ABSTRACT BOKHART, MARK THOMAS. Development And

ABSTRACT

BOKHART, MARK THOMAS. Development and Application of Matrix-Assisted Laser Desorption Electrospray Ionization Mass Spectrometry Imaging for Drug Distribution Studies. (Under the direction of Dr. David C. Muddiman.)

Infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) is a hybrid ionization technique combining resonant laser ablation and electrospray ionization. Analysis in a spatially resolved manner permits mass spectrometry imaging (MSI) analyses to be performed by displaying ion intensities in a 2-dimension array. IR-MALDESI has several unique advantages over conventional MSI techniques, for example matrix-assisted laser desorption/ionization (MALDI), such as operation at ambient conditions and the use of a biologically-compatible ice matrix. Additionally, the mid-infrared (IR) laser used for ablation has much greater ablation depth compared to ultraviolet lasers used in MALDI, usually completely ablating the sampled material at each rastered position. This attribute provides quantitative sampling of material at each position, giving high quality MSI data. The IR-MALDESI MSI source was optimized through a systematic investigation of factors effecting desorption of analytes including IR laser wavelength, geometric considerations and the inclusion of ice as a matrix. Additional optimization was performed by investigating trends in analyte response as the mass spectrometer C-trap injection time was varied. The shorter C-trap injection times lead to increases in ion abundance and identification of more species in untargeted analyses. A quantitative mass spectrometry imaging method was developed for use with IR- MALDESI, where a normalization compound is uniformly incorporated beneath the tissue to be quantified along with a spatial calibration curve incorporated on top of the same tissue. Normalization to the normalization compound reduced per-voxel variability, producing high quality MS images and accurate quantification from MSI data. This quantitative method was then applied to the analysis of 11 organs from a dosed, non-human primate study. IR- MALDESI MSI provided absolute quantification of an antiretroviral drug while displaying the heterogeneous distribution within sections of organs. IR-MALDESI MSI was also used to show drug incorporation into patient’s hair for monitoring drug regimen adherence. A method of normalization to hair melanin content was used to compare drug incorporation among several patients based on hair color. The spatial resolution of IR-MALDESI MSI was improved to 50-micron voxel size through the incorporation of a multi-element optical system. An adjustable iris, 3.75x beam expander and aspheric focusing lens were used to shrink the laser focal point compared to a single focusing lens. Important laser parameters were defined to characterize the laser beam caustic of both designs for comparison. The open-source MSI software, MSiReader, had several new tools added to facilitate the analysis of drug distribution studies. The software has the capability to load multiple files at once, allowing facile comparison of multiple data sets. An image overlay tool has been developed to allow comparison of the MSI data to any image file by displaying both concurrently. The quantification procedure has been streamlined into the MSiQuantification tool, where the user can define calibration and quantification parameters directly in MSiReader. Finally, MSiReader can now display mass measurement accuracy (MMA) heatmaps depicting the MMA for individual voxels in MSI data.

by Mark Thomas Bokhart

A dissertation submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the Degree of Doctor of Philosophy

Chemistry

Raleigh, North Carolina

2017

APPROVED BY:

______David C. Muddiman Edmond Bowden Committee Chair

______Gufeng Wang Balaji Rao

BIOGRAPHY

Mark Bokhart was born on June 6, 1990 in Bay City, Michigan to Dan and Rebecca Bokhart. His passion for science was evident early on in his education, taking a great interest in science and mathematics courses. He attended Garber High School in Essexville, Michigan where he took every science and math course available by the time he graduated in May 2008. Not quite sure what he wanted to do as a career after graduation from high school, Mark attended Delta Community College in University Center, Michigan beginning in August 2008. After 2 years of community college, Mark’s interest and success in chemistry courses led him to purse a bachelor’s degree in chemistry at Michigan State University starting in August 2010. While at MSU, he worked as an undergraduate researcher at MSU’s Diagnostic Center for Population and Animal Health in the toxicology section under the supervision of Drs. John Buchweitz and Andreas Lehner. While there, he developed an appreciation for analytical methodology and realized the great need to further development of analytical technologies to solve real-world problems. In May 2013, Mark graduated from MSU with a Bachelor’s of Science in Chemistry. He then went on to pursue a Ph.D. degree in Analytical Chemistry at North Carolina State University under the direction of Dr. David Muddiman.

ACKNOWLEDGMENTS

I would like to formally acknowledge the unwavering support and mentorship of Dr. David C. Muddiman throughout my Ph.D. studies. I am extremely grateful for the extensive resources and great opportunities provided by the Muddiman laboratory and North Carolina State University. My work was highly collaborative, spanning across departments and institutions, and my projects would not have been as successful without the guidance of everyone involved. In particular, thanks to Dr. Eli Rosen for always listening to my experimental ideas and Ken Garrard for helping me with a wide variety of data analysis and visualization techniques. My experiences in graduate school would not have been as enjoyable and productive without the support and companionship of the former and current Muddiman lab members. And finally, I would not have succeeded without the support of the friends I made while at NCSU, especially Rosa Castrejon Garcia.

iii

TABLE OF CONTENTS

List of Tables ……………………………...………………………………………...... xi List of Figures ………………………………...... …………………………………………...……. xii List of Publications ...... xv 1 Introduction ...... 1

1.1 Soft Ionization Methods ...... 1

1.1.1 Electrospray Ionization ...... 1

1.1.2 Matrix Assisted Laser Desorption Ionization (MALDI) ...... 3

1.1.3 Matrix Assisted Laser Desorption Electrospray Ionization (MALDESI) ...... 4

1.2 Fourier Transform Mass Spectrometry ...... 5

1.3 Mass Spectrometry Imaging (MSI) ...... 7

1.4 Human Immunodeficiency Virus (HIV) Treatment with Antiretroviral Drugs ...... 9

1.5 Synopsis of Completed Work ...... 10

1.6 References ...... 12

2 Influence of Desorption Conditions on Analyte Sensitivity and Internal Energy in Discrete Tissue or Whole Body Imaging by IR-MALDESI ...... 20

2.1 Introduction ...... 20

2.2 Experimental ...... 24

2.2.1 Chemicals and Materials ...... 24

2.2.2 Sample Preparation ...... 24

2.2.3 Instrumentation ...... 25

2.2.4 Data Analysis ...... 26

2.3 Results and Discussion ...... 27

2.3.1 Optimization of Desorption Conditions using an Ice Matrix ...... 27

2.3.2 Whole Body Imaging ...... 32

2.3.3 Response Factors and Internal Energy Distribution across Whole Body ...... 38

2.4 Conclusions ...... 42

2.5 Acknowledgments ...... 42

2.6 References ...... 43

3 Influence of C-Trap Ion Accumulation Time on the Detectability of Analytes in IR- MALDESI MSI ...... 50

3.1 Introduction ...... 50

3.2 Experimental Section ...... 52

3.2.1 Materials ...... 52

3.2.2 IR-MALDESI Source ...... 53

3.2.3 Mass Spectrometer ...... 54

3.2.4 IR-MALDESI Imaging ...... 55

3.2.5 Data Analysis ...... 56

3.3 Results and Discussion ...... 57

3.3.1 C-Trap Accumulation Time with Ions from Multiple Sources ...... 57

3.3.2 Untargeted Analysis of Endogenous Lipids ...... 59

3.3.3 Targeted Analysis of Xenobiotics ...... 63

3.4 Conclusions ...... 65

3.5 Acknowledgment ...... 66

3.6 References ...... 67

4 Quantitative Mass Spectrometry Imaging of Emtricitabine in Cervical Tissue Model using Infrared Matrix-Assisted Laser Desorption Electrospray Ionization ...... 72

4.1 Introduction ...... 72

4.2 Experimental ...... 75

4.2.1 Materials ...... 75

4.2.2 Tissue Samples...... 76

4.2.3 IR-MALDESI Imaging ...... 76

4.2.4 Electrospray Ionization Cationization Agents ...... 77

4.2.5 Sample Separation for Quantitative MSI ...... 77

4.2.6 Data Processing ...... 79

4.2.7 Quantitative IR-MALDESI Imaging ...... 79

4.2.8 LC-MS/MS Quantification ...... 80

4.3 Results and Discussion ...... 80

4.3.1 Sensitivity Enhancement and Interference Removal ...... 80

4.3.2 Normalization Strategy ...... 84

4.3.3 Accounting for Ionization Efficiency per Voxel ...... 88

4.3.4 Quantification of Emtricitabine using MALDESI QMSI ...... 90

4.3.5 Comparison to LC-MS/MS ...... 92

4.4 Conclusion ...... 93

4.5 Acknowledgments ...... 93

4.6 References ...... 94

5 Mass Spectrometry Imaging Reveals Heterogeneous Efavirenz Distribution within Putative HIV Reservoirs...... 101

5.1 Introduction ...... 101

5.2 Experimental ...... 102

5.3 Results and Discussion ...... 104

5.4 Conclusion ...... 109

5.5 Acknowledgements ...... 110

5.6 References ...... 110

6 Analysis of Antiretrovirals in Single Hair Strands for Evaluation of Drug Adherence with IR-MALDESI MSI ...... 114

6.1 Introduction ...... 114

6.2 Experimental Section ...... 116

6.2.1 Materials ...... 116

6.2.2 Hair Samples ...... 117

6.2.3 Sample Preparation ...... 118

6.2.4 IR-MALDESI Mass Spectrometry Imaging ...... 118

6.2.5 LC–MS/MS Analysis of Hair ...... 120

6.3 Results and Discussion ...... 120

6.3.1 Optimization of IR-MALDESI for Hair Analysis ...... 120

6.3.2 Calibration of IR-MALDESI Response to ARVs in Incubated Hair and Correlation to LC–MS/MS ...... 123

6.3.3 IR-MALDESI Response to ARVs in Dosed Hair...... 128

vii

6.3.4 Evaluating Hair Melanin Content by MSI ...... 131

6.4 Conclusions ...... 138

6.5 Acknowledgements ...... 138

6.6 References ...... 138

7 IR-MALDESI Mass Spectrometry Imaging at 50 Micron Spatial Resolution ...... 146

7.1 Introduction ...... 146

7.2 Experimental ...... 148

7.2.1 Materials ...... 148

7.2.2 Tissues...... 149

7.2.3 IR-MALDESI Source ...... 149

7.2.4 Safety ...... 150

7.2.5 Laser and Optics ...... 151

7.2.6 Laser Beam Caustic Visualization ...... 152

7.2.7 Fitting Real Data to Theoretical Real Laser Beam Focus ...... 155

7.2.8 Laser Ablation Diameter on Tissue ...... 155

7.2.9 MSI with High Spatial Resolution ...... 156

7.3 Results and Discussion ...... 156

7.3.1 Multi-element Optical System ...... 156

7.3.2 Visualization of Laser Beam Caustic ...... 158

7.3.3 Ablation Diameter on Mouse Liver Tissue ...... 162

7.3.4 High Spatial Resolution IR-MALDESI MSI ...... 164

7.4 Conclusion ...... 167

viii

7.5 Acknowledgments ...... 168

7.6 References ...... 168

8 MSiReader v1.0: Evolving Open-Source Mass Spectrometry Imaging Software for Targeted and Untargeted Analyses ...... 172

8.1 Introduction ...... 172

8.2 Experimental ...... 173

8.3 Results and Discussions ...... 174

8.3.1 Overview of MSiReader Features ...... 174

8.3.2 Loading and Processing of Imaging Data Sets ...... 175

8.3.3 Loading Multiple Data Sets ...... 177

8.3.4 Absolute Quantification in MSiReader for Mass Spectrometry Imaging ...... 178

8.3.5 Polarity Switching Mass Spectrometry Imaging Data Processing ...... 180

8.3.6 Image Overlay Tool ...... 182

8.3.7 Mass Measurement Accuracy (MMA) Heatmap and Histograms ...... 183

8.4 Conclusions ...... 186

8.5 Acknowledgements ...... 186

8.6 References ...... 186

APPENDIX ...... 190

Appendix A ...... 191

A. Infrared Matrix-Assisted Laser Desorption Electrospray Ionization Mass Spectrometry Imaging of Biospecimens ...... 191

A.1 Introduction ...... 191

A.2 Methods ...... 193

A.2.1 IR-MALDESI MSI Tissue Preparation ...... 194

A.2.2 Quantitative IR-MALDESI Imaging Preparation ...... 194

A.2.3 IR-MALDESI Tissue Imaging ...... 195

A.2.4 Formation of Ice Matrix Layer ...... 196

A.2.5 Data Analysis and Visualization in MSiReader ...... 197

A.3 Discussion ...... 198

A.3.1 Effects of Ice Matrix on IR-MALDESI Analysis ...... 198

A.3.2 Comparison of IR-MALDESI to MALDI MSI ...... 199

A.3.3 Electrospray Ionization Agent ...... 201

A.3.4 Metal Adduct Formation ...... 201

A.3.5 Drug Distribution Studies ...... 202

A.3.6 Untargeted Metabolomic MSI Analyses ...... 208

A.3.7 Limitations ...... 211

A.4 Conclusions ...... 212

A.5 Acknowledgements ...... 212

A.6 References ...... 212

LIST OF TABLES

Table 2-1. Summary of lipid classes identified by whole-body IR-MALDESI MSI ...... 35 Table 2-2. Summary of MS2I response to cocaine ...... 41 Table 4-1. TM sprayer conditions ...... 78 Table 4-2. Abundance and RSDs of ARVs in incubated tissue ...... 82 Table 4-3. Quantification summary of IR-MALDESI MSI and LC-MS/MS ...... 93 Table 5-1. Intratissue variability of EFV in dosed tissues...... 106 Table 5-2. Intertissue variability of EFV in dosed tissues...... 107 Table 6-1. ARV physiochemical properties ...... 125 Table 6-2. IR-MALDESI response for ARVs in incubated hair ...... 126 Table 6-3. Linear regression of IR-MALDESI calibration data in hair...... 128 Table 7-1. Defining beam characteristics...... 161 Table 8-1. Dataset loading time improvement comparison ...... 176 Table 8-2. MSiReader operation time improvement...... 177

LIST OF FIGURES

Figure 1-1. Schematic of electrospray ionization...... 2 Figure 1-2. Visualization of the MALDESI process...... 5 Figure 1-3. Schematic of the Q Exactive mass spectrometer...... 6 Figure 2-1. IR-MALDESI response while scanning from 2850 nm to 3100 nm...... 28 Figure 2-2. Design of Experiments factors and responses for desorption conditions ...... 30 Figure 2-3. Design of Experiments responses for endogenous metabolites ...... 31 Figure 2-4. Classification of metabolites based on wavelength dependent response ...... 34 Figure 2-5. MS images of selected lipids in a whole-body mouse section ...... 36 Figure 2-6. MS images of wavelength dependent lipid classes ...... 37 Figure 2-7. IR-MALDESI MS2I of cocaine...... 39 Figure 2-8. Breakdown curve for CID of cocaine ...... 40 Figure 3-1. Timing diagram of laser synchronization with the Q Exactive ...... 56 Figure 3-2. Dependence of ion abundance on C-trap injection time...... 58 Figure 3-3. Laser fluence optimization for the 20 Hz and 100 Hz laser...... 61 Figure 3-4. Untargeted lipid analysis of hen ovary...... 63 Figure 3-5. Comparison of the 100 Hz and 20 Hz laser for quantitative MSI ...... 65 Figure 4-1. Structures of the targeted analytes...... 75 Figure 4-2. Workflow for quantitative IR-MALDESI MSI...... 78 Figure 4-3. Optimization of ESI cationization agent ...... 82 Figure 4-4. Optimization of NaCl concentration in the ESI solvent ...... 84 Figure 4-5. Ion response for analytes placed above or below a tissue ...... 85 Figure 4-6. Normalization procedure for IR-MALDESI MSI ...... 86 Figure 4-7. Ion maps of TFV and RAL in tissue ...... 88 Figure 4-8. Normalization to account for ionization efficiency ...... 89 Figure 4-9. Summary of FTC quantification in a tissue section ...... 91

xii

Figure 5-1. EFV distribution in macaque reservoir sites...... 105 Figure 6-1. Optimization of IR-MALDESI analysis of hair strands ...... 122 Figure 6-2. Evaluation of sampling efficiency by IR-MALDESI MSI of hair strands ...... 123 Figure 6-3. Calibration of IR-MALDESI MSI response to ARV-incubated hair strands.. .. 126 Figure 6-4. Calibration of IR-MALDESI signal abundance to LC-MS/MS response ...... 127 Figure 6-5. Analysis of EFV in hair strands from three dosed patients...... 130 Figure 6-6. Calibration of IR-MALDESI ion abundance to LC-MS/MS response for EFV in dosed hair strands...... 130 Figure 6-7. Normalizaiton to hair melanin method...... 133 Figure 6-8. EFV normalization to melanin content in dose hair...... 135 Figure 6-9. PTCA response for dosed hair stands ...... 136 Figure 6-10. Mass excess of all hair-specific peaks...... 137 Figure 7-1. Schematic of IR-MALDESI using (a) a single spherical focusing lens, and (b) a beam expander with an iris and an aspherical focusing lens...... 152 Figure 7-2. Laser ablation burn spots recorded at specified z-positions for the single lens . 153 Figure 7-3. Laser ablation burn spots recorded at specified z-positions for the multi-element optical system...... 154 Figure 7-4. Visualization of laser beam caustic ...... 154 Figure 7-5. Beam profile of the opolette OPO IR laser at 2940 nm ...... 157 Figure 7-6. Fit of the laser beam caustics...... 160 Figure 7-7. Ablation diameters on mouse liver tissue with the multi-element system ...... 163 Figure 7-8. MSI of a lipid using oversampling down to 30 μm step size ...... 165 Figure 7-9. High spatial resolution IR-MALDESI MSI at 100, 75, and 50 μm ...... 166 Figure 7-10. Mass measurement accuracy of high resolution IR-MALDESI MSI ...... 167 Figure 8-1. Loading multiple imaging data sets...... 178 Figure 8-2. MSiQuantification tool for absolute quantification MSI experiments...... 179 Figure 8-3. Polarity switching and polarity filtering for MSI data ...... 181

xiii

Figure 8-4. MSiImage tool for overlaying an optical image with the ion image ...... 183 Figure 8-5. Mass measurement accuracy heat map for MSI ...... 185 Appendix A Figure A-1. Schematic of IR-MALDESI source...... 197 Figure A-2. Influence of ice as a matrix for a 2940 nm IR laser...... 199 Figure A-3. Comparison of MALDI and IR-MALDESI MSI analysis of serial sections from a lapatinib dosed liver ...... 200 Figure A-4. Summary of a quantitative MSI experiment using IR-MALDESI ...... 205 Figure A-5. MSI of efavirenz and CD3+ IHC imaging of (A) colon (B) ileum (C) inguinal lymph node (D) cerebellum (E) spleen ...... 206 Figure A-6. IR-MALDESI analysis of ARVs in hair...... 208 Figure A-7. Untargeted polarity-switching IR-MALDESI MSI of cancer and control hen ovary tissue ...... 210

xiv

LIST OF PUBLICATIONS

1. Bokhart MT†, Nazari M†, Garrard KP, Muddiman DC. MSiReader v1.0: Evolving Open-Source Mass Spectrometry Imaging Software for Targeted and Untargeted Analyses. Journal of The American Society for Mass Spectrometry. Submitted. †Authors contributed equally.

2. Bokhart MT, Manni JG, Garrard KP, Ekelof M, Nazari M, Muddiman DC. IR- MALDESI mass spectrometry imaging at 50 micron spatial resolution. Journal of The American Society for Mass Spectrometry. 2017. In press. doi: 10.1007/s13361-017- 1740-x

3. Bokhart MT and Muddiman DC. Infrared matrix assisted laser desorption electrospray ionization mass spectrometry imaging of biospecimens, Analyst, 2016, 141, 5236- 5245.

4. Rosen EP, Thompson CG, Bokhart MT, Prince H, Sykes C, Muddiman DC, Kashuba ADM. Analysis of antiretrovirals in single hair strands for evaluation of drug adherence with IR-MALDESI MSI. Analytical Chemistry. 2016, 88 (2), 1336-1344.

5. Nazari M, Bokhart MT, Muddiman DC. Whole-Body Mass Spectrometry Imaging by Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-MALDESI). Journal of Visualized Experiments. 2016, 109, e53942.

6. Rosen EP, Bokhart MT, Nazari M, Muddiman, DC. Influence of C-Trap Ion Accumulation Time on the Detectability of Analytes in IR-MALDESI MSI. Analytical Chemistry. 2015, 87 (20), 10483-10490.

7. Thompson CG, Bokhart MT, Sykes C, Adamson L, Fedoriw Y, Luciw P, Muddiman DC, Kashuba ADM, Rosen EP. Mass Spectrometry Imaging Reveals Heterogeneous Efavirenz Distribution within Putative HIV Reservoirs. Antimicrobial Agents and Chemotherapy. 2015, 59 (5), 2944-2948.

8. Rosen EP, Bokhart MT, Ghashghaei TG, Muddiman DC. Influence of desorption conditions on analyte sensitivity and internal energy in discrete tissue or whole body imaging by IR-MALDESI. Journal of The American Society for Mass Spectrometry. 2015. 26 (6), 899-910.

9. Bokhart MT, Rosen EP, Thompson CG, Sykes C, Kashuba ADM, Muddiman DC. Quantitative mass spectrometry imaging of emtricitabine in cervical tissue model using

infrared matrix-assisted laser desorption electrospray ionization. Analytical and Bioanalytical Chemistry. 2015, 407, 2073-2084.

10. Barry JA, Robichaud G, Bokhart MT, Thompson CG, Sykes C, Kashuba ADM, Muddiman DC. Mapping antiretroviral drugs in tissue by IR-MALDESI MSI coupled to the Q Exactive and comparison with LC-MS/MS SRM assay. Journal of The American Society for Mass Spectrometry. 2014, 25, 2038-2047.

xvi

1 Introduction 1.1 Soft Ionization Methods

The development of the soft ionization techniques of electrospray ionization [1] (ESI) and matrix-assisted laser desorption ionization [2,3] (MALDI) revolutionized the field of mass spectrometry (MS). Soft ionization refers to the ability of the ionization method to transfer charge onto an analyte without imparting excessive energy resulting in fragmentation of the analyte molecule. These methods found use in the analysis of large biomolecules leading to many exciting discoveries of their roles in biological systems. For their contributions to ionization methods, John Fenn and Koichi Tanaka, the pioneers of ESI and MALDI respectively, were awarded part of the Nobel Prize in Chemistry in 2002.

1.1.1 Electrospray Ionization

Electrospray ionization generates ions by the formation of charged droplets containing analyte and ultimately transfering charge onto the analyte molecule. Charged droplets are formed by applying an electric potential to a conductive solution flowing through an emitter tip. In positive ionization mode, chemical oxidation of water produces an excess of H+ ions, which migrate toward the counter electrode (MS inlet). The buildup of charge causes the elongation of the solvent droplet at the emitter tip forming a Taylor cone. When the charge density exceeds the surface tension of the solvent, charged droplets are expelled from the Taylor cone. These charged droplets are then drawn toward the MS inlet through the applied electric field, where evaporative loss of solvent reduces the volume of the charged droplets. When the electrostatic repulsion of the charges in the droplet exceeds the surface tension of the solvent droplet, Coulombic fission occurs resulting in the ejection of highly charged progeny droplets of reduced volume. These progeny droplets undergo subsequent desolvation and Coulombic fission, reducing size even further. An overview of the entire electrospray ionization process is shown in Figure 1-1.

Figure 1-1. Schematic of electrospray ionization. There are two commonly accepted mechanisms for analyte charging in ESI. The charged residue model (CRM) was proposed by Dole and coworkers in 1968 [4]. In this model, desolvation and fission occurs until there is a single analyte molecule within a charged droplet. When the solvent from this final droplet evaporates, charge is then attached to the single molecule in the droplet. Alternatively, the ion evaporation model proposed by Iribarne and Thomson in 1976 [5], proposed analytes are ejected from the surface of charged droplets, removing charge in the process, to reduce electrostatic repulsion in the droplet. Recent studies indicate that ionization in ESI is likely a result of both mechanisms occurring simultaneously, with small molecules more likely to follow the ion evaporation model and large biomolecules following charged residue model [6]. The use of ESI is particularly useful in coupling to liquid chromatography (LC) based separations in that ionization of analytes can occur directly online with the separation while at atmospheric pressure. ESI is a continuous ionization source, allowing constant ionization of the LC eluent. Additionally, ESI affords multiple charging of large biomolecules, facilitating detection by a variety of mass spectrometers that are limited in upper m/z dynamic range.

1.1.2 Matrix Assisted Laser Desorption Ionization (MALDI)

Matrix-assisted laser desorption ionization was introduced in 1989 by Tanaka [3], and Karas and Hillenkamp [2], both describing the soft ionization of molecules using an energy- absorbing matrix to facilitate desorption/ionization of an analyte. In MALDI, analyte molecules are cocrystalized with an ultraviolet (UV) absorbing matrix, typically a small organic acid, prior to irradiation by a laser. The absorption of laser radiation by the matrix causes desorption and ionization of the analytes in the sample. The method has found use analyzing a variety of large biomolecules including intact protein, DNA, as well as small metabolites. It is commonly accepted that analyte charging in MALDI occurs by one of two proposed models; the lucky survivor and gas phase protonation model. The lucky survivor model states that analyte charging occurs in solution and the crystallized matrix. At typical preparation conditions, analytes are likely incorporated into the matrix as precharged analytes with their respective counterions. Upon laser desorption, cluster dissociation can cause counterion neutralization leaving only the charged analyte molecule to be detected. Alternatively, counterion separation in the gas phase may occur during desorption. The gas phase protonation model describes analyte ionization by way of gas phase proton transfer reactions between matrix molecules and neutral analyte molecules. Gas phase collisions with protonated or deprotonated matrix molecules lead to charge transfer to analyte molecule. In both models, singly or lowly charged analyte molecules are produced. Based on the proposed ionization mechanisms at play in MALDI, it is clear that careful selection of matrix/analyte pair and application technique is essential for successful analyte ionization in MALDI. Matrix selection is often through empirical selection with matrix to analyte ratios typically 1000:1. Additionally, matrix application method has a significant impact on the ionization of analyte as the analyte molecules need to cocrystalize with matrix, a statement which will have significant impact on mass spectrometry imaging (MSI) using

MALDI (vide infra). Analysis is typically performed under vacuum, thus limiting the technique to nonvolatile molecules and requiring water to be removed from the sample prior to analysis. Despite these conditions, MALDI has proven itself as a sensitive analytical technique for the analysis of a wide variety of analytes including both large and small biomolecules, polymers, and small molecules.

1.1.3 Matrix Assisted Laser Desorption Electrospray Ionization (MALDESI)

Hybrid ionization techniques, those combining more than one established ionization method, can be coupled to offer advantages of both methods. Matrix-assisted laser desorption electrospray ionization (MALDESI) was first presented in 2006 by Muddiman and coworkers using a UV laser to resonantly excite a matrix in a manner analogous to MALDI with secondary post ionization by ESI to increase ionization yield [7]. A variety of techniques utilizing laser ablation coupled to electrospray post-ionization have been reported including non-resonant ablation in electrospray-assisted laser desorption/ionization (ELDI) [8] and laser electrospray mass spectrometry (LEMS) [9], and resonant ablation in laser electrospray ionization (LAESI) [10] and infrared laser-assisted desorption electrospray ionization (IR-LADESI) [11]. A visualization of the MALDESI process is shown in Figure 1-2. In a series of initial experiments, ionization of analytes in MALDESI were found to occur in an ESI-like manner. Analysis of stock protein solutions demonstrate multi-charged analytes with nearly identical charge state distributions to that of analytes introduced in the ESI solution [7]. Implementation of a remote analyte sampling, transfer and ionization relay (RASTIR) device allowed the desorption and electrospray ionization events to be decoupled in space. The use of deuterated solvents within the electrospray solution definitively showed [M+D+]+ species indicating ESI-like ionization [12].

Figure 1-2. Visualization of the MALDESI process. Laser radiation impinges on a sample causing desorption of material. The desorbed sample plume and electrospray plume overlap, where analytes partition into electrospray droplets and subsequent ionization in an ESI-like manner. The use of a mid-IR laser (2.94 µm) in MALDESI allows the use of endogenous and exogenous water to be used as a matrix thereby simplifying the sample preparation steps and providing spectra without matrix peak interference [13]. Additionally, the mid-IR laser allowed facile analysis of biological samples including thin fresh frozen tissue sections. The full potential of IR-MALDESI is realized in its application to mass spectrometry imaging (MSI) analysis of tissues [14]. The higher fluence of the mid-IR laser compared to the UV laser used in MALDI allows IR-MALDESI to completely ablate through a 10 μm thick tissue section at each rastered position with two laser pulses, resulting in analysis of a voxel of tissue [13]. Additionally, the complete ablation at each rastered position allows reproducible volumes of sample to be desorbed in an oversampling technique where the sample is translated a distance less than the desorption threshold of the laser [15,16].

1.2 Fourier Transform Mass Spectrometry

The charged molecules formed in the ionization process are subsequently measured by a variety of instruments, all of which measure an ions mass-to-charge ratio (m/z). There are

several mass spectrometer types to measure a molecules m/z ratio including magnetic and electric sector, time-of-flight, quadrupole mass filter, ion trap and Fourier transform ion cyclotron resonance (FTICR), each with specific advantages and disadvantages. All work presented within was conducted using a Q Exactive mass spectrometer, a hybrid instrument consisting of a low resolving power isolation quadrupole and a high resolving power orbitrap mass analyzer. A schematic of the instrument and annotation of important parts are given in Figure 1-3. The Q Exactive has many advantages including high resolving power mass analyzer, MS/MS fragmentation, high speed and sensitivity.

