trap. The LIT offers increased storage space within the trap as well as two detectors for ion detection, thus improving sensitivity and dynam ic range compared to the QIT. One limitation i n MS analysis with an ion trap is the limited ion storage available, thus limiting the dynamic range of the instrument. As the number of ions increases above a certain limit, an effect known as space charging is observed. This occurs as the ion density from the trapped ions shields the electric field from the voltages applied to the hyperbolic rods. The results can be mass shifts to high masses, decreased sensitivity, and decreased ma ss resolution. These ef fects are dramatically reduced in the LIT compared to the QIT due to the increased storage space.
24 MS/MS with a LIT MS/MS analysis on the LIT includes trapping the ions generated from the ion source, mass i solation of the parent ion, activation of the pare nt ion to induce fragmentation, and finally resonant ejection of the product ions for detection. In a typical MS/MS experiment, after the ions are trapped, isolation of the precursor ion is performed by applying RF and AC voltages to the quadrupole rods f or resonant ejection of all other ions from the trap. To isolate a particular ion a specific m/z ratio is chosen by the user along with a specific m/z range (isolation window). For instance, the isolation of caffeine which has m/z 195 corresponding to t he [M+H] + might be performed by isolation of m/z 195 1 m/z units. The isolation window for MS/MS analysis would be m/z 194 196. Narrow isolation windows increase specificity by isolating only the ion of interest for MS/MS analysis, but also may reduce sensitivity as fragile ions may fragment and be ejected from the trap from the resonant ejection frequencies applied. Wider isolation windows create a more stable environment for isolation, but also leaves a wider mass range of ions in the trap to be frag mented, and possibly com plicate the MS/MS spectrum. After isolation of the precursor ion, a resonant excitation RF voltage is applied to the X rods corresponding to the secular frequency of the precursor m/z This excitation voltage increases the kinetic energy of the precursor ions in the trap and through gas phase collisions with the He gas dissociation of the precursor ion occurs. This process is termed collision induced dissociation (CID) and results i n reproducible product ion fragments of the precursor ion. These product ions are subsequently ejected for the trap and detected producing an MS/MS spectrum. The product ions generated can be
25 used to provide structural information about the precursor ion or can be used to confirm the precursor m/z is free from contamination of other compounds at the same m/z On an ion trap instrument this process can be repeated by isolating and fragmenting one of the product ions, producing MS n spectra. 44, 46 Both full scan MS and MS/MS were used throughout this dissertation. Wide isolation MS/MS The ultimate goal of the presented research is to develop quantitative MALDI MSI methods which will use the ratio of the ion signal of the target analyte to the ion signal of an internal standard (deuterated analog of the target analyte). For MS applications, this is achieved simply by monitoring the [M+H] + ion signal of both the analyte and internal stand ard, and calculating the ratio. However, MS analyse s of tissue samples typically require sample cleanup steps, such a s extraction and chromatography prior to analysis. These steps help redu ce the occurrence of interfering peaks in the resulting mass spec tra. In MSI analysis, the tissue sample is directly analyzed, thus the ions generated by MALDI represent the entire molec ular makeup of the tissue, complicating the spectra and increasing the chances of interfering ions. The use of MS/MS and MS n can circ umvent these issues; provided the target analyte can be ionized sufficiently directly from a biological sample, and the resulting MS n can be verified to be analyte specific. Employing MS/MS for simultaneous analysis of an analyte and internal standard can be achieved with high precision by wide isolation MS/MS. 47 In wide isolation MS/MS, the center of isolation is moved from t he target precursor ion to an m/ z half way between the target and internal standard. The isolation window is then expanded to i nclude both the target analyte and internal standard. After isolation, fragmentation of isolated ions is induced through CID and the product ions are monitored, as illustrated
26 in Figure 1 7 In the product ion spectrum, the analyte and deuterated intern al standard can be differentiated allowing for the simultaneous MS/MS analysis of the analyte and internal standard. 46, 47 Wide isolation MS/MS has been shown to significantly improve the qua ntitative capabilities of MALDI for the analysis of cocaine from human brain tissue. 47 This technique will be evaluated and used throughout the presented research to generate quantitative MALDI MSI methods. MALDI Q TOF The QTOF instrument is a hybrid instrument coupling two mass analyzers, the quadrupole mass filter and a TOF and the instrument used here, was also equipped with a T wave ion mobility cell. A schematic of the instrument is shown in Figure 1 8 and has been previously described. 48, 49 The QTOF is equipped with an intermediate pressure MALDI source (0.07 Torr) and a 200 Hz, 335 nm, tripled Nd: YAG laser at a frequency of 200 Hz and spot diameter of 150 m Upon irradiation with the laser, the ions are transfer red through a series of lenses into the quadrupole for mass selection prior to fragmentation. T he ion mobility cell was not used in these e xperiments although the last portion of the cell (transfer cell) is used for CID. The quadrupole serves as a mass separating device prior to fragmentation for MS/MS applications. 50, 51 After fragm entation by CID, the product ions are accelerated into the TOF portion of the instrument for mass analysis. Mass selection with quadrupole The quadrupole serves as a mass filter prior to fragmentation in the ion mobility cell. A schematic is shown in Fi gure 1 9 with f our parallel rods with each opposing rod pair connected electrically. 51 An RF or DC voltage can be applied to produce an electric f i e l d within the four rods. By varying the applied voltages, ions of ce rtain m/z have
27 stable trajectories through the instrument and thus reach the detector. Constant voltages can be applied to allow a small m/z range to pass through, or the voltage can be continuously varied to produce a mass spectrum. Mass analysis with TOF A TOF mass analyzer measures amount of time it takes for an ion to travel a fixed distance to determine its m/z as shown in Figure 1 10 Ions are accelerated into the TOF region with an electric field of constant strength. Thus, ions of the same cha rge (usually the +1 state) have the same initial kinetic energy and the velocity of the ion is dependent upon the m/z As the ions travel to the detector, larger ions move more slowly and smaller ions move more quickly thus a time difference between the ions can be measured. This acceleration voltage may be applied continuously or a technique known as delayed extraction can be used to improve TO F sensitivity and resolution. On the QTOF, delayed extraction is not used; however, orthogonal injection into the TOF is used. This ensu res that ions arriving into the TOF region of the instrument are accelerated from the same initial velocity, irrespective of their velocities from the quadrupole and ion mobility portions of the instrument. MS/MS and wide isolation MS/MS with QTOF MS/MS analysis with the QTOF uses the quadrupole to allow a small m/z range to enter into the ion mobility portion of the instrument. Fragmentation of the ions is generated through CID (previously described). The resulting product ions are then software allows the user to control the mass resolution of the quadrupole upon setting a target mass for MS/MS analysis. This was presumably designed for target analytes with a wide isotopic peak envelope such as high molecular weight proteins with multiple
28 charge states For quantitative MSI analysis with an internal standard, this feature was used to allow both the target analyte and its deuterated a nalog to be simultaneously detected in a single MS/MS scan. The simultaneous detection of both the analyte and internal standard is imperative for quantitative MALDI MS n applications (further discussed in subsequent chapters). Research Motivation The overa ll goal of the presented research was to develop MSI techniques for the comparison of acetyl l carnitine (AC) concentration and distribution between piglet brain samples, one fed a ketogenic diet (KD) and one fed a normal diet. The ketogenic diet is used for the treatment of epilepsy cases which fail to respond to the current pharmaceuticals. Its effectiveness is often comparable to the best anti epileptic drugs available; however, the therapeutic mechanism remains unknown. MSI techniques would provide us eful information about any quantitative distribution differences of AC in the brain between piglets fed a ketogenic diet versus a normal diet. However, to use MSI to quantitatively compare endogenous metabolite concentration differences between tissues, n ew quantitative methods must be developed. Ketogenic D iet KDs were designed in the early 1900s to mimic the biochemical changes present during fasting. 52 Fasting has long been known to reduce the amount of se izure activity and has been described in the Hippocratic Corpus dated around 400 BC. 53, 54 The KD is an extremely strict and regulated diet in which the body derives much of its energy from fats rather than carbohydrates. The diet generally consists of 85% fat, 10% p rotein and 5% carbohydrates at only 75% of the recommended daily caloric intake. Fluids are also reduced to 80% of the normal level. Th e diet is initiated by fasting for 24 36 hours
29 to exhaust glucose reserves and force the body to rely on fatty acid oxidation as its main source of energy production resulting in a ketotic state or ketosis, which is a large buildup of ketone bodies resu lting from oxidation. 53, 54 These ketone bodies consist of hydroxybutyric acid and acetoacetic acid, which is spontaneously converted into acetone 53 and are used as energy sources in the brain S ome of the resulting metabolite s of oxidation and the breakdown of the ketone bodies for energy are hypothesized to aid in the anticonvulsant mechanism of the KD by controlling potassium channels in the ner ve 52, 54, 55 Acetyl l C arnitine Role in Treatment Of the metabolic changes resulting from the KD, changes in concentration of acetylcarnitine (AC) has been hypothesized by Dr. Peggy Borum from the Food Science and Human Nutrition Department at the University of Florida to play a key role in the therapeutic effects of the KD AC belongs to a class of compounds called carn i tines shown in Figure 1 1 1 that serve as the main transporters of fatty acids and acetyl groups across the mitochondrial membrane for oxidation 56 A n increase in carnitines and carnitine translocase proteins is necessary to accommodate for the increased fatty acid oxidation. Aside from the increased need for f atty acid transport, AC also has other metabolic fates such as the production of acetyl coenzyme A (acetyl CoA). The normal pathway for acety l CoA generation is slowed due to the decreased glu cose metabolism during the KD. 57 AC has also been shown to induce transient seizures in epileptic patients possibly by interrupting the GABA inhibitory system. 58 Although AC has been shown to increase in tissue during the ketogenic diet and fasting 59 it may be shunted from other metabolic
30 pathways and used primarily for fatty acid transpo rt and conversion to acetyl CoA 60 61 Methods to detect and quantify acyl carnitines in tissue have been developed using positron emission tomogr aphy (PET) 62 liquid chromatography (LC) / MS 63 and subcutaneous microdialysis (SMD) followed by MS detection. 59 PET provides real time imaging of the l abeled analyte but is limited by its lack of ch emical specificity. L C/ MS provides excellent quantitative data, but due to tissue homogenization and extraction methods, spatial distribution information is lost. SMD provides minimally inva sive sampling directly from tissue; however it is limited to the area of tissue the probe resides and no spatial distribution information is acquired. A method to quantify the distribution and concentration of acetylcarnitine and other suspected anticon vulsant metabolites in the brain w ould pro vide important information into the mechanism of the therapeutic effects of the KD. Brain tissues from p iglet test subject s, were used for the method development and establishing the effects of the KD on concentra tion and distribution of AC in brain tissue Overview of Research MALDI MSI Combining MALDI with MS and MS n has been used for numerous imaging applications. 19, 26, 46, 64 73 Specifical ly for smal l molecular applications, MSI has great potential in the area of pharmaceutical development. However, the limitation s of using MALDI for quantitative MSI is the main drawback preventing the technique from being used more readily. To highlight both the possibilities and the limitations of MSI, a whole body MSI e xperiment is shown in Figure 1 1 2 72 The experiment is designed to determine the distribution of a pharmaceutical in a thin layer whole body section of a rat
31 using MALDI MS/MS. From the optical image, Figure 1 12 A, different organs are identified and outlined. The resulting MS/M S image of the drug is produced by plotting (x,y) position sh own in Figure 1 12 B The MS/MS image of the dru g shows distribution differences between organs. Comparing signal from the brain to the bladder indicates the drug is present at a higher concentration in the bladder However, variations in ion signal may not be directly attributed to increases in concentration as numerous factors other than concentration may affect the detection ion signal. 7, 72 74 These factors include: tissue dependent matrix crystallization, ionization, and extraction variations. 75 Consider that the drug could be present in both the bladder and brain at equal concentrations. Th e bladder consists of mainly smooth hollow muscle tissue (which may contain frozen urine or crystallized uric acid) compared to complex, lipid rich tissue of the brain. As matrix is applied atop the two areas, the formation of crystals can vary signifi can tly. Also, during the matrix application, the matrix solvent plays a role in extracting the target drug from the tissue and subsequent co crystallization with the matrix. This extraction between the two tissue types may also vary. Finally, MALDI ionizat ion may also vary significantly between the tissue types, thus making any conclusions about the concen tration of the drug uncertain To verify the MS/MS imaging results, the tissues are typically analyzed via another analytical method which typically invo lves tissue excision and homogenization, analyte extraction, and separation via chromatography. It also typically involves the creation of matrix matched calibration samples via the quantitative addition of the drug to control tissue. The se steps not onl y increase analysis time and use more resources, they also lose any spatial distribution information within a given a
32 tissue typ e (due to homogenization). For MALDI MSI to be used more routinely a s an analytical method the quantitative limitations must be overcome T he overall motivation for this research is to develop quantitative methods for MALDI MSI through the application of an internal standard to account for MAL DI MSI ion signal variability. Quantitative MALDI MSI Method Development Nucleus accum bens (NA) excised d uring a post mortem autopsy was used for quanti tative MSI method development. Two samples were used for analysis, a control sample and one which tested positive for cocaine (COC) during the autopsy. COC is an extensive ly studied drug of abuse with numerous published method s on the extraction and quantitation from blood and tissue. 79 The ability to perform parallel experiments with well established methods to quantify COC from brain tissue ( as well as it being an exogenous compound ) made it an ideal candidate for quantitative MSI development. Deuterated analogs of COC are also readily available and were used an internal standard. The internal standard in any analytical experiment must behave identically to the target analyte, be unifor mly applied, and be detected separately. Isotopically labeled analogs provide near identical chemical characteristic and are considered the best option. The challenge in a MSI experiment is applying the internal standard in a uniform manner to the tissue sample and this can be overcome using a modified inkjet printer. 66, 76 This uniform application is imperative to control for ion signal variations resulting from variations in the matrix application and MALDI process. In an MSI experiment consisting of numerous spectra collected from a tissue sample, each raster spot can be considered a micro experiment. To better compare the ion signals within these micro experiments, normalization of the analyte by the internal standard ion signal
33 may be used to acco unt for ion signal variations. 39, 71, 77, 78 To perform ion signal normalization, the analyte and internal standard should be detected simultan eously in each MS scan at each raster step Wide isolation MS/MS allows for the simultaneous detection of both the analyte and internal standard and provides the increased specificity and sensitivity from tandem MS 47 The experiment s presented with COC serve primarily as method development experiments in which the quantitative results can be validated by another analytical method, LC MS/MS. For LC MS/MS analysis, the tissue samples were homogenized and extracted and the COC determine d by comparison to external calibration standards. 46, 79 These results were compared to quantitative MSI results. Wide isolation MS/MS was also evaluated for the quantitative evaluation of th e tissue extracts in parallel to the LC MS/MS analysis of the extracts. These experiments highlight the power of using an internal standard and wide isolation MS/MS for quantitative MALDI applications. Differences in the chemical and physical makeup of tissues can directly affect the resulting ion signal in a MSI experiment. The use of a deuterated internal standard was evaluated for the identification of tissue specific ion suppression and is discussed in Chapter 3. AC and acetyl d 3 carnitine (d 3 AC) w ere used as the target analyte and internal standard respectively. Both were applied beneath different tissue types and the resulting ion intensity of AC, d 3 AC, and the ratio of AC/d 3 AC were monitored. Piglet Brain Tissue Comparison The research presented in Chapter 4 a ppl ies methods developed in Chapters 2 and 3 to quantitatively determin e and compare AC distributions between two piglet brain sampl es. For the analysis of endogenous AC the standard additions method of
34 analysis was used for quant itative estimations. One piglet was fed a normal diet after birth consisting of an average ratio of fat, carbohydrate, and protein. These ratios can vary slightly, but typically the formula is rich in carbohydrates >50% by mass compared to protein and fat The carbohydrate ratio in the ketogenic formula is significantly reduced to keep the piglet in a state of ketosis, thus mimicking the metabolism of KD therapy. After one week of dieting, the brains were excised and AC content was compared with MSI. Es tablishing quantitative MALDI MSI methods to make these comparisons would be a significant improvement in the MSI methodology.