Figure 1-3. Schematic of the Q Exactive mass spectrometer. Figure adapted from http://planetorbitrap.com/q-exactive, accessed 8-29-17. A quadrupole mass filter consists of four parallel cylindrical or parabolic rods and acts to only transmit ions with a stable trajectory through the quadrupole. Opposing rods are connected electrically with a radiofrequency (RF) and direct current (DC) voltage applied.

Only ions with a stable trajectory based on the Mathieu stability diagram for the given RF and DC voltage will be transmitted. The coupling of a quadrupole mass analyzer to a high resolving power trapping analyzer such as the orbitrap allows only ions of interest to be injected into the high resolving power orbitrap mass analyzer [17]. The orbitrap mass analyzer is an electrostatic trapping device similar in principal to the Kingdon trap first reported in 1923 [18]. An orbitrap uses a central spindle-shaped electrode and two electrically isolated outer electrodes to electrostatically trap an ion packet. The axial oscillation of the ion packet within the orbitrap results in an image current to be detected on the outer electrodes of the trap. The complex waveform can then be decomposed into the individual sinusoidal wave components resulting from specific m/z ion packets using an enhanced Fourier transform. The axial frequencies of individual ion packets can then be given as their m/z by equation 1: ω=√(e/((m⁄z) )∙k) (1) where ω is the frequency of axial oscillation, e is the elementary charge of an electron, and k is a constant. High resolving power data is generated using a Q Exactive instrument, with experiments typically performed with 140,000 resolving power (FWHM, m/z 200) requiring a 512 ms transient. High mass measurement accuracy is maintained by use of a lock mass, in which ions of known exact mass are used as a single point recalibration parameter for each mass spectrum taken. The instrument collects approximately 4,000 ions of the lock mass prior to filling of the C-trap with analyte ions to ensure lock mass ion is detected in every scan [19].

1.3 Mass Spectrometry Imaging (MSI)

Mass spectrometry imaging (MSI) data is generated by recording mass spectra and the corresponding spatial location [20]. Construction of ion heat maps creates a visual representation of a compound relating the observed ion abundance with its location. MSI is quickly becoming an invaluable tool in drug distribution studies [21-23]. The efficacy of a

drug can be limited by the ability to reach its intended target; plasma concentrations of a drug are often used but may not accurately reflect the concentration of the drug at its site of action [24]. Thus, spatial distribution of a drug in tissue can provide vital information related to the efficacy of a drug. Moreover, the use of a MSI strategy can provide additional information about endogenous compounds and metabolites. Several ionization methods have been successfully used for MSI, with MALDI MSI being the most common. The pulsed nature of MALDI and desorption from discrete regions makes MSI data acquisition simple with the record of a mass spectrum and location from which it was generated. Since the initial report of peptides by MALDI MSI [25], MALDI has been used extensively for a wide variety of analytes, samples, and methods. Another ionization technique routinely used for MSI is desorption electrospray ionization (DESI) which utilizes electrospray ionization impinging on a translated sample surface [26,27]. The electrospray plume interacts with a surface, causing the ejection of charged secondary droplets from the surface which are then sampled by the MS. MALDESI MSI can provide the advantages of both MSI methodologies: pulsed nature of MALDI and resolution of small spatial features with the more efficient ionization of electrospray. MSI is an attractive technique for drug distribution studies as it is a label-free method capable of multiplexed and/or untargeted analysis. Simultaneous analysis of multiple species while retaining spatial specificity is a challenge for current quantitative methods. Selected reaction monitoring (SRM) assays using liquid chromatography tandem mass spectrometry (LC-MS/MS) for quantification of pharmaceuticals are commonly used for pharmacokinetic (PK) analysis of drugs in plasma and tissues [28]. These methods typically require extraction of the analyte from a tissue homogenate, and result in the loss of spatial information within a tissue or organ. Alternative approaches such as quantitative whole body autoradiography (QWBA) or positron emission tomography (PET) offer the ability to visualize drug distribution into tissues while providing quantitative information [23,29,30]. However, QWBA and PET require radiolabeled compounds which only give information about the radiolabel and are

therefore not able to distinguish between the parent compound and its metabolites. Further, these quantitative imaging techniques incur significant experimental cost especially in the analysis of multiple analytes in multi-drug therapies. Although MSI has proven its utility to provide important spatial information of drugs in tissue, quantitative information in MSI has proven difficult to achieve. MALDI has been used extensively in qualitative MSI and with some success of quantification [20,29,31-59]. The recent advances in quantification using MSI were the subject of recent review [60]. Quantitative MALDI MSI has several limitations including the need for organic matrix deposition for ionization of analytes; the analysis is also typically performed under vacuum. Ambient ionization mass spectrometry allows tissues to be sampled at conditions much closer to their natural state [61]. Pixel-to-pixel variability in MSI poses the biggest challenge for quantification [62]. This variability can arise from a range of sources, including morphological features, ionization efficiency, and detection efficiency. Increasing sensitivity and reducing variability are essential steps toward making MSI a routine quantitative technique. Several groups have reported various normalization methods to account for tissue specific signal response and reduced variation per pixel [39,47,48]. These studies exemplify the need for a suitable normalization compound to produce quantitative MSI data. The ideal normalization compound accounts for structure specific ablation and ionization efficiency for the analyte. Stable isotope labeled compounds are the best normalization compounds, as they are chemical and structural analogues of the target analyte.

1.4 Human Immunodeficiency Virus (HIV) Treatment with Antiretroviral Drugs

The need for a reliable method of characterizing spatial the distribution of drug therapies within tissues, and the advantages of MSI, is exemplified within the field of human

immunodeficiency virus (HIV). HIV replication has been shown to persist in certain anatomic sites such as the lymphatic system and reproductive tract [63]. These viral reservoirs represent a significant obstacle in the cure of HIV. Evaluations of ARV penetration into these reservoirs are critical for understanding whether current therapies will be sufficient to completely eliminate HIV from the body. Treatment of HIV commonly requires taking a combination of ARV drugs, a treatment strategy termed highly active antiretroviral therapy (HAART) [64,65]. The ability for MSI to detect multiple drugs simultaneously, without the need for chemical labeling as in fluorescence or QWBA, is desirable. MSI represents a promising technique to advance the understanding of within-tissue ARV distribution and concentration, and provide invaluable information toward the development of drug therapies targeting HIV reservoirs.

1.5 Synopsis of Completed Work

The work described within details the analytical advancement of the IR-MALDESI MSI over the course of my graduate studies. Upon joining the research laboratory, the technology to make IR-MALDESI a mass spectrometry imaging technique was being developed [14] and initial parameters were being optimized [13]. Although IR-MALDESI was capable of producing qualitative distributions of analytes, the quantitative abilities of the instrument had not been fully explored. Chapter 2 details the effects of desorption conditions on MSI analyses by IR-MALDESI. The laser ablation wavelength, fluence, geometric configuration and presence of ice matrix were all examined. The optimized fluence and geometric configuration were used with an ice matrix for a whole-body lipidomic investigation of a mouse specimen. Two whole body sections were analyzed with 2940 and 3100 nm laser wavelengths for the ablation process, with the detected lipids being categorized by laser wavelength preference. Chapter 3 examines the effects of shorter C-trap injection times on ion abundance made possible by a higher repetition rate laser. The Q Exactive mass spectrometer is operated with an experimental, fixed injection time to collect all sample related ions through an agreement

with the instrument manufacturer, Thermo Scientific. The shorter injection time more closely represents injection times of a typical LC-MS run that the instrument is commonly used and designed for. The effect of shorter C-trap injection time was defined by increases in ion abundance of analytes in targeted analyses, reduced variability through normalization and increased number of identifications in untargeted analyses. A quantitative mass spectrometry imaging method using IR-MALDESI is detailed in Chapter 4. The optimization of electrospray ionization modifiers was first presented, showing increased ion abundance and reduced variability with the use of metal cations. The incorporation of a normalization compound is shown to reduce variability on a per-voxel basis, which is necessary for absolute quantification using MSI. A calibration curve was generated using the stable isotope labeled version of emtricitabine, which allowed the calibration curve to be incorporated on the tissue to be quantified. The MSI quantification procedure was performed 5 times, and validated using LC-MS/MS. Chapter 5 uses the quantification procedure established in Chapter 4 to quantify efavirenz in a dosed, non-human primate study. Heterogeneous distribution of EFV was found in several tissue reservoirs that were analyzed by MSI. The MSI results were compared to traditional biochemical technique of H&E staining and immunohistochemistry to understand the heterogeneous EFV distribution. Chapter 6 uses IR-MALDESI MSI to investigate antiretroviral drug incorporation into patient’s hair as a way to measure drug regimen adherence. IR-MALDESI MSI was used to analyze the distribution of ARVs in hair without sample preparation steps. However, to compare across patient hair types, a normalization method involving peroxidation of hair melanin was developed to account for variations in drug incorporation based on hair color. The spatial resolution of IR-MALDESI MSI is inherently tied to the laser ablation spot size. Chapter 7 describes a method to reduce the laser focus size in an effort to perform MSI at higher spatial resolutions. Using an iris, 3.75x beam expander and aspheric focusing lens, the laser focus was dramatically reduced. Important laser characteristics were defined for a simple, single-optic design and the multi-element design to objectively compare the new

optical system. With the multi-element design, IR-MALDESI MSI was performed at 50- micron spatial resolution on a hen ovarian tissue. Chapter 8 details several new tools that have been incorporated into MSiReader for MSI data analysis. Tools that have been developed to facilitate drug distribution analyses include the ability to load multiple data sets at one time, a polarity filter to analyze polarity-switching MSI data, an image overlay tool, a pop-out quantification tool, and the ability to create mass measurement accuracy heat maps.

1.6 References

1. Fenn, J.B., Mann, M., Meng, C.K., Wong, S.F., Whitehouse, C.M.: Electrospray ionization for mass spectrometry of large biomolecules. Science. 246, 64-71 (1989)

2. Karas, M., Hillenkamp, F.: Laser desorption ionization of proteins with molecular masses exceeding 10,000 daltons. Analytical chemistry. 60, 2299-2301 (1988)

3. Tanaka, K., Waki, H., Ido, Y., Akita, S., Yoshida, Y., Yoshida, T., Matsuo, T.: Protein and polymer analyses up to m/z 100 000 by laser ionization time-of-flight mass spectrometry. Rapid communications in mass spectrometry. 2, 151-153 (1988)

4. Dole, M., Mack, L., Hines, R., Mobley, R., Ferguson, L., Alice, M.D.: Molecular beams of macroions. The Journal of Chemical Physics. 49, 2240-2249 (1968)

5. Iribarne, J., Thomson, B.: On the evaporation of small ions from charged droplets. The Journal of Chemical Physics. 64, 2287-2294 (1976)

6. Cole, R.B.: Some tenets pertaining to electrospray ionization mass spectrometry. Journal of Mass Spectrometry. 35, 763-772 (2000)

7. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Generation and detection of multiply-charged peptides and proteins by matrix-assisted laser desorption electrospray ionization (MALDESI) Fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry. 17, 1712-1716 (2006)

8. Shiea, J., Huang, M.-Z., Hsu, H.-J., Lee, C.-Y., Yuan, C.-H., Beech, I., Sunner, J.: Electrospray-assisted laser desorption/ionization mass spectrometry for direct ambient

analysis of solids. Rapid Communications in Mass Spectrometry. 19, 3701-3704 (2005)

9. Brady, J.J., Judge, E.J., Levis, R.J.: Mass spectrometry of intact neutral macromolecules using intense non-resonant femtosecond laser vaporization with electrospray post-ionization. Rapid communications in mass spectrometry : RCM. 23, 3151-3157 (2009)

10. Nemes, P., Vertes, A.: Laser ablation electrospray ionization for atmospheric pressure, in vivo, and imaging mass spectrometry. Analytical chemistry. 79, 8098-8106 (2007)

11. Rezenom, Y.H., Dong, J., Murray, K.K.: Infrared laser-assisted desorption electrospray ionization mass spectrometry. The Analyst. 133, 226-232 (2008)

12. Dixon, R.B., Muddiman, D.C.: Study of the ionization mechanism in hybrid laser based desorption techniques. The Analyst. 135, 880-882 (2010)

13. Robichaud, G., Barry, J.A., Muddiman, D.C.: IR-MALDESI mass spectrometry imaging of biological tissue sections using ice as a matrix. Journal of the American Society for Mass Spectrometry. 25, 319-328 (2014)

14. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) imaging source coupled to a FT-ICR mass spectrometer. Journal of the American Society for Mass Spectrometry. 24, 92-100 (2013)

15. Jurchen, J.C., Rubakhin, S.S., Sweedler, J.V.: MALDI-MS imaging of features smaller than the size of the laser beam. Journal of the American Society for Mass Spectrometry. 16, 1654-1659 (2005)

16. Nazari, M., Muddiman, D.: Cellular-level mass spectrometry imaging using infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) by oversampling. Analytical and bioanalytical chemistry. 407, 2265-2271 (2015)

17. Michalski, A., Damoc, E., Hauschild, J.-P., Lange, O., Wieghaus, A., Makarov, A., Nagaraj, N., Cox, J., Mann, M., Horning, S.: Mass Spectrometry-based Proteomics Using Q Exactive, a High-performance Benchtop Quadrupole Orbitrap Mass Spectrometer. Molecular & Cellular Proteomics. 10, (2011)

18. Kingdon, K.: A method for the neutralization of electron space charge by positive ionization at very low gas pressures. Physical Review. 21, 408 (1923)

19. Olsen, J.V., de Godoy, L.M.F., Li, G., Macek, B., Mortensen, P., Pesch, R., Makarov, A., Lange, O., Horning, S., Mann, M.: Parts per Million Mass Accuracy on an Orbitrap Mass Spectrometer via Lock Mass Injection into a C-trap. Molecular & Cellular Proteomics. 4, 2010-2021 (2005)

20. Caprioli, R.M., Farmer, T.B., Gile, J.: Molecular Imaging of Biological Samples: Localization of Peptides and Proteins Using MALDI-TOF MS. Analytical chemistry. 69, 4751-4760 (1997)

21. Prideaux, B., Staab, D., Stoeckli, M.: Applications of MALDI-MSI to pharmaceutical research. Methods in molecular biology. 656, 405-413 (2010)

22. Prideaux, B., Stoeckli, M.: Mass spectrometry imaging for drug distribution studies. Journal of proteomics. 75, 4999-5013 (2012)

23. Solon, E.G., Schweitzer, A., Stoeckli, M., Prideaux, B.: Autoradiography, MALDI- MS, and SIMS-MS imaging in pharmaceutical discovery and development. The AAPS journal. 12, 11-26 (2010)

24. Cohen, M.S., Gay, C., Kashuba, A.D.M., Blower, S., Paxton, L.: Narrative Review: Antiretroviral Therapy to Prevent the Sexual Transmission of HIV-1. Ann Intern Med. 146, 591-601 (2007)

25. Caprioli, R.M., Farmer, T.B., Gile, J.: Molecular Imaging of Biological Samples: Localization of Peptides and Proteins Using MALDI-TOF MS. Analytical Chemistry. 69, 4751-4760 (1997)

26. Takáts, Z., Wiseman, J.M., Gologan, B., Cooks, R.G.: Mass Spectrometry Sampling Under Ambient Conditions with Desorption Electrospray Ionization. Science. 306, 471-473 (2004)

27. Wiseman, J.M., Ifa, D.R., Song, Q., Cooks, R.G.: Tissue imaging at atmospheric pressure using desorption electrospray ionization (DESI) mass spectrometry. Angewandte Chemie International Edition. 45, 7188-7192 (2006)

28. Xu, R.N., Fan, L., Rieser, M.J., El-Shourbagy, T.A.: Recent advances in high- throughput quantitative bioanalysis by LC-MS/MS. Journal of pharmaceutical and biomedical analysis. 44, 342-355 (2007)

29. Drexler, D.M., Tannehill-Gregg, S.H., Wang, L., Brock, B.J.: Utility of quantitative whole-body autoradiography (QWBA) and imaging mass spectrometry (IMS) by

matrix-assisted laser desorption/ionization (MALDI) in the assessment of ocular distribution of drugs. Journal of pharmacological and toxicological methods. 63, 205- 208 (2011)

30. Rudin, M., Weissleder, R.: Molecular imaging in drug discovery and development. Nature reviews. Drug discovery. 2, 123-131 (2003)

31. Bunch, J., Clench, M.R., Richards, D.S.: Determination of pharmaceutical compounds in skin by imaging matrix-assisted laser desorption/ionisation mass spectrometry. Rapid communications in mass spectrometry : RCM. 18, 3051-3060 (2004)

32. Casadonte, R., Caprioli, R.M.: Proteomic analysis of formalin-fixed paraffin- embedded tissue by MALDI imaging mass spectrometry. Nature protocols. 6, 1695- 1709 (2011)

33. Djidja, M.C., Chang, J., Hadjiprocopis, A., Schmich, F., Sinclair, J., Mrsnik, M., Schoof, E.M., Barker, H.E., Linding, R., Jorgensen, C., Erler, J.T.: Identification of Hypoxia-Regulated Proteins Using MALDI-Mass Spectrometry Imaging Combined with Quantitative Proteomics. Journal of proteome research. 13, 2297-2313 (2014)

34. 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.: Utility of imaging mass spectrometry (IMS) by matrix-assisted laser desorption ionization (MALDI) on an ion trap mass spectrometer in the analysis of drugs and metabolites in biological tissues. Journal of pharmacological and toxicological methods. 55, 279-288 (2007)

35. Fehniger, T.E., Vegvari, A., Rezeli, M., Prikk, K., Ross, P., Dahlback, M., Edula, G., Sepper, R., Marko-Varga, G.: Direct demonstration of tissue uptake of an inhaled drug: proof-of-principle study using matrix-assisted laser desorption ionization mass spectrometry imaging. Analytical chemistry. 83, 8329-8336 (2011)

36. Goodwin, R.J., Mackay, C.L., Nilsson, A., Harrison, D.J., Farde, L., Andren, P.E., Iverson, S.L.: Qualitative and quantitative MALDI imaging of the positron emission tomography ligands raclopride (a D2 dopamine antagonist) and SCH 23390 (a D1 dopamine antagonist) in rat brain tissue sections using a solvent-free dry matrix application method. Analytical chemistry. 83, 9694-9701 (2011)

37. Goto, T., Terada, N., Inoue, T., Nakayama, K., Okada, Y., Yoshikawa, T., Miyazaki, Y., Uegaki, M., Sumiyoshi, S., Kobayashi, T., Kamba, T., Yoshimura, K., Ogawa, O.: The expression profile of phosphatidylinositol in high spatial resolution imaging mass spectrometry as a potential biomarker for prostate cancer. PloS one. 9, e90242 (2014)

38. Groseclose, M.R., Castellino, S.: A Mimetic Tissue Model for the Quantification of Drug Distributions by MALDI Imaging Mass Spectrometry. Analytical Chemistry. 85, 10099-10106 (2013)

39. Hamm, G., Bonnel, D., Legouffe, R., Pamelard, F., Delbos, J.M., Bouzom, F., Stauber, J.: Quantitative mass spectrometry imaging of propranolol and olanzapine using tissue extinction calculation as normalization factor. Journal of proteomics. 75, 4952-4961 (2012)

40. Handberg, E., Chingin, K., Wang, N., Dai, X., Chen, H.: Mass spectrometry imaging for visualizing organic analytes in food. Mass spectrometry reviews. 34, 641-658 (2014)

41. Hsieh, Y., Casale, R., Fukuda, E., Chen, J., Knemeyer, I., Wingate, J., Morrison, R., Korfmacher, W.: Matrix-assisted laser desorption/ionization imaging mass spectrometry for direct measurement of clozapine in rat brain tissue. Rapid communications in mass spectrometry : RCM. 20, 965-972 (2006)

42. Hsieh, Y., Chen, J., Korfmacher, W.A.: Mapping pharmaceuticals in tissues using MALDI imaging mass spectrometry. Journal of pharmacological and toxicological methods. 55, 193-200 (2007)

43. 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.: A quantitation method for mass spectrometry imaging. Rapid communications in mass spectrometry : RCM. 25, 503-510 (2011)

44. Kubo, A., Kajimura, M., Suematsu, M.: Matrix-Assisted Laser Desorption/Ionization (MALDI) Imaging Mass Spectrometry (IMS): A Challenge for Reliable Quantitative Analyses. Mass Spectrom (Tokyo). 1, A0004 (2012)

45. Lietz, C.B., Gemperline, E., Li, L.J.: Qualitative and quantitative mass spectrometry imaging of drugs and metabolites. Advanced drug delivery reviews. 65, 1074-1085 (2013)

46. Nilsson, A., Fehniger, T.E., Gustavsson, L., Andersson, M., Kenne, K., Marko-Varga, G., Andren, P.E.: Fine mapping the spatial distribution and concentration of unlabeled drugs within tissue micro-compartments using imaging mass spectrometry. PloS one. 5, e11411 (2010)

47. Pirman, D.A., Kiss, A., Heeren, R.M.A., Yost, R.A.: Identifying Tissue-Specific Signal Variation in MALDI Mass Spectrometric Imaging by Use of an Internal Standard. Analytical Chemistry. 85, 1090-1096 (2013)

48. Pirman, D.A., Reich, R.F., Kiss, A., Heeren, R.M., Yost, R.A.: Quantitative MALDI tandem mass spectrometric imaging of cocaine from brain tissue with a deuterated internal standard. Analytical chemistry. 85, 1081-1089 (2013)

49. Pirman, D.A., Yost, R.A.: Quantitative tandem mass spectrometric imaging of endogenous acetyl-L-carnitine from piglet brain tissue using an internal standard. Analytical chemistry. 83, 8575-8581 (2011)

50. Prideaux, B., Dartois, V., Staab, D., Weiner, D.M., Goh, A., Via, L.E., Barry, C.E., 3rd, Stoeckli, M.: High-sensitivity MALDI-MRM-MS imaging of moxifloxacin distribution in tuberculosis-infected rabbit lungs and granulomatous lesions. Analytical chemistry. 83, 2112-2118 (2011)

51. Shin, Y.G., Dong, T., Chou, B., Menghrajani, K.: Determination of loperamide in mdr1a/1b knock-out mouse brain tissue using matrix-assisted laser desorption/ionization mass spectrometry and comparison with quantitative electrospray-triple quadrupole mass spectrometry analysis. Archives of pharmacal research. 34, 1983-1988 (2011)

52. Signor, L., Varesio, E., Staack, R.F., Starke, V., Richter, W.F., Hopfgartner, G.: Analysis of erlotinib and its metabolites in rat tissue sections by MALDI quadrupole time-of-flight mass spectrometry. Journal of mass spectrometry : JMS. 42, 900-909 (2007)

53. Stoeckli, M., Staab, D., Schweitzer, A.: Compound and metabolite distribution measured by MALDI mass spectrometric imaging in whole-body tissue sections. International Journal of Mass Spectrometry. 260, 195-202 (2007)

54. Sugiura, Y., Konishi, Y., Zaima, N., Kajihara, S., Nakanishi, H., Taguchi, R., Setou, M.: Visualization of the cell-selective distribution of PUFA-containing phosphatidylcholines in mouse brain by imaging mass spectrometry. Journal of lipid research. 50, 1776-1788 (2009)

55. Sun, N., Walch, A.: Qualitative and quantitative mass spectrometry imaging of drugs and metabolites in tissue at therapeutic levels. Histochemistry and cell biology. 140, 93-104 (2013)

56. Takai, N., Tanaka, Y., Inazawa, K., Saji, H.: Quantitative analysis of pharmaceutical drug distribution in multiple organs by imaging mass spectrometry. Rapid communications in mass spectrometry : RCM. 26, 1549-1556 (2012)

57. Trede, D., Schiffler, S., Becker, M., Wirtz, S., Steinhorst, K., Strehlow, J., Aichler, M., Kobarg, J.H., Oetjen, J., Dyatlov, A., Heldmann, S., Walch, A., Thiele, H., Maass, P., Alexandrov, T.: Exploring three-dimensional matrix-assisted laser desorption/ionization imaging mass spectrometry data: three-dimensional spatial segmentation of mouse kidney. Analytical chemistry. 84, 6079-6087 (2012)

58. Velickovi, X.D., Ropartz, D., Guillon, F., Saulnier, L., Rogniaux, H.: New insights into the structural and spatial variability of cell-wall polysaccharides during wheat grain development, as revealed through MALDI mass spectrometry imaging. Journal of experimental botany. 65, 2079-2091 (2014)

59. Barry, J.A., Groseclose, M.R., Robichaud, G., Castellino, S., Muddiman, D.C.: Assessing drug and metabolite detection in liver tissue by UV-MALDI and IR- MALDESI mass spectrometry imaging coupled to FT-ICR MS. International Journal of Mass Spectrometry. 377, 448-455 (2015)

60. Ellis, S.R., Bruinen, A.L., Heeren, R.M.: A critical evaluation of the current state-of- the-art in quantitative imaging mass spectrometry. Analytical and bioanalytical chemistry. 406, 1275-1289 (2014)

61. Nemes, P., Vertes, A.: Ambient mass spectrometry for in vivo local analysis and in situ molecular tissue imaging. TrAC Trends in Analytical Chemistry. 34, 22-34 (2012)

62. Cohen, L.H., Gusev, A.I.: Small molecule analysis by MALDI mass spectrometry. Analytical and Bioanalytical Chemistry. 373, 571-586 (2002)

63. Chun, T.W., Nickle, D.C., Justement, J.S., Meyers, J.H., Roby, G., Hallahan, C.W., Kottilil, S., Moir, S., Mican, J.M., Mullins, J.I., Ward, D.J., Kovacs, J.A., Mannon, P.J., Fauci, A.S.: Persistence of HIV in gut-associated lymphoid tissue despite long-term antiretroviral therapy. The Journal of infectious diseases. 197, 714-720 (2008)

64. Behrens, G., Knuth, C., Schedel, I., Mendila, M., Schmidt, R.E.: Highly active antiretroviral therapy. The Lancet. 351, 1057-1058 (1998)

65. Launay, O., Tod, M., Tschope, I., Si-Mohamed, A., Bélarbi, L., Charpentier, C., Goujard, C., Taburet, A.-M., Lortholary, O., Leroy, V.: Residual HIV-1 RNA and HIV-

1 DNA production in the genital tract reservoir of women treated with HAART: the prospective ANRS EP24 GYNODYN study. Antiviral therapy. 16, 843 (2011)

2 Influence of Desorption Conditions on Analyte Sensitivity and Internal Energy in Discrete Tissue or Whole Body Imaging by IR-MALDESI

The following work was reprinted with permission from: Rosen, E.P., Bokhart, M.T., Ghashghaei, H.T., Muddiman, D.C.: Influence of Desorption Conditions on Analyte Sensitivity and Internal Energy in Discrete Tissue or Whole Body Imaging by IR-MALDESI. Journal of The American Society for Mass Spectrometry. 26, 899-910 (2015). DOI: 10.1007/s13361-015-1114-1. Copyright © 2015 American Society for Mass Spectrometry. The original publication may be accessed via the Internet at https://link.springer.com/article/10.1007%2Fs13361-015-1114-1

2.1 Introduction

Mass spectrometry imaging (MSI) is an analytical approach that provides the capability for simultaneously monitoring the spatial distribution within a biological sample of analytes ranging from small molecules, such as pharmaceutical drugs [1], to lipids [2, 3], to peptides and proteins [4]. The utility of MSI has been demonstrated for a variety of fields, with particular emphasis on biomedical applications like drug distribution [5] and biomarker identification [6]. Spatial scales of analysis can vary widely based on application from cellular (tumor margin identification) [7] to multi-organ (pathway analysis), to whole body (drug and metabolite distribution) imaging [8]. MSI sensitivity and selectivity toward different compound classes are dependent on sample preparation [9] and the methodology used to desorb and ionize an analyte from a sample. For the analysis of lipids and other labile small molecules, which have emerged as areas with relevance in drug studies and biomarker identification, the predominant MSI technique, matrix-assisted laser desorption/ionization (MALDI), has some limitations. This is largely due to the fact that MALDI requires the homogeneous application of a non-native organic matrix, which must be judiciously selected to target an analyte class of interest. For typical MALDI MSI using a time-of-flight (TOF) analyzer, the matrix and analytes must also be stable under high vacuum conditions. MALDI matrices also dominate the spectral response in the low mass range making the analysis of labile, small molecules and their metabolites

challenging because of isobaric interference, particularly when coupled to relatively low resolving power TOF MS. Ambient ionization techniques such as desorption electrospray ionization (DESI) [10], liquid extraction surface analysis (LESA) [11], and matrix-assisted laser desorption electrospray ionization (MALDESI) [12] offer alternatives to MALDI MSI with minimal sample preparation and none of the sample stability limitations of high vacuum. Unlike surface-based techniques that rely on liquid extraction of analytes, which can cause temporal response and selective extraction of analytes, desorption of materials by pulsed laser excitation is nearly instantaneous. At infrared wavelengths, incident radiation can excite and desorb a broad range of condensed phase biological targets [13-15]. While the ionization yield of the resulting desorbed plume is low, limiting the applicability of IR-LDI, MALDESI was the first ambient ionization technique to overcome this issue by coupling resonant laser desorption with electrospray post-ionization [16]. Like the subsequent and analogous methods LAESI [17] and LADESI [18], IR-MALDESI has utilized an incident wavelength of 2940 nm, corresponding to the absorption cross-section peak for the OH stretching band of liquid water. This wavelength excites species natively present in biological samples, such as endogenous water, thereby eliminating background interferences associated with non-native matrices. Pulsed infrared laser desorption of a water-rich biological target results in enhanced penetration depth relative to the UV excitation of a MALDI matrix, micrometers versus nanometers, respectively [14]. Specimen sections of standard thickness (10–50 μm) are completely ablated in 2–3 IR laser pulses. This rapid ejection of material associated with the probed volumetric element, or voxel, means that IR-MALDESI is agnostic to the location of internal standards, which can be placed readily above or below a tissue sample [19]. The sensitivity of IR-MALDESI is contingent on a high degree of spatial overlap between the material ejected from the sample surface and the orthogonal electrospray. The vertical distance from a sample surface traveled by a laser desorbed plume front is sensitive to both the laser fluence and the degree to which the wavelength of laser emission overlaps spectral absorption features of the analyte or matrix [20]. Recent IR-MALDESI results indicate

significant gains in sensitivity toward drugs, drug metabolites, and lipids following the addition of an exogenous ice layer to the surface of a tissue sample [21, 22]. Ice is deposited facilely and uniformly onto the surface of a tissue sample, and has an added benefit of maintaining cryopreservation of a sample throughout the imaging process. This is beneficial on multiple fronts since it preserves tissue structure and limits evaporative loss of volatile small molecules or their degradation attributable to enzymatic activity [23]. Although the role of ice as an IR- MALDI matrix remains uncertain [24, 25], we have recently investigated its effect on the IR- MALDESI desorption process [21]. Using shadowgraphy, we showed that the dynamics of an IR-MALDESI plume of desorbed material ablated at an incident wavelength of 2940 nm from a tissue section with an exogenous ice layer are dramatically altered relative to material desorbed from tissue without an ice layer. Based on these observations, differences in plume dynamics with the ice layer were attributed to either a reduction in laser fluence or a more efficient transfer of energy to the sample, or a combination of these factors. Ions were shown to be accumulated over multiple ablation events, suggesting that the excitation matrix is likely a mixture of water and ice. In this work, factors controlling ice-mediated IR-MALDESI laser desorption are systematically explored using the design of experiments (DOE) statistical approach [26, 27] and optimized desorption parameters for the detection of xenobiotics and lipids are identified. Previous IR-MALDESI method optimization using ice has fixed the incident excitation wavelength at 2940 nm, but the ice or ice/water matrix may be more resonantly excited at longer wavelengths. O-H stretch absorption by a thin film of ice can be significantly red-shifted to a wavelength of 3100 nm relative to liquid because of structural and morphologic changes that occur during freezing [28]. Increased ion yields have been observed at 3100 nm from an IR-MALDI analysis of an ice thin film [29, 30]. Both fluence and wavelength can influence IR-MALDI response to a biological analyte when using ice as a matrix [31]. Here, we examine the role of excitation fluence and wavelength in the ice-mediated analysis of biological samples using IR-MALDESI.