35 Figure 1 1. Workflow diagram of MSI
36 Figure 1 2. Example of MSI from analysis of piglet brain. The intensity of any ion in the mass spectrum can be plotted against (x,y) position to determine its distribution in the tissue sample
37 Figure 1 3. Illustration of the MALDI process.
38 Figure 1 4 Photomicrographs showing d ifferences in DHB crystals. DHB dissolved in different solvents, (a) DHB diluted in 70/30 methanol/water (v/v), (b) DHB diluted in 90/10 chloroform/ethanol (v/v) both applied atop a glass slide. DHB dissolved in different solvents (c) DHB diluted in 70/ 30 methanol/water (v/v), (d) DHB diluted in 90/10 chloroform/ethanol (v/v) both applied atop piglet brain tissue. Variations in DHB crystals directly a ffects resulting ion signal. Ideally, DHB would be applied in a uniform manner consisting of small cry stals. A B D C
39 Figure 1 5 Schematic of the Thermo MALDI LTQ X L. Samples are loaded onto an (x,y) stage and rastered beneath the MALDI laser. Positio n specific mass spectra are recorded and used to produce an ion specific distribution image. The LIT offers both MS and MS n capabilities 19, 44
40 Figure 1 6 Schematic of the LIT. MS or M S n spectra can be obtained from ions trapped in the center section by applying different RF and AC voltages. (A) LIT (B) Z axis view and voltages applied for trapping (C) Z axis v i ew and voltages applied for excitation. A B C
41 Figure 1 7 Spectra illustrating wide isolation MS/MS. (A) Narrow isolation window centered at m/z 204 1.5 Da employed for MS/MS analysis of AC. (B) Wide isolation window centered at m/z 205.5 3 Da employed for simultaneous detection of AC and d 3 AC in a single MS/MS scan indicated by the shaded window (C) Product ion spectrum of wide isolation MS/MS analysis. AC and d 3 AC can be differentiated due to the preserved deuterated label in the product ion A B C
42 Figure 1 8 Schematic of Waters Synapt QTOF instrument. Modified from Pringle, et al. 49 T Wave ion guide Quadrupole Dre Lens Trap Gate IMS Transfer Einzel Lens Intermediate pressure MALDI chamber Isolation valve Transfer Lens Reflection Pusher Detector
43 Figure 1 9 Schematic of quadrupole mass analyzer. The quadrupole on the QTOF was used for mass selection prior to CID.
44 Figure 1 10. Schematic of TOF mass analyzer. Ions are pulsed into the TOF region by an applied voltage on the pusher. All ions have the same kinetic energy, thus smaller ions travel the fixed distance to the detector faster than larger ions.
45 Figure 1 11. Examples of different carnitine compounds Carnitine s main function is to transport long chain acyl groups into the mitochondria for oxidation. The site of deuteration is circled in the AC structure.
46 Figure 1 12. Whole body image of olanzapine distribution in an entir e rat. (A) Optical image with identified tissues. (B) MS/MS image of main product ion of olanzapine. Ion intensity is plotted via color gradient, dark areas indicate areas of low ion intensity and white areas indicate areas of higher ion intensity. Mod ified from Caprioli, et al. 72 A B
47 Table 1 1. Summary of common MSI ionization techniques. Ionization Method m/z Range Lateral Resolution Advantages Disadvantages SIMS <1000 <50 nm Best spatial resolution Source fragmentation, low sensitivity LDI <1000 5 200 m Improved sensitivity over SIMS Source fragmentation MALDI 50 500,000 10 200 m Ionization of large intact biomolecules Application of MALDI matrix DESI <2000 ~200 m Simpler sample setup, possibly targeted ionization Poor l ateral resolution
48 CHAPTER 2 QUANTITATIVE IMAGING OF COCAINE FROM BRAI N TISSUE Introduction To Quantitative MSI Mass spectrometric imaging (MSI) has rapidly expanded from early application s to protein imaging 8 into numerous c ompound classes including lipids, 20 pharmaceuticals, 13 and endogenous metabolites. 72 The technique generates two d imensional ( or more recently three dimensional images ) of the distribution and concentration of analytes in a tissue section. 80, 81 The quantitative limitations of MALDI MSI have recently com e under scrutiny, 82 as MSI techniques suffer from two major limitations: the inability to control and correct for tissue specific ion suppression and the irreproducibility of ion signal from scan to scan. These limitations are caused by numerous factors, including tissue variation, matrix crystal heterogeneity, laser power fluctuations, as well as tissue specific ion suppression, a nd preclude in situ quantification. 47, 83 To date, quantitative MSI applications have been limited to relative determinations which require verification from other analytical techniques, such as liquid chro matography tandem mass spectrometry ( LC MS /MS ). 26 Many quantitative MS analytical techniques rely on internal standards to control for sources of error arising from sample preparation, injection volum e irreproducibility, or ion source variability. In most instances, an internal standard is added in a constant amount to the samples being analyzed prior to any additional sample preparation Applying an internal standard to control for the limitations pre viously listed may improve the quantitative capabilities of MSI. Utilizing an internal standard for traditional MALDI analysis has been shown to improve quantitation ; 41, 42, 78, 84 however, these methods simply add the internal standard to the sample droplet, which is not practical in an
49 imaging experiment. For MSI, the internal standard must be applied uniformly to the tissue. In this work, the internal sta ndard was applied via a modified inkjet printer 76 Application of an internal standard by inkjet printer has been shown to produce a uniform internal standard lay er at a given concentration (Chapter 4) To best mimi c the extraction of an analyte from the tissue, the internal standard was applied beneath the tissue (to the glass sample slide before placing the tissue) which then was extracted through the tissue and into the matrix crystals. For the quantitative analy sis of cocaine (COC) in brain tissue, deuterated cocaine (d 3 COC) was used as an internal standard. Deuterated analogs routinely perform best as an internal standard as they have nearly identical chemical characteristics as their unlabeled counterparts. A pplying an internal standard in this manner is shown in Chapter 4 to improve scan to scan reproducibility and calibration curve linearity for quanti tative imaging applications Recently, a n MSI experiment reported the quantitative tiotropium bromide distribution in lung tissue. 65 The quantitative estimations were made based on calibration standards pipetted atop a control tissue slice and comparing the signal of the calibration standard to the positive sample. Spotting standards atop a tissue section extraction efficiencies may vary. Also, comparing the signal from two separate tissues sections may be challenging due to differences in thickness, tissue composition and matrix coating variations. These experiments, however; do show progress in the area of quantitative MSI as the authors were able to produce a linear calibration curve from the spotted samples from whi ch they were able to determine a quantitative estimation of the administered drug.
50 Another recently published study by Murphy 18 examined the relationship between MALDI MSI signal and lipid concentrations in brain tissue and succeeded in making relative q uantitative comparisons between MSI and LC MS results. However, these comparisons were not made before extensive signal normalizat ion and baseline correction, indeed the authors concluded that MSI results are taken with caution due to the complex environme nt of direct tissue analysis. 74 To further improve MSI quantitation, normalizing the target analyte ion signal to an tissue spe cific ion suppression. 85 Scan by scan normalization by a deuterated internal standard for MALDI MSI is shown to improve single reproducibi lity (Chapter 4), improve image quality and significantly improve the quantita tive ability of MALDI MSI. 47 The internal standard must have similar extraction, co crystallization, and ionization properties as the target analyte for quantitation, thus a deuterated analog is an ideal option. The internal standard may be applied to the tissue in 3 ways: (1) apply a mixture of the internal standard and MALDI matrix to the tissue (2) apply an internal standard atop the tissue prior to applying the matrix 73 and (3) apply the internal standard beneath the tissue sample prior to placing the tissue and applying the matrix Applying the internal standard along with the matrix does not accurately reflect tissue specific ion suppression as the internal standard m ay not mimic the ionization efficiency of a target analyte being extract ed and detected from the tissue, as shown in Chapter 3 Also, if the matrix is applied manually, there may be uneven matrix/internal standard application resulting in unaccountable a nalyte to internal standard ratio differences. Applying an internal standard atop the sample may not accurately reflect the analyte
51 being extracted from within the tissue sample as the internal standard may reside atop the tissue. Applying the internal s tandard beneath the sample may best reflect the detected ion signal of the target analyte. Both analyte and internal standard must be extracted and co crystal lized from the tissue sample. This approach has been shown to help control for tissue variation as well as extraction differences from different tissue types shown in Chapter 3. Lastly, due to the significant amount of signal variation in MALDI, the internal standard and analyte should be detected in the same analytical MS n scan. Detecting the ana lyte and internal standard in separate MS scan has been shown to offer no quantitative advantage due to the signal variation form the MALDI process 47 For MS applications, this is easily done by detecting a mass range large enough to incorporate both ions; however, for tandem mass spectrometry (MS/MS) applications a wide isolation window must be utilized. 47 From the structure of COC shown in Figure 2 1, the location of the deuterated label allows for differentiation b etween COC and d 3 COC when using MS/MS. By combining the advantages of a uniformly applied internal standard beneath a tissue sample with the specificity of wide isolation MS/MS with scan by scan COC to d 3 COC normalization 47 absolute quantitative MSI was performed for the analysis of COC in a brain tissue section. These results were compared to LC MS/MS and MALDI MS/MS results from tissue extracts of the same brain sample. Experimental Chemicals Standards of COC (MW: 303) and d 3 COC (MW: 306) were purchased from Ceriliant ( Round Rock TX, USA ). S uperfrost glass plus microscope slides and LC grade methanol (MeOH) acetonitrile (ACN), and water ( H 2 O) were purchased from
52 Fisher Scientific (Pittsburgh, PA USA ). Stainless steel microscope slides were purchased from Thermo Scientific (San Jose, CA, USA). For MSI analysis, w orking standards of COC and d 3 COC were diluted to 10 ng/ L in ACN used for LC analysis and 10 ng/ L in 50/50 MeOH/ H 2 O (v/v) used for MSI Calibration standards of COC used for MALDI imaging were made to 0.030, 0.060, 0.13, 0.25, and 0.50 ng/ L in 50/50 ACN/H 2 O (v/v) and 1 L of each calibration standard were pipetted onto a stainless steel microscope slide prior to the d 3 COC application. Tissue concentrations were estimated from the area of the detected signal and the thickness of the tissue, assuming the density of wet tissue is ~1.0 g/mL. The MALDI matrix 2,5 dihydroxybenzoic acid (DHB) was purchased from Acros Organics (Geel, Belgium) and pr epared to a concentration of 20 mg/mL in 70/30 MeOH/H 2 O ( v/v). For LC working standards of COC were prepared to 0.50 and 5.0 ng/ L and working standards of d 3 COC were prepared to 0.50 ng/ L. Calibration standards were spiked into control brain tiss ue during tissue homogenization at concentrations of 50.0, 100, 200, 400, 1,000, and 2,000 ng COC /g wet tissue and the internal standard was spiked at 40 ng d3 COC /g wet tissue Tissue Preparatio n for MSI For quantitative MSI, calibration standards of COC in M eOH/H 2 O were pipette on to a microsco pe slide and allowed to dry under ambient conditions. An Epson (Long Beach, CA, USA) Stylus R280 was used to apply 3 coats of the d 3 COC in MeOH/H 2 O to the microscope slide s Positive and negative control sampl es of the nucleus accumbens Office in Colorado Springs, CO. The nucleus accumbens was extracted during autopsy,
53 ectioning of the br ain tissue was performed on a Microm HM 505E cryostat (Waldorf, Germany). Tissue was sliced at 25 C to a thickness of 20 m and thaw mounted onto the microscope slides previously coated with COC calibration standards and d 3 COC. DHB was then applied to the tissue samples using the inkjet printer. 76 MSI Quantitation For the MSI experiments, all mass spectrometric images were acquired with either a Synapt MALDI Q Ion Mobility (T wave cell) Time of Flight ( QTOF ) equipp ed with a 200 Hz, 335 nm, Nd: YAG laser at a frequency of 200 Hz and spot diameter of 15 0 m (Waters Corporation) or a MALDI LTQ XL equipped with a 60 Hz, 337 nm, nitrogen laser with a spot diameter of 100 m (Thermo Corp. San Jose, CA, USA) The QTOF instr ument was operated in MS/MS mode with the resolution of the quadrupole de creased to include the [M+H] + ions of both COC ( m/z 304) and d 3 COC ( m/z 307) prior to fragmentation through collisions with He Sample slides were in serted into the vacuum chamber and rastered beneath the laser at a step size of 250 m with a laser spot size of 100 m Wide isolation MS/MS spectra were summed from 200 laser shots at each raster step. For MS/MS imaging experiments on the LTQ, the isolation window centered at m/z 305.5 3 to trap the [M+H] + ions of both COC and d 3 COC prior to fragmentation. Upon insertion of the samples into the vacuum chamber, the sample was rastered beneath the laser at a step size of 250 m with a laser spot size of 100 m Wide isolation M S/MS spectra were collect using 8 las er shots at each raster step. The ions of interested were fragmented in both instruments by collision induced dissociation (CID) and the fragments of each precursor ion were monitored; COC m/z
54 304 182 and d 3 COC m/z 30 7 185. Resulting MS /MS data from the QTOF were processed using BioMap ( Novartis, Basel, Switzerland) and images from the LTQ were processed with QualBrowser (Thermo Scientific, San Jose, CA, USA) and Surfer 8 (Golden, CO, USA). Normalized images were generated using the divide function in BioMap to generate COC/d 3 COC ( m/z 182/185) images. By plotting the ratio of m/z 182/185 versus position, normalized imag es from the LTQ were generated in Surfer 8 from data extracted from QualBrowser. For quantitative determinations of COC, a standard additions method analysis was performed utilizing the product ion ratio of COC/d 3 COC from the spotted COC calibration stand ards beneath the positive tissue. Quantitation by external calibration was also performed with calibration standards spotted beneath a control tissue section and a quality control (QC) standard of COC (0.25 ng/ L) spotted beneath a positive tissue section to ensure ion signal was comparable between tissues. The resulting signal from the COC was compar ed to the calibration curve calculated from the control tissue. All quantitative MSI data was then compared to LC MS/MS and MALDI MS/MS data from extracts of the brain tissue Tissue Extraction/Quantitation N ucleus accumbens samples were extracted using manual tissue homogenization and solid phase extraction (SPE) according to the procedure published by Lowe et al. 79 Two 100 mg sections of the positive COC tissue samples were dissected and extracted. The extracts were compared to calibration sample extracts from 100 mg sections of the negative control sample. The total sample makeup was two positive samples, six calibra tion samples, and one negative control sample. Calibration samples were prepared at the concentrations previously stated and following SPE, the samples were
55 reconstituted in 100 L of H 2 O/ACN 90/10 (v/v). A 5 L aliquot was taken for wide isolation MALDI MS/MS analysis. The remaining 95 L of each sample was submitted for triplicate LC MS/MS analysis on a Thermo Finnigan TSQ 7000 triple quadrupole mass spectrometer (QqQ) equipped with an Agilent 1100 LC system (Santa Clara, CA, USA). Mobile phase A was composed of H 2 O with 0.1% formic acid and mobile phase B was composed of ACN with 0.1% formic acid. A gradient LC program was utilized starting with 5 minutes of 10% B, then a ramp up to 90% over 8 minutes with a 3 minute hold at 90%, and re equilibrated for 3 minutes at 10% B. A 5 L aliquot of each extract were loaded onto the LC column followed by MS/MS analysis by positive mode electrospray ionization (ESI) and m ultiple reaction monitoring for COC ( m/z 304 182) and d 3 COC ( m/z 307 185) for quantitative analysis. MALDI MS/MS of the 5 L aliquots was performed with a Thermo Scientific MALDI LTQ XL utilizing a wide isolation window centered at m/z 305.5 3 Da with scan by scan COC to d 3 COC From the 5 L aliquot, 1 L of each sample (2 p ositive samples, 6 calibration samples, and 1 negative control) was spotted in duplicate onto a glass plus microscope slide. 40 mg/mL of DHB dissolved in MeOH/H 2 O 70/30 (v/v) was applied uniformly to the slide utilizing a Meinhard nebulizer (Golden, CO, U SA) at a flow rate of 2.4 mL/min. Within each spotted sample, two areas of 50 MS scans were analyzed by MALDI MS/MS with a step size of 200 m, laser energy of 14 J, and 8 laser shots per step. This resulted in quadruplicate analysis of each sample. Fr om each 50 scan MS/MS experiment, the extracted ion current for COC ( m/z 304 182) and d 3 COC ( m/z 307 185) were plotted using data extracted to Microsoft Excel from QualBrowser.