The DOE-optimized desorption parameters have then been utilized to investigate IR- MALDESI lipid response across tissue types of a whole organism at two distinct incident wavelengths relevant to the water/ice matrix, 2940 and 3100 nm. Whole body imaging is critically important in ascertaining the localization and distribution of drug and its metabolites across an entire organism, or in discovery-based biomarker identification. It was anticipated that the attributes of IR-MALDESI MSI would complement efforts for whole body imaging by MALDI [32-35] or DESI [8]. Although the ultimate spatial resolution of IR-MALDESI cannot equal that of UV-MALDI because of fundamental diffraction limits of focused infrared light, its 100 μm voxel diameter is comparable to MALDI pixel dimensions selected for practical whole body analysis throughput [32] and equals that of DESI [8] and allows key anatomical features to be distinguished. The deposition and crystallization of an ice matrix is not expected to vary across different tissue types in the manner of MALDI matrices [9, 36, 37], which can lead to ion suppression that complicates interpretation of analyte response and localization across a tissue section encompassing multiple organs or a whole organism [32, 33]. Further, the combination of the IR-MALDESI source with a high resolution, high mass accuracy mass spectrometer [22] is expected to allow more ion peaks of interest to be accurately assigned without the additional necessity of selected reaction monitoring (SRM) to ensure proper peak identification, which better preserves the multiplexed capability that is one of the hallmarks of MSI as first conceived by Caprioli and coworkers [38]. Finally, the response factors of a standard analyte across the tissues of a whole organism have been evaluated to investigate the potential for ion suppression/response factors for different tissue types, as has been observed with MALDI [39]. Cocaine has been used as a standard, sprayed onto glass slides prior to mounting whole body cryosections. By imaging the tissue section using SRM for cocaine and determining its fragmentation ratios in collision induced dissociation, we attempt to evaluate the internal energy distribution of IR-MALDESI at 2940 and 3100 nm.

2.2 Experimental

2.2.1 Chemicals and Materials

HPLC grade methanol, water, and isopentane were purchased from Burdick and Jackson (Muskegon, MI, USA) and formic acid was purchased from Sigma-Aldrich (St. Louis, MO, USA). Emtricitabine (FTC) and raltegravir (RAL) were obtained from the NIH AIDS Reagent Program, directed by the Pathogenesis and Basic Research Branch, Basic Sciences Program, Division of AIDS (DAIDS), NIAID, NIH. Cocaine was purchased from Sigma- Aldrich. All materials were used as purchased without further purification. Cervical tissue was obtained from surgical waste via the University of North Carolina Tissue Procurement Facility through UNC IRB # 09-0921. Written informed consent was obtained from all patients. After harvest, cervical tissue was incubated in a solution of antiretroviral (ARV) therapies containing 100,000 ng/mL of both a nucleoside reverse transcriptase inhibitor FTC and an integrase strand transfer inhibitor RAL for 24 h at 37°C. Cervical tissue was then removed from drug solution and rinsed with fresh culture media before being frozen with dry ice vapor and stored at –80°C. Adult mouse liver and 2-d-old whole neonatal mouse pups were obtained according to Institutional Animal Care and Use Committee (IACUC) and North Carolina State University regulations approved for the Ghashghaei laboratory. Animals were sacrificed by hypothermia in an isopentane/dry ice bath to preserve tissue structure.

2.2.2 Sample Preparation

Each tissue was sectioned at –20°C using a Leica CM1950 cryomicrotome (Buffalo Grove, IL. USA) into slices of 10 or 25 μm for incubated cervical tissue or whole body neonate mouse, respectively. The sections were then thaw-mounted directly onto glass microscope slides for imaging. For evaluation of desorbed internal energy distribution across whole body

sections, cocaine was diluted as received in 50:50 methanol/water to a concentration of 411 μM, and then uniformly sprayed onto glass slides using a pneumatic sprayer (TM Sprayer, LEAP Technologies, Carrboro, NC, USA) prior to mounting a tissue sample resulting in an on-slide analyte distribution of approximately 180 fmol/voxel. Once mounted, the sample was placed on a liquid cooled thermoelectric stage that was cooled to –10°C while under nitrogen purge, after which the sample was exposed to the ambient environment in order to deposit a thin layer of ice over the surface of the tissue. Preservation of the ice layer throughout imaging was ensured by maintaining a stable relative humidity of approximately 10% within the IR- MALDESI source enclosure through the addition of dry nitrogen gas.

2.2.3 Instrumentation

A more detailed description of the IR-MALDESI imaging source and its optimization for tissue imaging has previously been reported [21, 40]. A tunable 20 Hz pulse rate, 7 ns pulse width mid-IR laser (IR-Opolette 2371; Opotek, Carlsbad, CA, USA) is used to resonantly excite the ice matrix layer and sample, facilitating the desorption of neutral molecules from the tissue section. On tissue, the laser beam diameter exceeding the ablation threshold is 150 μm. Incident laser energy is controlled by a custom-built rutile polarizer-based attenuator, and measured using a thermopile detector (Nova 2; Ophir, Jerusalem, Israel). Each voxel is subjected to two laser pulses, completely ablating the probed sample volume. Desorbed neutrals then partition into the charged solvent droplets of an electrospray plume where ions are generated through an ESI-like process. For the imaging experiments, 50% (v/v) aqueous methanol with 0.2% formic acid was used for the positive electrospray solvent, which has been shown to work well for small molecules and lipids [22]. All imaging experiments were performed with an oversampled spot-to-spot distance of 100 μm [41, 42]. The IR-MALDESI imaging source has been coupled to a Thermo Fisher Scientific Q Exactive (Bremen, Germany) such that ion accumulation was triggered to temporally overlap with the pulsing of the laser, and resulting in a single Orbitrap acquisition at each voxel.

Although this strategy precludes the use of automatic gain control (AGC) of the instrument because of the requirement for a fixed ion injection time (IT), mass accuracy was verified to be maintained within 2 ppm using two diisooctyl phthalate peaks (m/z 391.28428 [M + H+]+ and m/z 413.26623 [M + Na+]+) as lockmasses in the instrument control software [43, 44]. The mass range for the Orbitrap acquisition was set to m/z 150–600 for experiments involving incubated tissue, and otherwise set to m/z 250–1000. The mass resolving power was set to 140,000 at m/z 200. For MS2 imaging (MS2I) acquisition, a targeted MS2I method file was created using an inclusion list for isolating the protonated ion of cocaine (m/z 304.1550) with a 4 m/z window followed by ion accumulation in the C-trap. The accumulated ion packet was then fragmented in the higher-energy collisional dissociation (HCD) cell at a normalized collision energy of 25%. Direct infusion of cocaine validated assignment of the unique transitions. The normalized collision energy was optimized from the analysis of a cocaine standard evenly applied onto a glass microscope slide using the pneumatic sprayer. The mass resolving power was set to 140,000 at m/z 200 for the MS2I acquisition in the Orbitrap.

2.2.4 Data Analysis

To create ion heat maps, the raw data (.RAW) from the Thermo Q Exactive was converted to the mzXML format using the MSConvert software from Proteowizard [45]. For concatenated ion images, the raw files were converted to mzML files using the MSConvert software from Proteowizard and were then converted to individual imzML files using imzMLConverter [46]. The imzML Converter was then used to stack the individual imzML files into one master imzML file. The mzXML or imzML files were then loaded into the standalone version of MSiReader, which is freely available software developed in our lab for processing MSI data [47]. In order to demonstrate the quality of the raw data, ion images were neither interpolated nor normalized (unless otherwise specified). Dimension scales associated with all presented ion heat maps reflect length in millimeters. Supervised analysis of the MSI

data was undertaken with MSiReader for untargeted discovery of ions associated with tissue. In this approach, the software averages voxel spectra over a region of interest specified by the user and identifies unique peaks associated with this region relative to a user-specified reference region, chosen to be off-tissue. Putative identification of ions selected by MSiReader based on exact mass has been performed where possible using the LIPID MAPS [48] and METLIN structure databases [49]. No further targeted SRM was undertaken to confirm these assignments and, as a result, no attempt is made to distinguish structural isomers. These ions are identified by their molecular formula and compound class.

2.3 Results and Discussion

2.3.1 Optimization of Desorption Conditions using an Ice Matrix

Initial efforts to evaluate the influence of excitation wavelength on IR-MALDESI response were performed by scanning the OPO emission wavelength while imaging a homogeneous 10 μm thick tissue slice of mouse liver. Imaging of the tissue was conducted in a standard scan pattern proceeding from the top left to the bottom right of the prescribed region of interest, with the incident wavelength ranging from 2850 to 3100 nm and increasing in 5 nm increments every two scan lines across the tissue. Ion abundance of many endogenous lipids within mouse liver such as cholesterol increased with wavelength, as shown in Figure 2-1. Unattenuated OPO pulse energy is wavelength-dependent over this spectral region, peaking at 2940 nm as shown in Figure 2-1, and while the trend in lipid response does not match that of OPO pulse energy, the influence of fluence and wavelength could not be entirely decoupled when interpreting these results. An optical attenuator was added to the beam path in order to match fluence while scanning emission wavelength, and a fluence scan (data not shown) was conducted using wavelengths of 2940 and 3100 nm on a serial tissue slice in a similar manner to Figure 2-1 with pulse energies varied from maximum OPO output to the desorption

threshold. The fluence scan suggested that optimal ion abundance was achieved at each incident wavelength for laser fluences below the maximum output of the OPO.

Figure 2-1. IR-MALDESI ion map of cholesterol response in mouse liver while scanning the incident wavelength from 2850 nm to 3100 nm. OPO wavelength was incremented 5 nm after the completion of every two scan lines. The position of key wavelengths, 2940 nm and 3100 nm, are highlighted on the sample. Unattenuated OPO power is also shown as a function of emission wavelength. To systematically explore variables hypothesized to influence desorption plume dynamics, a full factorial screening design of experiments was conducted. This two-level statistical design investigated four parameters (laser fluence, excitation wavelength, absorption matrix, sample height) resulting in 16 total experiments. Selection of relevant parameter values for the screening experiment was informed by the wavelength and fluence scans described above, and these settings are summarized in Figure 2-2a. Experiments were conducted on antiretroviral (ARV)-incubated cervical tissue to investigate the influence of desorption conditions on IR-MALDESI response to both endogenous and xenobiotic compounds from tissue samples. Each of four quadrants of a morphologically homogeneous tissue cryosection were imaged with a unique combination of wavelength and matched fluence, as depicted in Figure 2-2b. This procedure was repeated four times with and without addition of an ice matrix for each sample height using serial slices, accounting for all combinations of variables in the

16 screening experiments. Also shown in Figure 2-2b are concatenated ion maps from each quadrant of FTC and RAL distributions within the incubated cervical tissue, analyzed with an ice matrix and with the sample stage 5 mm below the plane of the ESI emitter and the MS inlet. In addition to the ARV drugs, ions unique to the cervical tissue were identified from the IR- MALDESI imaging results using the MSiReader software. Representative ion maps of these species and, where available, their putative identification based on the METLIN database [49] can be seen in Figure 2-3. The average ion abundance in each of the 16 investigated quadrants was used as the figure of merit to evaluate the effect, or contrast, of each variable on the resulting response for all ions or each ion individually through statistical analysis using JMP software. Results of the full factorial screening experiment can be seen in Figure 2-2c. While the aggregate response for all detected ions does not indicate universal statistical significance between variable settings, the DOE screening results show common trends in variable dependence. Ion abundance is cumulatively enhanced when ice is present as an absorption matrix, when OPO laser pulse energy is reduced, and, to a lesser extent, at greater sample height and incident laser wavelength. The significance of each variable is analyte-dependent, as shown for the ARV drugs in Figure 2-2c and for the other representative components in Figure 2-3, and the significance of the cross-terms suggests coupled effects.

Figure 2-2. (a) Parameters tested in screening design of experiments. (b) Pictogram of approach for quadrant imaging and representative IR-MALDESI ion maps with ice matrix for selected ARV therapies. (c) DOE screening results for the average response of all identified ions unique to tissue, as well as targeted results for individual ARV drugs indicating some specificity of parameter importance to individual analytes. The importance of the ice matrix is consistent with previous work using IR-MALDESI. Shadowgraphy of the IR-MALDESI laser desorption process using an ice matrix has indicated complex desorption plume dynamics that involve absorption of energy by, and interaction between, tissue, ice, and liquid water [21]. The presence of an ice matrix reduces the ejection of large tissue fragments that are not likely to contribute to mass spectral response, while also allowing the accumulation of ion signal over multiple ablation events [21].

Figure 2-3. IR-MALDESI ion maps of representative endogenous on-tissue ions and matching DOE screening results. This study is the first demonstration of the role of laser fluence on the IR-MALDESI mechanism and sensitivity. While signal abundance of IR-MALDI using an ice matrix has been observed to increase exponentially with laser fluence at low fluence values (<0.2 J/cm2) [31], higher fluences may lead to a reduction in the effective absorption cross-section of the

matrix as has been observed and modeled for liquid water [50]. Based on measurements of incident OPO energy and beam diameter measured on thermal paper, the laser fluence over the on-tissue ablation threshold diameter of 150 μm can be estimated for this work assuming a Gaussian beam profile. Pulse energies utilized in this work correspond to fluences of approximately 2.8–3.8 J/cm2, which are above the fluence level at which Shori et al. predict a “blue shift” in the absorption cross-section of water due to thermal excitation [50]. Though the OPO pulse duration of 7 ns limits the thermal diffusion of incident energy, a reduction in the fluence likely results in more efficient absorption of energy by the matrix or simply better overlap between the desorbed plume and the orthogonal electrospray. Analyte specificity toward desorption variables may arise from multiple sources. Infrared photosensitivity of compounds within the biological sample may contribute to wavelength- dependent differences in energy deposition, which have been demonstrated with IR-MALDI for matrices possessing different functionalities [25]. Pirkl et al. have suggested that the preferential IR-MALDI response at 2940 nm to a single analyte, Substance P, using an ice matrix is due to transient melting of the ice during ablation and predominant absorption of the laser light by liquid water. This certainly may be true, particularly for spectra acquired over 100 s of laser pulses [31], but may also have been further influenced by analyte absorption. The excitation wavelengths investigated here correspond to absorption of free O–H and N–H stretching modes around 3000 nm present in many endogenous compounds. Any changes to the chemical composition or dynamics of the neutral plume because of energy absorption by the ice matrix and sample will alter the complexion of analytes ionized by electrospray, particularly for a complex biological sample.

2.3.2 Whole Body Imaging

To extend the results of the DOE experiments and to consider differences in response across different tissue types, a lipidomic MSI analysis of a neonatal mouse pup was performed. To preserve the integrity of all organs during cryosectioning, sagittal cross-sections of the

mouse (roughly 1.5 cm × 2.5 cm) were cut to 25 μm thickness. Serial sections were imaged by IR-MALDESI using a full MS scan for untargeted lipidome analysis at incident wavelengths of both 2940 and 3100 nm while using the other optimized MSI conditions from the DOE experiments (ice matrix, reduced fluence, 5 mm stage height). Images of the whole-body sections typically comprised 35,000–40,000 voxels, and were collected over 6–7 h of imaging time at a scan rate of 1.6 scans/s. IR-MALDESI whole body imaging data was analyzed using MSiReader, which yielded 686 and 613 ions associated with tissue for imaging conducted with incident wavelengths of 2940 and 3100 nm, respectively. Of these detected compounds, 237 were detected preferentially at 2940 nm, 449 compounds were detected with similar relative ion abundance at both wavelengths, and 164 compounds were detected preferentially at 3100 nm, corresponding to classes I–III in Figure 2-4a. Mass excesses of all tissue-related peaks, which represent the difference between the nominal and monoisotopic ion mass, indicate a high degree of overlap with the mass excess distribution for lipid classes from the LIPID MAPS structural database [48] (Figure 2-4b). While full MS acquisition does not eliminate the detection of isobaric species, the high resolution exact mass of ions measured by the Q Exactive significantly reduces the number of species to be considered during peak assignment when + + + + + + searching a lipid database for anticipated cations ([M + H ] , [M + Na ] , [M + H – H2O] ) with 5 ppm mass accuracy. As such, lipid subclasses associated with many of the detected ions were identified unambiguously. IR-MALDESI is sensitive to a range of lipid classes (Table 2-1), which include fatty acyls, sterol lipids, glycerophospholipids, glycerolipids, and sphingolipids. Representative ion maps from each of these lipid classes can be seen in Figure 2-5, along with an anatomically annotated cryosection. These images demonstrate the capability of IR-MALDESI to evaluate the discrete biodistribution of ions across a whole body organism, exhibiting localization within individual or functionally related regions. Differences in ion response factors between tissue types will be addressed in the next section, and the ion distributions shown here do not account for such effects. Nonetheless, IR-MALDESI ion maps

show similar patterns of lipid subclasses across tissues to a recent systematic lipid survey in 17 different mouse tissues by LC-MS [51]: fatty acylcarnitines appear to concentrate in the heart and liver; sterols like cholesterol are concentrated in the brain; glycerophospholipids like PS 40:1 are distributed across many tissues; diradyl and triradylglycerolipids were observed to concentrate in the brown fat and intestines; and sphingomyelins like SM 45:1 are concentrated in kidney and stomach.

Figure 2-4. (a) Diagram of tissue-related ions identified by MSiReader during whole body imaging of a neonatal mouse using an excitation wavelength of λ = 2940 and 3100 nm, indicating: (I) ions detected preferentially at 2940 nm; (II) ions detected equally at both wavelengths; and, (III) ions detected preferentially at 3100 nm. (b) Mass excess of all tissue related peaks overlapped with mass excess distribution for lipids from LIPID MAPS Structure Database. Location of the peaks found on tissue indicate that most of the 686 and 613 peaks found for λ = 2940 and 3100 nm, respectively, correspond to lipids.

Table 2-1. Summary of putative lipid class and subclass assignments for the measured exact masses of on-tissue ions during whole body imaging of a neonate mouse based on comparison to the METLIN database. *Where there is ambiguity in lipid class assignment due to the identification of isobaric species from multiple classes within the 5 ppm database search tolerance, ions have been grouped accordingly (i.e., PC/PE or PC/PE/PS).

Lipids IR-MALDESI response

Class Subclass Annotation Class I Class II Class III

2940 nm Both 3100 nm

Fatty acyls Fatty acyl carnitine 0 0 2

Glycerophospholipids Glycerophosphocholine PC 0 5 3

Glycerophosphoethanolamine PE 4 4 2

PC/PE* 0 56 0

Glycerophosphoserine PS 3 17 1

PC/PE/PS* 24 0

Glycerophosphate PA 6 21 10

Glycerophosphoglycerols PG 0 9 2

Glycerophosphoinositols PI 0 1 0

Glycerolipids Diradylglycerol DG 5 27 31

Triradylglycerol TG 0 51 6

Sphingolipids Sphingosine 0 0 1

Ceramide Cer 0 6 3

Sphingomyelin SM 1 5 1

Sulfatide 0 0 0

Sterol lipids Steryl esters CE 2 3 1

Total 18 221 59

Figure 2-5. Selected ion maps representing lipid classes identified by IR-MALDESI in whole body sections of neonate mouse, along with putative ion identification from Lipid Maps where available. An abbreviated mouse anatomy is included at top of the image to highlight regions where ion-specific localization of IR-MALDESI response has been observed. A comparison of IR-MALDESI response to serial sections imaged at different wavelengths can be seen in Figure 2-6. Ion distributions exhibit localization within individual or functionally related regions including: the whole brain (m/z 695.462) or discrete brain regions such as the Rostral migratory system (m/z 837.569); the musculature (m/z 835.579); the heart and lungs (m/z 617.180); and the stomach (m/z 732.609 and m/z 872.771). Class II ion maps show a similar magnitude of response at both 2940 and 3100 nm, but may exhibit differences in response to varying tissue types as seen in the liver (m/z 577.519) and the olfactory epithelium (m/z 748.528). Both class I and class III ion maps illustrate differences in the ion abundance between the experiments conducted at different wavelengths.

Figure 2-6. Selected ion maps of serial 25 μm thick whole mouse corresponding to each of the three classes of ions defined in Figure 2-4. Pairs of ion maps corresponding to the matching MSI experiments are shown with matching intensity scales to illustrate differences in ion abundances between the sections imaged at λ = 2940 and 3100 nm. Changes in the response to a given analyte observed between the cryosections imaged at 2940 and 3100 nm may be the result of differences in analyte concentration or differences in the desorption conditions. Each cryosection examined was separated by 25 μm, such that changes in tissue composition or morphology between the cryosections used for each experiment are expected to be minimal. The amount of incident laser energy coupled to an analyte via the ice matrix and through direct absorption is wavelength-dependent, which influences threshold fluence [25] and resulting plume height [52] of an analyte. Given a finite ionization capacity of the electrospray, alterations to the composition of the desorbed plume interacting with the orthogonal electrospray are likely to lead to differences in MS response to a complex sample. While the untargeted nature of the whole body MSI precludes distinguishing whether the observed differences arise more strongly from selective direct excitation of analytes or from differences to the plume behavior as a result of the tissue type and/or ice matrix, it is clear that imaging whole body sections at these two discrete excitation wavelengths has led to enhanced coverage of lipid response.

Although whole body lipidome studies are limited in number, this work represents a significantly more comprehensive positive ion lipid response across whole body mouse sections than previously reported with MALDI MSI [32], where the MS response between m/z 650 and 900 was characterized as being typical of phospholipids but no attempt was made to identify lipid subclasses. IR-MALDESI shows a strong response to phospholipids such as phosphatidylcholines (PC), which have been the focus of many mouse MSI studies [53], presumably attributable not only to their abundance and ionization efficiency but also due to their suppression of other ions [54]. Additionally, IR-MALDESI is capable of response to a greater breadth of species like glycerolipids and sphingolipids than MALDI [3] and DESI [55] without requiring multiple matrices [56] or additional tissue preparation such as washing [57] or enzymatic digestion [58]. Simultaneous acquisition of these species has great potential for future lipid imaging studies of disease pathologies, for example.

2.3.3 Response Factors and Internal Energy Distribution across Whole Body

Differences in the ionization efficiency across tissue types were evaluated by monitoring IR-MALDESI MS2I response to an analyte standard, cocaine, sprayed onto the glass mounting slide beneath a whole body cryosection. Homogeneity of analyte deposition onto the slide and variability of IR-MALDESI response to it beneath biological tissue has recently been characterized [19]. While that study utilized the analyte to normalize ion response as part of a quantitative MSI method, here it is employed in an effort to evaluate the relative internal energy distribution of ions generated at the two different incident wavelengths, 2940 and 3100 nm, across a plethora of tissue types. Cocaine’s simple fragmentation pattern [59] offers facile interpretation of its collision-induced dissociation (CID) spectra for monitoring ion fragmentation ratios during MS2I analysis using the “survival yield” method [60], which has been used to evaluate internal energy deposition of DESI [61] and LAESI [62]. Discrete regions of a sprayed slide were imaged at varying normalized collision energies (NCE), shown in Figure 2-7a, to determine the IR-MALDESI breakdown diagram of cocaine

in the QE HCD cell (Figure 2-8). The breakdown diagram indicated that ion internal energy in the HCD cell is similar for both + ESI direct infusion and IR-MALDESI. These experiments were repeated using the in-source CID capability of the QE to account for any collisional cooling within the HCD cell, yielding statistically the same results.

Figure 2-7. IR-MALDESI MS2I monitoring of cocaine: (a) evaluation of cocaine survival yield breakdown diagram by varying the CID normalized collision energy over discrete regions on a glass slide; (b) cocaine [M + H+]+ response during imaging of a 25 μm-thick whole mouse cryosection, with discrete anatomical regions highlighted that correspond to differences in response as summarized in text and in Table 2-2; (c) survival yield of cocaine during whole-body imaging (bottom panels show a narrower range of response and include overlaid optical images of analyzed sections).

Figure 2-8. Breakdown diagrams for cocaine analyzed by direct infusion (red symbols: M, filled circles; F, open circles) or by MALDESI (gray symbols) demonstrating similar distribution of internal energy among ions generated by either technique. Whole body sections of neonatal mouse were then thaw-mounted on top of glass slides sprayed with the cocaine analyte. MS2I of cocaine was performed across whole body cryosections using the incident wavelengths 2940 and 3100 nm with a fixed value of NCE, 25%, which corresponded to a cocaine survival yield of ~50% based on the breakdown diagram. IR-MALDESI response to cocaine was apparent across the entire cryosection and varied beneath different tissue types corresponding to anatomical features, as can be seen in Figure 2-7. Unlike previous MALDI MSI analysis of a homogeneous standard sprayed onto a whole-body mouse section [35], there are no imaged regions exhibiting complete ion suppression. Differences in the cocaine [M + H+]+ ion abundance across discrete regions of the cryosection highlighted in Figure 2-7 are summarized in Table 2-2. Voxel-to-voxel variability of response below tissue is similar to, and in some cases better than, the response from off-

tissue regions of the coated slide. Average voxel response varies by less than a factor of five between all virtual micro-dissected regions. Since MALDESI is not subject to tissue-specific issues regarding matrix application or analyte extraction efficiency [35, 36, 39], the observed differences in response factor are more likely to result from desorption conditions. While tissue response factors are expected to be analyte-specific, this approach provides a pathway to quantitative whole body imaging with IR-MALDESI for targeted small molecules [63] or untargeted analysis of lipids with adroit selection of representative lipid standards [64].

Table 2-2. Summary of MS2I response to cocaine [M + H+]+, m/z 304.155, beneath different tissue types of a whole body mouse cryosection.

Region 1 2 3 4 5 6

Tissue Whole Brain Adipose Heart and Gut Off-tissue Type mouse tissue Lung

Average 9.0E + 05 1.6E + 06 1.4E + 06 6.6E + 05 3.4E + 05 2.67E + 06

Max 1.4E + 07 8.7E + 06 5.4E + 06 6.6E + 06 2.2E + 06 2.15E + 07

Min 1.7E + 03 1.7E + 04 2.5E + 04 3.8E + 03 2.3E + 03 6.77E + 03

%RSD 92 63 54 101 99 87

The survival yield of cocaine at each wavelength is shown in Figure 2-7c. These images were obtained by normalizing the cocaine response to the sum of the response for all molecular ion and fragment ions (m/z values of 304.1550, 182.177, and 150.092) on a voxel-by-voxel basis using MSiReader. Survival yields at each wavelength across the entire cryosections

(SY2940nm = 46.5% ± 9.2% and SY3100nm = 51.3% ± 11.5%) are similar on average to the value of 50% derived from IR-MALDESI analysis of the sprayed slide alone. These results indicate complete collisional cooling of desorbed neutral molecules in the gas phase [15] or during uptake into ESI droplets, and show no evidence for direct ionization. As such, they support the mechanistic view of IR-MALDESI neutrals that are collisionally cooled rapidly during their partitioning into ESI droplets as has been posited previously [16, 65], but do not allow

conclusions regarding initial differences in internal energy attributable to excitation wavelength. Although an MS2I approach was selected to eliminate the possibility of endogenous isobaric species interfering with calculation of the survival yield, further information regarding wavelength dependence in the IR-MALDESI mechanism may be gained by full MS imaging of biological tissues using traditional benzylpyridinium thermometer ions with lower bond dissociation energies [61, 62] as a sprayed analyte.