56 From each MS/MS experiment the ratio of m/z 182/185 was calculated, averaged, and used for quantitation. Results and Discussion For successful quantitation of COC from brain tissue by MSI, results from imaging experiments must be validated with quantitative results using accepted quantitative anal ytical techniques such as LC MS/MS. Four requirements have been suggested 86 for successful quantitation of small molecules by MALDI: a high repetition rate laser (for analysis with a QqQ), tandem MS (for increased sele ctivity), sample cleanup (for proper co crystallization), and an internal standard (for increased precision). In these MSI experiments, these requirements were addressed by employing wide isolation MS/MS imaging experiments that allow for scan by scan MS/ MS ion signal normalization of the COC ion signal by the internal standard ion signal, d 3 COC With wide isolation MS/MS, the advantages of MS/MS are preserved (improved selectivity and sensitivity) while also improving MALDI quantitation and ion signal re producibility through the scan by scan normalization of the analyte to internal standard. 47 Scan by scan ion signal normalization circumvents the need for a high repetition rate laser and has been shown to improve si gnal reproducibility five fold (Chapter 2) Co crystallization of the target analyte (COC) and internal standard (d 3 COC) has been enhanced by applying the MALDI matrix in a uniform manner with an inkjet printer. 76 In MSI, co crystallization of the target analytes can also be enhanced by improved extraction of the analyte from the tissue with the matrix solvent. Improving the extraction and co crystallization of the target analytes can lead to improved sensitivity. I n this study, we show extraction of spiked COC and d 3 COC from beneath the tissue; however, this extraction could be
57 further optimized by evaluating other matrix solvents and the effect of tissue thicknesses. Quantitative D etection of COC by MS/MS Imaging Adapting a wide isolation MS/MS approach from the LTQ to the QTOF allowed for simultaneous MS/MS detection of COC and d 3 COC as shown in Figure 2 2. For the QTOF this was accomplished by decreasing the resolution of the resolving quadrupole to include both COC ( m/z 304) and d 3 COC ( m/z 307) after CID, product ions from both are detected in the MS/MS spectrum (COC, m/z 304 182; d 3 COC, m/z 307 185). Prior to performing quantitative imaging on a tissue experiment, the instrumental parameters, laser energy (LE) and laser shots (LS) were optimized for wide isolation MS/MS analysis of a mixture of COC and d 3 COC. More important than the optimization of LE and LS, was the identification that the internal standard can account for signal variation arising from l aser energy fluctuations. Ion signal can be altered with variations in both the number of LS and the amount of LE at each raster position. Plots of these changes are shown for varying LS and LE are shown in Figure 2 3 and Figure 2 4; respectively. Altho ugh the signal may change significantly with these changes in instrumental parameters, the detected A/I is preserved, indicating the d 3 COC is similar in ionization behavior to that of COC. The tissue imaging experimental setup for the quantitative detect ion of COC is illustrated in Figure 2 5. Calibration standards were spotted onto a MALDI target followed by the uniform application of the internal standard, d 3 COC. Tissue is then placed over that standards and coated with MALDI matrix. Images and quant itative analysis were produced from either the intensity of m/z 182 or the ratio of m/z 182/185 intensity determined from the MS/MS spectrum, illustrated in Figure 2 6 The MS/MS
58 images of the product ion intensity of m/z 182 ( Figure 2 6 A), the normalize d COC/d 3 COC ( m/z 182/185) image ( Figure 2 6 C), and the corresponding standard additions plots (Figures 2 6 B and 2 6 D, respectively) were generated from the wide isolation MS/MS experiment performed on th e positive COC section of NA. Normalizing the COC/d 3 COC ion signal ratio for quantitative analysis improved the linear correlation coefficient ( R 2 ) while decreasing the width of the 95% confidence limits by a factor of two and decreasing the uncertainty in the best fit slope from 100% to 40%. These improvements are a reflection of the improved ion signal reproducibility observed when monitoring the ion signal ratio of analyte to internal standard Analysis of three serial sections with the QTOF by standard additions exemplified in Figure 2 6, yielde d an average COC concentration of 240 100 ng/g tissue The improvements in calibration curve linearity and signal reproducibility from normalizing by the internal standard is attributed to an improvement in scan to scan reproducibility as well as contro lling for extraction and ionization variability resulting from tissue makeup differences across the sample Scan to scan signal variation and back extrapolation to the x axis contribute significantly to the calculated error. Also, error in the calculated concentration of the spotted COC standards can affect the resulting quantitative COC determination. The calculated COC tissue concentration of standards relies on two assumptions: (1) the spotted COC is distributed evenly throughout the 20 m thick tissue section and (2) the diameter of the COC remains consistent for the calibration spots (minimal lateral diffusion). How evenly the spotted COC standard is distributed through the tissue has proven difficult to determine; however, shown in Figure 2 6 signal was detected in areas of the spotted COC standards indicating extraction. To estimate
59 this extractions, d 3 COC signal detected from the tissue was 1 2% of the detected d 3 COC signal from the slide alone. However, this reduction in ion signal may not corre late well to extraction efficiency due to ionization efficiency differences from the slide and tissue. Standards of COC could also be spotted atop the tissue; however, this also may not accurately reflect extraction efficiency of standards spotted beneath tissue as the spotted COC atop tissue may reside atop the tissue and not distribute evenly through the thickness of the tissue sample. Scan by scan normalization by a deuterated internal standard that has also been extracted through the tissue is the bes t method of signal normalization and quantitation, as the internal standard behaves similarly to the target analyte in terms of extraction, co crystallization, and ionization efficiency Concentration of COC was also measured by comparing the COC/d 3 COC s ignal ratio from the positive tissue to the spotted COC standards beneath a control section of tissue (external calibration). The resulting calibration plot with the MS/MS ion ratio of m/z 182/185 is shown in Figure 2 7 The spotted concentration of the applied QC fro m the positive sample was determined to be ~4000 ng COC /g tissue This concentration was determined from the area of the applied COC standard assuming even extraction into the 20 m thick tissue. Calculating the concentration of the QC sample from calibration curve generated from the control tissue by using the m/z 182/185 ion ratio averaged over the entire area of the QC COC spotted standard resulted in a calculated concentration of 4100 400 ng COC /g tissue indicating that the MS/MS ion signal ratios from the two tissues are comparable. This is important in quantitative MSI as different tissue composition, variations in tissue thickness, inconsistent matrix application, and
60 uneven internal standard concent rations can all affect the detected ion signal. Four areas from the positive sample were integrated and compared to the calibration curve generated from the control tissue. This resulted in an average concentration of ~400 50 ng COC /g tissue Error in c alculation of the spotted COC standard concentrations in tissue may also contribute to error in the determined COC concentration from the positive sample, as stated previously. The MSI results from the QTOF were compared to similar experiments performed on the LTQ. The same sample preparation protocol was followed and similar instrument settings were utilized, except that the isolation window on the LTQ was centered at m/z 305.5 3 AMU which allowed for trapping the [M+H] + ions of both COC and d 3 COC simultaneously An example of the resulting ratio MS/MS image of COC to d 3 COC ( m/z 182/185) is shown in Figure 2 8 A and the resulting standard additions plot in Figure 2 8 B which yielded a COC concentration of 550 1000 ng/g tissue The large error in th e 95% confidence interval, compared to Figure 2 8B is due to the high signal variation as well as the error in back extrapolation to the x axis. However, we report fairly good linearity (R 2 = 0.9853) from the spiked calibration standards. Analysis of thr ee serial sections by standard addition from the MSI experiments with the LTQ y ielded an average concentration of 880 300 ng/g tissue : A summary of the MSI analyses from both the LTQ and QTOF is shown in Table 2 1 and Table 2 2 respectively The MS/MS images from the LTQ were similar to those obtained with the QTOF ; however, a significantly large amount of error was observed in these measurements. The differences in error between the two instruments may be attributed primarily to the
61 ion source. MSI analysis with the QTOF was performed with a significantly more powerful solid state 200 Hz, frequency tripled Nd:YAG laser fired 200 times at each raster step. This resulted in nearly complete ablation of the 20 m thick sample, resulting in all the tissue ablated and analyzed as indicated by the visual ablation of tissue at each raster position. MSI analysis with the LTQ equipped with a 60 Hz, nitrogen laser on the LTQ and was fired only 8 times per raster step and, resulted in no visible damage to the tissue after MSI analysis and indicated a thin layer of sample was interrogated. More laser shots and energy could not be used on the LTQ for numerous reasons. Since the LTQ is an ion trap, space charging within the trap is always a limitation; howev er, previous experiments in our lab have shown this effect to be analysis. Another reason is the inability to sum spectra over hundreds of laser shots in a practical manner. Summing spectra with TOF instruments is routinely performed to increase ion signal and reduce background noise; however, the lower repetition rate of the LTQ laser combined w ith a slower scan speed of the ion trap inhibit this summing ability. The last and arguably the most limiting factor limiting the sample size consumption on the LTQ is the ablation of the MALDI matrix Previous analysis in our lab has shown that a limit exists on the number of analyse s which can be performed on a given tissue with matrix applied 66 As the matrix i s ablated away, ion signal is decreased signi ficantly over subsequent analyse s. More MALD I matrix can be applied and ion signal restored; however, this is not a practical experiment for tissue imaging. This effect was not observed with the QTOF instrument as the spectra were summed
62 from the 200 laser shots at each raster position and subsequ ent analysis was impossible due to the complete ablation of the sample LC and MALDI Quantitation From Tissue Extracts Sections of the nucleus accumbens (~100 mg) were dissected and extracted according the procedures previously outlined. The extracts were analyzed by both MALDI wide isolation MS/MS with scan by scan COC to d 3 COC normalization and LC MS/MS with a QqQ. MALDI has been previously employed successfully to quantify small molecules 41, 42, 78, 84 ; however, these experiments were not performed with scan by scan normalization of the analyte and internal standard signal. Employing wide isolation MS/MS for analysis of the tissue extracts with scan by scan COC to d 3 COC normalization on the LTQ resulted in a significant improvement in signal reproducibility for each calibration standard. The average relative standard deviation (RSD) of the 50 MS/MS scans for the sample spots analyzed was 45% for COC ( m/z 18 2) ion intensity. By normalizing the COC intensity of each MS/MS scan by the d 3 COC signal ( m/z 185), the average RSD was reduced to 17% across the spotted tissue extracts. We have previously shown this signal variation in both the COC and COC/d 3 COC ion s ignal is primarily due to matrix crystal differences across the area of the sample being analyzed shown in Chapter 4 Table 2 3 compares the analytical performance of MALDI wide isolation MS/MS to LC MS/MS of the same tissue extracts as well as the quan titative values determined via MSI. It is informative to compare MALDI quantitative results (MSI and tissue extracts) to LC MS/MS experiments with MRM from a QqQ, as this is a well characterized quantitative analytical method An example of the LC MS/MS chromatogram emplolying MRM is shown in Figure 2 9. The calibration curves produced form the two
63 ex periments are shown in Figure 2 10. A nearly ideal slope (~1.0) was observed from the LC MS/MS experiment, indicating a linear i nstrument response factor to increasing analyte concentration. The deviation of the slope from 1.0 with MALDI may be attributed to background signa l from the matrix or interfering compounds from the extracts, si nce no chromatographic method was used prior to MS analysis of the extracts. Another factor contributing to the non unity slope with MLADI may be the varying matrix to analyte ratio (M/A) of the analyte within the calibration standard concentrations. Varying the M/A ratio has been previously shown to have non linear effects on ion signal across a broad range of concentrations 34 This effect was evaluated by applying different amounts of matrix to the samples, as shown in Figure 2 11 The amount of matrix could not be precisely determined due to spray coating; however, the time for spray coating was doubled from 30 seconds to one minute and double again to two minutes to increase the amount of DHB applied. This resulted in a calibration curve diffe rences attributed to varying M/A of the calibration standards of COC. There appears to be an optimal range of the M/A ratio to produce adequate quantitative results. Standardizing the matrix application with automated matrix addition instruments (e.g., a nd inkjet printer or Bruker ImagePrep) would allow for a more reproducible matrix addition in a uniform manner. More importantly, using the best linear MALDI calibration curve, COC concentration from the tissue extracts with MALDI MS/MS were similar t o co ncentrations (within error) of COC from the brain samples determined by LC MS/MS Including spotting the samples (two spots of 1 L) and quadruplicate MALDI MS/MS analysis of each scan, the total experiment time was less than 1 hr. Comparing
64 this to LC MS/MS experiments, the LC gradient for each sample was 18 min with a total triplicate analysis time of ~13 hr including the analysis of solvent blanks to determine carryover. The total sample size for each MALDI analysis was 2 L of the tissue extract com pared to the 15 L injection volume for LC (triplicate analysis of 5 L each). There is also a significant reduction in solvent use for MALDI analysis since the LC separation step is avoided. Improvements in MALDI sensitivity for the tissue extract could also be gained by reducing the reconstitution volume of the tissue extracts as well as reducing the area of the spotted extract solution on the MALDI target, both leading to a greater concentration of the analyte per area for analysis by MALDI. The quan titative results from the MSI experiments are compared to the LC MS/MS results for the tissue extracts is also shown in Table 2 3 The quantitative MSI results compare favorably to analysis by LC specifically MSI quantitation by a standard addition method The erro r in quantitative MSI results were significantly larger, which is to be expected when performing direct tissue analysis with no sample cleanup. However, normalizing the wide isolation MS/MS to t he internal standard signal led to significant imp rovements in quantitative performance. These results show a significant improvement in quantitative MSI, which may lead to the use of the technique for relative or absolute quantitative determinations, particularly for whole body imaging studies Conclusions Of Quantitative MSI Here we have improved quantitative MSI by incorporating a uniformly applied internal standard combined with scan by scan COC and d 3 COC ion signal normalization. This approach overcomes the challenges associated with quantitative MALDI by accounting for significant ion signal variability. For MSI, employing an
65 internal standard applied beneath a tissue sample for MSI not only improv ed quantitation but may also identify instances of tissue specific ion suppression resulting from either poor analyte extraction or ionization. This may lead to conclusive MS specific organ or entire body without the need for further analysis. These techniques are applicable to numerous types of analysis, but may be most useful in the area of pharmaceutical development specifically for determining a new drug and its metabolit distribution in a whole body sample with the specificity advantages of MS. Further development of quantitative MSI would allow the technique to be used universally for physiological and pharmaceutical determinations. Employing an internal standard n ot only allowed for quantitative MSI, but quantitative MALDI analysis of tissue extracts as well. Analysis with MALDI MS/MS reduced the analysis time from 13 hr to 1 hr, a significant reductio n in analysis time per sample. In some instances, MALDI with w ide isolation MS/MS could replace LC MS/MS analysis with little reduction in analytical performance.
66 Figure 2 1 Structure of cocaine and the two main product ions observed in the MS/MS spectrum. The site of the deuterium label is circled and is preserved after fragmentation allowing for discrimination of the labeled versus unlabeled standard.
67 Figure 2 2 W ide isolation MS/MS spectra of COC and d 3 COC analyzed by the LTQ A) and QTOF ( B) instruments. The main product ions preserve the deuter ated label allowing for differentiation of the COC and d 3 COC in a single MS/MS spectrum. The ratio of m/z 182/185 was calculated at each image position and used for quantitation and MS images.
68 Figure 2 3 Optimization of LS with constant LE using wide isolation MS/MS. A) Intensity of COC and d 3 COC with varying laser shots. B) Ratio of COC/d 3 COC ( m/z 182/185) with varying LS With significantly changing signal, the detected A/I ratio is preserved. A B
69 Figure 2 4 Optimization of LE with 4 LS using wide isolation MS/MS. (A) Intensity of COC and d 3 COC with varying LE. (B) Ratio of COC/d 3 COC ( m/z 182/185) with varying LE. With significantly changing signal, the detected A/I ratio is preserved. A B
70 Figure 2 5 Experimental setup for quantitative imaging. Standards are pipette first and allowed to dry, followed by the uniform application of the internal standard via inkjet printing. Thin tissue sections are placed atop the standards followed by MALDI matrix.
71 Figure 2 6 Wide isolation MS/MS imaging experiment for the quantitative analysis of COC from nucleus accumbens. A, C) Ion intensity of m/z 182 and the ratio of m/z 182/185 (C) were used to generate MS/MS images. Calibration spots increase in concent ration from 200 (spot 5) to 2200 ng/g tissue (spot 1). B, D) The corresponding standard addition plots for quantitation with m/z 182 ion intensity and the ratio of m/z 182/185 are shown respectively. Normalization to the d 3 COC improved the linearity and er ror of the calibration plot, leading to improved quantitative estimation of COC in the tissue. Confidence bands are calculated from the 95% confidence interval and error bars represent the standard deviation of the mean
72 Figure 2 7 Normalized calibrat ion plot produced from COC spotted beneath a control section of brain tissue. Product ion signal ratio ( m/z 182/185) from the positive sample was compared to the curve to estimate the concentration of COC. The QC sample served as a means to ensure the si gnal between the two tissues was comparable. The calculated concentration of the QC sample from the calibration curve was within 5% of the spotted concentration. Error bars represent the standard error from the variation in the ~40 MS/MS spectra averaged for each calibration point
73 Figure 2 8. Product ion ratio image generated from wide isolation MS/MS (A) from MSI experiment performed on the LTQ. The corresponding standard additions plot is shown in (B). From this wide isolation MS/MS experiment, the concentration of COC was calculated to be 550 ng/g tissue
74 Figure 2 9. Sample LC extracted ion chromatogram of COC (A) and d 3 COC (B). Chromatograms were plotted from monitoring the transition of 304 182 for COC (C) and 307 185 for d 3 COC (D) by SIM. B A C D
75 Figure 2 10. Calibration curves of the tissue extracts used to quantify COC in the nucleus accumbens determined by LC MS/MS (A) and wide isolation MALDI MS/MS on the LTQ (B). Analysis by MALDI MS/MS resulted in comparable COC concentrations to that of th e LC MS/MS experiment A B
76 Figure 2 11 Calibration plots of the product ion signal ratio with different amounts of DHB applied to the tissue extracts analyzed by wide isolation MS/MS on the LTQ. Since DHB was spray coated, the amount of DHB applied was varied by time sprayed atop the spotted extracts. Non linear increases in the ratio of COC/d3COC are observed with a non ideal M/A ratio
77 Table 2 1 Summary of calibration plots and coca ine concentration determined by MALDI MSI analysis with the LTQ. Concentrations were determined by standard additions analysis Analysis Line equation (y=) R 2 Cocaine (ng/g tissue ) 95% CI 1 = (0.00027 0.00002)x +(0.29 0.05) 0.9853 550 1000 2 = (0.00055 0.00002)x +(0.30 0.17) 0.9979 1000 700 3 = (0.00027 0.00002)x +(0.16 0.01) 0.9973 1100 400
78 Table 2 2. Summary of calibration plots and cocaine concentration determined by MALDI MSI analysis with the QTOF Concentrations were determined by both standard additions and external calibration. Analysis Line equation (y=) R 2 Cocaine (ng/g tissue ) 95% CI 1 = (0.00019 0.00004)x + (0.071 0.017) 0.9993 370 300 2 = (0.00055 0.000 03 )x + (0. 096 0. 003 ) 0.99 31 170 150 3 = (0.00104 0.000 03 )x + (0. 194 0.0 34 ) 0.9973 19 0 1 00 External standard = (0.00015 0.000 01 )x + (0. 016 0.0 18 ) 0.9922 400 50
79 Table 2 3 Tissue COC concentrations determined by v arious analytical methods. Method Cocaine Concentration (ng/g tissue ) Std Deviation of Replicates (ng/g tissue ) LC MS/MS of Extracts 1 295 18 254 14 MALDI MS/MS of Extracts 1 315 34 266 26 QTOF Standard Additions 240 110 LTQ Standard Additions 880 30 0 External Calibration QTOF 415 5 0 1 Extracts consisted of two homogenized sections of the brain, analyzed in triplicate.
80 CHAPTER 3 USING AN INTERNAL ST ANDARD TO IDENTIFY T ISSUE SPECIFIC ION SIGNAL VARIATION IN MALDI MASS SPECTR OMETRIC IMAGING Introduction To Tissue Specific Ion Signal Variation The ability of mass spectrometric imaging (MSI) techniques to generate analyte specific distribution maps is becoming a useful tool in the identification of new pharmaceutical candidates. D irect tissue analysis by matrix assisted laser/ desorption (MALDI) coupled with mass spectrometry (MS) provides molecular composition analysis dir ectly from a tissue section. By employing multiple stages of MS analyte specificity is further enhanced and can allow improved drug and metabolite imaging capabilities. Earlier imaging techniques such as magnetic resonance spectroscopy imaging 87 (MRI) or positron emission tomography 62 (PET) can be used to image distribution throughout a tissue or even an entire animal. These techniques, however; lack the chemica l specificity to differentiate between the pa rent drug and its metabolites, since the labels attached to the drug may or may not be preserved through metabolism. This lack of chemical specificity necessitates the need for further characterization, usually by tissue extraction and chromatography coupled with MS n ultimately resulting in the nearly complete loss of spatial information. These methods can also be time consuming and cost prohibitive especially if the use of radioactive labels is necessary. M SI however, provides a label free, direct tissue analysis of the molecular makeup of a tissue sample which can be targeted (tandem MS imaging) or untargeted (full scan MS imaging). Both endogenous and exogenous compounds can be analyzed and compared betwe en tissue samples such as a comparing disease and normal states 8, 19, 72, 73, 88 92 or determining the distribution of pharmaceuticals. 71 73, 90
81 MALDI based MSI does, however, have limitations. Inherently, MALDI is a poor quantitative ionization source, although with an appropriate internal standard, adequate quant itative results can be obtained, as shown in Chapter 2. 41, 42, 47, 84 The poor quantitative ability is generally attributed to poor ion signal reproducibility resulting from heterogeneous matrix crystallization, sample variation, and laser energy fluctuation. 39 MALDI signal variability leads to a source of ion signal irreproducibility in MSI applications, thus limiting the quantitative ability of MSI. The MALDI matrix must be applied in a manner that minimizes crystal size while enhancing extraction and co crystallization of the target analyte from the tissue and into the matrix. To minimize crystal growth rapidly evaporating solvents may be used ; however; this can limit extraction and co crystallization. Aside from matrix crys tal variability, the wi de variety in the tissue types may also have major effect s on the detected ion signal, especially for whole body imaging applications. Variation between tissue types can lead to unaccountable differences in analyte extraction and ionization efficiency leading to differences in ion signal. 73 This effect was exemplified in the MSI analysis of a whole body sample for the distribution of administered olanzapine (OLZ) and its m etabolies 72 These experiments produce d remarkable images of the distribution of OLZ related compounds but it was impossible to discern whether increase s in MS signal directly corresponded to increases in conc entration without further analysis or what Stoeckli et tissue specific suppression effects in MSI. 73 Stoeckli et al. a ddress tissue de pend e nt ion signal variation in experiments by coating a solution containing the target compound over a whole body section using a pneumatic sprayer until the tissue sample was completely
82 wetted ( and the compound presumably incorporated into the tissue ) f ollowed by MSI. From the MSI data, ionization differences were determined between tissue types and these were used to normalize images from previous experimen ts. Areas of ion suppression wer e observed and accounted for by this approach ; however, the appl ication of the compound atop the tissue may not accurately reflect detecting the compound from within the tissue. These experiments d id provide an excellent exam ple of the ion suppression effect and m a y lead to protocols for applications of quantitative MSI. An alternative to identifying tissue specific ionization variability as described above would be to apply a uniform layer of a deuterated labeled compound, to be used as an internal standard, across the MALDI target prior to placing the tissue samp le on the target for MSI analysis. Utilizing an internal standard in this manner has le d to improved quantitative MALDI MSI on tissues and standards by ge nerating and analyzing analyte to internal standard ratio images. Applying the internal standard wit h an inkjet printer resulted in improvements both ion signal reproducibility and MS/MS images shown in Chapter 4 Two requirements for using an internal standard in MSI are that the internal standard must be applied uniformly and the analyte and internal standard are det ected in a single analytical MS n scan. This allows for scan by scan normalization of the resulting images and has been shown to improve quantitation and ion signal reproducibility (Chapter 2) Alternate MS/MS scan ap proaches have been shown to yield minimal improvements on ion signal reproducibility and typically do not lead to improvements in quantitation. 47 In order to detect both the analyte and internal standard in a single MS/MS scan a wide isolation or multi isolatio n approach must be
83 used. 47 Wide isolation MS/MS is a simpler approach typically requiring re centering and expansion of the mass isolation window of the ion trap or lowering the resolution of a quadrupole (for hybrid quadruple time of flight instruments) To utilize wide isolation MS/MS, the deuterated label attached to the analyte must remain in the product ion The other requirement for an internal standard is that internal standard must be as chemically similar as possible to the target analyte. In t his study, acetyl d 3 carnitine (d 3 AC) was utilized as an internal standard for the quantitative detection of acetyl l carnitine (AC) from piglet brain tissue and for the identification of tissue specific ionization differences. The structure of AC is show n in Figure 3 1 with the location of the deuterated label circled. The deuterated label is preserved after fragmentation allowing for wide isolation MS/MS. d 3 AC was first evaluated as an appropriate internal standard for improving MSI results when employ ing wide isolation MS/MS Optimal cutting temperature compound (OTC), a well known ion suppressant 93 was utilized to generate a controlled area of ion suppression. Multiple tissue types were also analyzed for th e detection of spike d AC to determine if analyzing the ratio of d 3 AC and AC could account for instances ion suppression. Materials and Methods Chemicals AC and d 3 AC were purchased from Sigma Aldrich (St. Louis, MO). Superfrost glass plus microscope slides and LC grade methanol, ethanol, water, and chloroform were purchased from Fisher Scientific (Pittsburgh, PA). Stainless steel microscope slides were purchased from Thermo Scientific (San Jose, CA). Optimal cutting temperatur e compound was provided by the EM Laboratory at the University of Florida. Working calibration standards of AC were diluted in 50/50 v/v methanol/water (v/v) at
84 concentrations of 0.10, 0.20, 0.040, 0.60, 0.80 and 1.0 ng/ L. d 3 AC was diluted in 50/50 meth anol/water (v/v) at a concentration of 10 ng/ L. An Epson Stylus R280 (Long Beach, CA) inkjet printer 86 was used to apply d 3 AC and AC to microscope slides. 2,5 dihydroxybenzoic acid (DHB) was purchased from Acros Organics (Geel, Belgium) and prepared at a concentration of 20 mg/mL in 90/10 chloroform/ethanol (v/v). Tissue Preparation lab from the Food Science and Human Nutrition Department at the University of Florida (Gainesville, FL) and stored at 80 C. During excision, the brain was dissected into hemispheres along the center sagittal plane and then along the coronal plane resul ting in half of a hemisphere dissected into the frontal lobe and parietal/occipital lobe. The second cut was performed to reduce tissue sample size all owing for simpler sectioning; the frontal lobe was used for analysis. Piglet brain tissue was utilized for the quantitative analysis of endogenous AC u sing a standard additions method. The internal standard, d 3 AC, was utilized to identify tissue specific ionization differences. Brain, liver, testes, kidney, and heart from Sprague Dawley rats were provide d by Dr. Dieter Drexler at Bristol Meyer Squibb ( Wallingford, CT, USA) ) and stored at 80 C. Transverse sections of each tissue were applied to microscope slides spotted with AC and d 3 AC. These tis sues were used to mimic a whole body imaging experiment. One section of each tissue was placed on one pre coated microscope slide and the detected AC and d 3 AC signa ls w ere compared. Sectioning of the brain tissue was performed on a Microm HM 505E cryostat (Waldorf, Germany). Tissue was sliced at 25 C to a thickness of 20 m and thaw
85 mounted onto the previously coated microscope slide with d 3 AC and stored at 80 C. To minimize condensation, the tissue and slide were first warmed to room temperature in a vacuum dessicator for ~30 minutes, then DHB was appl ied with a Meinhard nebulizer (Golden, CO) at a flow rate of 2.4 mL/min. Ion Suppression Studies with St andards and Tissue Prior to performing ion suppression studies, t he method of applying the internal standard was evaluated. Internal standards can be mixed with the MALDI matrix prior the matrix application step or they can be applied separately, either beneath (shown in Chapter 2) or atop a tissue sample To evaluate this, 1 ng/ L of AC was a pplied to a micro scope slide using an inkjet printer. One mL of 10 ng/ L d 3 AC was added to 10 mL of 20 mg/mL DHB. The mixture was then spray coated atop the applied AC. This was compared to a glass microscope slide on which both AC and d 3 AC were first applied separately with an inkjet printer, followed by the a pplication of DHB. Wide isolation MS/MS imaging centered at m/z 205.5 3 AMU was performed on 2 mm squares of the applied standards. Images were then generated using the product ion intensity of m/z 145 and m/z 148. Previous MS/MS studies using a 1 AMU isolation window across the m/z range 202 209 have shown that the product ions m/z 145 and m/z 148 almost exclusively arise from the precursor ions m/z 204 and m/z 207, corresponding to AC and d 3 AC, respectively. To determine if an internal standard could identify and account for areas of ion suppression, controlled experiment s mimicking areas o f ion suppression were created with a well known ion suppresser OCT. 93 First, 1 L of a 1% solution of OCT was spotted onto a microscope slide followed by the addition of a 1: 2 mixture of AC: d 3 AC
86 (v/v) by the inkjet printer. DHB was t hen applied via pneumatic spray coating and both MS and wide isolation MS/MS analyses were performed covering the entire OCT spot. For multiple tissue analysis, 1 L spots of a 1:2 mixture of AC: d 3 AC (v/v) was applied atop a microscope slide. Brain, liver, testes, kidney, and heart sec tions were each placed over one of the spots. One spot was left uncovered and analyzed as a standard to determine the detected m/z 145/148 ratio. DHB was spray coated over the entire slide and each tissue was analyzed via wide isolation MS/MS. Final ly, the distribution of AC was determined in a coronal section of piglet brain tissue. Calibration standards of AC were first applied to a glass slide (1 L) with a micropipette and allowed to dry. d 3 AC was then applied covering the entire slide via an inkjet printer. Coronal sections of piglet brain tissue were then sectioned to 20 m and thaw mounted atop the calibration standards. Instrumentation For the MSI experiments of AC in piglet brain tissue, all mass spectra were acquired with a Synapt MALDI Q Ion Mobility (T wave cell) Time of Flight ( QTOF ) equipped with a 200 Hz, 335 nm, Nd: YAG laser at a frequency of 200 Hz and spot diameter of 150 m (Waters Corporation). The QTOF instrument was operated in MS mode. All experiments evaluat ing ion suppression were performed on a MALDI LTQ XL equipped wi th a 20 Hz, 337 nm, N 2 laser with a spot size of 100 m (Thermo Corp. San Jose, CA). For MS/MS imaging experiments on the LTQ, a 6 AMU wide isolation window centered at m/z 205.5 The compounds were fragmented in both instruments using collision induced di ssociation and the fragments of each compounds were monitored; AC m/z 204 145 and d 3 AC m/z 207 148. Sample slides were inserted
87 into the vacuum chamber and rastered beneath the laser at a step size of 100 m. Resulting data from the QTOF were processed u sing BioMap (Novartis, Basel, Switzerland ) and images from the LTQ were processed with QualBrowser (Thermo Scientific San Jose, CA, USA ) and Surfer 8 (Golden, CO USA ). BioMap was utilized func tion built into the software. Normalized images from the LTQ were generated in Surfer 8 from extracted data. For quantitative determinations of endogenous AC, a standard additions method analysis was performed utilizing the signal from the spotted c alibration standards and the uniformly applied internal standard. For ion suppression determinations, the product ion intensities of m/z 145, m/z 148, and the ratio of m/z 145/148 were analyzed. Discussion Internal Standard A pplication Prior to evaluated tissue dependent ion signal variation, methods for the application of the internal standard d 3 AC were evaluated. It has previously been shown that scan by scan normalization with wide isolation MS/MS spectra of AC by d 3 AC resulted in significant improvement s in ion signal reproducibility, calibration cu rve linearity and image quality (Chapter 2 and 4). To evaluate the methods of applying d 3 AC for imaging studies the d 3 AC was applied separately and combined with the MALDI matrix. W ide isolation MS/MS was performed over a 10 X 10 raster step square (100 scans with a 200 m step size) on the applied s tandards. For normalized imag es, each MS/MS scan, the intensity of m/z 145 was divided by the intensity of m/z 148 for each MS/MS scan. The mean ce ntered resu lts are plotted in Figure 3 2 Figures 3 2 plot the ion intensity image generated from m/z 145, m/z 148, and m/z 145/148