2.4 Conclusions

This work has shown that the IR-MALDESI response to individual analytes within a complex biological system is sensitive to factors controlling analyte desorption, such as laser wavelength, fluence, stage height, and matrix. Overall trends of pharmaceutical drugs and endogenous lipids indicated enhanced response when reducing laser fluence and utilizing an ice matrix. Although the influence of incident wavelength on the internal energy of desorbed neutrals could not be determined, it was shown that altering the excitation wavelength resulted in enhanced lipid response to whole body sections of neonate mouse. IR-MALDESI was shown to be responsive to a broad range of lipid classes present across an organism without any requirement for sample manipulation. Through the analysis of a standard deposited beneath the whole body sections, tissue-specific response factors were determined that will be a critical component for quantitative MSI on a multi-organ scale.

2.5 Acknowledgments

The authors acknowledge Professor Angela Kashuba and Corbin Thompson of the Eshelman School of Pharmacy, UNC Chapel Hill for providing ARV-incubated tissue samples. The authors gratefully acknowledge the financial support received from the National Institutes of Health R01GM087964 and R01AI111891.

2.6 References

1. Prideaux, B., Stoeckli, M.: Mass spectrometry imaging for drug distribution studies. Journal of Proteomics. 75, 4999-5013 (2012)

2. Murphy, R.C., Hankin, J.A., Barkley, R.M.: Imaging of lipid species by MALDI mass spectrometry. Journal of lipid research. 50, S317-S322 (2009)

3. Gode, D., Volmer, D.A.: Lipid imaging by mass spectrometry–a review. Analyst. 138, 1289-1315 (2013)

4. Burnum, K.E., Frappier, S.L., Caprioli, R.M.: Matrix-assisted laser desorption/ionization imaging mass spectrometry for the investigation of proteins and peptides. Annu. Rev. Anal. Chem. 1, 689-705 (2008)

5. Greer, T., Sturm, R., Li, L.: Mass spectrometry imaging for drugs and metabolites. Journal of proteomics. 74, 2617-2631 (2011)

6. Schwamborn, K.: Imaging mass spectrometry in biomarker discovery and validation. Journal of proteomics. 75, 4990-4998 (2012)

7. Calligaris, D., Caragacianu, D., Liu, X., Norton, I., Thompson, C.J., Richardson, A.L., Golshan, M., Easterling, M.L., Santagata, S., Dillon, D.A.: Application of desorption electrospray ionization mass spectrometry imaging in breast cancer margin analysis. Proceedings of the National Academy of Sciences. 111, 15184-15189 (2014)

8. Liu, J., Gingras, J., Ganley, K.P., Vismeh, R., Teffera, Y., Zhao, Z.: Whole‐body tissue distribution study of drugs in neonate mice using desorption electrospray ionization mass spectrometry imaging. Rapid Communications in Mass Spectrometry. 28, 185- 190 (2014)

9. Goodwin, R.J.: Sample preparation for mass spectrometry imaging: small mistakes can lead to big consequences. Journal of proteomics. 75, 4893-4911 (2012)

10. Takats, Z., Wiseman, J.M., Gologan, B., Cooks, R.G.: Mass spectrometry sampling under ambient conditions with desorption electrospray ionization. Science. 306, 471- 473 (2004)

11. Eikel, D., Vavrek, M., Smith, S., Bason, C., Yeh, S., Korfmacher, W.A., Henion, J.D.: Liquid extraction surface analysis mass spectrometry (LESA‐MS) as a novel profiling

tool for drug distribution and metabolism analysis: the terfenadine example. Rapid Communications in Mass Spectrometry. 25, 3587-3596 (2011)

12. Sampson, J.S., Murray, K.K., Muddiman, D.C.: Intact and top-down characterization of biomolecules and direct analysis using infrared matrix-assisted laser desorption electrospray ionization coupled to FT-ICR mass spectrometry. Journal of the American Society for Mass Spectrometry. 20, 667-673 (2009)

13. Berkenkamp, S., Kirpekar, F., Hillenkamp, F.: Infrared MALDI mass spectrometry of large nucleic acids. Science. 281, 260-262 (1998)

14. Dreisewerd, K., Berkenkamp, S., Leisner, A., Rohlfing, A., Menzel, C.: Fundamentals of matrix-assisted laser desorption/ionization mass spectrometry with pulsed infrared lasers. International Journal of Mass Spectrometry. 226, 189-209 (2003)

15. Levis, R.J.: Laser desorption and ejection of biomolecules from the condensed phase into the gas phase. Annual review of physical chemistry. 45, 483-518 (1994)

16. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Generation and detection of multiply-charged peptides and proteins by matrix-assisted laser desorption electrospray ionization (MALDESI) Fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry. 17, 1712-1716 (2006)

17. Nemes, P., Vertes, A.: Laser ablation electrospray ionization for atmospheric pressure, in vivo, and imaging mass spectrometry. Analytical chemistry. 79, 8098-8106 (2007)

18. Rezenom, Y.H., Dong, J., Murray, K.K.: Infrared laser-assisted desorption electrospray ionization mass spectrometry. Analyst. 133, 226-232 (2008)

19. Bokhart, M.T., Rosen, E., Thompson, C., Sykes, C., Kashuba, A.D.M., Muddiman, D.C.: Quantitative mass spectrometry imaging of emtricitabine in cervical tissue model using infrared matrix-assisted laser desorption electrospray ionization. Analytical and Bioanalytical Chemistry. 407, 2073-2084 (2015)

20. Fan, X., Murray, K.K.: Wavelength and time-resolved imaging of material ejection in infrared matrix-assisted laser desorption. The Journal of Physical Chemistry A. 114, 1492-1497 (2009)

21. Robichaud, G., Barry, J.A., Muddiman, D.C.: IR-MALDESI mass spectrometry imaging of biological tissue sections using ice as a matrix. Journal of the American Society for Mass Spectrometry. 25, 319-328 (2014)

22. Barry, J.A., Robichaud, G., Bokhart, M.T., Thompson, C., Sykes, C., Kashuba, A.D., Muddiman, D.C.: Mapping antiretroviral drugs in tissue by IR-MALDESI MSI coupled to the Q Exactive and comparison with LC-MS/MS SRM assay. Journal of The American Society for Mass Spectrometry. 25, 2038-2047 (2014)

23. Goodwin, R.J., Iverson, S.L., Andren, P.E.: The significance of ambient‐temperature on pharmaceutical and endogenous compound abundance and distribution in tissues sections when analyzed by matrix‐assisted laser desorption/ionization mass spectrometry imaging. Rapid Communications in Mass Spectrometry. 26, 494-498 (2012)

24. Berkenkamp, S., Karas, M., Hillenkamp, F.: Ice as a matrix for IR-matrix-assisted laser desorption/ionization: mass spectra from a protein single crystal. Proceedings of the National Academy of Sciences. 93, 7003-7007 (1996)

25. Menzel, C., Dreisewerd, K., Berkenkamp, S., Hillenkamp, F.: Mechanisms of energy deposition in infrared matrix-assisted laser desorption/ionization mass spectrometry. International Journal of Mass Spectrometry. 207, 73-96 (2001)

26. Fisher, R.A.: The design of experiments. Hafner Publishing Company, New York, NY., (1966)

27. Riter, L.S., Vitek, O., Gooding, K.M., Hodge, B.D., Julian, R.K.: Statistical design of experiments as a tool in mass spectrometry. Journal of mass spectrometry. 40, 565-579 (2005)

28. Schriver-Mazzuoli, L., Schriver, A., Hallou, A.: IR reflection–absorption spectra of thin water ice films between 10 and 160K at low pressure. Journal of Molecular Structure. 554, 289-300 (2000)

29. Williams, P.: Time of flight mass spectrometry of DNA laser-ablated from frozen aqueous solutions: applications to the Human Genome Project. International Journal of Mass Spectrometry and Ion Processes. 131, 335-344 (1994)

30. Focsa, C., Mihesan, C., Ziskind, M., Chazallon, B., Therssen, E., Desgroux, P., Destombes, J.: Wavelength-selective vibrationally excited photodesorption with tunable IR sources. Journal of Physics: Condensed Matter. 18, S1357 (2006)

31. Pirkl, A., Soltwisch, J., Draude, F., Dreisewerd, K.: Infrared matrix-assisted laser desorption/ionization orthogonal-time-of-flight mass spectrometry employing a cooling stage and water ice as a matrix. Analytical chemistry. 84, 5669-5676 (2012)

32. Chaurand, P., Cornett, D.S., Angel, P.M., Caprioli, R.M.: From whole-body sections down to cellular level, multiscale imaging of phospholipids by MALDI mass spectrometry. Molecular & Cellular Proteomics. 10, O110. 004259 (2011)

33. Khatib-Shahidi, S., Andersson, M., Herman, J.L., Gillespie, T.A., Caprioli, R.M.: Direct molecular analysis of whole-body animal tissue sections by imaging MALDI mass spectrometry. Analytical chemistry. 78, 6448-6456 (2006)

34. Trim, P.J., Henson, C.M., Avery, J.L., McEwen, A., Snel, M.F., Claude, E., Marshall, P.S., West, A., Princivalle, A.P., Clench, M.R.: Matrix-assisted laser desorption/ionization-ion mobility separation-mass spectrometry imaging of vinblastine in whole body tissue sections. Analytical chemistry. 80, 8628-8634 (2008)

35. Stoeckli, M., Staab, D., Schweitzer, A.: Compound and metabolite distribution measured by MALDI mass spectrometric imaging in whole-body tissue sections. International Journal of Mass Spectrometry. 260, 195-202 (2007)

36. Crossman, L., McHugh, N.A., Hsieh, Y., Korfmacher, W.A., Chen, J.: Investigation of the profiling depth in matrix‐assisted laser desorption/ionization imaging mass spectrometry. Rapid communications in mass spectrometry. 20, 284-290 (2006)

37. Reyzer, M., Chaurand, P., Caprioli, R.: In: Rubakhin, S.S., Sweedler, J.V. (eds.) Mass Spectrometry Imaging, pp. 285-301. Humana Press, New York, NY, vol. 656, chap. 17 (2010)

38. Caprioli, R.M., Farmer, T.B., Gile, J.: Molecular imaging of biological samples: localization of peptides and proteins using MALDI-TOF MS. Analytical chemistry. 69, 4751-4760 (1997)

39. Tomlinson, L., Fuchser, J., Fütterer, A., Baumert, M., Hassall, D.G., West, A., Marshall, P.S.: Using a single, high mass resolution mass spectrometry platform to investigate ion suppression effects observed during tissue imaging. Rapid Communications in Mass Spectrometry. 28, 995-1003 (2014)

40. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) imaging source coupled to a FT-ICR mass spectrometer. Journal of the American Society for Mass Spectrometry. 24, 92-100 (2013)

41. Nazari, M., Muddiman, D.C.: Cellular-level mass spectrometry imaging using infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) by oversampling. Analytical and Bioanalytical Chemistry. 407, 2265-2271 (2015)

42. Jurchen, J.C., Rubakhin, S.S., Sweedler, J.V.: MALDI-MS imaging of features smaller than the size of the laser beam. Journal of the American Society for Mass Spectrometry. 16, 1654-1659 (2005)

43. Barry, J.A., Robichaud, G., Muddiman, D.C.: Mass recalibration of FT-ICR mass spectrometry imaging data using the average frequency shift of ambient ions. Journal of the American Society for Mass Spectrometry. 24, 1137-1145 (2013)

44. Olsen, J.V., de Godoy, L.M., Li, G., Macek, B., Mortensen, P., Pesch, R., Makarov, A., Lange, O., Horning, S., Mann, M.: Parts per million mass accuracy on an Orbitrap mass spectrometer via lock mass injection into a C-trap. Molecular & Cellular Proteomics. 4, 2010-2021 (2005)

45. Kessner, D., Chambers, M., Burke, R., Agus, D., Mallick, P.: ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics. 24, 2534- 2536 (2008)

46. Race, A.M., Styles, I.B., Bunch, J.: Inclusive sharing of mass spectrometry imaging data requires a converter for all. Journal of proteomics. 75, 5111-5112 (2012)

47. Robichaud, G., Garrard, K.P., Barry, J.A., Muddiman, D.C.: MSiReader: an open- source interface to view and analyze high resolving power MS imaging files on Matlab platform. Journal of the American Society for Mass Spectrometry. 24, 718-721 (2013)

48. Sud, M., Fahy, E., Cotter, D., Brown, A., Dennis, E.A., Glass, C.K., Merrill Jr, A.H., Murphy, R.C., Raetz, C.R., Russell, D.W.: Lmsd: lipid maps structure database. Nucleic acids research. 35, D527-D532 (2006)

49. Smith, C.A., O'Maille, G., Want, E.J., Qin, C., Trauger, S.A., Brandon, T.R., Custodio, D.E., Abagyan, R., Siuzdak, G.: METLIN: a metabolite mass spectral database. Therapeutic drug monitoring. 27, 747-751 (2005)

50. Shori, R.K., Walston, A.A., Stafsudd, O.M., Fried, D., Walsh, J.T.: Quantification and modeling of the dynamic changes in the absorption coefficient of water at lambda = 2.94 um. IEEE Journal of selected topics in quantum electronics. 7, 959-970 (2001)

51. Jain, M., Ngoy, S., Sheth, S.A., Swanson, R.A., Rhee, E.P., Liao, R., Clish, C.B., Mootha, V.K., Nilsson, R.: A systematic survey of lipids across mouse tissues. American Journal of Physiology-Endocrinology and Metabolism. 306, E854-E868 (2014)

52. Fan, X., Little, M.W., Murray, K.K.: Infrared laser wavelength dependence of particles ablated from glycerol. Applied Surface Science. 255, 1699-1704 (2008)

53. Hankin, J.A., Murphy, R.C.: Relationship between MALDI IMS intensity and measured quantity of selected phospholipids in rat brain sections. Analytical chemistry. 82, 8476-8484 (2010)

54. Petković, M., Schiller, J., Müller, M., Benard, S., Reichl, S., Arnold, K., Arnhold, J.: Detection of individual phospholipids in lipid mixtures by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry: phosphatidylcholine prevents the detection of further species. Analytical biochemistry. 289, 202-216 (2001)

55. Eberlin, L.S., Ferreira, C.R., Dill, A.L., Ifa, D.R., Cooks, R.G.: Desorption electrospray ionization mass spectrometry for lipid characterization and biological tissue imaging. Biochimica et Biophysica Acta (BBA)-Molecular and Cell Biology of Lipids. 1811, 946-960 (2011)

56. Ellis, S.R., Brown, S.H., in het Panhuis, M., Blanksby, S.J., Mitchell, T.W.: Surface analysis of lipids by mass spectrometry: more than just imaging. Progress in lipid research. 52, 329-353 (2013)

57. Angel, P.M., Spraggins, J.M., Baldwin, H.S., Caprioli, R.: Enhanced sensitivity for high spatial resolution lipid analysis by negative ion mode matrix assisted laser desorption ionization imaging mass spectrometry. Analytical chemistry. 84, 1557-1564 (2012)

58. Jones, E.E., Dworski, S., Canals, D., Casas, J., Fabrias, G., Schoenling, D., Levade, T., Denlinger, C., Hannun, Y.A., Medin, J.A.: On-tissue localization of ceramides and other sphingolipids by MALDI mass spectrometry imaging. Analytical chemistry. 86, 8303-8311 (2014)

59. Dixon, S.J., Brereton, R.G., Carter, J.F., Sleeman, R.: Determination of cocaine contamination on banknotes using tandem mass spectrometry and pattern recognition. Analytica Chimica Acta. 559, 54-63 (2006)

60. Kenttämaa, H.I., Cooks, R.G.: Internal energy distributions acquired through collisional activation at low and high energies. International journal of mass spectrometry and ion processes. 64, 79-83 (1985)

61. Nefliu, M., Smith, J.N., Venter, A., Cooks, R.G.: Internal energy distributions in desorption electrospray ionization (DESI). Journal of the American Society for Mass Spectrometry. 19, 420-427 (2008)

62. Nemes, P., Huang, H., Vertes, A.: Internal energy deposition and ion fragmentation in atmospheric-pressure mid-infrared laser ablation electrospray ionization. Physical Chemistry Chemical Physics. 14, 2501-2507 (2012)

63. Hamm, G., Bonnel, D., Legouffe, R., Pamelard, F., Delbos, J.-M., Bouzom, F., Stauber, J.: Quantitative mass spectrometry imaging of propranolol and olanzapine using tissue extinction calculation as normalization factor. Journal of proteomics. 75, 4952-4961 (2012)

64. Lanekoff, I., Thomas, M., Laskin, J.: Shotgun approach for quantitative imaging of phospholipids using nanospray desorption electrospray ionization mass spectrometry. Analytical chemistry. 86, 1872-1880 (2014)

65. Dixon, R.B., Muddiman, D.C.: Study of the ionization mechanism in hybrid laser based desorption techniques. Analyst. 135, 880-882 (2010)

3 Influence of C-Trap Ion Accumulation Time on the Detectability of Analytes in IR-MALDESI MSI

The following work was reprinted with permission from: Rosen, E.P., Bokhart, M.T., Nazari, M., Muddiman, D.C.: Influence of C-Trap Ion Accumulation Time on the Detectability of Analytes in IR-MALDESI MSI. Analytical Chemistry. 87, 10483-10490 (2015). DOI: 10.1021/acs.analchem.5b02641. Copyright © 2015 American Chemical Society. The original publication may be accessed via the Internet at http://pubs.acs.org/doi/10.1021/acs.analchem.5b02641

3.1 Introduction

Mass spectrometry imaging (MSI) entails the systematic acquisition of mass spectral information from an array of discrete positions across a sample, allowing the distribution of analytes to be visualized. Initially conceived of by Castaing and Slodzian in the 1960s [1], MSI began as a relatively crude technique for biological ion imaging [2] that through technological advancements [3-9] now combines the high specificity and sensitivity of mass spectrometric detection with high spatial resolution for unparalleled measurement of analyte distribution within biological samples. MSI has been shown to be sensitive for analysis of metabolomics [10, 11], proteomics [10, 12-14], lipidomics [15-18], as well as targeted analysis of pharmaceutical drugs [19-23]. For MSI analysis of small molecules and lipids within biological matrices, often including analytes that are labile and/or semivolatile, ambient ionization approaches that preserve the native state of the sample for in situ analysis of intact tissue are desired. Ambient MSI techniques such as infrared matrix-assisted laser desorption electrospray ionization (IR- MALDESI) [24] have the added benefit of requiring minimal sample preparation, which for more preparation-intensive MSI approaches like MALDI can dictate analyte sensitivity [25]. The mechanism of MALDESI involves a two-step process wherein material is first desorbed from a sample surface by an impinging laser pulse, with the resulting plume of desorbed neutrals then gently ionized by entrainment in an orthogonal electrospray [26]. By operating

at mid-infrared wavelengths, IR-MALDESI laser stimulation of biological tissue samples leads to the resonant excitation of species natively present, such as endogenous water or an exogenous layer of ice [27], eliminating background interferences associated with organic matrices as used in MALDI. The thin layer of exogenous ice also accounts for variation in water content in different tissue compartments, and has been shown to greatly influence tissue desorption [27] and improve ion abundance [28]. MSI analysis of complex samples such as biological tissue specimens can lead to the simultaneous generation of hundreds to thousands of unique ions. These ions are sampled into a mass spectrometer in concert, demanding high instrument mass resolving power to distinguish ions that are nearly isobaric. The capability of assigning peak identities based on exact mass provided by high mass resolving power and accuracy mass spectrometry reduces the need for additional analysis steps for identification like tandem mass spectrometry that can either reduce sample throughput or limit the breadth of multiplexed sample analysis. To this end, the IR-MALDESI source has recently been integrated with a hybrid quadrupole Orbitrap mass spectrometer, the Q Exactive, and the utility of this combination has been demonstrated for sensitive detection of lipids and metabolites [28, 29] as well as xenobiotics [28, 30, 31] in targeted or untargeted MSI methods. Two pulses of the mid-IR OPO laser are required to achieve complete and reproducible desorption and optimal response from IR-MALDESI analysis of tissue cryosections tens of micrometers thick [27]. To capture ions generated from each voxel probed during IR- MALDESI MSI, the ion accumulation time must be sufficiently long to encompass both laser desorption events with additional time for ion transfer and synchronization overhead. Hence, requisite ion inject time is inversely proportional to laser repetition rate, with a 20 Hz repetition rate mid-IR laser requiring more than 100 ms to accumulate ions generated from two laser pulses. As with other successful pulsed infrared MSI approaches [9], it is desirable to fix the number of pulses and the inject time during MSI acquisition at individual voxels, and as a result methods to control trapped ion populations such as automatic gain control (AGC) cannot

be utilized. AGC uses a short prescan to dynamically set an ion accumulation time in order to maintain the trapped ion population at a target number and reduce space charging and saturation effects. In the Q Exactive, ions are accumulated in an RF-only storage trap (C-trap) before being injected into the Orbitrap for a single FT acquisition. The C-trap can be filled with roughly 106 charges [32], below the capacity of the Orbitrap itself [33] and typical ion accumulation times with AGC on tend to be on the order of milliseconds for a typical LC-MS analysis. While measurement mass accuracy (MMA) of the mass spectrometer can be maintained during MSI in the absence of AGC through the use of lock-mass correction [32], the duration of ion accumulation required for MSI using typical mid-IR lasers is expected to degrade targeted ion abundance either as a result of reduced trapping efficiency or due to losses associated with space-charging of accumulated ions as the flux of sample-derived and ambient ions exceed the charge limits of the trap. In this work, we have developed a next-generation IR-MALDESI imaging source coupled to a Q-Exactive Plus mass spectrometer and investigate the gains in IR-MALDESI analyte response that can be attained by reduction of ion accumulation time offered by a higher repetition rate (100 Hz) pulsed mid-IR laser source. A comparison is made between response to ions introduced continuously and those that are generated through the pulsed IR-MALDESI mechanism. Results are presented for both targeted and untargeted studies of biological tissues.

3.2 Experimental Section

3.2.1 Materials

HPLC grade methanol and water were purchased from Burdick and Jackson (Muskegon, MI, U.S.A.). Formic acid, acetic acid, and lamivudine (3TC) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Emtricitabine (FTC) was obtained through the NIH AIDS Reagent Program. All materials were used without further purification.

Human cervical tissues were obtained from the University of North Carolina Tissue Procurement Facility through UNC IRB #09–0921 with written informed consent and incubated in a solution of 100 μg/mL FTC for 24 h, as previously described [28]. Mouse liver and hen ovaries were obtained from the NCSU College of Veterinary Medicine and School of Poultry Science, respectively. Animals were managed in accordance with the Institute for Laboratory Animal Research Guide and all husbandry practices were approved by North Carolina State University Institutional Animal Care and Use Committee (IACUC). All tissues were sectioned into 10 μm thick sections using a Leica CM1950 cryomicrotome (Buffalo Grove, IL, U.S.A.) and thaw mounted onto precleaned glass microscope slides or slides uniformly coated with an internal standard [30].

3.2.2 IR-MALDESI Source

A more advanced IR-MALDESI source for the mass spectrometry imaging system has been developed for the determination of analyte distribution within biological matrices. On the basis of the design from previous work [24], this next-generation ionization source incorporates a new high repetition IR laser, proportional-integral-derivative (PID) control of temperature and relative humidity (RH), and a high-precision, high-speed translation stage. Details of the new design components are described below. Within the fully enclosed source, a matrix layer of ice is carefully deposited onto the sample by adjusting the relative humidity within the source and the sample temperature. Incorporation of two new devices now allows for better control over ice deposition. First, power applied to the Peltier thermoelectric cooling element is automatically regulated using a TC-48–20 (TE Technology, Inc., Traverse City, MI) controller using pulse-width modulation. Second, relative humidity within the source chamber is monitored by an iSeries Humidity and Temperature controller (CNiTH-i32; Omega

Engineering) that can provide PID control to an actuated valve supplying a flow of dry N2 gas. Two-dimensional sample translation is achieved by stacked, motion controlled linear translation stages. A high-precision and high-speed linear stage (GTS70; Newport

Corporation, Irvine, CA, U.S.A.) is used along the x-axis to allow rapid transit between voxel positions at 100 nm resolution, with a slower motion controller (LTA-HS; Newport Corporation, Irvine, CA, U.S.A.) used for translation along the y-axis. A high repetition optical parametric oscillator (OPO) laser (IR Opolette HR; Opotek, Carlsbad, CA) is used to generate λ = 2.94 μm mid-infrared laser emission with a pulse duration of 9 ns for excitation of the asymmetric O—H stretch in water. The diode-pumped IR Opolette HR can operate at a 100 Hz repetition rate, whereas the previously employed Q-switched IR Opolette was limited to 20

Hz [24]. Two metallic silver mirrors steer the output beam through a pair of CaF2 lenses (PCCM2506CI-075 and PCX-25.4CI-400; Lambda Research Optics, Costa Mesa, CA) for beam expansion and collimation before it is focused with an uncoated CaF2 plano-convex lens (PCX-25.4CI-75) to a diameter of ∼200 μm, as evaluated by burn spots on thermal paper. Pulse energy at the sample stage was measured to be 1.05 mJ/pulse (Nova 2; Ophir, Jerusalem, Israel), corresponding to a laser fluence of 3.35 J/cm2 distributed over the measured beam waist, and can be adjusted with a motorized variable attenuator. All experiments were conducted at full laser fluence unless otherwise noted.

3.2.3 Mass Spectrometer

The IR-MALDESI imaging source was fully synchronized with a Thermo Fisher Scientific Q Exactive Plus mass spectrometer (Bremen, Germany). The Q Exactive Plus was operated in full scan mode with a mass range m/z 150–600 for small molecule analysis and m/z 250–1000 for lipid analysis, with mass resolving power set to 140,000 (fwhm at m/z 200). Since AGC is turned off during IR-MALDESI experiments, lock masses were utilized in the control software to achieve parts per million mass accuracy [32]. For positive ionization mode two peaks of an ambient ion, diisooctyl phthalate, at m/z 391.2843 [M + H+]+ and m/z 413.2662 [M + Na+]+ were used as lock masses. The peaks of palmitic acid at m/z 255.2329 [M – H+]− and stearic acid at m/z 283.2643 [M – H+]− were used as lock masses in negative ionization mode.

3.2.4 IR-MALDESI Imaging

As has been described previously [27], slide-mounted tissue sections were placed onto the Peltier stage and a controlled layer of ice was deposited on the surface. Source geometry was matched to parameter values previously found to yield optimal response from biological tissues when using an ice matrix [27] and no further optimization of this geometry for the next- generation source has yet been undertaken. Two laser pulses were used to completely ablate the 10 μm thick tissue cryosection for each voxel. Samples were translated 100 μm between laser ablation events in an oversampling method such that the stage was moved a distance less than the desorption threshold of the laser resulting in a consistent volume of tissue desorbed [34]. Imaging of a rectangular region of interest is performed in a left–right, top–down flyback pattern. The plume of neutral tissue material interacts with an orthogonal electrospray plume, where analytes partition into the electrospray droplets and ionize in an ESI-like fashion [29, 35]. For targeted, positive polarity experiments, a 50/50 (v/v) solution of methanol/water with 0.2% formic acid was used as the electrospray solvent. Untargeted lipid analysis was performed using a polarity switching method, where spectra are adjacent voxels are analyzed with alternating polarities [36, 37]. For polarity switching experiments, a 50/50 (v/v) solution of methanol/water with 1 mM acetic acid was used as the electrospray solvent, which has been found to yield the best untargeted analyte detection in acquisition from each polarity in comparison to other solvent systems [37]. The IR-MALDESI MSI acquisition GUI allows for the laser repetition rate to be changed by selecting fractions of the 100 Hz diode pump duty cycle for OPO excitation. On the basis of selected laser repetition rate, C-trap inject time is fixed according to t (ms) = 2n + 10 to allow collection of ions generated from two laser pulses with a laser period of n milliseconds and an additional 10 ms overhead to ensure synchronization of stage, laser, and mass spectrometer for each scan. To evaluate the effect of laser frequency and trap time on ion abundance, OPO excitation was performed using IR OPO repetition rates of 100, 50, 33, 25,

and 20 Hz with corresponding C-trap accumulation times of 30, 50, 70, 90, and 110 ms. A timing schematic for 100 and 20 Hz scenarios illustrating temporal overlap between signaling of laser firing and C-trap accumulation can be seen in Figure 3-1.

Figure 3-1. Timing diagram of key IR-MALDESI events during MSI associated with pulsed sample desorption by the mid-IR laser and analysis of ions by the Q Exactive Plus mass spectrometer. For each acquisition event, C-trap accumulation time is sufficiently long to accumulate ions generated from two laser pulses and scales inversely with laser repetition rate.

3.2.5 Data Analysis

The .RAW files obtained from the Q Exactive Plus instrument were converted to the mzML file format with the MSConvert software from Proteowizard [38], using a polarity filter to separate the positive and negative acquisitions for polarity switching experiments. The mzML files were then converted to the imzML file format using imzMLConverter [39] and processed using MSiReader [40]. In order to demonstrate the quality of the raw data, ion images were neither interpolated nor normalized (unless specified). All ion maps were created with a mass tolerance of 5 ppm, and sample dimensions are indicated by scale bars. Supervised analysis of the MSI data was performed by MSiReader for untargeted discovery of ions

associated with tissue. In this approach, the software averages voxel spectra over a region of interest specified by the user and identifies unique peaks associated with this region relative to a user-specified reference region, chosen to be off-tissue. In a left-to-right and top-to-bottom oversampling regime, the tissue sample volume of the first row and column imaged is larger than that of the remainder of the region of interest. These data have been excluded from analysis. Putative identification of ions selected by MSiReader based on exact mass has been performed where possible using the LIPID MAPS [41] and METLIN structure databases [42]. No further targeted MS/MS analysis was undertaken to confirm these assignments, and as a result no attempt is made to distinguish structural isomers. These ions are identified by their molecular formula and lipid class.

3.3 Results and Discussion

3.3.1 C-Trap Accumulation Time with Ions from Multiple Sources

In MSI analysis of biological materials by IR-MALDESI, analytical targets may include endogenous metabolites and therapeutic xenobiotics. Additionally, the mass spectra also include ions from ambient compounds, despite the IR-MALDESI source enclosure being purged with dry nitrogen to a constant RH. The effect of laser repetition rate and corresponding C-trap accumulation time on analyte ion abundance was evaluated initially for a representative ion from each of these classes during IR-MALDESI MSI of cervical tissue incubated in the antiretroviral drug FTC. Composed of a single stratified squamous epithelium cell type, cervical tissue represents a near-uniform lipid substrate and drug penetration was assumed to be uniform in tissue following 24 h incubation. Tissue cryosections from this model were analyzed by IR-MALDESI in discrete regions of 2 mm × 2 mm (20 × 20 voxels) using five laser ablation repetition rates (100, 50, 33, 25, and 20 Hz) with matching C-trap accumulation times (30, 50, 70, 90, 110 ms). Resulting ion maps, shown in Figure 3-2A, illustrate that analyte ion abundance increased as C-trap accumulation time was reduced for the following: (1) the

total ion current (TIC) over m/z 150–600 range; (2) cholesterol, an endogenous sterol; and (3)

FTC, incubated xenobiotic. Polydimethylcyclosiloxane (PDMS, [(CH3)2SiO]5), a ubiquitous ambient compound used in many personal products and considered a common interferent in mass spectrometry [43], decreased as C-trap accumulation time was reduced.

Figure 3-2. Dependence of Q Exactive Plus C-trap accumulation time on ambient and tissue-specific ion abundance during IR-MALDESI MSI analysis of ARV-incubated human cervical tissue. (A) MSI ion maps indicate increased ion abundance as C-trap inject time is reduced for: total ion current (150– 600 m/z); endogenous cholesterol; and xenobiotic emtricitabine (FTC). (B) Normalized ion abundance for each analyte, including 95% confidence limit, showing 2–100 fold increase when reducing accumulation time from 110 ms (previous limiting conditions with 20 Hz laser) to 30 ms (100 Hz laser). Average abundances for these ions were evaluated for each region of interest using MSiReader and normalized by the response at 100 Hz for each compound to allow the comparison of trends with C-trap accumulation time (Figure 3-2B). PDMS and other ambient compounds remain present in the IR-MALDESI source during imaging where they are ionized by the electrospray and sampled into the Q Exactive Plus continuously over the period of C- trap ion accumulation. The average abundance of this ambient compound increases nearly linearly over the accumulation time range. Analytes associated with tissue that are ablated and ionized via the pulsed IR-MALDESI mechanism exhibit a greater dependence on the laser repetition rate and trap time. Response to cholesterol, which commonly represents the most abundant tissue-related peak evaluated by IR-MALDESI, increases 6-fold as accumulation

time is reduced from 110 to 30 ms. The average ion abundance of FTC, an order of magnitude lower than that of cholesterol, increases by 2 orders of magnitude over the range of accumulation times. The TIC follows a similar response to cholesterol rather than ambient ions, suggesting that TIC is significantly comprised of tissue-derived ions. The relative contribution of C-trap accumulation time, based on laser repetition rate, to the observed gains in ion abundance is considered next. With typical thermal relaxation time in tissue on the order of milliseconds for length scales of the laser spot [44, 45], the laser ablation of tissue is expected to take place under a condition of total thermal confinement with no diffusion of heat over the laser pulse duration of 9 ns. The confined interaction preferentially imparts laser energy to explosive tissue ablation, leaving minimal remnant heat to contribute to improved ablation efficiency over the duty cycles under consideration in this work [32]. It is conjectured that the gains in ion abundance observed may be a result of minimizing ion loss due to space charging within the C-trap. This hypothesis is supported indirectly by the fact that the instrumental response to ambient ions injected continuously to the trap during ion accumulation decays to a lesser extent than the response to ions created by pulsed laser ablation/postionization, which must be trapped for longer periods of time before analysis. While it is beyond the scope of the current work to formally decouple effects attributable separately to laser duty cycle and ion inject time, it is nevertheless clear that their combination results in ion abundance and sensitivity gains during IR-MALDESI MSI that are anticipated to translate to lower limits of detection and better visualization of low abundance analyte distributions. The following sections summarize evaluation of the effect of C-trap inject time on IR-MALDESI response in both untargeted and targeted analyses, comparing laser repetition rates of 100 and 20 Hz.

3.3.2 Untargeted Analysis of Endogenous Lipids

Lipid profiles provide valuable information for understanding the biological basis of disease. Alterations in lipid metabolism have been linked to several diseases such as

hypertension, diabetes, and cancer [46, 47]. MSI lipidomics have been proposed to detect region-specific lipid signature of cancerous cellular regions in tissue [48] and in the determination of tumor margins [49]. A homogeneous tissue sample, mouse liver, was initially used to investigate untargeted analysis of biological materials. Lipids natively present were evaluated from a tissue section in positive ion mode over the mass range m/z 150–600. For each of the two laser repetition rates considered, nine 1 mm × 1 mm (10 × 10 voxels) regions of interest were analyzed at laser powers ranging from 10% to 100% to optimize MSI response of the next-generation IR- MALDESI source, although it should be stressed that a systematic optimization of experimental parameters remains to be performed. Representative ion maps associated with carnitine and cholesterol are shown in Figure 3-3A, systematically indicating higher lipid ion abundance at the higher laser repetition rate and shorter ion accumulation window. While these maps indicate that optimal laser fluence exhibits some analyte specificity, as was recently demonstrated [29], the highest response across all tissue-related analytes is achieved at 100% laser power. This condition has been used in all subsequent MSI experiments. Mass excess plots of all tissue-related peaks discovered by MSiReader relative to an off-tissue reference region can be seen in Figure 3-3B for all regions imaged at 100 and 20 Hz. Untargeted analysis of lipids associated with tissue over this mass range resulted in almost a 2-fold increase in analytes identified using the 100 Hz repetition rate than when imaging at 20 Hz (416 ions versus 246 ions, respectively). Ion abundance of these species ranged from approximately 1 to 3 orders of magnitude higher at 100 Hz than 20 Hz.

Figure 3-3. Laser fluence and C-trap accumulation dependence on endogenous lipid response from + + mouse liver tissue. (A) Representative ion maps of cholesterol (m/z 369.3516, [M – H2O + H ] ) and carnitine (m/z 162.1125, [M + H+]+). (B) Mass excess of all tissue-specific peaks (+ESI, black dots) overlaid on mass excess distribution of lipids from LIPID MAPS Structure Database [41], with greater discovery of tissue related peaks at 100 Hz repetition rate. The hen ovary, representing the only animal model for the spontaneous development of ovarian cancer [50, 51], was selected as a highly heterogeneous substrate for untargeted lipid analysis. An immature follicle in a hen ovarian tissue sample was analyzed in an untargeted manner on serial tissue sections using 100 and 20 Hz repetition rates (Figure 3-4A). Experiments were conducted in a polarity switching mode over the mass range m/z 250–1000 using a prescribed acquisition method alternating the acquisition polarity between adjacent

voxels to increase instrumental response across lipid classes consisting of a range of polar headgroup compositions that can be preferentially ionized in either positive or negative ionization mode. The electrospray solvent composition and the concentration of modifier added were optimized in order to minimize the bias of the solvent for one polarity over the other and also to ensure a comprehensive lipid coverage in both polarities [37]. Representative ion maps of the follicle in positive mode IR-MALDESI (m/z 369.3516, cholesterol) and negative mode IR-MALDESI (m/z 289.2174, putatively assigned nonadecadiynoic acid) are shown in Figure 3-4, parts B and C, respectively. As with the mouse liver, IR-MALDESI ion abundances for tissue-related ions were higher when imaging at the higher laser repetition rate and shorter C-trap accumulation time. The heterogeneous distribution of less abundant ions such as the fatty acid (Figure 3-4C) around the follicle nucleus relative to other regions of the ovarian tissue is more readily discerned under these conditions since the low analyte abundance remains above the limit of detection (LOD). At a repetition rate of 20 Hz, the fatty acid can only be detected in the follicular wall, whereas at 100 Hz the analyte can be resolved in the surrounding tissue structure. Hence, the added sensitivity of the higher laser repetition rate and shorter C-trap accumulation time offered a more accurate mapping of analyte distribution. This sensitivity gain also resulted in a 2-fold increase in the number of identified tissue-related peaks: A total of 368 tissue-specific ions (272 in positive, and 96 in negative) were identified at 20 Hz, whereas 777 tissue-specific ions (567 in positive, and 210 in negative) were identified at 100 Hz. All tissue-specific ions detected using a repetition rate of 20 Hz were present in the data set collected using a repetition rate of 100 Hz and their ion abundance was lower by at least an order of magnitude.

Figure 3-4. Untargeted lipid analysis of hen ovary. (A) Optical images of the tissue sections and the ROIs analyzed using 100 Hz (left) and 20 Hz (right) repetition rates, along with ion maps of (B) + + + − cholesterol (m/z 369.3516, [M – H2O + H ] ) and (C) Nonadecadiynoic acid (m/z 289.2174, [M – H ] ). (D) Mass excess of all tissue-specific peaks (+ESI, black dots; −ESI, white dots) overlaid on mass excess distribution of lipids generated from LIPIDMAPS Structure Database [41]. The plots indicate that all of the tissue-specific peaks using 100 and 20 Hz repetition rates correspond to lipids across different classes, with greater discovery of tissue related peaks at 100 Hz repetition rate. By improving the accuracy of analyte distribution for compounds with low ion abundance and increasing the number of potential biomarker candidates identified, IR- MALDESI MSI using the higher repetition rate and reduced ion accumulation time is expected to increase the likelihood for successful biomarker discovery from untargeted analysis of lipids.

3.3.3 Targeted Analysis of Xenobiotics

To evaluate drug efficacy, it is necessary to quantify drug concentration at key anatomical sites of action. In order to elucidate areas of analyte concentration that may arise due to biological variability of a sample, minimizing instrumental variability is imperative. As

has been demonstrated recently [30], normalization strategies relating the response of an analyte of interest to that of a uniformly deposited internal standard with similar desorption and ionization efficiency can significantly reduce the IR-MALDESI voxel-to-voxel variability. The FTC-incubated cervical tissue model was used to determine the analytical variability associated with MSI of a uniformly distributed analyte to determine the best experimental conditions for evaluating xenobiotic distributions. Serial sections of the FTC- incubated cervical tissue were mounted onto a glass slide uniformly sprayed with the FTC analogue 3TC, which is also an HIV antiretroviral drug, and then imaged consecutively at 100 and 20 Hz. As in the discrete regions of interest shown in Figure 3-2, the ion abundance of FTC distributed over an entire tissue section increased an order of magnitude when using the 100 Hz laser repetition rate and 30 ms accumulation time relative to imaging with the laser operating at a 20 Hz repetition rate (Figure 3-5, top). Additionally, the higher ion abundance of the internal standard, 3TC, in each voxel measured at 100 Hz imaging yields a more effective normalization of FTC ion abundances as quantified by the reduced variability per-voxel. With this normalization, relative standard deviations (%RSD) of incubated FTC to 3TC decreased from 103% to 28% for 110 and 30 ms inject time, respectively (Figure 3-5, bottom). This reduced variability per-voxel is critically important in determining accurate absolute concentrations of analyte within small tissue microenvironments encountered in heterogeneous tissues.

Figure 3-5. Targeted analysis of emtricitabine (FTC). Top: Ion maps of FTC during imaging at 100 and 20 Hz showing 100-fold increased absolute ion abundance under the latter conditions. Bottom: Normalized ion maps depicting the ratio of FTC:3TC, which was distributed uniformly onto the mounting slide as an internal standard. The higher abundance and reduced variability per spectrum of analyte and internal standard allows for normalization on a per-voxel level, improving %RSD from 103% to 28%. 3.4 Conclusions

A next-generation IR-MALDESI source has been designed and constructed, incorporating a mid-infrared OPO laser capable of operating at a higher repetition rate than the IR-MALDESI prototype. The higher repetition rate reduces the duration of time required to fire multiple laser shots required for complete desorption of material from each voxel of a biological sample, thereby shrinking the window of ion accumulation time in the C-trap of the

Q Exactive Plus during mass spectrometry imaging. Reducing the ion inject time was shown to result in gains in measured ion abundance for ions derived from continuous and pulsed mechanisms. Enhancement is significantly greater for ions generated by IR-MALDESI associated with biological material, resulting in up to 3 orders of magnitude greater ion abundance for certain analytes. This approach has multiple advantages for both untargeted and targeted analysis of biological tissues by MSI. For untargeted studies, the increased analyte response results in identification of greater number of tissue-specific metabolites that may represent key identifiers and biomarkers for evaluation of discrete tissue health or systems biology. Additionally, the heterogeneous distribution of less abundant analytes can be more precisely evaluated. As shown for ARV-incubated tissue, increased instrumental response toward both an incubated xenobiotic and a matched internal standard was achieved with the higher repetition rate laser, facilitated by a reduction in ion accumulation time. This significantly reduced the voxel-to-voxel variability when normalizing the xenobiotic response by that of the internal standard, which improves per-voxel analyte quantification. All gains in detectability observed in untargeted and targeted analyses are expected to increase as higher repetition rate mid-IR lasers become available, since the current ion accumulation time of 30 ms still represents a window that is an order of magnitude longer than those for dynamically controlled ion populations such as in typical LC–MS/MS analysis. The advantages detailed by this work are expected to be applicable to other MSI approaches utilizing pulsed infrared lasers.

3.5 Acknowledgment

The authors would like to thank Dr. Troy Ghashghaei, Dr. James Petitte, and Dr. Angela Kashuba for facilitating access to biological samples; A.K. is also thanked for access to the 100 Hz mid-IR laser for these investigations. The authors gratefully acknowledge financial support received from National Institutes of Health (R01GM087964, P30AI50410,

U19AI096113, and R01AI111891), the W. M. Keck Foundation, and North Carolina State University.

3.6 References

1. Castaing, R., Slodzian, G.: Microanalyse par émission ionique secondaire. J. Microscopie. 1, 395-410 (1962)

2. Spengler, B., Hubert, M., Kaufman, R.: In Proceedings of the 42nd Annual Conference on Mass Spectrometry and Allied Topics., Chicago, Illinois, p. 1041 (1994)

3. Chaurand, P., Schwartz, S.A., Reyzer, M.L., Caprioli, R.M.: Imaging mass spectrometry: principles and potentials. Toxicologic pathology. 33, 92-101 (2005)

4. McDonnell, L.A., Heeren, R.: Imaging mass spectrometry. Mass spectrometry reviews. 26, 606-643 (2007)

5. Heeren, R.M.A., Smith, D.F., Stauber, J., Kükrer-Kaletas, B., MacAleese, L.: Imaging mass spectrometry: Hype or hope? Journal of the American Society for Mass Spectrometry. 20, 1006-1014 (2009)

6. Balluff, B., Schöne, C., Höfler, H., Walch, A.: MALDI imaging mass spectrometry for direct tissue analysis: technological advancements and recent applications. Histochemistry and Cell Biology. 136, 227 (2011)

7. Chaurand, P.: Imaging mass spectrometry of thin tissue sections: A decade of collective efforts. Journal of Proteomics. 75, 4883-4892 (2012)

8. Jungmann, J.H., Heeren, R.M.A.: Emerging technologies in mass spectrometry imaging. Journal of Proteomics. 75, 5077-5092 (2012)

9. Römpp, A., Schäfer, K.C., Guenther, S., Wang, Z., Köstler, M., Leisner, A., Paschke, C., Schramm, T., Spengler, B.: High-resolution atmospheric pressure infrared laser desorption/ionization mass spectrometry imaging of biological tissue. Analytical and Bioanalytical Chemistry. 405, 6959-6968 (2013)

10. Reyzer, M.L., Caprioli, R.M.: MALDI-MS-based imaging of small molecules and proteins in tissues. Current Opinion in Chemical Biology. 11, 29-35 (2007)

11. Svatoš, A.: Mass spectrometric imaging of small molecules. Trends in Biotechnology. 28, 425-434 (2010)

12. Burnum, K.E., Frappier, S.L., Caprioli, R.M.: Matrix-Assisted Laser Desorption/Ionization Imaging Mass Spectrometry for the Investigation of Proteins and Peptides. Annual Review of Analytical Chemistry. 1, 689-705 (2008)

13. Goodwin, R.J.A., Pennington, S.R., Pitt, A.R.: Protein and peptides in pictures: Imaging with MALDI mass spectrometry. PROTEOMICS. 8, 3785-3800 (2008)

14. Cillero-Pastor, B., Heeren, R.M.A.: Matrix-Assisted Laser Desorption Ionization Mass Spectrometry Imaging for Peptide and Protein Analyses: A Critical Review of On- Tissue Digestion. Journal of Proteome Research. 13, 325-335 (2014)

15. Fernández, J.A., Ochoa, B., Fresnedo, O., Giralt, M.T., Rodríguez-Puertas, R.: Matrix- assisted laser desorption ionization imaging mass spectrometry in lipidomics. Analytical and Bioanalytical Chemistry. 401, 29-51 (2011)

16. Murphy, R.C., Merrill Jr, A.H.: Lipidomics and Imaging Mass Spectrometry. Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids. 1811, 635-636 (2011)

17. Goto-Inoue, N., Hayasaka, T., Zaima, N., Setou, M.: Imaging mass spectrometry for lipidomics. Biochimica et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids. 1811, 961-969 (2011)

18. Touboul, D., Brunelle, A., Laprévote, O.: Mass spectrometry imaging: Towards a lipid microscope? Biochimie. 93, 113-119 (2011)

19. Rubakhin, S.S., Jurchen, J.C., Monroe, E.B., Sweedler, J.V.: Imaging mass spectrometry: fundamentals and applications to drug discovery. Drug Discovery Today. 10, 823-837 (2005)

20. Rudin, M., Rausch, M., Stoeckli, M.: Molecular Imaging in Drug Discovery and Development: Potential and Limitations of Nonnuclear Methods. Molecular Imaging and Biology. 7, 5-13 (2005)

21. Rudin, M., Weissleder, R.: Molecular imaging in drug discovery and development. Nat Rev Drug Discov. 2, 123-131 (2003)

22. Castellino, S., Groseclose, M.R., Wagner, D.: MALDI imaging mass spectrometry: bridging biology and chemistry in drug development. Bioanalysis. 3, 2427-2441 (2011)

23. Greer, T., Sturm, R., Li, L.: Mass spectrometry imaging for drugs and metabolites. Journal of Proteomics. 74, 2617-2631 (2011)

24. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-MALDESI) Imaging Source Coupled to a FT-ICR Mass Spectrometer. Journal of The American Society for Mass Spectrometry. 24, 92-100 (2013)

25. Goodwin, R.J.A.: Sample preparation for mass spectrometry imaging: Small mistakes can lead to big consequences. Journal of Proteomics. 75, 4893-4911 (2012)

26. Dixon, R.B., Muddiman, D.C.: Study of the ionization mechanism in hybrid laser based desorption techniques. Analyst. 135, 880-882 (2010)

27. Robichaud, G., Barry, J.A., Muddiman, D.C.: IR-MALDESI Mass Spectrometry Imaging of Biological Tissue Sections Using Ice as a Matrix. Journal of The American Society for Mass Spectrometry. 25, 319-328 (2014)

28. Barry, J.A., Robichaud, G., Bokhart, M.T., Thompson, C., Sykes, C., Kashuba, A.D.M., Muddiman, D.C.: Mapping Antiretroviral Drugs in Tissue by IR-MALDESI MSI Coupled to the Q Exactive and Comparison with LC-MS/MS SRM Assay. Journal of The American Society for Mass Spectrometry. 25, 2038-2047 (2014)

29. Rosen, E.P., Bokhart, M.T., Ghashghaei, H.T., Muddiman, D.C.: Influence of Desorption Conditions on Analyte Sensitivity and Internal Energy in Discrete Tissue or Whole Body Imaging by IR-MALDESI. Journal of The American Society for Mass Spectrometry. 26, 899-910 (2015)

30. Bokhart, M.T., Rosen, E., Thompson, C., Sykes, C., Kashuba, A.D.M., Muddiman, D.C.: Quantitative mass spectrometry imaging of emtricitabine in cervical tissue model using infrared matrix-assisted laser desorption electrospray ionization. Analytical and Bioanalytical Chemistry. 407, 2073-2084 (2015)

31. Thompson, C.G., Bokhart, M.T., Sykes, C., Adamson, L., Fedoriw, Y., Luciw, P.A., Muddiman, D.C., Kashuba, A.D.M., Rosen, E.P.: Mass Spectrometry Imaging Reveals Heterogeneous Efavirenz Distribution within Putative HIV Reservoirs. Antimicrobial Agents and Chemotherapy. 59, 2944-2948 (2015)

32. Olsen, J.V., de Godoy, L.M.F., Li, G., Macek, B., Mortensen, P., Pesch, R., Makarov, A., Lange, O., Horning, S., Mann, M.: Parts per Million Mass Accuracy on an Orbitrap Mass Spectrometer via Lock Mass Injection into a C-trap. Molecular & Cellular Proteomics. 4, 2010-2021 (2005)

33. Makarov, A., Denisov, E., Lange, O., Horning, S.: Dynamic range of mass accuracy in LTQ orbitrap hybrid mass spectrometer. Journal of the American Society for Mass Spectrometry. 17, 977-982 (2006)

34. Jurchen, J.C., Rubakhin, S.S., Sweedler, J.V.: MALDI-MS imaging of features smaller than the size of the laser beam. Journal of the American Society for Mass Spectrometry. 16, 1654-1659 (2005)

35. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Generation and detection of multiply-charged peptides and proteins by matrix-assisted laser desorption electrospray ionization (MALDESI) fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry. 17, 1712-1716 (2006)

36. Cochran, K.H., Barry, J.A., Robichaud, G., Muddiman, D.C.: Analysis of trace fibers by IR-MALDESI imaging coupled with high resolving power MS. Analytical and Bioanalytical Chemistry. 407, 813-820 (2015)

37. Nazari, M., Muddiman, D.C.: Polarity switching mass spectrometry imaging of healthy and cancerous hen ovarian tissue sections by infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). Analyst. 141, 595-605 (2016)

38. Kessner, D., Chambers, M., Burke, R., Agus, D., Mallick, P.: ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics. 24, 2534- 2536 (2008)

39. Race, A.M., Styles, I.B., Bunch, J.: Inclusive sharing of mass spectrometry imaging data requires a converter for all. Journal of Proteomics. 75, 5111-5112 (2012)

40. Robichaud, G., Garrard, K.P., Barry, J.A., Muddiman, D.C.: MSiReader: an open- source interface to view and analyze high resolving power MS imaging files on Matlab platform. Journal of the American Society for Mass Spectrometry. 24, 718-721 (2013)

41. Sud, M., Fahy, E., Cotter, D., Brown, A., Dennis, E.A., Glass, C.K., Merrill Jr, A.H., Murphy, R.C., Raetz, C.R., Russell, D.W.: Lmsd: lipid maps structure database. Nucleic acids research. 35, D527-D532 (2006)

42. Smith, C.A., O'Maille, G., Want, E.J., Qin, C., Trauger, S.A., Brandon, T.R., Custodio, D.E., Abagyan, R., Siuzdak, G.: METLIN: a metabolite mass spectral database. Therapeutic drug monitoring. 27, 747-751 (2005)

43. Keller, B.O., Sui, J., Young, A.B., Whittal, R.M.: Interferences and contaminants encountered in modern mass spectrometry. Analytica Chimica Acta. 627, 71-81 (2008)

44. Levis, R.J.: Laser Desorption and Ejection of Biomolecules From the Condensed Phase into the Gas Phase. Annual Review of Physical Chemistry. 45, 483-518 (1994)

45. Dreisewerd, K.: The Desorption Process in MALDI. Chemical Reviews. 103, 395-426 (2003)

46. Menendez, J.A., Lupu, R.: Fatty acid synthase and the lipogenic phenotype in cancer pathogenesis. Nat Rev Cancer. 7, 763-777 (2007)

47. Zaidi, N., Lupien, L., Kuemmerle, N.B., Kinlaw, W.B., Swinnen, J.V., Smans, K.: Lipogenesis and lipolysis: The pathways exploited by the cancer cells to acquire fatty acids. Progress in Lipid Research. 52, 585-589 (2013)

48. Veselkov, K.A., Mirnezami, R., Strittmatter, N., Goldin, R.D., Kinross, J., Speller, A.V.M., Abramov, T., Jones, E.A., Darzi, A., Holmes, E., Nicholson, J.K., Takats, Z.: Chemo-informatic strategy for imaging mass spectrometry-based hyperspectral profiling of lipid signatures in colorectal cancer. Proceedings of the National Academy of Sciences. 111, 1216-1221 (2014)

49. Wiseman, J.M., Puolitaival, S.M., Takáts, Z., Cooks, R.G., Caprioli, R.M.: Mass Spectrometric Profiling of Intact Biological Tissue by Using Desorption Electrospray Ionization. Angewandte Chemie International Edition. 44, 7094-7097 (2005)

50. Johnson, P.A., Giles, J.R.: The hen as a model of ovarian cancer. Nat Rev Cancer. 13, 432-436 (2013)

51. Hakim, A.A., Barry, C.P., Barnes, H.J., Anderson, K.E., Petitte, J., Whitaker, R., Lancaster, J.M., Wenham, R.M., Carver, D.K., Turbov, J., Berchuck, A., Kopelovich, L., Rodriguez, G.C.: Ovarian Adenocarcinomas in the Laying Hen and Women Share Similar Alterations in p53, ras, and HER-2/neu. Cancer Prevention Research. 2, 114- 121 (2009)

4 Quantitative Mass Spectrometry Imaging of Emtricitabine in Cervical Tissue Model using Infrared Matrix-Assisted Laser Desorption Electrospray Ionization

The following work was reprinted with permission from: Bokhart, M.T., Rosen, E., Thompson, C., Sykes, C., Kashuba, A.D.M., Muddiman, D.C.: Quantitative mass spectrometry imaging of emtricitabine in cervical tissue model using infrared matrix-assisted laser desorption electrospray ionization. Analytical and Bioanalytical Chemistry. 407, 2073-2084 (2015). DOI: 10.1007/s00216-014-8220-y. Copyright © 2014 Springer-Verlag Berlin Heidelberg. The original publication may be accessed via the Internet at https://link.springer.com/article/10.1007%2Fs00216-014-8220-y

4.1 Introduction

Mass spectrometry imaging (MSI) data is generated by recording mass spectra and the corresponding spatial location [1]. Construction of ion heat maps creates a visual representation of a compound relating the observed ion abundance with its location. MSI is quickly becoming an invaluable tool in drug distribution studies [2, 3, 4]. The efficacy of a drug can be limited by the ability to reach its intended target; plasma concentrations of a drug are often used but may not accurately reflect the concentration of the drug at its site of action [5]. Thus, spatial distribution of a drug in tissue can provide vital information related to the efficacy of a drug. Moreover, the use of a MSI strategy can provide additional information about endogenous compounds and metabolites. Simultaneous analysis of multiple species while retaining spatial specificity is a challenge for current quantitative methods. Selective reaction monitoring (SRM) assays using liquid chromatography tandem mass spectrometry (LC-MS/MS) for quantification of pharmaceuticals are commonly used for pharmacokinetic analysis of drugs in plasma and tissues [6]. These methods typically require extraction of the analyte from a tissue homogenate and result in the loss of spatial information within a tissue or organ. Alternative approaches such as quantitative whole body autoradiography (QWBA) or positron emission tomography (PET) offer the ability to visualize drug distribution into tissues while providing quantitative

information [4, 7, 8]. However, QWBA and PET require radiolabeled compounds which only give information about the radiolabel and are therefore not able to distinguish between the parent compound and its metabolites. Further, these quantitative imaging techniques incur significant experimental cost especially in the analysis of multiple analytes in multidrug therapies. Although MSI has proven its utility to provide important spatial information of drugs in tissue, quantitative information in MSI has proven difficult to achieve. Matrix-assisted laser desorption ionization (MALDI) has been used extensively in qualitative MSI and with some success of quantification [1, 7, 9-37]. The recent advances in quantification using MSI were the subject of recent review [38]. Quantitative MALDI MSI has several limitations including the need for organic matrix deposition for ionization of analytes; the analysis is also typically performed under vacuum. Ambient ionization mass spectrometry allows tissues to be sampled at conditions much closer to their natural state [39]. Pixel-to-pixel variability in MSI poses the biggest challenge for quantification [40]. This variability can arise from a range of sources, including morphological features, ionization efficiency, and detection efficiency. Increasing sensitivity and reducing variability are essential steps toward making MSI a routine quantitative technique. Several groups have reported various normalization methods to account for tissue-specific signal response and reduced variation per pixel [18, 26, 27]. These studies exemplify the need for a suitable normalization compound to produce quantitative MSI data. The ideal normalization compound accounts for structure-specific ablation and ionization efficiency for the analyte. Stable isotope-labeled compounds are the best normalization compounds, as they are chemical and structural analogs of the target analyte. Matrix-assisted laser desorption electrospray ionization (MALDESI) was first presented in 2006 using a UV laser to resonantly excite an organic matrix with secondary post ionization by electrospray ionization (ESI) [41]. Several other reports demonstrate the use of ESI for post ionization of a laser ablation plume [42, 43]. The use of a mid-infrared (IR) laser (2.94 μm) in

MALDESI allows the use of endogenous and exogenous ice matrix to be used as a matrix, thus simplifying the sample preparation steps and providing spectra without matrix peak interference [44]. High fluence of the mid-IR laser allows IR-MALDESI to completely ablate through a 10 μm thick tissue section at each rastered position in two laser pulses, resulting in the analysis of a voxel of tissue [44]. Recently, the quantitative measurement abilities of the secondary ionization source laser electrospray mass spectrometry (LEMS) were reported for equimolar and nonequimolar solutions of multiple analytes [45]. LEMS was demonstrated to have a monotonic signal response as a function of concentration which was similar to ESI. The need for a reliable method of characterizing spatial distribution of drug therapies within tissues, and the advantages of MSI, is exemplified within the field of human immunodeficiency virus (HIV). HIV replication has been shown to persist in certain anatomic sites such as the lymphatic system and reproductive tract [46, 47]. These viral reservoirs represent a significant obstacle in the cure of HIV. Evaluations of antiretroviral (ARV) penetration into these reservoirs are critical for understanding whether current therapies will be sufficient to completely eliminate HIV from the body. MSI represents a promising technique to advance the understanding of within-tissue ARV distribution and to provide invaluable information toward the development of drug therapies targeting the HIV reservoir. Herein, we present a quantitative mass spectrometry imaging (QMSI) technique for the quantification of emtricitabine, a commonly used ARV [48], in human cervical tissue. Prior to quantification, several approaches for enhancing detection of multiple ARV drugs via IR- MALDESI MSI were evaluated. Tissue or morphologic specific ionization variability was then specifically addressed by using normalization compound-coated slides. Increased sensitivity and reduced voxel-to-voxel variability subsequently allowed for quantification of therapeutically relevant concentrations of emtricitabine in tissue by IR-MALDESI MSI. Cross- validation was performed using a validated LC-MS/MS method for adjacent tissue slices.