88 when applying the d 3 AC mixed with the matrix. The average variation in product ion intensity (defined as the % relativ e standard deviation of the detected ion signal) was 57% for m/z 145, 54% for m/z 148, and 60% for m/z 145/148. The lack of improvement in image quality and signal variation indicates that applying the inter nal standard with the matrix does not lead to improved analytical performance. This may be due to differences in co crystallization between the AC an d d 3 AC with the DHB. Figures 3 2 m/z 145, m/z 148, and m/z 145/148 when applying the d 3 AC s epara te from the matrix and AC. This signal intensity patterns in Figure s 3 2 D and 3 2 E are very similar to each to other and the normalized image in Figure 3 2 F resulted in a relatively smooth ion product images. The % relative signal deviation (%RSD) in product ion intensity was 42% for m/z 145, 41% for m/z 148 and 8% for m/z 145/148, indicating the application of the internal standard separate from the matrix solution is superior. Both the AC and d 3 AC appear to co crystallize with the MALDI matrix in a similar manner indicated by the similar pattern observed in Figure 3 2 D and 3 2 E It is unlikely the exact pattern was replicated by the inkjet printer multiple times and is due more likely to differences in crystallization of the DHB. From these res ults, it was concluded that the internal standard be added separately from the MALDI matrix for all further experiments. Ion Suppression with Standards To determine if an internal standard could help identify and account for areas of ion suppression, ana lysis of AC and d 3 AC was performed using OCT as a n ion suppressant. MSI was performed over spots of applied OCT (1 mm across) AC and d 3 AC (3 mm across) in MS and wide isolation MS/MS mode. Images from the MSI experiments utilizing M S are shown in Figure 3 3 Three methods of image
89 normalization reported in the literature were compared, i ncluding normalizing the analyte signal to the matrix signal, the analyte signal to the total ion current (TIC), and the analyte signal to an externally applied in ternal standard. 77 These results, compare images produced using the ion intensity of the compound of interest ( Figure 3 3 to different methods of normalization ( Figure 3 3 The area of the spotted OCT is clearly visible from plotting m/z 332, which corresponds to the additive benzyl alkonium chloride ( Figure 3 3 A ). Areas of ion suppression resulting from OCT are sho wn in the MS images of AC and d 3 AC. Both AC and d 3 AC (Figure 3 3B and 3 3C, respect ively) were applied atop the slide homogeneously with an inkjet printer, which should lead to uniform signal distribution throughout the area analyzed; however, the presence of OCT results in drastic signal reduction. Without prior knowledge of ion suppre ssion, one may conclude an uneven distribution of AC using the ion intensity alone. These areas of ion suppression are not fully identified when normalizing AC signal to the TIC ( Figure 3 3 D) or to the matrix ( Figure 3 3 E). These normalization methods still produce images showing uneven AC signal distribution although normalization by the TIC is more successful than normalization by the matrix Normalizing to the d 3 AC, however, does result in an image showing nearly uniform distribution of AC ; the dem onstrates that normalizing to the internal standard (which has nearly identical chemical properties as the target analyte ) can successfully correct for ion suppression Tissue Specific Ion Signal Variability To test whether normalizing images by an inter nal standard can be used with sectioned an d applied atop a pipetted spot of a 1:2 mixture of AC and d 3 AC. Shown in Fig 3 4 A is a photograph of the slide with tissues o utlined and identified. A black spot
90 was applied to the backside of the slide to mark where the spots of AC/d 3 AC were applied For analysis by wide isolation MSI, a rectangle was drawn surround the applied AC/d 3 m step size The resulting data w ere extracted into Excel for ratio determination and Surfer was used to produce the MS/MS images. MS/MS images of AC from the spots applied beneath the brain and liver sections are shown in Figure 3 4 B and 3 4 D, respectively. T he ion intensity of m/z 145 shows a 90% reduction in the liver section compared to the brain section. This signal reduction could result from either poor extraction or ion suppression from an endogenous compound in the liver tissue; however, we were unabl e to make a definite conclusion on which effect was dominant. A summary of the analysis of all the tissues is shown in Figure 3 5 which plots the summed intensity covering the entire AC/d 3 AC spot. The histogram has been scaled to the maximum signal or ra tio detected across the tissue types for basis of comparison. Tissue specific signal attenuation is observed significantly in both the liver and kidney and as well as slightly in the testes. As state d previously, the c ause of the signal decrease cannot b e determined but analyzing the data using only the ion in tensity would lead to a n incorrect conclusion that there is a l ower concentration of AC in these tissues. After scan by scan normalization with the d 3 AC internal standard the detected AC/d 3 AC rati os from tissue to tissue varied 13% RSD Clearly, using an isotopically labeled internal standard is critical for a comparison of the endogenous concentration of AC between the tissue types Apart from creating known instances of ion suppression or tis variation, this phenomenon was also observed in experiments designed to quantify AC
91 in a coronal section of piglet brain tissue. These data were taken using a QTOF instrument Figure 3 6 A plots the MS intensity of m/z 204 corres ponding to the molecular ion of AC. A clear d ifference between the signal in white and grey matter is observed (lower signal was observed in the white matter) Calibration spots were applied beneath the tissue section and are als o visible in the MS image Figure 3 6 B is the MS image of the d 3 AC internal standard, which was uniformly applied beneath the tissue by an inkjet printer. A distinct reduction in signal is also observed in white compared to grey matter, even though the d 3 AC was applied uniformly observed in Figure 3 6B. This suggests the differe nce in the AC signal may not reflect differences in the AC concentration, but rather variations in e ither extraction or ionization efficiency Analyzing the ratio of AC/d 3 AC (Figure 3 6C) gives a better indication of the distribution of AC between grey and white matter in this sample by accounting for spot to spot signal variation, ionization variability and extraction differences shown in Figure 3 6 C Furthermore, comparison of the calibration curves generated from the MSI experiment show a dramatic improvement in linearity when performing scan by scan normalization to the deuterated internal standard Figure 3 7 (intensity of m/z 204 ) and Figure 3 8 (ratio of m/z 204 / 207 ) A dramatic improvement in c alibration curve linearity and precision is observed from the m/z 204/207 ratio plot compared to the plot generated from the intensity of m/z 204 alone. Although it is not possible to determine which factor, extraction or ionization, is the dominant source of signal variation, it could well be a combination of the two. F or protein analysis, multiple washing steps are routinely performed to remove the majority of the lipids prior to MSI because they tend to suppress protein ion signal s Such
92 washes a re not typically performed for small molecule MSI due to the solubility of small molecules in washing solutions, which may result in analyte migration or loss Differences in the amount of lipids and salts in various tissue type s as well as within a singl e tissue may also create significant differences in the extractability of a pa rticular compound. T he extraction can be optimizing for each tissue type; however, within certain tissue s such as brain, there may be regional variations, as between grey and wh ite matter A single tissue type may vary significantly on the millimeter scale and applying this across multiple tissue type s, these effects can be exacerbated. Applying an internal standard, preferably a deuterated analog, may help control for these ex traction and ionization issues which limit otherwise limit the potential for quantitative MSI Conclusions To Ion Specific Ion Suppression These results demonstrate that performing an MSI experiment without an internal standard may lead to inaccurate co nclusions about a particular compound s distribution within a tissue and more so across multiple tissue types i.e. in a whole body MSI experiment. Applying an internal standard corrected for tissue specific signal variability resulting from either ion s uppression or analyte extraction The method s developed overcome these tissue specific variations leading to more conclusive results about concentration differences in a given sample. The ability to quantitatively map a drugs time resolved distribution t hroughout an organ 65 or entire animal could provide made about metabolites distribution and possibly changes these compounds are inducing on the endogenous compounds present in the sample, i.e. metabolites, l ipids, and proteins.
93 Figur e 3 1 Structure of AC and the three main product ions observed in the MS/MS spectrum. The site of the deuterium label is circled and is preserved on the 145 product ion after fragmentation allowing for discrimination of the labeled versus unlabeled standard.
94 Figure 3 2 MS/MS images comparing the application of the internal standard d 3 AC. d 3 AC was mixed with the MALDI matrix solution and spray coated atop AC, which was applied with an inkjet printer (A C). This was compared to applying AC and d 3 AC separately (D F) with the inkjet printer, then spray coating the MALDI matrix solution atop. Images were generated from the intensity of the product ions m/z 145 and m/z 148 (A, B, D, E) compared to normalizing the intensi ty of AC by d 3 AC at each raster step. The intensities are mean centered. Thus, the average intensity is 0, and the intensities vary from 1 (zero absolute intensity) to +1 (2 times the average intensity).
95 Figure 3 3 MS images of AC mixed with d 3 AC (each ~3 mm across) applied atop a spot of OCT (~1 mm across) (A). Ion suppression is visible in MS images for AC and d 3 AC (B,C). Multiple methods of normalization are compared in D F. The MS image for AC should appear uniform resulting for the uniform application. This is produced only by normalizing to the internal standard The color scales for A C are determined from the plotted ion intensity values. The color scales for D F are determined from the ratio of the m/z 204 to the respective value.
96 Figure 3 4 Multiple tissue analysis of AC ion signal variability using wide isolation MS/MS. (A) Optical image of the glass slide with multiple tissue sections applied over spots of 1:2 mixture of AC:d 3 AC ( marked with black spots). (B) MS/MS image of product ion for AC applied under the brain section compared to MS/MS image from the liver tissue. The detected signal for AC was d ecreased by 90% from brain to liver tissue. The intensity scale reflects absolute counts of m/z 145.
97 Figure 3 5 Scaled MS/MS intensities of m/z 145 and m/z 148 summed over the raster area. Data were scaled as a percentage of the maximum detected for basis of comparison. A significant decrease in ion signal is observed for both the AC and d 3 AC in the kidney and liver sections compared to the brain, heart, and testes. However, the ratio of AC to d 3 AC remains constant, varying only 13% RSD between the 5 tissue types, whereas the variation of AC and d 3 AC 70% RSD between tissue types.
98 Figure 3 6 MS image of piglet brain tissue for th e detection of endogenous AC. A) Image of the [M+H] + ion corresponding to AC ( m/z 204) shows a clear differentiation between white and grey matter of the brain. B) The applied d 3 AC shows a similar pattern implying there is e ither decreased extraction or ionization from the two tissue types. C) The ratio image of AC/d 3 AC results in a better representation of the distribution of AC within the brain sample as well as improves the linearity of calibration curve Figure 3 7 and 3 8
99 Figure 3 7 Calibration curve of spotted AC standard plotted from the intensity of m/z 204 correspondi ng to the MS image in Figure 3 6 (A).
100 Figure 3 8 Calibration curve of spotted AC standard plotted from the ratio of m/z 204 /207 corresponding to the MS image shown in Figure 3 6 B
101 CHAPTER 4 QUANTITATIVE TANDEM MASS SPECTROMETRIC IMAGING OF ENDOGENOUS ACETYL L CARNITINE FROM PIGLET BRAIN TISSUE U SING AN INTERNAL STANDARD Introduction to Quantitative MSI for Tissue To Tissue Comparison Matrix assisted laser desorption / ionization mass spectrometric imaging (MALDI MSI) has become an in valuable tool for evaluating molecular changes of tissue samples, especially for d ifferentiation between tissue states. Ever s ince MALDI MSI was first used to map pharmaceutical compounds in intact tissues 13 and to generate MS images of proteins 8 MSI has rapidly expanded to include small molecu les, lipids 20 pharmaceuticals 94 and recently endogenous metabolites. 68, 95 Although instrumenta l advancements allow for increased selectivity of a target analyte by implementing MS n 20, 68, 96, 97 ion signal ir reproducibility limits MALDI MSI to qualitative applications Currently, MALDI MSI has be en used for relative quantitative applications 66, 73, 75 ; however, these results typically must be verified by another method such as high pressure liquid chromatography ( LC ) MS n of tissue extracts. 26 Ion signal ir reproducibility results from numerous factors : inhomogeneous matrix crystallization and tissue sample heterogeneity leading to inefficient analyte extractio n, ionization, and matrix application 25, 28, 76 Applying the MALDI matrix to a tissue sample is an important step in producing quality images. Significant care must be taken to apply the matrix in an even layer of minute crystals. Currently, matrix is applied using numerous methods including : ro botic micro spotting 22 sublimation 25 inkjet printing 76 and pneumatic spray coating. 8, 19, 98 Spray coating used in this study, generally produces a more homogenous matrix layer in a shorter amount of time compared to robotic spotting; however, this method may
102 result in widely varying matrix crystal sizes as well as the possibility of over wetting the tissue, thus resulting in analyte migration 98, 99 Crystal sizes of the matrix layer can also significantly vary based on their rate of formation. Spray coating was used with different solvents to create matrix crystals varying in size to determine their effects on signal reproducibility. Improving the quality of the matrix coating has recently been shown to minimize signal variation which could lead to improved quantitative MSI 27 Other quanti tative methodologies such as L C MS n employ an internal standard (IS) to correc t for signal variability arising from sample extraction, separation, and detection. Implementing an IS in MSI presents the unique challenge of applying the internal standard. A previous study utilized a pneumatic nebulizer to apply the IS by completely w etting the tissue sample then allowing it to dry prior to MSI. 73 Applying an internal standard in this manner may result in undesirable analyte migration as well as possible partitioning of the IS into different tissue types. An alternative to applyi ng an IS atop the tissue sample is to apply a uniform layer of the IS beneath the sample (to the sample plate) and extract the IS into the tissue during ma trix application. We utilize an inkjet printer to apply an i sotopically labeled IS to a glass slide prior to mounting a tissue section to account for the significant signal variability with MALDI. Utilizing an inkjet printer has been successfully used to apply an internal lipid standard to estimate the lipid conce ntrati on in rat spinal cord with full scan MSI. 66 76 In this study, quantitative data were determined by averaging the signal from a large area within the tissue. By averaging ion intensities from large sections of tissue, the resolution of the MS image is lost. To generate quantitative images at a resolution near the laser spot
103 size, improvements in the signal reproducibility from spot to spot must be achieved in order to produce quantitative MS images. In the present study, two methods were employed to improve signal reproducibility: 1) optimizing matrix application via spray coating with different matrix solvents and 2) utilizing an IS placed beneath the sample to account for spot to spot signal variation. An intermediate pressure MALDI linear ion trap utilizing wide isolation MS/MS was employed for the detection of acetyl l carnitine (AC) from piglet brain tissue using acetyl d 3 carntine (d 3 AC) as the IS. Wide isolation MS/MS allows for simultaneous detection of both the analyte and IS in a single MS n scan, and it has been previously shown to be superior to using alternate scans or multiple imaging experiments 47 The abil ity to detect both the analyte and IS in a single MS/MS scan allows for the ratio of the two analytes to be analyzed and not the signal intensities of the target analyte. Upon generating a successful method of quantitation, the endogenous levels of AC we re compared between two piglet brain samples. One piglet was fed a ketogenic diet during the first week after birth compared to a piglet fed a normal, carbohydrate rich diet during the same time. Changes in carnitine concentrations were hypothesized to o ccur as a result of the diet. By identifying endogenous metabolite concentration differences, further analysis were to be performed by our collaborator Dr. Peggy Borum of the Food and Nutritional Sciences Department at the University of Florida, to determ ine if these compounds were responsible for the therapeutic mechanism of the ketogenic diet for the treatment of epilepsy.
104 Materials and Methods Chemicals AC and d 3 AC were purchased from Sigma Aldrich (St. Louis, MO). S uperfrost glass plus microscope sl ides and LC grade methanol, ethanol, water, and chloroform were purchased from Fisher Scientific (Pittsburgh, PA). Working standards of AC were diluted in 50/50 v/v methanol/water (v/v) at concentrations of 10, 8 .0 4 .0 2 .0 1 .0 0.8 0 0.4 0 0.2 0 and 0.1 0 ng/ L. d 3 AC was diluted in 90/10 methanol/water (v/v) at a concentration of 10 ng/ L. 2,5 dihydroxybenzoic acid (DHB) was purchased from Acros Organics (Geel, Belgium) and prepared at a concentration of 40 mg/mL in either 90/10 chloroform/ethanol (v/v) or 70/30 methanol/water (v/v) Analysis of AC Standards Prior to tissue MSI analysis, the viability of using d 3 AC as an internal standard for quantitative MSI analysis was performed. To determine if using the AC/d 3 AC ion ratio to account for signal var iation, AC and d 3 AC were applied separately to a glass slide using the inkjet printer. DHB was then spray coated with a Meinhard nebulizer model number TR 30 A3, (Golden, CO) dissolved in 90/10 chloroform/ethanol (v/v) or 70/30 methanol/water (v/v) to p roduce matrix crystals of varying sizes from the differing evaporation rates of the solvents. Wide isolation MS/MS was performed over the areas of the applied AC and d 3 AC and the resulting images were evaluated in terms of ion signal variation. Crystal s izes were analyzed using a Leica DFC 295 (camera) coupled with a Leica DMLP microscope (Wetzlar, Germany). Matrix homogeneity was measured using percent relative standard deviation (RSD) of the detected matrix ion signal from an MS imaging experiment on a glass slide. A selected area of 50 MS/MS scans was analyzed 8 times monitor ing the signal variation over multiple analysis.