4.2 Experimental

4.2.1 Materials

HPLC grade methanol and water were purchased from Burdick and Jackson (Muskegon, MI, USA). Sodium chloride (>99.5 %), silver nitrate (>99 %), potassium chloride (>99.5 %), formic acid (LC/MS grade), lamivudine (>98 %), acyclovir (>99 %), and prednisolone (>99 %) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Emtricitabine (FTC), tenofovir (TFV), and raltegravir (RAL) were obtained from the NIH AIDS Reagent Program, directed by the Pathogenesis and Basic Research Branch, Basic 13 15 13 Sciences Program, Division of AIDS (DIADS), NIAID, NIH. C N2-FTC and C5-TFV were purchased from Moravek Biochemicals (Brea, CA, USA). The structures of the ARV drugs and prednisolone are shown in Figure 4-1. All materials were used as received without further purification.

Figure 4-1. Structures of the targeted analytes. Emtricitabine (FTC) and lamivudine (3TC) are nucleoside reverse transcriptase inhibitors (NRTIs) and are cytosine analogs. Tenofovir (TFV) and acyclovir (ACV) are also NRTIs and are adenosine and guanosine analogs, respectively. Raltegravir (RAL) belongs to the HIV integrase inhibitor class of ARV drugs. Although not structurally similar, prednisolone (PRED) was used as an internal standard for RAL

4.2.2 Tissue Samples

Human cervical tissues were obtained from the University of North Carolina Tissue Procurement Facility through UNC IRB #09-0921. The tissues were prepared and incubated in FTC, TFV, and RAL at a concentration of 100 μg/mL as previously described [49]. Tissues were cryosectioned using a Leica CM1950 cryomicrotome (Buffalo Grove, IL, USA) to a thickness of 10 μm.

4.2.3 IR-MALDESI Imaging

The imaging MALDESI source coupled to the Q Exactive mass spectrometer (Thermo Scientific, Bremen, Germany) has been previously described in greater detail [44, 49, 50]. Briefly, a tissue section was thaw mounted onto a microscope slide and placed on a XYZ- controlled Peltier stage. The enclosure surrounding the source was purged with nitrogen (MWSC, Raleigh, NC, USA) to less than 3 % relative humidity (RH) prior to the Peltier stage being cooled to −9 °C for approximately 10 min. The cooled stage and sample were then exposed to the ambient RH in the laboratory air to form a thin ice layer on the sample. After a sufficient amount of ice was deposited, the enclosure was purged with nitrogen to an approximate RH of 10 % to balance the rate of ice formation and sublimation. Two pulses at a repetition rate of 20 Hz from a mid-IR laser λ = 2.94 μm (IR-Opolette 2371, Opotek, Carlsbad, CA, USA) were used to resonantly excite endogenous and exogenous water on the sample causing complete ablation of tissue material. The ablated material from the laser pulse is ejected normal to the surface where it interacts with an intersecting ESI plume. The neutral species generated in the laser ablation event partition into the charged electrospray droplets where they undergo secondary ionization in an ESI-like mechanism [41, 51]. The tissue ablation diameter of the laser is approximately 150 μm [52]. The laser was rastered across the sample at a spot-to-spot distance of 100 μm using an oversampling technique [44, 53]. The Q Exactive mass analyzer was fully integrated with IR-MALDESI source and synchronized to

accumulate ions in the C-trap from both laser pulses prior to a single Orbitrap acquisition. The mass range was set to m/z 150–600, with the mass resolving power set to RPFWHM = 140,000 at m/z 200. High mass measurement accuracy (MMA) within 2 ppm was achieved using protonated and sodiated adducts of diisooctyl phthalate as two internal lock masses at m/z 391.28428 and 413.26623 [54].

4.2.4 Electrospray Ionization Cationization Agents

The effects of several cationization agents were investigated using cervical tissue incubated in ARV drug solution. The incubated tissue was analyzed using 0.2 % formic acid or 30 μM solutions of sodium chloride, potassium chloride, or silver nitrate in 50/50 (v/v) 13 15 methanol/water as electrospray solvents. Subsequently, 100 nL spots of 4 μg/mL C N2-FTC, 13 C5-TFV, TFV, FTC, RAL, lamivudine (3TC), acyclovir (ACV), and prednisolone (PRED) solution were pipetted on top of a blank cervical tissue. The analyte spots were analyzed in duplicate with electrospray solvents containing 0.2 % formic acid or 0, 10, 20, 30, 40, and 50 μM sodium chloride in 50/50 (v/v) methanol/water.

4.2.5 Sample Separation for Quantitative MSI

An overview of the quantification workflow using IR-MALDESI MSI is presented in Figure 4-2. For quantitative MSI, microscope slides were evenly coated with 3TC, ACV, and PRED using an automated pneumatic sprayer (TM Sprayer, LEAP Technologies, Carrboro, NC, USA). 3TC, ACV, and PRED were selected as internal standards based on their structural similarities to the ARV drugs (Figure 4-1). The conditions used for evenly coating a microscope slide using the TM sprayer are summarized in Table 4-1. Cryosectioned tissue slices were subsequently thaw mounted onto the coated slides. For absolute quantification, a 13 15 dilution series of C N2-FTC in 50/50 (v/v) methanol/water at concentrations of 0, 0.25, 0.50, 1.0, 2.0, 4.0, 6.0, and 8.0 μg/mL was pipetted on top of the tissue using a modified microliter

syringe (Hamilton Company, Reno, NV, USA). One hundred nanoliters of each standard solution was used for each calibration spot which covered an area approximately 1.0 mm in diameter on tissue. The absolute quantification was performed with five replicates.

Figure 4-2. Workflow for quantitative IR-MALDESI MSI. Table 4-1. TM sprayer conditions.

Temperature 45 ºC Number of passes 4

Sheath gas pressure 10 psi Line spacing 3.0 mm

Solvent 50/50 methanol/water Rotate 90º Pass 2 and 4 composition

Solvent flow 10 μl/min Offset 1.5 mm on pass 2 and 4

Stage speed 500 mm/min Nozzle height 20 mm

TM Sprayer conditions were optimized using rhodamine 6G fluorescent dye in the solvent and imaged using a BioRad PharosFX fluorescence imager to estimate coating homogeneity. An area of 35.4 mm x 65 mm was covered in 2.5 minutes per coat. Four coats were applied at a solvent flow rate of 10 μl per minute. 10 μl/min*2.5 min*4 coats= 100 μl of 200 μM ARV solution to give 87 fmol/voxel with the 100 μm laser step size.

4.2.6 Data Processing

For generating ion images, the Thermo Fisher .RAW files were converted to mzXML files using MSConvert software from Proteowizard [55]. Composite images containing data sets from multiple acquisitions were created by converting the .RAW file to an mzXL file using MSConvert followed by conversion into an imzML file using imzML converter [56]. The individual imzML files were then merged into a single master imzML file containing the stacked images from multiple experiments. mzXML or imzML file format was loaded into and processed in MSiReader, an open source, vendor neutral imaging software [57]. Ion abundance maps were created with a bin width of 5 ppm. Centroid values for analytes used in determining MMA were obtained using RawMeat 2.1 (VAST Scientific, Cambridge, MA, USA).

4.2.7 Quantitative IR-MALDESI Imaging

13 15 All observed FTC and C N2-FTC ion abundances were normalized to the internal standard 3TC in MSiReader unless otherwise noted. The concentration of each calibration spot 13 15 was calculated by dividing the total amount of C N2-FTC in picograms by the corresponding spot area in square millimeters as determined in MSiReader. The resulting calibration points are given as a concentration in picograms per square millimeter. This method accounts for variability in spot size from manually spotting the standard on the tissue. The concentration of 13 15 each calibration point was plotted against the corresponding normalized average C N2- FTC/3TC ion abundance ratio. The resulting calibration points were fitted with an unweighted

linear regression. The average ratio of incubated FTC to 3TC over the entire tissue was calculated. The concentration of FTC in tissue was determined using the average normalized 13 15 ion abundance and the C N2-FTC linear regression equation. This value represents incubated FTC as a concentration in picograms per square millimeter. The FTC concentration was converted to microgram per gram tissue assuming a tissue density of 1.0 mg/mm3 for comparison to LC-MS/MS results.

4.2.8 LC-MS/MS Quantification

For each tissue slice analyzed by IR-MALDESI, an adjacent 10 μm thick tissue slice was homogenized and analyzed by a validated LC-MS/MS method as previously described [49]. LC-MS/MS calibration standards were prepared at 0.3, 0.6, 1.5, 6, 15, 30, 75, 150, 255, and 300 ng/mL with quality control (QC) samples at 0.9, 21, and 240 ng/mL. The tissue slices were homogenized and extracted with 1.00 mL of solvent; therefore, the final concentration in nanograms per milliliter was equivalent to nanogram per tissue slice. The tissue area of the adjacent section analyzed by MALDESI QMSI was used to calculate the FTC concentration as microgram per gram tissue, assuming a tissue density of 1.0 mg/mm3. All calibration and QC samples were within 10 % of their nominal concentrations.

4.3 Results and Discussion

4.3.1 Sensitivity Enhancement and Interference Removal

A single incubated cervical tissue was analyzed in four quadrants to compare different cationization agents, namely formic acid, NaCl, KCl, and AgNO3. We hypothesized that different cationization agents may increase the ionization efficiency and simultaneously reduce the variability between spectra. Moreover, the corresponding m/z shift associated with different cations can be particularly advantageous in instances where an isobaric interference overlaps with the [M+H+]+ analyte peak. Figure 4-3A shows the optical image of the incubated tissue

with the cationization agent labeled. The corresponding ion maps of FTC, TFV, and RAL with each respective adduct for each quadrant are summarized in Figure 4-3B–D. The sodium adduct of the three incubated drugs is detected with higher ion abundance and with less variability than any of the other cationization agents (Table 4-2). The largest gain in detected ion abundance is observed for the detection of RAL, with nearly an order of magnitude increase. The high affinity of RAL toward sodium cations is attributed to its molecular structure (Figure 4-1). RAL is classified as an integrase inhibitor, all of which share two common structural features: a hydrophobic benzyl moiety and a chelating triad to bind to two Mg2+ ions [58]. RAL is suspected to chelate sodium ions in a similar manner since Na+ and Mg2+ have a similar ionic radius. The formation of two Na+ adducts was negligible at the concentrations studied here (data not shown). Based on the improved response across all ARVs, NaCl was doped into the electrospray solvent for all subsequent experiments.

Figure 4-3. Cervical tissue incubated in 100 μg/mL ARV drug solution was analyzed using 0.2 % formic acid [M+H+]+, 30 μM sodium chloride [M+Na+]+, 30 μM potassium chloride [M+K+]+, or 30 μM silver nitrate [M+107Ag+]+ in 50:50 methanol water ESI solution. (A) Optical image of the tissue before exogenous ice matrix formation and ion maps for incubated (B) emtricitabine, (C) tenofovir, and (D) raltegravir Table 4-2. Average abundances and relative standard deviations of the incubated drugs seen in Figure 4-3. Sodium cationization provided higher average abundance and lower RSD values. nd= not detected.

Cation H+ Na+ K+ 107Ag+ 109Ag+

Average abundance, % relative standard deviation

FTC 2051, 132% 4599, 86% 2229, 82% 264, 229% 237, 214%

TFV 925, 160% 1250, 117% 199, 212% nd nd

RAL 5886, 130% 73518, 62% 6221, 73% 269, 233% 250, 242%

To further optimize the cationization conditions, the concentration of the sodium chloride in the electrospray solvent was systematically varied from 0 to 50 μM in 10 μM increments using a series of ARV calibration spots of the same concentration. In addition, two spots were analyzed with 0.2 % formic acid for comparison to the previously optimized 13 15 conditions. Ion maps of the protonated and sodiated molecules of C N2-FTC are shown in Figure 4-4. The [M+H+]+ to [M+Na+]+ ratio decreased with increasing concentration of NaCl 13 15 + + in solution as expected from competing ionization. However, a maximum [ C N2-FTC+Na ] ion abundance occurred at 30 μM. This maximum at 30 μM NaCl was consistent for all compounds in the solution (data not shown). Figure 4-4 also highlights the removal of an isobaric interference from the analyte. The protonated form of stable isotope-labeled (SIL) 13 15 + + FTC, [ C N2-FTC+H ] , had an isobaric interference which makes identification and quantification at low concentrations difficult. In contrast, the sodium adduct of the compound was free from this interference. Based on these results, 30 μM NaCl in 50/50 methanol/water was used for the quantification of the incubated tissues.

Figure 4-4. Composite image demonstrating the ionization effects of several concentrations of sodium chloride and 0.2 % formic acid in the ESI solvent. A solution containing all the ARV in the study was 13 15 pipetted onto a blank tissue to determine optimum selectivity and sensitivity. The ion map of C N2- FTC is shown to exemplify the use of 30 μM sodium chloride in the ESI solvent to remove isobaric interference and increase sensitivity.

4.3.2 Normalization Strategy

The QMSI calibration curve presented used the stable isotope-labeled analog to quantify the analyte. In the present study, a different but structurally similar compound was used as an internal standard to account for voxel-to-voxel variation because the SIL-FTC is being used for quantification. Fortunately, several structural analogs mimicking a specific

nucleotide have been synthesized for use in ARV therapy. 3TC and FTC are cytosine analogs, while ACV and TFV are guanosine and adenosine analogs, respectively. The analytes and their respective normalization compounds are shown in Figure 4-1. The incorporation of a structurally similar normalization compound allows for the normalization of analyte ion abundances on a per voxel basis. It is important to note that the laser in IR-MALDESI experiments completely ablates all tissue material from each location to ensure the same amount of material is sampled at each position in the oversampling method used. This complete ablation also allows the use of internal standard to be placed under the tissue in the analysis and still be effectively sampled. Variability between IR-MALDESI response to standards spotted above and below tissue has been demonstrated to be less than 17 % (Figure 4-5). 3TC, ACV, and PRED were evenly sprayed on a microscope slide at approximately 90 fmol/voxel prior to thaw mounting a 10 μm thick incubated tissue section.

Figure 4-5. Three calibration spots of FTC, TFV, and RAL were placed on the microscope slide prior to thaw mounting blank cervical tissue followed by three calibration spots placed directly on top of the tissue. The application of the standard solution on the microscope slide vs. on tissue resulted in different size calibration spots. As a result, the ion abundance for each spot (in triplicate) was summed (arbitrary units) ± for direct comparison of spots under and on top of the tissue. The sum ion abundance differed by 3%, 11%, and 17% for FTC, TFV, and RAL, respectively. This result supports the near complete ablation of tissue related material that is visually observed when performing the analysis with IR- MALDESI.

Figure 4-6 shows the ion maps used in the normalization strategy presented. Figure 4-6A shows the absolute ion abundance for the sodium adduct of the incubated FTC, while Figure 4-6B depicts the absolute ion abundance for the sodium adduct of the normalization compound, 3TC. The normalized ion map in Figure 4-6C is created by taking the ratio of analyte to normalization compound at each voxel. The inclusion of a normalization compound has been previously shown to account for tissue-specific analyte response in MSI [26, 27, 28].

Figure 4-6. Visualization of the normalization process used to reduce variability at a per voxel basis of tissue replicate 2. Ion maps of (A) the absolute ion abundance of analyte [FTC+Na+]+, (B) absolute ion abundance of normalization compound [3TC+Na+]+, (C) ratio of analyte to internal standard [FTC+Na+]+ / [3TC+Na+]+ The complete ablation of the sample at each position is a major advantage over MALDI-based MSI techniques where compounds placed below the tissue may be incorporated

at a lesser extent due to extraction into the tissue and then into the organic matrix used in MALDI analyses. Similarly, compounds placed on top of the tissue may be incorporated to a greater extent than compounds within the tissue, representing a surface concentration and not an in-tissue concentration. The extraction of compounds into the organic matrix is seen to be one of the biggest sources of variability in MALDI MSI analysis. IR-MALDESI proves to circumvent most of these issues with the complete ablation of tissue at each spatial location. Therefore, each pixel in the ion maps actually represents a volume element of tissue and should be referred to as a voxel. This work here is performed on a single tissue type, cervical, which was assumed to be relatively homogeneous in composition. Also, the incubation method used in distributing the compound within the tissue is assumed to be homogeneous. Finally, the internal standard was evenly applied using an automatic pneumatic sprayer and assumed to be homogeneously distributed. These assumptions allow us to develop a normalization strategy in order to reduce signal variability across the entire tissue. A consistent detected abundance would be optimal. The use of sodium cationization along with normalization to an evenly incorporated internal standard was able to reduce the voxel %RSD for the entire tissue from 56 % for [FTC+H+]+ (Figure 4-6B) to 32 % for [FTC+Na+]+ / [3TC+Na+]+ (Figure 4-6C). Although RAL and PRED are not structurally related, normalization of RAL to PRED produces an image with reduced variability per voxel, indicating similar extraction and ionization efficiencies. The images generated for TFV and RAL using the absolute abundance and with the normalization to their internal standard applied are shown in Figure 4-7.

Figure 4-7. The ion maps of the two ARV drugs in tissue but not quantified, tenofovir and raltegravir. Application of normalization to their normalization compound (Figure 4-1) is shown to reduce variability per voxel. Normalization of RAL to PRED is unable to account for analyte redistribution associated with tissue rewetting during application of the calibration solutions on the tissue.

4.3.3 Accounting for Ionization Efficiency per Voxel

The normalization method presented in Figure 4-6 shows a significant improvement in the observed analyte heat map and a corresponding reduction in voxel %RSD. The 30 μM NaCl electrospray solvent produces both [M+H+]+ and [M+Na+]+ ions (Figure 4-4) through competing ionization mechanisms. Since we know the amount of material ablated at each voxel is consistent, we are able to compare the competing protonation and sodiation processes on a per spectrum basis in a single experiment. Figure 4-8A, B shows the absolute ion abundances associated with the sodiated and protonated adducts, respectively. The ion maps show localized areas of higher ion abundance commonly referred to as hot spots in the MSI community. An interesting result occurs when the analyte ion abundance is normalized to the competing cationization of the normalization compound (Figure 4-8C, D). The images are visibly more heterogeneous, a statement that is supported by the increased corresponding voxel %RSD

values for the normalized heat maps. The resulting differences in the ion maps are the direct result of the competing ionization efficiencies of protonation and sodiation. If the analyte is normalized to the normalization compound having the same adduct, the voxel-to-voxel %RSD values are significantly improved. This result is independent of which ionization mechanism, protonation or sodiation, is applied to the analyte and internal standard. This is evidence demonstrating that the internal standard is truly accounting for ionization efficiency per spectrum (voxel), which is crucial to reducing variability that prevents per voxel quantification in MSI.

Figure 4-8. Ion maps comparing the ionization efficiencies of protonation and sodiation adducts of FTC. The ion map and %RSD for the entire tissue are shown for (A) [FTC+Na+]+ and (B) [FTC+H+]+ representing the detected ion abundance of FTC. The normalization of FTC to the competing ionization mechanism of internal standard for (C) [FTC+Na+]+ / [3TC+H+]+ and (D) [FTC+H+]+ / [3TC+Na+]+. The %RSD per pixel increases in both cross normalizations and the ion maps are visually more heterogeneous. Normalization of the analyte and internal standard with the same adduct produces the lowest pixel %RSD and improved visual quality for both the (E) sodiated and (F) protonated adducts.

4.3.4 Quantification of Emtricitabine using MALDESI QMSI

The calibration curves for the MSI experiment were generated using the SIL analog of the analyte. This method allows for correction of tissue-specific response by placing the calibration curve directly on the analyzed tissue. The use of SIL calibration spots also allows a direct comparison of analyte to the calibration curve, avoiding the use of a standard addition calibration curve. The accuracy and sensitivity of a standard addition curve may be adversely affected by the high voxel-to-voxel variability in MSI. Calibration spots were placed on top of the tissue using the SIL analog of FTC. To account for variations in spot size, the concentration of the standard per area was calculated for each calibration level separately. This was seen to effectively reduce variability due to the area of the calibration spots. The concentration was then plotted against the average ratio of SIL-FTC to 3TC to generate the calibration curves for the five QMSI replicates. Construction of interday calibration curves leads to differences in sensitivity as represented by variability in the slope of the calibration curve. This may be due to a variety of interday differences including the amount and composition of ambient ions, electrospray stability, or long-term stability of analyte and normalization compounds. Nonetheless, the average ratio of incubated FTC to internal standard 3TC of the entire tissue can be taken, related back to the SIL-FTC calibration curves generated to give a tissue average concentration in picograms per square millimeter. The total area of the tissue is known from the MSI analysis in MSiReader, which in conjunction with the calculated concentration can give the amount of FTC in the tissue slice. A representative calibration curve is shown in Figure 4-9.

Figure 4-9. Summary of FTC quantification of tissue section 4. (A) Ion map of [FTC+Na+]+ / [3TC+Na+]+ representing abundance of incubated FTC in the tissue section. The average 13 15 + + + + ratio was 0.728. (B) Ion map of [ C N2-FTC+Na ] / [3TC+Na ] representing the calibration curve 13 15 at 0, 0.25, 0.5, 1, 2, 4, 6, and 8 μg/mL solution. (C) Resulting calibration curve generated from C N2- FTC showing good linearity with R2 = 0.9973. The calculated tissue concentration was near the center of the calibration range. (D) Summary of values used to generate the total amount of drug present in tissue section. Using the average ratio and the equation of the calibration curve returns a value of the FTC concentration in tissue in picograms per square millimeter. Using the area of the tissue, the total amount of FTC in the section was determined to be 24.0 ng The average tissue concentration calculated from the calibration curves is given per 1 mm2, which allows for the direct comparison of the five samples. The average FTC tissue concentration of 172.0 ± 17.4 pg/mm2 (±95 % CI, n = 5) calculated from the IR-MALDESI QMSI analysis indicates good agreement between QMSI replicates. The total amount of drug in the tissue slice was calculated using the total area and its corresponding concentration. To allow comparison of QMSI to LC-MS/MS data for all replicates, the concentration of FTC was given as micrograms per gram tissue. The average FTC tissue concentration was 17.2 ± 1.8

μg/gtissue for IR-MALDESI QMSI data. Relative to LC-MS/MS, IR-MALDESI detection of FTC on a per voxel basis requires significantly higher sensitivity since the sample volume is greatly reduced. This spatial information provides important insight into the distribution of the compound but challenges instrument detection limits. The QMSI data presented represents the detection of the compound at approximately 7 fmol/voxel.

4.3.5 Comparison to LC-MS/MS

Ultimately, the IR-MALDESI QMSI experiment was cross-validated using a SRM LC- MS/MS assay on adjacent slices of the incubated tissue. The overall size of the tissue changed by approximately 20 % over the course of sectioning the irregularly shaped tissue. This change in tissue size can be correlated to a change in absolute amount of tissue analyzed and must be compensated for in the comparison of intra-LC-MS/MS analyses and the IR-MALDESI QMSI analyses. The average tissue FTC amount was 28.4 ± 2.8 μg/gtissue based on LC-MS/MS data. Summary of the QMSI and LC-MS/MS results are presented in Table 4-3. FTC concentrations obtained from IR-MALDESI QMSI and LC-MS/MS analyses are in good agreement with each other; however, they are not within the error of either measurement. As previously discussed, IR-MALDESI completely ablates the entire volume of tissue present at a given ablation spot, such that the difference is not likely to be solely attributable to a surface vs. volume concentration difference as often seen in MALDI MSI. Further work will be conducted to understand the source of any systematic error between these two methods.

Table 4-3. Quantification summary of the five tissue sections and the respective adjacent section analyzed by LC-MS/MS. Replicate 1 2 3 4 5 Average Tissue area ± 95% CI 110.85 122.15 128.33 134.22 134.83 (mm2)

2 R 0.995 0.9978 0.9973 0.9973 0.9994 17.2 ± MALDESI pg/mm2 171.4 148.0 179.4 177.8 184.2 1.8 µg/g μg/g tissue 17.1 14.8 17.9 17.8 18.4 tissue FTC 28.4 ± LC- 28.7 37.0 36.5 41.6 35.8 (ng/tissue) 2.8 µg/g MS/MS µg/g tissue 25.9 30.3 28.4 31.0 26.6 tissue 4.4 Conclusion

A new approach for IR-MALDESI QMSI was demonstrated for the HIV drug FTC present in incubated cervical tissue. Sodium cationization was used to increase ion abundance and remove isobaric interference. A normalization compound was selected based on molecular structure and implemented during QMSI. Normalization led to reduced signal variability providing higher quality images and demonstrated improvements toward a per voxel quantification strategy. The total amount of FTC in tissue was 17.2 ± 1.8 μg/gtissue for IR-

MALDESI QMSI and 28.4 ± 2.8 μg/gtissue based on LC-MS/MS data. The IR-MALDESI QMSI analyses represent the detection of FTC at approximately 7 fmol/voxel.

4.5 Acknowledgments

The authors gratefully acknowledge the financial support from the National Institutes of Health (R01GM087964), the W.M. Keck Foundation, and North Carolina State University.

4.6 References

1. Caprioli, R.M., Farmer, T.B., Gile, J.: Molecular Imaging of Biological Samples: Localization of Peptides and Proteins Using MALDI-TOF MS. Analytical Chemistry. 69, 4751-4760 (1997)

2. Prideaux, B., Staab, D., Stoeckli, M.: Applications of MALDI-MSI to Pharmaceutical Research. In: Rubakhin SS, Sweedler JV (eds.). Humana Press, Totowa, NJ, (2010)

3. Prideaux, B., Stoeckli, M.: Mass spectrometry imaging for drug distribution studies. Journal of Proteomics. 75, 4999-5013 (2012)

4. Solon, E.G., Schweitzer, A., Stoeckli, M., Prideaux, B.: Autoradiography, MALDI- MS, and SIMS-MS Imaging in Pharmaceutical Discovery and Development. The AAPS Journal. 12, 11-26 (2010)

5. Cohen, M.S., Gay, C., Kashuba, A.M., Blower, S., Paxton, L.: Narrative review: Antiretroviral therapy to prevent the sexual transmission of hiv-1. Annals of Internal Medicine. 146, 591-601 (2007)

6. Xu, R.N., Fan, L., Rieser, M.J., El-Shourbagy, T.A.: Recent advances in high- throughput quantitative bioanalysis by LC–MS/MS. Journal of Pharmaceutical and Biomedical Analysis. 44, 342-355 (2007)

7. Drexler, D.M., Tannehill-Gregg, S.H., Wang, L., Brock, B.J.: Utility of quantitative whole-body autoradiography (QWBA) and imaging mass spectrometry (IMS) by matrix-assisted laser desorption/ionization (MALDI) in the assessment of ocular distribution of drugs. Journal of Pharmacological and Toxicological Methods. 63, 205- 208 (2011)

8. Rudin, M., Weissleder, R.: Molecular imaging in drug discovery and development. Nat Rev Drug Discov. 2, 123-131 (2003)

9. Barry, J.A., Groseclose, M.R., Robichaud, G., Castellino, S., Muddiman, D.C.: Assessing drug and metabolite detection in liver tissue by UV-MALDI and IR- MALDESI mass spectrometry imaging coupled to FT-ICR MS. International journal of mass spectrometry. 377, 448-455 (2015)

10. Bunch, J., Clench, M.R., Richards, D.S.: Determination of pharmaceutical compounds in skin by imaging matrix-assisted laser desorption/ionisation mass spectrometry. Rapid Communications in Mass Spectrometry. 18, 3051-3060 (2004)

11. Casadonte, R., Caprioli, R.M.: Proteomic analysis of formalin-fixed paraffin- embedded tissue by MALDI imaging mass spectrometry. Nat. Protocols. 6, 1695-1709 (2011)

12. Djidja, M.-C., Chang, J., Hadjiprocopis, A., Schmich, F., Sinclair, J., Mršnik, M., Schoof, E.M., Barker, H.E., Linding, R., Jørgensen, C.: Identification of hypoxia- regulated proteins using MALDI-mass spectrometry imaging combined with quantitative proteomics. Journal of proteome research. 13, 2297-2313 (2014)

13. 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.: Utility of imaging mass spectrometry (IMS) by matrix-assisted laser desorption ionization (MALDI) on an ion trap mass spectrometer in the analysis of drugs and metabolites in biological tissues. Journal of Pharmacological and Toxicological Methods. 55, 279-288 (2007)

14. Fehniger, T.E., Végvári, Á., Rezeli, M., Prikk, K., Ross, P., Dahlbäck, M., Edula, G., Sepper, R., Marko-Varga, G.: Direct Demonstration of Tissue Uptake of an Inhaled Drug: Proof-of-Principle Study Using Matrix-Assisted Laser Desorption Ionization Mass Spectrometry Imaging. Analytical Chemistry. 83, 8329-8336 (2011)

15. Goodwin, R.J.A., Mackay, C.L., Nilsson, A., Harrison, D.J., Farde, L., Andren, P.E., Iverson, S.L.: Qualitative and Quantitative MALDI Imaging of the Positron Emission Tomography Ligands Raclopride (a D2 Dopamine Antagonist) and SCH 23390 (a D1 Dopamine Antagonist) in Rat Brain Tissue Sections Using a Solvent-Free Dry Matrix Application Method. Analytical Chemistry. 83, 9694-9701 (2011)

16. Goto, T., Terada, N., Inoue, T., Nakayama, K., Okada, Y., Yoshikawa, T., Miyazaki, Y., Uegaki, M., Sumiyoshi, S., Kobayashi, T., Kamba, T., Yoshimura, K., Ogawa, O.: The Expression Profile of Phosphatidylinositol in High Spatial Resolution Imaging Mass Spectrometry as a Potential Biomarker for Prostate Cancer. PLOS ONE. 9, e90242 (2014)

17. Groseclose, M.R., Castellino, S.: A Mimetic Tissue Model for the Quantification of Drug Distributions by MALDI Imaging Mass Spectrometry. Analytical Chemistry. 85, 10099-10106 (2013)

18. Hamm, G., Bonnel, D., Legouffe, R., Pamelard, F., Delbos, J.-M., Bouzom, F., Stauber, J.: Quantitative mass spectrometry imaging of propranolol and olanzapine using tissue extinction calculation as normalization factor. Journal of proteomics. 75, 4952-4961 (2012)

19. Handberg, E., Chingin, K., Wang, N., Dai, X., Chen, H.: Mass spectrometry imaging for visualizing organic analytes in food. Mass spectrometry reviews. 34, 641-658 (2015)

20. Hsieh, Y., Casale, R., Fukuda, E., Chen, J., Knemeyer, I., Wingate, J., Morrison, R., Korfmacher, W.: Matrix-assisted laser desorption/ionization imaging mass spectrometry for direct measurement of clozapine in rat brain tissue. Rapid Communications in Mass Spectrometry. 20, 965-972 (2006)

21. Hsieh, Y., Chen, J., Korfmacher, W.A.: Mapping pharmaceuticals in tissues using MALDI imaging mass spectrometry. Journal of Pharmacological and Toxicological Methods. 55, 193-200 (2007)

22. 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.: A quantitation method for mass spectrometry imaging. Rapid Communications in Mass Spectrometry. 25, 503-510 (2011)

23. Kubo, A., Kajimura, M., Suematsu, M.: Matrix-assisted laser desorption/ionization (MALDI) imaging mass spectrometry (IMS): a challenge for reliable quantitative analyses. Mass Spectrometry. 1, A0004-A0004 (2012)

24. Lietz, C.B., Gemperline, E., Li, L.: Qualitative and quantitative mass spectrometry imaging of drugs and metabolites. Advanced Drug Delivery Reviews. 65, 1074-1085 (2013)

25. Nilsson, A., Fehniger, T.E., Gustavsson, L., Andersson, M., Kenne, K., Marko-Varga, G., Andrén, P.E.: Fine Mapping the Spatial Distribution and Concentration of Unlabeled Drugs within Tissue Micro-Compartments Using Imaging Mass Spectrometry. PLOS ONE. 5, e11411 (2010)

26. Pirman, D.A., Kiss, A., Heeren, R.M.A., Yost, R.A.: Identifying Tissue-Specific Signal Variation in MALDI Mass Spectrometric Imaging by Use of an Internal Standard. Analytical Chemistry. 85, 1090-1096 (2013)

27. Pirman, D.A., Reich, R.F., Kiss, A., Heeren, R.M.A., Yost, R.A.: Quantitative MALDI Tandem Mass Spectrometric Imaging of Cocaine from Brain Tissue with a Deuterated Internal Standard. Analytical Chemistry. 85, 1081-1089 (2013)

28. Pirman, D.A., Yost, R.A.: Quantitative Tandem Mass Spectrometric Imaging of Endogenous Acetyl-l-carnitine from Piglet Brain Tissue Using an Internal Standard. Analytical Chemistry. 83, 8575-8581 (2011)

29. Prideaux, B., Dartois, V., Staab, D., Weiner, D.M., Goh, A., Via, L.E., Barry Iii, C.E., Stoeckli, M.: High-Sensitivity MALDI-MRM-MS Imaging of Moxifloxacin Distribution in Tuberculosis-Infected Rabbit Lungs and Granulomatous Lesions. Analytical Chemistry. 83, 2112-2118 (2011)

30. Shin, Y.G., Dong, T., Chou, B., Menghrajani, K.: Determination of loperamide in Mdr1a/1b knock-out mouse brain tissue using matrix-assisted laser desorption/ionization mass spectrometry and comparison with quantitative electrospray-triple quadrupole mass spectrometry analysis. Archives of Pharmacal Research. 34, 1983-1988 (2011)

31. Signor, L., Varesio, E., Staack, R.F., Starke, V., Richter, W.F., Hopfgartner, G.: Analysis of erlotinib and its metabolites in rat tissue sections by MALDI quadrupole time-of-flight mass spectrometry. Journal of Mass Spectrometry. 42, 900-909 (2007)

32. Stoeckli, M., Staab, D., Schweitzer, A.: Compound and metabolite distribution measured by MALDI mass spectrometric imaging in whole-body tissue sections. International Journal of Mass Spectrometry. 260, 195-202 (2007)

33. Sugiura, Y., Konishi, Y., Zaima, N., Kajihara, S., Nakanishi, H., Taguchi, R., Setou, M.: Visualization of the cell-selective distribution of PUFA-containing phosphatidylcholines in mouse brain by imaging mass spectrometry. Journal of Lipid Research. 50, 1776-1788 (2009)

34. Sun, N., Walch, A.: Qualitative and quantitative mass spectrometry imaging of drugs and metabolites in tissue at therapeutic levels. Histochemistry and Cell Biology. 140, 93-104 (2013)

35. Takai, N., Tanaka, Y., Inazawa, K., Saji, H.: Quantitative analysis of pharmaceutical drug distribution in multiple organs by imaging mass spectrometry. Rapid Communications in Mass Spectrometry. 26, 1549-1556 (2012)

36. Trede, D., Schiffler, S., Becker, M., Wirtz, S., Steinhorst, K., Strehlow, J., Aichler, M., Kobarg, J.H., Oetjen, J., Dyatlov, A., Heldmann, S., Walch, A., Thiele, H., Maass, P., Alexandrov, T.: Exploring Three-Dimensional Matrix-Assisted Laser Desorption/Ionization Imaging Mass Spectrometry Data: Three-Dimensional Spatial Segmentation of Mouse Kidney. Analytical Chemistry. 84, 6079-6087 (2012)

37. Veličković, D., Ropartz, D., Guillon, F., Saulnier, L., Rogniaux, H.: New insights into the structural and spatial variability of cell-wall polysaccharides during wheat grain development, as revealed through MALDI mass spectrometry imaging. Journal of experimental botany. 65, 2079-2091 (2014)

38. Ellis, S.R., Bruinen, A.L., Heeren, R.M.A.: A critical evaluation of the current state- of-the-art in quantitative imaging mass spectrometry. Analytical and Bioanalytical Chemistry. 406, 1275-1289 (2014)

39. Nemes, P., Vertes, A.: Ambient mass spectrometry for in vivo local analysis and in situ molecular tissue imaging. TrAC Trends in Analytical Chemistry. 34, 22-34 (2012)

40. Cohen, L.H., Gusev, A.I.: Small molecule analysis by MALDI mass spectrometry. Analytical and Bioanalytical Chemistry. 373, 571-586 (2002)

41. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Generation and detection of multiply-charged peptides and proteins by matrix-assisted laser desorption electrospray ionization (MALDESI) fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry. 17, 1712-1716 (2006)

42. Nemes, P., Vertes, A.: Laser ablation electrospray ionization for atmospheric pressure, in vivo, and imaging mass spectrometry. Analytical chemistry. 79, 8098-8106 (2007)

43. Brady, J.J., Judge, E.J., Levis, R.J.: Mass spectrometry of intact neutral macromolecules using intense non-resonant femtosecond laser vaporization with electrospray post-ionization. Rapid Communications in Mass Spectrometry. 23, 3151- 3157 (2009)

44. Robichaud, G., Barry, J.A., Muddiman, D.C.: IR-MALDESI Mass Spectrometry Imaging of Biological Tissue Sections Using Ice as a Matrix. Journal of The American Society for Mass Spectrometry. 25, 319-328 (2014)

45. Flanigan, P.M., Perez, J.J., Karki, S., Levis, R.J.: Quantitative Measurements of Small Molecule Mixtures Using Laser Electrospray Mass Spectrometry. Analytical Chemistry. 85, 3629-3637 (2013)

46. Chun, T.-W., Nickle, D.C., Justement, J.S., Meyers, J.H., Roby, G., Hallahan, C.W., Kottilil, S., Moir, S., Mican, J.M., Mullins, J.I., Ward, D.J., Joseph A, K., Mannon, P.J., Fauci, A.S.: Persistence of HIV in Gut-Associated Lymphoid Tissue despite Long- Term Antiretroviral Therapy. The Journal of Infectious Diseases. 197, 714-720 (2008)

47. Launay, O., Tod, M., Tschope, I., Si-Mohamed, A., Bélarbi, L., Charpentier, C., Goujard, C., Taburet, A.-M., Lortholary, O., Leroy, V.: Residual HIV-1 RNA and HIV- 1 DNA production in the genital tract reservoir of women treated with HAART: the prospective ANRS EP24 GYNODYN study. Antiviral therapy. 16, 843 (2011)

48. Stanley, S.K.: Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents. Morbidity and Mortality Weekly Report: Recommendations and Reports. 39-82 (1998)

49. Barry, J.A., Robichaud, G., Bokhart, M.T., Thompson, C., Sykes, C., Kashuba, A.D.M., Muddiman, D.C.: Mapping Antiretroviral Drugs in Tissue by IR-MALDESI MSI Coupled to the Q Exactive and Comparison with LC-MS/MS SRM Assay. Journal of The American Society for Mass Spectrometry. 25, 2038-2047 (2014)

50. Barry, J.A., Muddiman, D.C.: Global optimization of the infrared matrix-assisted laser desorption electrospray ionization (IR MALDESI) source for mass spectrometry using statistical design of experiments. Rapid Communications in Mass Spectrometry. 25, 3527-3536 (2011)

51. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Development and Characterization of an Ionization Technique for Analysis of Biological Macromolecules: Liquid Matrix- Assisted Laser Desorption Electrospray Ionization. Analytical Chemistry. 80, 6773- 6778 (2008)

52. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-MALDESI) Imaging Source Coupled to a FT-ICR Mass Spectrometer. Journal of The American Society for Mass Spectrometry. 24, 92-100 (2013)

53. Jurchen, J.C., Rubakhin, S.S., Sweedler, J.V.: MALDI-MS imaging of features smaller than the size of the laser beam. Journal of the American Society for Mass Spectrometry. 16, 1654-1659 (2005)

54. Olsen, J.V., de Godoy, L.M.F., Li, G., Macek, B., Mortensen, P., Pesch, R., Makarov, A., Lange, O., Horning, S., Mann, M.: Parts per Million Mass Accuracy on an Orbitrap Mass Spectrometer via Lock Mass Injection into a C-trap. Molecular & Cellular Proteomics. 4, 2010-2021 (2005)

55. Kessner, D., Chambers, M., Burke, R., Agus, D., Mallick, P.: ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics. 24, 2534- 2536 (2008)

56. Schramm, T., Hester, A., Klinkert, I., Both, J.-P., Heeren, R.M.A., Brunelle, A., Laprévote, O., Desbenoit, N., Robbe, M.-F., Stoeckli, M., Spengler, B., Römpp, A.: imzML — A common data format for the flexible exchange and processing of mass spectrometry imaging data. Journal of Proteomics. 75, 5106-5110 (2012)

57. Robichaud, G., Garrard, K.P., Barry, J.A., Muddiman, D.C.: MSiReader: an open- source interface to view and analyze high resolving power MS imaging files on Matlab platform. Journal of the American Society for Mass Spectrometry. 24, 718-721 (2013)

58. Wang, Z., Tang, J., Salomon, C.E., Dreis, C.D., Vince, R.: Pharmacophore and structure–activity relationships of integrase inhibition within a dual inhibitor scaffold of HIV reverse transcriptase and integrase. Bioorganic & Medicinal Chemistry. 18, 4202-4211 (2010)

100

5 Mass Spectrometry Imaging Reveals Heterogeneous Efavirenz Distribution within Putative HIV Reservoirs.

The following work was reprinted with permission from: Thompson, C.G., Bokhart, M.T., Sykes, C., Adamson, L., Fedoriw, Y., Luciw, P.A., Muddiman, D.C., Kashuba, A.D.M., Rosen, E.P.: Mass Spectrometry Imaging Reveals Heterogeneous Efavirenz Distribution within Putative HIV Reservoirs. Antimicrobial Agents and Chemotherapy. 59, 2944-2948 (2015). DOI: 10.1128/AAC.04952-14. Copyright © 2015 American Society for Microbiology. The original publication may be accessed via the Internet at http://aac.asm.org/content/59/5/2944.full

5.1 Introduction

Human immunodeficiency virus (HIV) replication has been shown to persist in certain anatomic sites, known as active viral reservoirs, despite treatment with highly active antiretroviral (ARV) therapy (HAART) [1, 2]. Understanding the factors that contribute to the formation and propagation of these active viral reservoirs is essential to the design of targeted therapies for HIV eradication. It has been suggested that subtherapeutic drug concentrations in certain tissues resulting from poor drug penetration may provide a favorable environment for reservoir formation and drug-resistant viral variants [3]. Several groups, including our own, have assessed ARV penetration of tissues by directly measuring drug concentrations by liquid chromatography-mass spectrometry (LC-MS) of homogenized whole tissue [4] or isolated mononuclear cells [3, 5]. Though these methods can provide useful quantitative data, they do not have the ability to spatially define the distribution of the drug within the tissue, as either the entire sample is consumed in the homogenization process or spatial information is lost during cellular isolation. This is a critical limitation of these methodologies, as our preliminary data have shown that ARV distribution across tissue is not uniform [6]. MS imaging (MSI) offers an alternative strategy for quantifying ARV distribution in tissues and cells that maintains the sensitivity and specificity of LC-MS while preserving the spatial distribution of analytes within tissue. Through stepwise interrogation of discrete sample locations, MSI simultaneously collects information that can be concatenated into images of

101

multiple molecules and their respective metabolites. This attribute is an important advantage for the combinatorial nature of HAART and has already led to the implementation of MSI in the drug development process [7]. One approach to MSI that is particularly well suited to the analysis of small molecules is infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) [8], which allows the detection of ARVs in human tissue, as we have previously demonstrated [9, 10]. Here, we used IR-MALDESI to characterize the ARV distribution in 11 non-human primate tissues implicated as viral reservoirs [11–14]. Further, we quantified the variability in ARV exposure between tissues and compared this to LC-MS and immunohistochemistry (IHC) data, allowing for absolute quantification of observed ARV signal abundance and identification of the tissue compartments or cellular populations where a drug may be concentrating. These data are the first quantitative images of the ARV distribution in a macaque, an important species for studies of HIV/simian immunodeficiency virus (SIV) therapy, and show that MSI is a promising approach for evaluating ARV disposition in HIV reservoirs [15].

5.2 Experimental

One healthy male rhesus macaque (Macaca mulatta) was given 7 daily oral doses of 200 mg of efavirenz (EFV). This dose of EFV equates to roughly 60 mg/kg and is consistent with standard treatment doses for SIV [16, 17]. Prior to necropsy, blood plasma and cerebrospinal fluid were collected. The animal was euthanized by pentobarbital overdose 24 h after the final dose of EFV, and necropsy was performed by the pathology staff at the California National Primate Research Center. Tissue samples from the gastrointestinal (GI) tract (ileum, colon, rectum), central nervous system (CNS; cerebellum, basal ganglia), lymph nodes (axillary, iliac, mesenteric, inguinal), and spleen were snap-frozen on dry ice and stored at −80°C until analysis [18]. Calibration of the IR-MALDESI response to EFV from the dosed tissue was

102

conducted by MSI of tissues from non-dosed (“blank”) macaques (Bioreclamation IVT, Baltimore, MD), matching dosed tissue samples where possible, upon which a set of EFV standards were pipetted. Prior to imaging, 10 μm sections of each tissue (dosed and non-dosed) were sliced and thaw mounted on a single glass microscope slide uniformly coated with internal standards and the tissue sections were spotted with 100 nl containing 0 to 5,000 pg of EFV before the sample slide was placed in the IR-MALDESI imaging source. Serial 10 μm sections were set aside for LC-MS/MS and IHC analyses. The IR-MALDESI MSI approach for analysis of tissue samples has been described previously [8, 9]. Briefly, tissue samples maintained at −10°C in the source chamber were ablated at a spot-to-spot distance of 100 μm by two pulses of an IR laser (IR-Opolette 2371; Opotek, Carlsbad, CA, USA) that resulted in the complete desorption of neutral molecules for a given volume element or voxel. The desorbed neutral molecules were then ionized by an orthogonal electrospray plume and sampled into a high-resolving-power Thermo Fisher Scientific Q Exactive (Bremen, Germany) mass spectrometer for synchronized analysis [9]. To generate images from mass spectrometry data, raw data from each voxel were converted to the mzXML format with MSConvert software [19]. These mzXML files were interrogated with MSiReader, a free software developed for processing of MSI data, from which measurements such as tissue surface area can be made and images of analyte distribution can be generated [20]. For LC-MS/MS analysis of EFV concentrations, serial 10 μm tissue sections were homogenized in 1 ml of 70:30 acetonitrile 1 mM ammonium phosphate (pH 7.4) with a Precellys 24 tissue homogenizer (Bertin Technologies, Montigny-le-Bretonneux, France). A Shimadzu high-performance liquid chromatography system was used for separation, and an AB SCIEX API 5000 mass spectrometer (AB SCIEX, Foster City, CA, USA) equipped with a turbo spray interface was used as the detector. The samples were analyzed with a set of calibration standards (0.02 to 20 ng) and quality control (QC) samples. The precision and accuracy of the calibration standards and QC samples were within the acceptable range of 15%.

103

LC-MS/MS quantification of EFV in the homogenate of each tissue section was compared to the summed MSI response on a per-mass-of-tissue basis by using the MSI-derived tissue surface area, the known section thickness, and an assumed tissue density of 1.06 g/cm3. MSI quantitation and LC-MS/MS analysis were performed by different individuals at separate institutions, and no data were shared before analyses were completed. The LC-MS/MS data underwent QC by a designated individual not directly involved in this study to ensure accuracy. To verify tissue quality and assess architecture for comparison with EFV distribution by MSI, serial sections of frozen tissue were sliced at a 10 μm thickness, thaw mounted on glass slides, and fixed in 100% ethanol for 10 min. After fixation, the tissues were stained with hematoxylin and eosin by standard histological techniques. IHC analysis of similarly prepared frozen tissue slices was performed with human primary antibodies for CD3 (clone LN10; Leica Biosystems, Buffalo Grove, IL), followed by staining with secondary antibodies. All staining was performed with the Leica Bond automated tissue stainer (Leica Biosystems).

5.3 Results and Discussion

MSI revealed heterogeneous intratissue EFV distribution into several anatomic sites. Figure 5-1 showcases these findings for representative tissues. When MSI images were compared with IHC staining, interesting spatial distributions were noted. For example, EFV was concentrated in the mucosa and lamina propria of the colon (Figure 5-1A), which corresponds to a high CD3+ cell density on IHC analysis. However, this distribution was not observed in the ileum (Figure 5-1B). The inguinal lymph node showed EFV in some, but not all, primary follicles (Figure 5-1C). EFV concentrated in the gray matter of the cerebellum (Figure 5-1D) and showed a homogeneous distribution in the spleen, testes, and axillary lymph nodes (Figure 5-1E). The heterogeneity of EFV distribution is quantified in Table 5-2 by the dynamic range of the MSI response (expressed in the base 10 logarithmic units decibels [dB]) in each tissue type that can be observed in the images in Table 5-1. The dynamic range of the

104

EFV response was lower in tissues such as the basal ganglia and lymph nodes, reflecting a more homogeneous EFV distribution, whereas tissues such as the colon (37.6 dB) and rectum (26.8 dB) had much larger differences between minimum and maximum concentrations that suggest greater biological differences in drug uptake.

Figure 5-1. EFV distribution in macaque reservoir sites. Representative MSI images are shown on the left, with adjacent CD3+ cell staining of serial colon (A), ileum (B), inguinal lymph node (C), cerebellum (D), and spleen (E) tissue slices. MSI signal intensity is shown next to each image on a concentration-dependent scale. The bottom of the scale (0) represents the presence of no EFV, while the top of the scale reflects the highest per-voxel EFV signal observed within each slice. Brighter colors represent higher EFV concentrations.

105

Table 5-1. Variability of EFV MSI responses within dosed macaque tissues.

No. of ng/voxela Tissue type Maximum Median Minimum DR (dB)b CNS Cerebellum 1.8E+04 5.0E+03 6.5E+02 14.5 Basal ganglion 1.8E+03 9.2E+02 4.8E+02 5.8 Lymph node Axillary 3.0E+04 4.2E+03 2.0E+03 11.8 Mesenteric 9.8E+03 2.6E+03 1.1E+03 9.5 Inguinal 2.7E+04 1.6E+03 8.1E+02 15.2 Iliac 4.0E+03 9.3E+02 3.4E+02 10.7 Spleen 4.2E+04 5.1E+03 1.4E+03 14.6 GI tract Ileum 1.4E+04 3.7E+03 2.5E+03 7.5 Colon 8.7E+06 1.4E+04 1.5E+03 37.6 Rectum 1.6E+06 1.2E+04 3.4E+03 26.8 Testis 2.7E+03 5.9E+02 3.7E+02 8.6 a The EFV concentration within each voxel across the entire tissue slice was quantified by using calibration standards. b Dynamic range (DR), expressed in the logarithmic units decibels, was calculated as DR =

10log10(maximum/minimum).

Intertissue EFV quantitation is summarized in Table 5-2. LC-MS/MS analysis demonstrated a 20-fold variability in total tissue EFV exposure, with concentrations ranging from 1.2 μg/g in the testes to 20.8 μg/g in the colon. A similar trend was observed in the MSI quantification, though agreement varied between tissue types. EFV concentrations determined by MSI and LC-MS/MS were found to be in agreement (<30% difference) for half of the tissues after correction for tissue size. In tissues such as the lymph nodes, concentrations varied by as little as 8%. Tissues of the GI tract demonstrated less agreement between techniques, with variations of up to −70%. Table 5-2 also compares EFV exposures in tissue and plasma. EFV achieved high exposure in the CNS, where tissue drug concentrations were 6.8 to 7.6 log units higher than in the CSF. EFV exposure was consistent among the lymph nodes, with 1.7- to 2.2-log increases over plasma observed. In the GI tract, EFV exposure was 3.6 log units higher than in plasma in the colon and rectum and 2.7 log units higher in the ileum.

106

Table 5-2. Comparison of EFV quantifications in macaque tissues by MSI and LC-MS/MS.

LC-MS/MS MSI Concn (μg/g Log increase over plasma Concn (μg/g Log increase over plasma Tissue type tissue)a or CSFb tissue) or CSFb Differencec(%) CNS Cerebellum 6.86 7.6 3.09 6.8 −54.89 Basal ganglion 2.01 6.4 1.67 6.2 −16.80 Lymph node Axillary 3.91 2.0 3.33 1.8 −14.91 Mesenteric 3.82 2.0 3.12 1.8 −18.48 Inguinal 4.80 2.2 2.86 1.7 −40.38 Iliac 2.82 1.7 3.06 1.7 8.40 Spleen 5.01 2.2 3.61 1.9 −27.83 GI tract Ileum 8.41 2.7 3.20 1.8 −61.94 Colon 20.77 3.6 6.12 2.4 −70.54 Rectum 20.69 3.6 8.22 2.7 −60.26 Testis 1.22 0.8 2.91 1.7 138.94 a On day 8, the concentrations in plasma and CSF were 541 and 3.30 ng/ml, respectively, as measured by LC-MS/MS. b To compare tissue drug concentrations to plasma or CSF drug concentrations, tissue drug concentrations in μg/g were converted to ng/ml, assuming a tissue density of approximately 1 g/ml, and then divided by the plasma or CSF drug concentration and converted to log units. c The percent difference between methods was calculated by subtracting LC-MS/MS concentrations from MSI concentrations, dividing by the LC-MS/MS concentration, and multiplying by 100.

The persistence of HIV replication within anatomic reservoirs necessitates the use of tissue pharmacology to inform the design of effective treatment strategies. This requires knowledge of tissue penetration to sites of action, as underscored by recent findings that the 50 to 90% reduction of the EFV concentration in mononuclear cells isolated from reservoir tissues relative to that in peripheral blood mononuclear cells was associated with persistent viral replication in these tissues [3]. This finding, in combination with the fact that EFV receives widespread clinical use as a component of Atripla (a fixed-dose combination of tenofovir, emtricitabine, and EFV dosed once daily) and is frequently included in HIV treatment and cure research regimens for macaques, led us to choose EFV for our evaluations.

107

The observed ARV drug distribution within these putative viral reservoirs reveals important information regarding tissue pharmacology that can inform treatment strategy. The heterogeneous penetration of the lymphoid follicles by EFV suggests that further quantification of effective drug exposure in these tissues is required. Conversely, the abundance of EFV signal in the CD3+ cell populations of the gut is evidence that adequate EFV concentrations are likely reached in this compartment. Both of these findings are consistent with previous studies that have examined tissue EFV concentrations by LC-MS [5]. The EFV distributions observed here would not have been possible with traditional LC-MS of tissue homogenates or isolated mononuclear cells; the heterogeneity of EFV distribution within tissue slices as measured by the dynamic range of response (Table 5-1) is only measurable by MSI. Moreover, our MSI analysis provides evidence that the use of plasma or CSF as a surrogate for tissue drug concentrations may be inappropriate without detailed quantification of these relationships. The higher CNS tissue EFV concentrations than CSF EFV concentrations (Table 5-2) and the concentration of EFV within the gray matter of the cerebellum (Figure 5-1) agree with brain microdialysis data showing that CNS drug concentrations are higher than CSF drug concentrations (21, 22). The variability in the extent of EFV distribution between tissue types suggests that biological processes, more than the cellular populations present, drive the movement of EFV into tissues. The nonhomogeneous distribution of EFV in tissues such as the colon may be attributable to the physicochemical properties of EFV or to active transport mechanisms. Our previous work identifying variables affecting ARV exposure in the female genital tract (another putative viral reservoir) found that the efflux transporters MRP1 and MRP4 were associated with ARV penetration of this compartment [23]. While EFV is not a known substrate of these transporters, other drug transporters such as MDR1 or BCRP may affect its disposition and explain the areas of EFV concentration seen here [24, 25]. There are several limitations of this analysis that should be addressed, the most important of which is our limited sample size. As this study was conducted with a single

108

animal, the variability in tissue drug distribution between animals remains unknown and remains to be evaluated. Further, the assessment of EFV distribution shown in Figure 5-1 is based on individual slices of tissue under steady-state conditions. Repeated sectioning may reveal additional biological variability. Although EFV has a long plasma half-life and a relatively flat blood plasma concentration-versus-time curve, EFV exposure over the dosing interval could not be determined because sampling was performed only at the end of the dosing interval. Additionally, we were unable to determine the relationship between drug and viral dynamics in this uninfected animal, though we selected tissues with previous evidence supporting persistent HIV infection [11–14]. Finally, only CD3 was used to correlate IHC analysis with drug distribution. Though visualization of the overall T cell compartment is informative, future work will relate ARV localization to CD4+ T cell distribution, as these cells are the most relevant for HIV infection.