105 For quantitative analysis, calibration standards of AC were spotted onto a glass slide at concentrations of: 0.1, 0.2, 0.4, 0.8, 1, 4, 8 ng/ L followed by the application of d 3 AC via the inkjet printer. DHB diluted in 90/10 chloroform/ethanol (v/v) was applied by spray coating with a Meinhard nebulizer. Wide isolation MS/MS was performed covering the entire calibration curve area. MS/MS intensity from m/z 145 and the ratio of m/z 145/148 were used to produce calibration plots. Tissue Preparation from the Food Science and Human Nutrition Department at the University of Florida (Gainesville, FL, USA) and stored at 80 C. During excision, the brain was dissected into hemispheres along the center sagittal plane and then along the coronal plane resu lting in half of a hemisphere dissected into the frontal lobe and parietal/occipital lobe. The second cut was performed to reduce tissue size for simpler sectioning; the frontal lobe was used for method development Prior to tissue sectioning, d 3 AC was ap plied to a standard glass microscope slide with an Epson Stylus Photo R220 (Long Beach, CA) inkjet printer and a modified compact disc tray as a glass slide holder d 3 AC diluted in 50/50 methanol/water (v/v) to a concentration of 10 ng/ L was used to fill a clean ed, unused ink cartridge. Adobe Illustrator (San Jose, CA) was used to created the IS patterns to be applied to the glass slide. The glass slides where then inserted into the printer and the d 3 AC was deposited covering the entire glass slide. Th is procedure was repeated 3 times to ensure adequate and uniform coverage. For quantitative MSI, c alibration standards (1 L) of AC were micropipetted onto the glass slide at concentrations of 0.10, 0.20, 0.40, 0.80,
106 1.0, 2.0, 4.0 ng/ L. Each spot was mea sured to be 2.5 mm in diameter which was verified using the detected ion signal and raster step size of the MSI experiments For tissue to tissue comparison experiments, calibration standards were spotted at concentrations of: 0.1, 0.2, 0.4, 0.6, 0.8, 1 ng/ L beneath each tissue, followed by the application of d 3 AC with the inkjet printer. These calibration standards were used to produce standard additions plots to estimate the endogenous concentration of AC. Sectioning of the brain tissue was performed on a Microm HM 505E cryostat (Waldorf, Germany). Tissue was sliced at 25 C to a thickness of 20 m and thaw mounted onto the glass microscope slide previously coated with d 3 AC and/or AC standards and stored at 80 C. To minimize condensation the tis sue was first warmed to room temperature in a vacuum dessicator for ~30 minutes then DHB diluted to 40 mg/mL in 90/10 ethanol/chloroform (v/v) was applied with a Meinhard nebulizer model number TR 30 A3 at a flow rate of 2.4 mL/min Mass Spectrometry A ll MS images and spectra except for the tissue to tissue comparison, were acquired with a MALDI LTQ XL (ThermoFinn i gan, San Jose, CA) equipped with a 337 nm nitr ogen laser at a frequency of 6 0 Hz and spot diameter of 100 m. F or full scan MS imaging experiments, spectra were collected using four laser shots at a laser energy of 2 J. All MS/MS experiments utilized a 6 amu wide isolation window centered at m/z 205.5 unless otherwise noted. MS/MS i maging experiments were acquir ed with four laser shots with a laser energy of 16 J to achieve optimal ion signal. At this laser energy and number of laser shots space charging was evident in the full MS spectrum; however, space charging was not observed using wide isolation MS/MS
107 an d, thus, not a concern Furthermore, we were unable to observe space charging effects at higher laser energies (40 J) combined with a higher number of laser shots (50) durin g wide isolation MS/MS mode. Sample slides were inserted into the intermediate va cuum chamber (0. 0 7 Torr) and rastered with respect to the laser at varying step sizes depending on the experiment (100 300 m). Total and extracted ion signal were analyzed with Qualitative Browser from Thermo Fisher Scientific A M ALDI data extract ion program from Thermo Fisher Scientific was used to extract the ion intensity at each corresponding MS/MS scan. Extracted data was imported into Microsoft Excel to produce analyte to internal standard ion intensity ratios at each position. All images were created using Surfer 8 (Golden, CO) by gridding the extracted data using triangulation with linear interpolation. From the improved quantitative results shown in Chapter 3, t he tissue to tissue comparison studies were analyzed with a Synapt MALDI Q Ion Mo bility (T wave cell) Time of Flight ( QTOF ) equipped with a 200 Hz, 355 nm, Nd:YAG laser at a frequency of 200 Hz and spot diameter of 150 m (Waters Corporation; Milford, MA, USA The QTOF instrument was operated in MS mode Sample slides were inserted int o the vacuum chamber and rastered beneath the laser at a step size of 250 m and MS spectra were summed from 200 laser shots at each raster step. Resulting MS data from the QTOF were processed using BioMap (Novartis, Basel, Switzerland) Normalized image s were generated using the divide function in BioMap to generate AC /d 3 AC ( m/z 204/207 ) images. For quantitative determinations of AC in treated and
108 untreated tissue a standard additions analysis was performed utilizing the product ion ratio of AC /d 3 AC fr om the spotted AC calibration standards beneath the tissues Results and Discussion MS/MS C haracterization Prior to imaging experiments, the [M+H] + of AC ( m/z 204) and d 3 AC ( m/z 207) standards were characterized by MS/MS. Collision induced dissociation of the [M+H] + ion of AC produced three main product ions at m/z 145, 85, 60 and d 3 AC produced product ions at m/z 148, 85, 60 as shown previously in Chapter 3, Figure 3 1 Product ions m/z 145 and 148 from AC and d 3 AC respe ctively, resulted from the loss of the trimethyl amine group and were monitored for quantitative analysis. Narrow isolation experiments were performed to ensure neither the AC n or d 3 AC produced product ions at m/z 148 and m/z 145 respectively. It is imp ortant to note that for wide isolation MS/MS spectra, the observed product ions m/z 145 and 148 are generated from the range of m/z 202.5 208.5. Pre liminary studies performed on tissue utilizing a 1 amu narrow isolation MS/MS experiment at each nominal mass from m/z 202 209 showed no significant contribution of m/z 145 and 148 signal from endogenous compoun ds within the isolation range The isotopic purity of the d 3 AC was stated to be > 98% from the manufacturer, which was verified by comparing the sign al intensity of the product ions at m/z 145 to that at m/z 148 when analyzing d 3 AC applied to a glass slide with wide isolation MS/MS. Three layers of 10 ng/ L d 3 AC were applied using the inkjet printer and then coated with DHB diluted to 40 mg/mL in 90/10 chloroform/ethanol ( v/v ) Analyzing 50 MS/MS scans of the d 3 AC, t he contribution from the impure isotope to the product ion signal at m/z 145 was determine d to be < 1.2% 0.6%
109 Matrix Application for Solvent Comparisons To determine matrix application effects, standards of AC and d 3 AC were applied to a glass slide with an inkjet printer and analyzed using wide isolation MS/MS. One inch squares AC of were applied atop two previously d 3 AC coated glass microscope slides. DHB crystals that were formed from the two solvent mixtures 90/10 chloroform/ethanol (v/v) and 70/30 methanol/water (v/v) are compared in Figure 4 1. The chloroform/ethanol solvent mi xture resulted in a more uniform matrix coating with significantly smaller crystal sizes In Figure 4 1A the water/methanol solvent produced crystals of varying sizes that approached upwards of 250 m in length, whereas the chloroform/ethanol solvent mixt ure resulted in crystal size s consistently smaller than 60 m The smaller crystal sizes in the chloroform/ethanol solvent can be attributed to the rapid evaporation of the matrix solvent, Figure 4 1B MSI was used to analyze matrix uniformity between th e solvent systems by selecting an area of each DHB coated glass slide and comparing the percent RSD of 500 MS scans of signal intensity of m/z 137 [ DHB H 2 O] + As reported in Table 4 1, the r eproducibility of the matrix ion signal was improved from 66% (methanol/water) to 31% (chloroform/ethanol) utilizing a more volatile solvent that enhances rapid crystal formation The improved signal reproducibility of the matrix ion signal reflects a more uniform coating of smaller matrix crystals resulting from ra pid evaporation of the solvent. Conversely, t he observed increase percent RSD of the signal from methanol/water is attributed to larger crystal formation r esulting in significant sample differences when probed on a 100 m scale.
110 To determine if improved m atrix signal reproducibility correlates to improved analyte signal reproducibility, wide isolation MS/MS experiments were performed over areas of the slide s w here AC and d 3 AC were applied Within each 1 inch square, 50 MS/MS scans were acquired and the pe rcent RSD of the product ion signal of m/z 145, 148 and the ratio of m/z 145/148 were calculated and reported in Table 4 1. T he improved matrix coverage from chloroform/etha nol mixture resulted in better ion signal reproducibility for the product ions of both AC and d 3 AC The improved matrix coverage produced a more homogenous sample when analyzed on a 100 m scale. Matrix c rystal s that were sign ificantly larger than the laser spot diameter result ed in irreproducible signal resulting from the large variation in crystallization at each raster step. This larger variation of the ion signal intensity across a sample may only be due to the differences in matrix crystallization. Laser energy fluctuation may also cause signal variation; however, th ( 4% ) is too low to account for the observed signal variability. Furthermore, the percent RSD for both AC and d 3 AC are comparable, which indicat es that d 3 AC is acting as an adequate IS To further account for signal variability normalizing the AC by the d 3 AC intensity on a scan by scan basis resulted in a four to five fold improvement in percent RSD. To further determine if d 3 AC is behaving as an adequate IS wide i solation MSI experiments of the AC standard were performed on the chloroform/ethanol coated slide. Within the 1 inch square of AC, an 1 1 1 1 step raster pattern was created using a 100 m step size. After wide isolation MSI, separate MS images were generated using the normalized intensities of the product ions m/z 145 (AC), 148 (d 3 AC), and 145/148 versus scan position, shown in Figure 4 2. To produce the normalized images on t he
111 same color scale, the mean ion intensity or ratio was calculated for each image. At each raster step, the percent difference from the average intensity was calculated for the detected ion signal, which produced images showing the relative variation in signal. The images produced from the ion i ntensity of AC and d 3 AC ( Figure s 4 2A and 4 2 B) show similar ion intensity distribution s indicating that d 3 AC is co crystallizing and ionizing mutually with AC This effect is further exhibited in Figure 4 2C where the ratio of the two ions exhibits a un iform distribution across the image, suggesting t he signal variability observed in Figure s 4 2A and 4 2B is attributed to differences in matrix crystallization and not the application of the standards by the inkjet printer. The viability of the IS is furt her exemplified in Figure 4 3, showing normalized intensities of a line scan of 11 spectra across the image indicated in Figure 4 2A For each scan, the intensities of both the AC and d 3 AC vary similarly; however, the ratio of the two ions is conserved resulting in improved signal reproducibility. The statistical range of ion intensities may also be important for determining concentration differences on an area scale approaching that of the la ser diameter. T he range of absolute signal intensities for the product ions of AC and d 3 AC (AC = 52 00 17 0000 ; d 3 AC = 62 00 26 0000 ) were separated by a factor of 34 and 43 respectively; however the range of the ratio (0.57 0.97) was reduced to a factor of 1.7 Clearly utilizing an internal standard to control for ion signal variability may be necessary for quantitative MALDI MSI application. As the data in Figures 4 2 and 4 3 show quantitative MSI from tissue, where matrix crystallization differences ar e compounded by sample heterogeneit y, spot to spot compariso ns of intensities alone may be
112 statistically irre levant; in contrast, the use of an internal standard to correct for sample heterogeneity and matrix crystallization variability can result in infor mative images. The signal variability over multiple analyses was also determined by reanalyzing a given raster area eight times. The variation in AC and d 3 AC signal is plotted in Figure 4 4. As the matrix is ablated away, thus vary ing the M/A ratio, the detected ion signal varies non linearly as the matrix is ablated away A l though the absolute ion signal varies dramatically, the detected AC/d 3 AC ratio does not vary significantly as illustrated by the relatively flat plot. This suggests the internal sta ndard may also account for differing M/A ratios in a given imaging experiment and can be especially important for imaging a particular compound that may vary significantly in concentration in a given sample. Calibration Curve Experiments For calibration linearity comparisons; 1 L of AC at concentrations of 0.1 8 ng/ L were spo tted using a micropipette onto a glass microscope slide previously coated with d 3 AC (3 layers, 10 ng/ L) as shown in Figure 4 5. DHB diluted in chloroform/ethanol was then applied and each spot was analyzed using wide isolation MS/MS. W ide isolation MS/MS signal intensity for m/z 145 (AC) and the m/z 145/148 (AC/d 3 AC) ratio from 35 scans within each calibration spot were averaged and used to cal culate two calibration curves. Signal variability of the ion signal for m/z 145, 148 and 145/148 within each calibration spot is listed in Table 4 2. Shown in Figure 4 6A, the signal intensity of m/z 145 versus concentration resulted in l ine of best fit of y (counts 145 ) = (12 700 25 00 ) ( g/mL) + (12 800 86 00 ) with an R 2 = 0.8332. The curve plotted using the ratio of m/z 145/148 Figure 4 6B, resulted in a line of best fit of y (ratio 145/148 )
113 = 0.4 70 (0.018) ( g/mL) 0.06 5 (0.06 1 ) with an R 2 = 0.9929. Linearity and scan to scan reproducibility were significantly improved when normalizing by the internal standard The average percent RSD from the calibration standards using the m/z 145 ion intensity was 84% with a range of 56 122 c ompared to averaging the percent RSD of the ratio of m/z 145/148, which had a percent RSD of 22% with a range of 9.2 45.1%. N o correlation was observed between the percent RSD and an increase in concentration monitoring the intensity of m/z 145; whereas t he percent RSD of the averaged ratio improved with an increase in concentration of AC. Utilizing a uniformly applied IS coupled with wide isolation MS/MS showed considerable improveme nt in precision using a scan by scan analysis. Imaging of Piglet Brain Tissue w ith Spotted Standards Three layers of 10 ng/ L d 3 AC followed by spotting 1 L of calibration standards (0.1 8 ng/ L ) of AC in 50% methanol were applied to a glass slide. A 20 m thick tissue section was placed over the calibration standards and t he entire slide was coated with DHB diluted in chloroform/ethanol. The entire tissue was analyzed using a 6 amu wide isolation window centered at m/z 205.5. The imaging experiment was performed using four laser shots with a raster step size of 200 m an d the laser energy was set at 14 J to ensure optimum MS/MS signal without space charging the ion trap. The estimated concentration (moles/kg tissue ) of the spotted standards in estimated weight of the wet tissue was calculated using the area of the spotted standards incorporated into a 20 m thick tissue section For the experiments discussed in this paper, significant care was taken in the spotting of the AC standards to ensure reproducible droplets measuring ~ 2.5 mm in diameter. Upon extraction through a 20
114 m section of tissue, a 1 L spot of 1 ng/ L results in a tissue concentration 0.02 mol/kg tissue assuming the density of brain tissue is nearly that of water 1.05 g/mL. 100 In Figure 4 7 A C, Surfer 8 was used to produce images of the AC (Figure 4 7 A ) and d 3 AC (Figure 4 7 B ) using the intensity of m/z 145 and 148 respectively as well as an image of the ratio of the two ions, m/z 145/148 (Figure 4 7 C ) in each sca n. Areas wh ere standards were applied are circled and increase in concentration from 0.1 to 4 ng/ L starting with the lower left spot moving clockwise around the tissue Signal intensity is observed in three main locations from the image of m/z 145: 1) the area where no tissue is present; 2) the area of spike d AC standards and 3) the area on tissue where no AC had been added The detected m/z 145 signal on the glass slide is due to the isotopic impurity of AC in the d 3 AC ( > 98% isotopic purity) standard used to coat the entire slide. The signal ratio of AC to the d 3 AC in this area was calculated to be ~1.2% 0.2% Assuming the same contribution for the tissue area, the signal contribution would be insignificant as the average signal intensity from d 3 A C is ~500 counts. Averaged signal from ~20 scans within the area of the spiked AC standards was used to produce two calibration curves from 40 200 nmol/kg tissue using the intensity of m/z 145 and resulted in a curve with the equation: y (counts 145 ) = 7. 70 ( 0.74) ( g/mL) 113. 0 ( 85.9); R 2 = 0.982, and from the ratio of m/z 145/148: y (ratio 145/148 ) = 0.00972 ( 0.00029) ( g/mL) 0.116 ( 0.033); R 2 = 0.998. The improved R 2 value for the calibration curve using the ratio of 145/148 may be attributed to the d 3 AC extraction and co crystallization in a similar manner to AC as well as improved matrix coating crystallization producing a more reproducible signal from each of the spotted standards
115 The detected m/z 145 signal from areas within the tissue where AC was not spiked is thought to arise from endogenous AC with minimal or no contribution from the impure d 3 AC This is further supported by t he image in Figure 4 7 C, which shows the m/z 145 signal on the glass slide to be reduced to nearly z ero due to the comparatively low ratio of AC to d 3 AC present. Thus, ratio increases greater than the contribution from the impure standard may be from the endogenous metabolite. Using the calibration equation from the ratio of m/z 145/148, the concentrat ion of AC detected at each raster step can be estimated and a quantitative image generated, shown in F igure 4 8 From this image, the concentration of AC is estimated to be in the range of 20 100 nmol/kg tissue and falls within the concentration expected. A limitation of this type of quantitative estimation assumes that the standards spotted beneath the sample will extract, co crystallize, and ionize in the same manner as the endogenous compound, which is difficult to determine via MALDI MSI. However, ana lyte concentration differences may be observed within a tissue sample or from tissue to tissue These concentration differences may be determined with improved confidence due to the control for numerous factors including matrix application heterogeneity, t issue specific ion suppression 73 or any other cause of signal variation. Quantitative Comparison Between Treated and Untreated Tissue For the quantitative comparison of the endogenous AC content between treated ( ketogenic diet) piglets and untreated piglets (normal diet), MSI was performed using MS imaging on the QTOF instrument. The resulting AC/d 3 AC images from the MSI experiments are shown in Figure 4 9. Figure 4 9A is the MS images of the ketogenic treated s ample and Figure 4 9B is the MS image of the untreated sample. Concentrations of endogenous AC were estimated usin g a standard additions analysis
116 shown in Figure 4 10. Estimations of the endogenous AC concentrations were determined to be 170 50 ng AC / g tissue for the treated (ketogenic) sample and 100 70 ng AC / g tissue for the untreated sample. These values corresponded to an estimated upper limit of concentration of 600 molar for the untreated sample and 500 molar for the untreated sample. These va lues represent the upper limit as the integrated areas to generate the y intercept on the standard additions plots were taken from areas of higher signal intensity. From the generated MS images, most of the endogenous signal was between 50 100% of the max imum ratio of m/z 204/207 indicating a range of endogenous concentration of AC between 300 600 molar for the treated sample and 250 500 molar for the untreated sample. However, t he error in these measurements is large preventing the accurate differentiation of endogenous AC concentration between the two samples. Although we detected no significant difference in AC concentration, the ability to perform quantitative tissue to tissu e comparison is consider ed an advancement of MALDI MSI. Quantitative MSI Of Endogenous AC Conclusions Developing quantitative MALDI MSI methods is critical for its expansion into new applications. Two approaches were successfully utilized that leads to be tter signal reproducibility for MSI: 1) optimizing the matrix application method 2) implementing an isotopically labeled internal standard to improve signal reproducibility. A source of signal variation was shown to be caused by the matrix crystallization layer applied during MSI. Improving this matrix layer leads to better quantitative mass spectral images as well as better signal reproducibility. Utilizing an isotopically labeled internal standard resulted in a significant reduction of signal variabili ty leading to quantitative
117 tissue imaging. The quantitative information generated from an imaging experiment may reveal regional distribution of an analyte or may be used to determine overall tissue concentrations. This may be significant for whole body tissue imaging experiments, as outlined in Chapter 1, to determine concentration differences of a drug between different tissue types. Although, we did not detect a significant change in AC concentration between the treated and untreated piglet brain sam ples, the differences we did detect may be proven with a much larger sample size. Our preliminary experiments consisted of one brain sample from each piglet type; however, with numerous biological errors combined with the large error in our measurement, a n estimation of ~20 samples would need to be analyzed of each piglet type.
118 Figure 4 1 Microscope images (20 ) of formed DHB crystal s applied with a Meinhard nebulizer onto a glass slide using two different solvents: (A) diluted to 40 mg/mL in 70/30 me thanol/water (v/v) and (B) diluted to 40 mg/mL in 90/10 chloroform/ethanol (v/v) The largest crystals observed in A were measured to be <250 m compared to the largest crystals in B, measured at <60 m.
119 Figure 4 2 Extracted wide isolation MS/MS images of the normalized product ions of: AC, m/z 145 (A), d 3 AC m/z 148 (B), and the ratio of m/z 145/148 (C) at each position. The color scale is determined from calculating the percent difference from the mean. AC and d 3 AC exhibit significant signal variation within the area imaged; however, this variation is significantly reduced by plotting the ratio of m/z 145/148 (C) for each analytical scan
120 Figure 4 3 Line scan intensities normalized to the mean for each ion and ratio. Intensities of m/z 145, 148 and 145/148 of a line scan right (scan #1) to left (scan #11) from area indicated in Figure 4 2A.
121 Figure 4 4 Signal variation over subsequent MALDI MSI analysis.
122 Figure 4 5 Experimental setup for quantitative analysis of AC standards spotted onto a glass slide. AC was applied with a pipette and d 3 AC was applied uniformly with an inkjet printer.
123 Figure 4 6 Calibration plots from standard of AC spotted onto a glass microscope slide. (A) Calibration plot from the intensity of m/z 145 (AC) versus concentration. (B) Calibration plot of the ratio of m/z 145/148 (AC/d 3 AC ) showing improved linearity and signa l reproducibility. Error bars represent the standard e rror of the average MS/MS intensities. A B
124 Figure 4 7 W ide isolation MS/ MS images quantifying AC in piglet brain tissue. A) d 3 AC, absolute intensity of m/z 148. B) AC absolute intensity of m/z 145. C) T he ratio of AC/d 3 AC (C) at each po sition. The internal standard image A exhibits intense signal outlining the tissue sample. In images B and C, red circles indicate the location of the spotted AC standards, starting with the lower left as 0.1 ng/ L and moving clockwise ending with 4 ng/ L. The green square indicates a 4 ng/ L standard spotted atop the tissue sample for comparison with the standards spotted b eneath. The color scale for A and B is determined from the absolute intensities of the ions m/z 145 and 148, respectively. The color scale for C is determined from the ion intensity ratio of m/z 145/148 A B C Intensity ( m/z 14 8 ) Intensity ( m/z 145) Intensity ( m/z 145 /148 )
125 Figure 4 8 Quantitative image of AC in the frontal lobe of a piglet brain slice. Concentration of AC was calculated from standards placed beneath the tissue using a calibration curve normalized by an internal standard. The color scale is determined from the calculated concentration of endogenous AC from the calibration curve of the spotted standards of AC and is plotted in nmol AC /kg tissue
126 Figure 4 9 Full MS images of AC distribution in piglet brain tissue obtained from the QTOF Images were normalized by the d 3 AC signal by determining the m/z 204/207 ratio at each raster position (scale bar indicates ratio of m/z 204/207) The endogenous AC was quantitatively compared between piglets fed a ketogenic diet (A) to piglets fed a normal diet (B). A B
127 Fig ure 4 10 Standards addition plots for the quantitative determination of AC in piglet brain thin tissue sections. A) AC concentration is compared between a piglet fed a ketogenic diet and (B) a piglet fed a normal diet From these experiments, the differences in endogenous AC concentration were unable to be detected using quantitative MSI. A B
128 Table 4 1 Matrix dependence on signal MALDI ion signal variation. Ion Intensity Signal (%RSD) DHB Solvent AC m/z 204 145 d 3 AC m/z 207 148 AC/d 3 AC m/z 145/148 DHB H 2 O m/z 137 70/30 MeOH/H 2 O 63 72 16 66 90/10 CHCl 3 /EtOH 46 46 7.0 31
129 Table 4 2. Signal variation of MS/MS ion signal within calibration spots. Variation is reported in %RSD. %RSD of MS/MS Scans Within Each Spot Concentration (ng/ L) AC m/z 204 145 d 3 AC m/z 207 148 AC/d 3 AC m/z 145/148 0.1 55 62 59 0.2 49 52 10 0.4 38 39 10 0.8 51 47 8 1 49 42 7 4 42 40 7 8 25 29 8
130 CHAPTER 5 CONCLUSIONS AND FUTU RE WORK Conclusions The goal of this research was to develop quantitative MALDI MS I methods and to evaluate these methods for the analysis of acetyl l carnitine (AC) in piglet brain tissue. Initial method development of a quantitative MALDI MSI method was performed on the analysis of cocaine (COC) from a human brain sample. Qu antitative determinations made from the analysis by MALDI MSI were compared to quantitative results from the well established analytical technique LC MS /MS The need for a deuterated internal standard was identified during this method development phase. Further analysis employing a deuterated internal standard for MALDI MSI was performed with AC and d 3 AC. It was shown that the internal standard not only improves the quantitative ability of MALDI, but also can control for other limitations of MSI, such as tissue specific ion signal variation. Finally, quantitative estimat es of the AC con tent in piglet brain samples were made using the methods developed and analyzed in Chapters 2 and 3. The methods and analysis presented in this research are not limited to COC and AC, but can be generally employed for any small molecule target analyte, provided a n isotopically labeled internal standard is available. Initial experiments with cocaine showed that quantitative analysis could be performed with MALDI with the uniform application of an internal standard. The internal standard is us ed to normalize the target analytes ion signal at each MS scan thus accounting for the significant ion signal variation associated with MALDI. Quantitative results from MALDI MSI e xperiments by standard additions analysis compared very well to LC MS/MS results from tissue extracts. COC brain concentrations from the MSI
131 analysis were within 20% of the concentration determined by LC MS/MS; h owever, the MSI results were significantly less precise primarily due to the lack of sample cleanup steps such as tissue extraction and chromatography. MALDI MSI allows for directs sampling of the tissue performed in a two dimensional manner which preserves the physical integrity of the sample. Quantitative MALDI MSI was also compared to LC MS/MS for the analysis of tissue extracts. With an appropriate internal standard, wide isolation MS/MS was used in place of chromatography to separate interfering ions. Analysis of the tissue extracts resu lted in quantitative COC values within 5% of each other when analyzed by MALDI and LC MS/MS. The LC analysis was performed with a triple quadrupole mass spectrometer employing selected reaction monitoring, which is generally considered the best MS /MS approach for quantitative analysis. However, employing chromatography significantly increases analysis time and the consumption of resources. A typical LC analysis can last between 5 and 20 minutes for each sample analyzed. Including analysis of solvent blanks a calibration experiment can take many hours to complete. For the triplicate LC MS/MS analysis presented here, the total analysis time was approximately 15 hours. Analysis by MALDI shortens the analysis time significantly. For the experiments per formed, 50 MS/MS scans were recorded in quadruplicate resulting in analysis time of only 30 minutes for the entire quantitative experiment. A disadvantage of the MALDI approach, however, is the potential for non linear response thus limiting the linear dynamic range. These non linear effects result from differences in the matrix to analy te ratio across the calibration curve. Either too little or too much matrix may result in attenuation of ion signal. A utomated matrix application methods
13 2 such as the B ruker ImagePrep may be used to apply a constant amount of matrix from analysis to analysis. Using an internal standard for MALDI MSI also allowed for the identification of tissue specific ion signal variation. This is a well known limitation in MALDI MS I applications preventing the technique from being used as a standalone technique to draw biological conclusions. Differences in ion signal were monitored: by spotting a constant concentration of AC beneath multiple tissue types. These differences in io n signal are a result of numerous factors including differences in matrix crystallization atop the tissues, variations in analyte extraction from the tissue, and ionization variability resulting from the presence of potential ion suppressing compounds. H owever, when monitoring the analyte to internal standard ratio these differences in ion signal were accounted for. Ion signal of AC varied by a factor of 10 from tissue to tissue for the same concentration of AC applied beneath the sample. This variation in ion signal could lead to false conclusions about the concentration of AC in the tissue without normalization by an internal standard The AC/d 3 AC ratio, however, varied only by a factor of 1.3, producing a far more accurate representation of the spott ed AC concentration. Using an internal standard in this manner should also lead to better confidence in resulting MS images and provide a better quantitative representation of analyte concentrations between tissues. Lastly, MSI was used to make a quant itative comparison of the endogenous level s of AC between two samples. Since quantitative MALDI is challenging on any given sample, quantitative comparison between two samples is difficult. Ion signal can vary based on differences in tissue makeup, matri x application, and tissue thickness.
133 Analysis with an internal standard was shown to account for many of these differences thus allowing for tissue to tissue comparisons. The method does contain a significant amount of error in the measurements due to t he lack of sample cleanup s uch as extraction and chromatography. This error associated with quantitative measurements means that small changes in concentration will be difficult to determine. For this application, the concentration of AC was expected to change very little due to homeostasis, and indeed, the differences in AC concentration between these two samples was too small to determine with reasonable confidense However, developing methods to make tissue to tissue comparisons has the potential to b e used in numerous other applications. Future Work With the development of a quantitative MALDI MSI method the reproducibility of the experimental setup should also be improved. Currently calibration standard are manually spotted onto the MALDI target This produces error from spot to spot based on pipette delivery differences and uneven deposition as the spot dries. Developing and employing a more uniform spot delivery system would produce calibration standards applied in a uniform manner with a def ined spot area. Keeping the spot area consistent between samples is crucial for quantitative analysis as the area is used in calculating tissue concentrations. Other deposition techniques may be employed to produce a uniform layer of the internal standard onto the MALDI target. The modified inkjet printer produces picoliter droplet s ; however, these may merge to produce large r droplets formed on the target. This effect is solvent dependent and the inkjet printer is only compatible with methanol and water. Employing other spotting techniques such as
134 the LabCyte Portrait spotter may result in more reprodu cibl e internal and calibration standard applications. Sensitivity is also a potential limitation of the technique. This work shows quantitative estimations in the microm olar range; however, many applications require detection limits at least an order of magnitude less. MSI quantitation was improved with the QTOF instrument. Complete ablation of the tissue sample from spot to spot resulted in a more reproducible sample size analyzed at each raster position. Taking advantage of the complete ablation of the sample, thicker tissue sections may be used to increase the amount of the target analyte for detection. Employing MS n in the ion trap may also improve sensitivity through the reduction of background interferences; however, CID efficiency and ion loss from increased storage time can limit this approach. Space charging in the ion trap also limits the dynamic range of quantitation. This effect is not observed in the QTOF instrument; however, the large noise levels in the low mass range require a large number of spectra to be summed. This requires a large numb er of laser shots (>100) at each raster position. Comparing each instrument, the LTQ was more sensitive on a laser shot to ion signal basis; however, space charging and differences in samples size at each raster step result in variable ion signal. The QTOF results in better quantitative MSI data, but is limited in sensitivity on a laser shot to ion signal basis. Developing MSI instrumentation that could offer the sensitivity of the LTQ without th e space charging issues may result in significant improvements in the overall sensitivity of MALDI MSI. In summary generating more reproducible sample preparation methods and improvin g sensitivity are the main areas of quantitative MALDI MSI to be improved
135 upon. This would advance the methodology and allow it to be used in a wide array of applications.
136 LIST OF REFERENCES (1) Rontgen, W. K. Science. 1896 3 726 729. (2) Damadian, R. Science. 1971 171 1151 1153. (3) Raylman, R. R.; Majewski, S.; Lemieux, S. K.; Velan, S. S.; Kross, B.; Popov, V.; Smith, M. F.; Weisenberger, A. G.; Zorn, C.; Marano, G. D. Phys.Med.Biol. 2006 51 6371 6379. (4) Yuan, C.; Kerwin, W. S.; Ferguson, M. S.; Polissar, N.; Zhang, S.; Cai, J.; Hatsukami, T. S. J.Magn.Reson.Imaging. 2002 15 62 67. (5) Chalfie, M.; Tu, Y.; Euskirchen, G.; Ward, W. W.; Prasher, D. C. Science. 1994 263 802 805. (6) Helveg, S.; Lopez Cartes, C.; Sehested, J.; Hansen, P. L.; Clausen, B. S.; Rostrup Nielsen, J. R.; Abild Pedersen, F.; Norskov, J. K. Nature. 2004 427 426 429. (7) Svatos, A. Trends Biotechnol. 2010 28 425 434. (8) Caprioli, R. M.; Farmer, T. B.; Gile, J. Anal.Chem. 1997 69 4751 4760. (9) Castaing, R.; S lodzian, J. J. Microsc. 1962 1 395. ( 1 0) Kaufmann, R.; Hillenkamp, F.; Wechsung, R. Med.Prog.Technol. 1979 6 109 121. ( 1 1) Kaufmann, R.; Hillenkamp, F.; Wechsung, R.; Heinen, H. J.; Schurmann, M. Scan.Electron Microsc. 1979 2 279 290. ( 1 2) Perchalski, R. J. Characteristics and Application of a Laser Ionization/Evaporation Source for Tandem Mass Spectrometry. Ph.D. Dissertation, University of Florida, Gainesville, FL, 1985 ( 1 3) Troendle, F. J.; Reddick, C. D.; Yost, R. A. J.Am.Soc.Mass Spec trom. 1999 10 1315 1321. ( 1 4) Reddick, C. D. The Detection of Pharmaceutical Drug Compounds from Intact Biological Tissue by Matrix Assisted Laser Desorption Ionization (MALDI) Quadrupole Ion Trap Mass Spectrometr y. Ph.D. Dissertation, University of Flo rida, Gainesville, FL, 1997. ( 1 5) Cheng, J.; Winograd, N. Anal.Chem. 2005 77 3651 3659.
137 ( 1 6) Luxembourg, S. L.; McDonnell, L. A.; Duursma, M. C.; Guo, X.; Heeren, R. M. Anal.Chem. 2003 75 2333 2341. (17) Wiseman, J. M.; Ifa, D. R.; Zhu, Y.; Kissinge r, C. B.; Manicke, N. E.; Kissinger, P. T.; Cooks, R. G. Proc.Natl.Acad.Sci.U.S.A. 2008 105 18120 18125. (18) Wiseman, J. M.; Ifa, D. R.; Song, Q.; Cooks, R. G. Angew.Chem.Int.Ed Engl. 2006 45 7188 7192. (19) Garrett, T. J.; Prieto Conaway, M. C.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford, G.; Yost, R. A. Int. J. Mass spectrom. ,. 2007 260 166 176. (20) Garrett, T. J. Imaging Small Molecules in Tissue by Matrix Assisted Laser desorption/ionization Tandem Mass Spectrometry Ph.D. Dissertation [ Online], University of Florida, Gainesville, FL 2006. http://etd.fcla.edu/UF/UFE0013807/garrett_t.pdf (Accessed Jan 2011) (21) Chaurand, P.; Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. Toxicol.Pathol. 2005 33 92 101. (22) Aerni, H. R.; Cornett, D. S.; Caprioli, R. M. Anal.Chem. 2006 78 827 834. (23) Klerk, L. A.; Altelaar, A. F. M.; Froesch, M.; McDonnell, L. A.; Heeren, R. M. A. Int. J. Mass spectrom. 2009 285 19 25. (24) Chaurand, P.; Norris, J. L.; Cornett, D. S.; Mobley, J. A.; Caprioli, R. M. J.Proteome Res. 2006 5 2889 2900. (25) Hankin, J. A.; Barkley, R. M.; Murphy, R. C. J.Am.Soc.Mass Spectrom. 2007 18 1646 1652. (26) Goodwin, R. J.; Scullion, P.; Macintyre, L .; Watson, D. G.; Pitt, A. R. Anal.Chem. 2010 24 1682 1686 (27) Trimpin, S.; Herath, T. N.; Inutan, E. D.; Wager Miller, J.; Kowalski, P.; Claude, E.; Walker, J. M.; Mackie, K. Anal.Chem. 2010 82 359 367. (28) Puolitaival, S. M.; Burnum, K. E.; Corne tt, D. S.; Caprioli, R. M. J.Am.Soc.Mass Spectrom. 2008 19 882 886. (29) Wei, H.; Nolkrantz, K.; Powell, D. H.; Woods, J. H.; Ko, M. C.; Kennedy, R. T. Rapid Commun.Mass Spectrom. 2004 18 1193 1200. (30) Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Int. J. Mass spectrom. and Ion Processes,. 1987 78 53 68. (31) Karas, M.; Bachmann, D.; Hillenkamp, F. Anal.Chem. 1985 57 2935 2939.
138 (32) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T. Rapid Commun.Mass Spectrom. 1988 2 151. (33) Karas, M.; Hillenkamp, F. Anal.Chem. 1988 60 2299 2301. (34) Knochenmuss, R.; Karbach, V.; Wiesli, U.; Breuker, K.; Zenobi, R. Rapid Commun.Mass Spectrom. 1998 12 529. (35) Zenobi, R.; Knochenmuss, R. Mass Spectrom.Rev. 1998 17 337 366. (36 ) Puretzky, A. A.; Geohegan, D. B. Chem. Phys. Lett. 1998 286 425 432. (37) Hillenkamp, F. and Peter Katalinic, J. MALDI MS: A Practical Guide to Instrumentation, Methods and Applications, ed.; Wiley VCH:2007. (38) Karas, M.; Gluckmann, M.; Schafer, J. J.Mass Spectrom. 2000 35 1 12. (39) Gusev, A. I.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M. Anal Chem. 1995 67 1034. (40) Yu, H.; Lopez, E.; Young, S. W.; Luo, J.; Tian, H.; Cao, P. Anal.Biochem. 2006 354 182 191. (41) Hatsis, P.; Brombache r, S.; Corr, J.; Kovarik, P.; Volmer, D. A. Rapid Commun.Mass Spectrom. 2003 17 2303 2309. (42) Bucknall, M.; Fung, K. Y.; Duncan, M. W. J.Am.Soc.Mass Spectrom. 2002 13 1015 1027. (43) Knochenmuss, R. J.Mass Spectrom. 2002 37 867 877. (44) Schwartz, J. C.; Senko, M. W.; Syka, J. E. P. J. Am. Soc. Mass. Spectrom. 2002 13 659 669. (45) Paul, W.; Steinwedel, H. June 7, 1960 U.S. Patent# 2,939,952 (46) Reich, R. Quantitative Imaging of Cocaine and its Metabolites in Brain Tissue by Matrix A ssisted Laser desorption/ionization Linear Ion Trap Tandem Mass Spectrometry Ph.D. Dissertation [Online], University of Florida, Gainesville, FL 2010 http://purl.fcla.edu/fcla/etd/UFE0041855 (Accessed Jan 2011) (47) Reich, R. F.; Cudzilo, K.; Levisky, J. A.; Yost, R. A. J.Am.Soc.Mass Spectrom. 2010 21 564 571. (48) Setou, M. Imaging Mass Spectrometry: Protocols for Mass Microscopy 1st ed.; Springer:Tokyo, Japan; 2010.
139 (49) Pringle, S. D.; Giles, K.; Wildgoose, J. L.; Williams, J. P.; Slade, S. E.; Thalassinos, K.; Bateman, R. H.; Bowers, M. T.; Scrivens, J. H. Int. J. Mass spectrom. 2007 261 1 12. (50) Yost, R. A.; Enke, C. G. Anal.Chem. 1979 51 1251 1264. (51) Paul, W.; Helmut, S. 1960 U.S. Patent # 2 939 952 (52) Freeman, J. M., Freeman, J. B., Freeman,John Mark Epilepsy diet treatment and Kelly, M. T. The Ketogenic Diet : A Treatment for Epilepsy, 3rd ed.; Demos:New York; 2000. (53) Hartman, A. L.; Gasior, M.; Vining, E. P.; Rogawski, M. A. Pediatr.Neurol. 2007 36 281 292. (54) Kalapos, M. P. Med.Hypotheses. 2007 68 1382 1388. (55) Ma, W.; Berg, J.; Yellen, G. J.Neu rosci. 2007 27 3618 3625. (56) Borum, P. R. Annu.Rev.Nutr. 1983 3 233 259. (57) Nalecz, K. A.; Miecz, D.; Berezowski, V.; Cecchelli, R. Mol.Aspects Med. 2004 25 551 567. (58) Fariello, R. G.; Zeeman, E.; Golden, G. T.; Reyes, P. T.; Ramacci, T. Ne uropharmacology. 1984 23 585 587. (59) Hack, A.; Busch, V.; Pascher, B.; Busch, R.; Bieger, I.; Gempel, K.; Baumeister, F. A. Pediatr.Res. 2006 60 93 96. (60) Aureli, T.; Di Cocco, M. E.; Puccetti, C.; Ricciolini, R.; Scalibastri, M.; Miccheli, A.; M anetti, C.; Conti, F. Brain Reas 1998 796 75 81. (61) Brass, E. P.; Hoppel, C. L. J. Biol. Chem. 1978 253 2688. (62) Kuratsune, H.; Watanabe, Y.; Yamaguti, K.; Jacobsson, G.; Takahashi, M.; Machii, T.; Onoe, H.; Onoe, K.; Matsumura, K.; Valind, S.; Kitani, T.; Langstrom, B. Biochem.Biophys.Res.Commun. 1997 231 488 493. (63) Minkler, P. E.; Stoll, M. S.; Ingalls, S. T.; Yang, S.; Kerner, J.; Hoppel, C. L. Clin.Chem. 2008 54 1451 1462. (64) Hsieh, Y.; Li, F.; Korfmacher, W. A. Methods Mol.Biol. 2010 656 147 158. (65) Nilsson, A.; Fehniger, T. E.; Gustavsson, L.; Andersson, M.; Kenne, K.; Marko Varga, G.; Andren, P. E. PLoS One. 2010 5 e11411.
140 (66) Landgraf, R. R. Analysis of Lipids In Nerve Tissue by MALDI Tandem Mass Spectrometric Imaging Ph.D. Dissertation [Online], University of Florida, Gainesville, FL, 2009. http://purl.fcla.edu/fcla/etd/UFE00 24156 (Accessed Jan 2011) (67) Quia, H.; Piyadasa, G.; Spicer, V.; Ens, W. Int. J. Mass S pectrom. 2009 281 41. (68) Zaima, N.; Hayasaka, T.; Goto Inoue, N.; Setou, M. J.Oleo Sci. 2009 58 415 419. (69) Willmann, J. K.; van Bruggen, N.; Dinkelborg L. M.; Gambhir, S. S. Nat.Rev.Drug Discov. 2008 7 591 607. (70) Wiseman, J. M.; Ifa, D. R.; Zhu, Y.; Kissinger, C. B.; Manicke, N. E.; Kissinger, P. T.; Cooks, R. G. Proc.Natl.Acad.Sci.U.S.A. 2008 105 18120 18125. (71) Signor, L.; Varesio, E.; Staa ck, R. F.; Starke, V.; Richter, W. F.; Hopfgartner, G. J.Mass Spectrom. 2007 42 900 909. (72) Khatib Shahidi, S.; Andersson, M.; Herman, J. L.; Gillespie, T. A.; Caprioli, R. M. Anal.Chem. 2006 78 6448 6456. (73) Stoeckli, M.; Staab, D.; Schweitzer, A. Int. J. Mass S pectrom. 2007 260 195 202. (74) Hankin, J. A.; Murphy, R. C. Anal.Chem. 2010 (75) Koeniger, S. L.; Talaty, N.; Luo, Y.; Ready, D.; Voorbach, M.; Seifert, T.; Cepa, S.; Fagerland, J. A.; Bouska, J.; Buck, W.; Johnson, R. W.; Spanton S. Rapid Commun. Mass Spectrom. 2011 25 503 510. (76) Baluya, D. L.; Garrett, T. J.; Yost, R. A. Anal.Chem. 2007 79 6862 6867. (77) Naofumi, H.; Yuki, S.; Mitsutoshi, S. J. Mass Spectrom. Soc. Jpn. 2008 56 77. (78) Gobey, J.; Cole, M.; Janiszewski, J.; Covey, T.; Chau, T.; Kovarik, P.; Corr, J. Anal.Chem. 2005 77 5643 5654. (79) Lowe, R. H.; Barnes, A. J.; Lehrmann, E.; Freed, W. J.; Kleinman, J. E.; Hyde, T. M.; Herman, M. M.; Huestis, M. A. J.Mass Spectrom. 2006 41 175 184. (80) Chen, R.; Hui, L.; Sturm, R. M.; Li, L. J.Am.Soc.Mass Spectrom. 2009 20 1068 1077. (81) Andersson, M.; Groseclose, M. R.; Deutch, A. Y.; Caprioli, R. M. Nat.Methods. 2008 5 101 108. (82) Solon, E. G.; Schweitzer, A.; Stoeckli, M.; Prideaux, B. AAPS J 2010 12 11 26.
141 (83) Heeren, R. M. A.; Smith, D. F.; Stauber, J.; Kkrer Kaletas, B.; MacAleese, L. J.Am.Soc.Mass Spectrom. 2009 20 1006 1014. (84) Sleno, L.; Volmer, D. A. Rapid Commun.Mass Spectrom. 2005 19 1928 1936. (85) Rohner, T. C.; Staab, D.; Stoeckli, M. Mech.Ageing Dev. 2005 126 177 185. (86) Gobey, J.; Cole, M.; Janiszewski, J.; Covey, T.; Chau, T.; Kovarik, P.; Corr, J. Anal.Chem. 2005 77 5643 5654. (87) Beckmann, N.; Laurent, D.; Tigani, B.; Panizzutti, R.; Rudin, M. Drug Discov .Today. 2004 9 35 42. (88) Shimma, S.; Sugiura, Y.; Hayasaka, T.; Hoshikawa, Y.; Noda, T.; Setou, M. J.Chromatogr.B.Analyt Technol.Biomed.Life.Sci. 2007 855 98 103. (89) Cornett, D. S.; Frappier, S. L.; Caprioli, R. M. Anal.Chem. 2008 (90) Hsieh, Y.; Chen, J.; Korfmacher, W. A. J.Pharmacol.Toxicol.Methods. 2007 55 193 200. (91) Reyzer, M. L.; Caprioli, R. M. Curr.Opin.Chem.Biol. 2007 11 29 35. (92) Landgraf, R. R.; Garrett, T. J.; Calcutt, N. A.; Stacpoole, P. W.; Yost, R. A. Anal.Chem 2007 79 8170 8175. (93) Schwartz, S. A.; Reyzer, M. L.; Caprioli, R. M. J.Mass Spectrom. 2003 38 699 708. (94) Reyzer, M. L.; Hsieh, Y.; Ng, K.; Korfmacher, W. A.; Caprioli, R. M. J.Mass Spectrom. 2003 38 1081 1092. (95) Nemes, P.; Woods, A. S.; Vertes, A. Anal.Chem. 2010 (96) Garrett, T. J.; Yost, R. A. Anal.Chem. 2006 78 2465 2469. (97) Drexler, D. M.; Garrett, T. J.; Cantone, J. L.; Diters, R. W.; Mitroka, J. G.; Prieto Conaway, M. C.; Adams, S. P.; Yost, R. A.; Sanders, M. J.Pharmacol.To xicol.Methods. 2007 55 279 288. (98) Chen, Y.; Allegood, J.; Liu, Y.; Wang, E.; Cachon Gonzalez, B.; Cox, T.; Merrill, A.; Sullards, M. Anal Chem. 2008 80 2780. (99) Puolitaival, S. M.; Burnum, K. E.; Cornett, D. S.; Caprioli, R. M. J.Am.Soc.Mass Sp ectrom. 2008 19 882 886. ( 1 00) Adelina Pellicer; Francisco Gay; Rosario Madero; Jos Quero, M.; Fernando Cabaas. Pediatrics. 2002 109 434.
142 BIOGRAPHICAL SKETCH David Andrew Pirman, Jr. was the first of two children to parents Brenda Carsey Henderson and David Andrew Pirman, Sr. From the his first introduction to chemistry freshman year while attending Gulf High School in New Port Richey, FL, David has been continuously intrigued with the subje ct. Ms. Margaret Peek was his first chemistry instructor and challenged him from the first day of class. She still serves a source of motivation and inspiration, not only in his ongoing chemistry education but his life as well. Something is to be said a bout a teacher who set in motion the ultimate path in life, and David uses this opportunity to recognize and thank one of the most influential teachers in his life. After his initial interest in chemistry developed in high school, David enrolled at the University of Florida and imm ediately selected his major as c hemistry. During his junior year, David began doing undergraduate research under the late Dr. Merle Battiste in synthetic organic research. David also spent many hours volunteering in the G ainesville community with local and campus organizations. Upon graduation, David worked at the University of Florida Racing Laboratory doing method development for instrumental screening techniques for the analysis of drugs of abuse in the horse and dog r acing industries. This position initiated his genuine interested in field of analytical chemistry. David then began his graduate career at the University of Florida in August of 2006, pursuing a Ph.D. in chemistry under the advisement of Dr. Richard A. Y ost. His graduate research focused on method development for quantitative MALDI and its application to tissue imaging. David thoroughly enjoys the work he does and hopes to
143 continue a similar scientific research in the field of analytical research in the biomedical field