5.4 Conclusion

This is the first study to apply MSI to ARV distribution in potential tissue reservoirs of HIV infection. Using IR-MALDESI, we have confirmed that ARV tissue distribution is heterogeneous and that the distribution of a single ARV can vary greatly between tissues within an individual. By comparison to the gold standard of tissue quantification, LC-MS/MS, our analysis confirms the importance of MSI for drug quantification. Future work will address the existing limitations of our approach. For MSI, this will entail a systematic exploration of factors, such as matrix effects or electrospray ionization capacity, which may influence the quantitative agreement with LC-MS for different tissue types and drug exposures. IR- MALDESI is sensitive to a wide variety of endogenous lipids (the profiles of which vary between tissue types) that are ablated and analyzed simultaneously with EFV. Any suppression of the EFV response as a result of tissue-specific ablation and ionization conditions is intended to be taken into account by performing EFV calibrations with matching or closely related blank

109

tissue types and evaluating the IR-MALDESI response to an internal standard. However, a more thorough investigation of these effects must be undertaken to improve analytical agreement. Additionally, lower limits of detection of all ARVs and their active metabolites within a drug regimen must be attained in order to link tissue drug exposure and suppression of viral replication. We will also evaluate ARV distribution in SIV/HIV-infected samples to determine the effect of ARV disposition on viral expression. Despite these limitations, these data show that MSI is a critical tool for the disposition of ARVs within putative active HIV reservoirs, which is an important step toward understanding how to eradicate HIV infection.

5.5 Acknowledgements

This work was supported by NIH grants R01AI111891 (E.P.R., C.G.T., C.S., Y.F., P.A.L., D.C.M., and A.D.M.K.), R01GM087964 (E.P.R., M.T.B., and D.C.M.), and U01AI095031, UNC Center for AIDS Research grant P30AI050410 (C.G.T., C.S., Y.F., A.D.M.K., and E.P.R.), and Collaboratory of AIDS Researchers for Eradication (CARE) grant U19AI096113. We also acknowledge additional support from GlaxoSmithKline, the William R. Kenan, Jr. Fund for Engineering, Technology, and Science, and the W. M. Keck Foundation (E.P.R., M.T.B., and D.C.M.).

5.6 References

1. Palmer, S., Josefsson, L., Coffin, J.M.: HIV reservoirs and the possibility of a cure for HIV infection. Journal of Internal Medicine. 270, 550-560 (2011)

2. Chun, T.-W., Nickle, D.C., Justement, J.S., Meyers, J.H., Roby, G., Hallahan, C.W., Kottilil, S., Moir, S., Mican, J.M., Mullins, J.I., Ward, D.J., Joseph A, K., Mannon, P.J., Fauci, A.S.: Persistence of HIV in Gut-Associated Lymphoid Tissue despite Long- Term Antiretroviral Therapy. The Journal of Infectious Diseases. 197, 714-720 (2008)

3. Fletcher, C.V., Staskus, K., Wietgrefe, S.W., Rothenberger, M., Reilly, C., Chipman, J.G., Beilman, G.J., Khoruts, A., Thorkelson, A., Schmidt, T.E., Anderson, J., Perkey, K., Stevenson, M., Perelson, A.S., Douek, D.C., Haase, A.T., Schacker, T.W.:

110

Persistent HIV-1 replication is associated with lower antiretroviral drug concentrations in lymphatic tissues. Proceedings of the National Academy of Sciences. 111, 2307- 2312 (2014)

4. Dumond, J.B., Yeh, R.F., Patterson, K.B., Corbett, A.H., Jung, B.H., Rezk, N.L., Bridges, A.S., Stewart, P.W., Cohen, M.S., Kashuba, A.D.: Antiretroviral drug exposure in the female genital tract: implications for oral pre-and post-exposure prophylaxis. AIDS (London, England). 21, 1899 (2007)

5. Podany, A.T., Winchester, L.C., Robbins, B.L., Fletcher, C.V.: Quantification of Cell- Associated Atazanavir, Darunavir, Lopinavir, Ritonavir, and Efavirenz Concentrations in Human Mononuclear Cell Extracts. Antimicrobial Agents and Chemotherapy. 58, 2866-2870 (2014)

6. Thompson, C.G., Rosen, E.P., Sykes, C., Fedoriw, Y., Luciw, P.A., Muddiman, D.C., Kashuba, A.D.: Characterizing antiretroviral distribution within active ciral reserviors using mass spectrometry imaging, abstr p_09, p 36-37. Abstract Book 15th International Workshop on Clinical Pharmacology of HIV & Hepatitis Therapy, Washington DC. http://regist2.virology-education.com/abstractbook/2014_4.pdf

7. Prideaux, B., Stoeckli, M.: Mass spectrometry imaging for drug distribution studies. Journal of Proteomics. 75, 4999-5013 (2012)

8. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-MALDESI) Imaging Source Coupled to a FT-ICR Mass Spectrometer. Journal of The American Society for Mass Spectrometry. 24, 92-100 (2013)

9. Barry, J.A., Robichaud, G., Bokhart, M.T., Thompson, C., Sykes, C., Kashuba, A.D., Muddiman, D.C.: Mapping antiretroviral drugs in tissue by IR-MALDESI MSI coupled to the Q Exactive and comparison with LC-MS/MS SRM assay. Journal of The American Society for Mass Spectrometry. 25, 2038-2047 (2014)

10. Bokhart, M.T., Rosen, E., Thompson, C., Sykes, C., Kashuba, A.D.M., Muddiman, D.C.: Quantitative mass spectrometry imaging of emtricitabine in cervical tissue model using infrared matrix-assisted laser desorption electrospray ionization. Analytical and Bioanalytical Chemistry. 407, 2073-2084 (2015)

11. Gratton, S., Cheynier, R., Dumaurier, M.-J., Oksenhendler, E., Wain-Hobson, S.: Highly restricted spread of HIV-1 and multiply infected cells within splenic germinal centers. Proceedings of the National Academy of Sciences. 97, 14566-14571 (2000)

111

12. Burton, G.F., Keele, B.F., Estes, J.D., Thacker, T.C., Gartner, S.: Follicular dendritic cell contributions to HIV pathogenesis. Seminars in Immunology. 14, 275-284 (2002)

13. Avettand-Fenoel, V., Hocqueloux, L., Müller-Trutwin, M., Prazuck, T., Melard, A., Chaix, M.-L., Agoute, E., Michau, C., Rouzioux, C.: Greater diversity of HIV DNA variants in the rectum compared to variants in the blood in patients without HAART. Journal of Medical Virology. 83, 1499-1507 (2011)

14. Zink, M.C., Clements, J.E.: A novel simian immunodeficiency virus model that provides insight into mechanisms of human immunodeficiency virus central nervous system disease. Journal of neurovirology. 8, (2002)

15. Van Rompay, K.K.: The use of nonhuman primate models of HIV infection for the evaluation of antiviral strategies. AIDS research and human retroviruses. 28, 16-35 (2012)

16. North, T.W., Van Rompay, K.K.A., Higgins, J., Matthews, T.B., Wadford, D.A., Pedersen, N.C., Schinazi, R.F.: Suppression of Virus Load by Highly Active Antiretroviral Therapy in Rhesus Macaques Infected with a Recombinant Simian Immunodeficiency Virus Containing Reverse Transcriptase from Human Immunodeficiency Virus Type 1. Journal of Virology. 79, 7349-7354 (2005)

17. Hofman, M.J., Higgins, J., Matthews, T.B., Pedersen, N.C., Tan, C., Schinazi, R.F., North, T.W.: Efavirenz Therapy in Rhesus Macaques Infected with a Chimera of Simian Immunodeficiency Virus Containing Reverse Transcriptase from Human Immunodeficiency Virus Type 1. Antimicrobial Agents and Chemotherapy. 48, 3483- 3490 (2004)

18. North, T.W., Higgins, J., Deere, J.D., Hayes, T.L., Villalobos, A., Adamson, L., Shacklett, B.L., Schinazi, R.F., Luciw, P.A.: Viral Sanctuaries during Highly Active Antiretroviral Therapy in a Nonhuman Primate Model for AIDS. Journal of Virology. 84, 2913-2922 (2010)

19. Kessner, D., Chambers, M., Burke, R., Agus, D., Mallick, P.: ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics. 24, 2534- 2536 (2008)

20. Robichaud, G., Garrard, K.P., Barry, J.A., Muddiman, D.C.: MSiReader: an open- source interface to view and analyze high resolving power MS imaging files on Matlab platform. Journal of the American Society for Mass Spectrometry. 24, 718-721 (2013)

112

21. Liu, X., Van Natta, K., Yeo, H., Vilenski, O., Weller, P.E., Worboys, P.D., Monshouwer, M.: Unbound Drug Concentration in Brain Homogenate and Cerebral Spinal Fluid at Steady State as a Surrogate for Unbound Concentration in Brain Interstitial Fluid. Drug Metabolism and Disposition. 37, 787-793 (2009)

22. de Lange, E.C.M., Danhof, M.: Considerations in the Use of Cerebrospinal Fluid Pharmacokinetics to Predict Brain Target Concentrations in the Clinical Setting. Clinical Pharmacokinetics. 41, 691-703 (2002)

23. Thompson, C.G., Sedykh, A., Nicol, M.R., Muratov, E., Fourches, D., Tropsha, A., Kashuba, A.D.: Cheminformatics analysis to identify predictors of antiviral drug penetration into the female genital tract. AIDS research and human retroviruses. 30, 1058-1064 (2014)

24. Peroni, R.N., Di Gennaro, S.S., Hocht, C., Chiappetta, D.A., Rubio, M.C., Sosnik, A., Bramuglia, G.F.: Efavirenz is a substrate and in turn modulates the expression of the efflux transporter ABCG2/BCRP in the gastrointestinal tract of the rat. Biochemical Pharmacology. 82, 1227-1233 (2011)

25. Fellay, J., Marzolini, C., Meaden, E.R., Back, D.J., Buclin, T., Chave, J.-P., Decosterd, L.A., Furrer, H., Opravil, M., Pantaleo, G., Retelska, D., Ruiz, L., Schinkel, A.H., Vernazza, P., Eap, C.B., Telenti, A.: Response to antiretroviral treatment in HIV-1- infected individuals with allelic variants of the multidrug resistance transporter 1: a pharmacogenetics study. The Lancet. 359, 30-36 (2002)

113

6 Analysis of Antiretrovirals in Single Hair Strands for Evaluation of Drug Adherence with IR-MALDESI MSI

The following work was reprinted with permission from: Rosen, E.P., Thompson, C.G., Bokhart, M.T., Prince, H.M.A., Sykes, C., Muddiman, D.C., Kashuba, A.D.M.: Analysis of Antiretrovirals in Single Hair Strands for Evaluation of Drug Adherence with Infrared-Matrix- Assisted Laser Desorption Electrospray Ionization Mass Spectrometry Imaging. Analytical Chemistry. 88, 1336-1344 (2016). DOI: 10.1021/acs.analchem.5b03794. Copyright © 2015 American Chemical Society. The original publication may be accessed via the Internet at http://pubs.acs.org/doi/abs/10.1021/acs.analchem.5b03794

6.1 Introduction

Adherence to antiretroviral (ARV) therapy is critical for achieving HIV RNA suppression in HIV-infected patients [1-3] and for preventing HIV acquisition in uninfected individuals using pre-exposure prophylaxis (PrEP) [4]. Yet a high level of adherence is challenging for HIV-infected individuals on life-long ARVs and for HIV-negative individuals using daily PrEP who are not at daily risk for HIV acquisition. Poor adherence was primarily responsible for a lack of drug effectiveness in multiple recent double-blind, placebo-controlled PrEP studies [5-7]. These studies found that counting product returns and using patient self- report significantly overpredicted adherence as measured by ARV concentrations in blood plasma or cells. Since the consequences of poor or intermittent adherence are significant, valid measures of adherence are critical for optimizing the effectiveness of both HIV treatment and prevention, in both the clinic and research settings. Hair has long been a targeted analysis matrix for forensic monitoring of drug abuse [8, 9] because it’s temporal record (weeks– months) provides a longer term for retrospective analysis than other human matrixes and is being increasingly utilized for therapeutic drug monitoring [10, 11]. Currently, the disposition of ARVs in hair is measured by LC–MS/MS analysis [12, 13]. This approach represents the current “gold standard” and has been correlated with adherence in PrEP [14] and with health outcomes in treatment [15, 16]. While LC–MS/MS methods are sensitive and specific, these techniques require multiple (e.g., 25–100) strands of at least 1 cm

114

in length to be combined for a single measurement. This approach effectively averages drug response over a month’s time based on growth rates for hair in the active, anagen, phase of hair growth [9] and cannot differentiate strict adherence to therapy at regular dosing intervals from heterogeneous adherence. The admixture of various exposures within such a homogenate can lead to hybrid hair drug concentration which may over- or underestimate drug exposure at any one point in time. Given that periodic self-imposed cessation from treatment (“drug holidays”) for periods greater than 48 h can be associated with rebounding viral loads [17] and treatment failure due to development of drug resistance [18], a smaller temporal window of adherence monitoring would be beneficial to inform evaluation of drug efficacy and strategy for care. Additionally, LC–MS/MS methods have a potentially burdensome requirement for quantity of donated hair and necessitate a time intensive sample preparation. Mass spectrometry imaging (MSI) of biological samples offers an alternative analytical approach providing the capability for simultaneously monitoring the spatial distribution of analytes ranging from small molecules, such as pharmaceutical drugs [19], to lipids [20, 21], to peptides and proteins [22]. The utility of MSI has been demonstrated for a variety of fields, with particular emphasis on biomedical applications like drug distribution [23] and biomarker identification [24]. While such studies predominately target biological tissue as a sample matrix, recent work has investigated the distribution of pharmaceuticals [25], drugs of abuse [26-29], and endogenous biomolecules [30] in hair using multiple MSI techniques. In this work, we extend the application of infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) [31], which we have demonstrated to be a sensitive MSI technique for the analysis of ARVs in biological tissue [32-34], to the morphologically distinct matrix of hair. The technique couples resonant laser desorption with electrospray postionization for a soft, ESI-like ionization mechanism. Unlike MALDI, the traditional approach for MSI, IR-MALDESI does not require an organic matrix to promote analyte desorption and ionization, which can interfere with the analysis of small molecules such as ARVs. Instead, a layer of ice deposited on the sample is used as an energy-absorbing matrix

115

for infrared radiation that is easily applied and does not contribute to the background MS response. An evaluation of variables controlling the desorption of material from a single hair strand was performed to maximize analyte response by IR-MALDESI. These conditions were then used to determine the IR-MALDESI response to a broad range of ARVs representing several drug classes, which has been correlated to LC–MS/MS analyses for concentrations ranging 3 orders of magnitude. The distribution of ARVs within single hair strands collected from HIV positive patients on active ARV therapy is evaluated here for the first time. Additionally, an MSI method for determining relative hair melanin content is demonstrated, providing a critical means of comparing drug exposure in hair strands collected from different patients. During the hair growth process, averaging 1 cm per month on the scalp, three factors predominately contribute to the incorporation of drug in hair strands: drug basicity, drug lipophilicity, and hair melanin content [9]. The former two physicochemical attributes of the drug control its ability to penetrate cellular membranes of hair, where the drug then binds to melanin. Hence, preferential accumulation has been observed for lipophilic and basic drugs in pigmented hair strands, with drug concentration correlating with melanin concentration [35-37] or otherwise present among race groups possessing higher melanin in hair [38]. Here, we examine a MSI hair sample preparation step to oxidize and chemically degrade polymeric melanin into smaller molecules used as direct markers of pigmentation [39- 41]. We show that normalizing drug response in hair based on melanin content can facilitate comparison of drug response in hair between patients.

6.2 Experimental Section

6.2.1 Materials

Antiretroviral reference compounds emtricitabine (FTC), tenofovir (TFV), efavirenz (EFV), dolutegravir (DTG), raltegravir (RAL), darunavir (DRV), maraviroc (MRV), and lamivudine (3TC) were obtained through the NIH AIDS Reagent Program. Pyrrole-2,3,5-

116

tricarboxylic acid (PTCA) acetic acid salt was purchased from Toronto Research Chemicals 13 15 13 (Toronto, Ontario, CAN). Internal standard components FTC- C N2 and TFV- C5 were 15 purchased from Moravek Biochemicals (CA), 3TC- N-d2 and DRV-d9 from Aptochem

(Montreal, Canada), and RAL-d3 and MVC-d6, from Toronto Research Chemicals (Toronto, Canada). HPLC grade methanol and formic acid were purchased from Fisher Scientific (PA). Hydrogen peroxide (30%) was purchased from VWR International (PA). All materials were used without further purification.

6.2.2 Hair Samples

Sample collection occurred in accordance with Good Clinical Practice procedures, all applicable regulatory requirements, and the guiding principles of the Declaration of Helsinki. The study protocol was approved by the Biomedical Institutional Review Board at the University of North Carolina at Chapel Hill (IRBIS 08-0047), and all subjects provided written informed consent before study entry. Hair strands collected from a healthy donor not on HIV therapy were used both as negative controls and also as a blank hair matrix for incubation in a solution of ARVs. Blank hair was incubated in 1 mL of tissue culture media [Iscove’s Modified Dulbecco’s Media (Gibco, Grand Island, NY), 10% fetal bovine serum (Gibco), 240 units/mL nystatin (Sigma, St. Louis, MO), 100 units/mL penicillin-streptomycin (Gibco), and MEM vitamin solution (Sigma)] containing 6 ARVs (EFV, DTG, RAL, DRV, MRV, 3TC) ranging in concentration over 3 logs (1, 10, 100 μg/mL) for 24 h at 37 °C, allowing penetration of drug into the hair matrix, before the incubated strands were rinsed surficially with methanol/water and dried. ARV-incubated hair strands were used to optimize IR-MALDESI desorption conditions and calibrate instrument response to ARVs in hair. Hair strands were also collected at the UNC Clinical and Translational Research Center from HIV positive, virologically suppressed subjects (<50 copies/mL RNA by PCR) receiving Atripla (a fixed dose combination of EFV, FTC, and prodrug form of TFV) for ≥1 year. Hair was collected from the occipital region of the scalp by lifting the top layer of hair and clipping 20–30 fibers of hair at

117

least 1 cm long from as close to the scalp as possible. All hair samples were stored in aluminum foil and sealed in a plastic bag with desiccant until analysis.

6.2.3 Sample Preparation

Hair samples were adhered to glass microscope slides using double-sided adhesive tape without any prewashing. No solvent extraction step or sample manipulation was undertaken prior to MSI analysis when targeting disposition of ARVs in hair. For analysis of PTCA as a marker of eumelanin content in hair, sample-mounted slides were sprayed with a solution of

15% hydrogen peroxide and 1 M NH4OH in 50/50 (v/v) methanol/water using a thermal- assisted pneumatic sprayer (TM Sprayer, LEAP Technologies) and allowed to incubate on a lab bench at room temperature for 24 h prior to analysis.

6.2.4 IR-MALDESI Mass Spectrometry Imaging

The components and working principles of the IR-MALDESI MSI source used in this work have been detailed elsewhere [42], based on the original source design [31] and will be described here briefly. Sample slides for analysis were placed onto a Peltier-cooled sample stage where a controlled layer of ice was allowed to deposit onto the sample surface, which has been shown to enhance observed IR-MALDESI ion abundances [43] The ice layer thickness was maintained throughout analysis by holding the relative humidity in the source enclosure at 10% using a low flow of dry nitrogen gas. A 100 Hz IR Opolette HR infrared OPO laser (Opotek, Carlsbad, CA) tuned to the asymmetric stretching vibrational mode of liquid water (λ = 2.94 μm) was used to desorb sample material. The plume of neutral material generated by laser desorption of the sample expanded upward from the stage and was ionized with an orthogonal electrospray plume in an ESI-like fashion [44, 45]. Positive polarity experiments were conducted using a 50/50 (v/v) solution of methanol/water with 0.2% formic acid as the electrospray solvent. For polarity switching experiments, a 20 mM ammonium

118

hydroxide modifier was used with the 50/50 (v/v) solution of methanol/water, as has been used previously [46]. A recent evaluation of solvent modifiers for polarity switching experiments, conducted after the work presented here was performed, indicates that increased ion abundance is gained when using an acetic acid modifier without bias of response between positive and negative ionization modes [47] and will be utilized in future work. Spatial resolution of 100 μm was achieved from the ∼250 μm diameter laser spot by translating the two-axis stage in 100 μm increments based on an oversampling approach [48]. Resulting voxel dimensions for positive polarity experiments (100 μm × 100 μm) represent a theoretical temporal window of approximately 7 h based on an average growth rate of 1 cm per month. The method of data acquisition for polarity switching elongates the voxel dimension along one axis (200 μm × 100 μm) since the measurement polarity is alternated between sampling points across a scanline. The IR-MALDESI imaging source was fully synchronized with a Thermo Fisher Scientific Q Exactive Plus mass spectrometer (Bremen, Germany). The Q Exactive Plus was operated in full scan mode with a mass range m/z 150–600, with mass resolving power set to 140,000

(RPFWHM at m/z 200). Since AGC is turned off during IR-MALDESI experiments, lock masses are utilized in the control software to achieve parts per million mass accuracy [49]. For positive ionization mode two peaks of an ambient ion, diisooctyl phthalate, at m/z 391.2843 [M + H]+ and m/z 413.2662 [M + Na]+ were used as lock masses. The peaks of palmitic acid at m/z 255.2329 [M – H]− and stearic acid at m/z 283.2643 [M – H]− were used as lock masses in negative ionization mode. MS2 imaging (MS2I) of PTCA ([M – H]− m/z 198.0039) was conducted with a 0.5 m/z window centered at m/z 198.0 followed by ion accumulation in the C-trap. The accumulated ion packet was fragmented in the HCD cell at a normalized collision energy of 10% and then analyzed with mass resolving power of 140,000 (RPFWHM at m/z 200). All Thermo .RAW files were converted first to mzML[50] and then imzML [51] file formats before data processing and analysis of MSI data using MSiReader [52]. Ion maps were created with a mass tolerance of 5 ppm, and sample dimensions are indicated by scale bars.

119

6.2.5 LC–MS/MS Analysis of Hair

Extraction of analytes from snippets (1–2 cm) of incubated hair strands and from intact (3–3.5 cm) dosed hair strands was performed at room temperature by 10 min of sonication and 20 min of vortexing in 0.300 mL of 70:30 methanol/water. Following extraction, the samples 13 15 15 13 were mixed with isotopically labeled internal standards (FTC- C N2, 3TC- N-d2, TFV- C5,

RAL-d3 (for RAL and DTG quantitation), MVC-d6, and DRV-d9 (for DRV and EFV quantitation)). Sample extracts were then evaporated to dryness and reconstituted in 90:10 water/methanol. A Shimadzu HPLC system performed chromatographic separation with a Waters Atlantis T3 (50 mm × 2.1 mm, 3 μm) analytical column under gradient conditions. An AB SCIEX API 5000 mass spectrometer (AB SCIEX, Foster City, CA) equipped with a turbo spray interface was used as the detector. The samples were analyzed with a set of calibration standards and QC samples prepared at 1 mg/mL in DMSO (DTG), methanol (DRV, EFV, MRV), and water (RAL, 3TC) and serially diluted in 70:30 methanol/water to cover an analytical range of 0.06–150 ng/sample. Extraction efficiency was evaluated based on analyte recovery from serial extraction of hair strands [12]. Precision and accuracy of the calibration standards and QC samples were within acceptance criteria of 20%.

6.3 Results and Discussion

6.3.1 Optimization of IR-MALDESI for Hair Analysis

The analytical response of IR-MALDESI is sensitive to a range of variables in its experimental geometry controlling sample desorption and ionization, which have been previously optimized for interrogation of tissue cryosections [43]. Intact hair strands, roughly 80 μm in diameter and consisting of three distinct layers (cuticle, cortex, and medulla), present a distinct morphology relative to tissue cryosections. While incorporation of drug into hair can occur from the exterior cuticle layer through exposure to sweat and sebum, incorporation from

120

blood occurs during follicular cell development and concentrates in the cortex [9]. MSI analysis of hair strands using UV lasers, with depth penetration on the order tens of nanometers [53], typically includes preprocessing steps to facilitate probing analytes localized within the cortex either through solvent extraction [25] or by longitudinally sectioning [27] and abraiding [28] individual strands to expose the cortex directly. These methods add both labor and time to the analysis process, while also introducing the potential for sample contamination [28] and analyte delocalization [25] that may compromise the temporal record of drug use. Given the greater penetration depth (micrometers) of IR lasers into biological samples [53], we sought to develop an IR-MALDESI analysis method for hair that did not require sample manipulation. We first investigated the extent to which the unique morphology of hair strands required different IR-MALDESI desorption conditions (sample height, laser fluence, and number of laser shots per voxel) than tissue to optimize response to ARVs. Hair strands incubated in EFV were used as a test matrix. Although the radial distribution of drug in hair may vary between incubated strands (where external drug solution is absorbed through the exterior of the strand) and physiologically incorporated strands, incubated strands have the advantage of representing a controlled reference sample in which drug is expected to be uniformly distributed for calibration of MSI response in quantitative hair analysis [25]. A representative optical image of a hair strand analyzed by IR-MALDESI can be seen in Figure 6-1A. The image shows a region of the strand that has not been analyzed, in which the pigmented, optically translucent strand with darker medulla layer is clearly visible. While profilometry measuring the laser penetration depth was not performed, the change in opacity of the medulla and surrounding strand suggests that infrared laser desorption penetrates through the cuticle layer and potentially the cortex layer as well. Figure 6-1B shows ion maps of cholesterol ([M + H – + + H2O] , m/z 369.3516) and EFV ([M + H] , m/z 316.0347) from IR-MALDESI analysis of incubated hair strands analyzed with varying numbers of laser shots (1–5) acquired at each volumetric element, or voxel, probed by the laser. Average ion abundances of EFV and endogenous lipids indicated that two laser shots per voxel, with a laser pulse of 0.6 mJ/pulse,

121

yielded the best IR-MALDESI response to analytes in hair. Since the strand is not completely ablated with <5 laser pulses, the preference for two laser pulses may reflect the concentration of incorporated EFV closer to the surface of the strand or may also be a result of unfavorable effects on analyte ion abundance recently shown to occur as the C-trap accumulation time is extended to accommodate a greater number of laser pulses [42]. Reanalysis of a hair strand by IR-MALDESI, profiling a region in which the cuticle and some of the cortical layer had been previously removed by laser desorption, indicated a ∼60% reduction in signal abundance of EFV (Figure 6-2). This suggests that approximately 80%, as an upper bound, of accumulated EFV is extracted from incubated hair strands during initial IR-MALDESI analysis of intact hair.

Figure 6-1. Optimization of IR-MALDESI analysis of hair strands. (A) Optical image of hair strand indicating laser desorption of material in region analyzed by IR-MALDESI. (B) Ion maps of an endogenous lipid, cholesterol, and incubated ARV efavirenz collected with varying numbers of laser pulses (1–5) at each voxel.

122

Figure 6-2. IR-MALDESI depth profiling of EFV in incubated hair strands. Re-analysis of a hair strand (right panel) that has already undergone laser desorption by IR-MALDESI (left panel) indicated a reduction of a 61% reduction in average EFV signal abundance. Under the assumption that all EFV is removed from the strand during re-analysis, this suggests that ~80% (4.9 x 104/ (4.9+1.9) x 104) of EFV is extracted from an incubated strand during initial IR-MALDESI analysis. This value represents an upper bound on analyte extraction efficiency.

6.3.2 Calibration of IR-MALDESI Response to ARVs in Incubated Hair and Correlation to LC–MS/MS

Using conditions yielding the best response to incubated strands, a calibration was performed of IR-MALDESI response to hair strands incubated in six ARVs: EFV, DTG, DRV, RAL, MRV, and 3TC. These ARVs, summarized in Table 6-1, represent a range of drug classes often utilized in HIV treatment and which possess different basicity and lipophilicity controlling their incorporation into hair. Hair strands were incubated in a 3-log concentration range (1, 10, 100 μg/mL solutions) of the six ARVs as described above. Representative hair IR-MALDESI ion maps of ARV distribution in hair for EFV can be seen in Figure 6-3A for each of the three incubated drug concentrations. The averaged ion abundance of the ARVs from five individual strands measured by IR-MALDESI was compared to LC–MS/MS

123

response, where the amount of drug in fixed segment lengths of individual strands is estimated based on hair mass using an average hair density (0.633 g/cm3) and hair diameter (80 μm) [25]. A complete summary of calibration data including linear plots of IR-MALDESI response to each of the six ARVs and blank samples can be found in Table 6-2 and Figure 6-4. These data are shown in Figure 6-3B, plotted on a log–log scale to allow IR-MALDESI response to all ARVs to be seen easily. Data are color-coded by available information on drug pKa [54-59], increasing in shade from 3TC (pKa = 4.3) [59] to EFV (pKa = 10.2) [55]. Focusing on the cross-hatched data points, representing the lowest incubation concentration for each of the ARVs, it is observed that the amount of drug penetrating into hair strands as quantified by LC– MS/MS tends to increase with reported pKa. The range of ARV concentrations in incubated strands scales in a similar manner, increasing from 3TC (1.00–61.76 ng/mg hair) to EFV (24.34–1,772.58 ng/mg hair). LC–MS/MS extraction efficiencies for each ARV from incubated strands were estimated to be 86% or higher. A linear relationship (R2 > 0.962) between IR-MALDESI response and LC–MS/MS exists over the entire range of concentrations for each ARV with the exception of DTG (R2 = 0.645), as summarized in Table 6-3. Intercept values for linear regressions exceed IR-MALDESI response to blank samples, particularly for RAL, EFV, and DTG. Fits for these higher accumulating ARVs may be influenced by ion losses due to space charging [49] at the high end of the calibration range, where high IR- MALDESI ion abundance (>106) and variability are seen, at incubated hair concentrations greater than 300 ng/mg hair. Accumulation of ARVs in dosed hair is typically below 100–150 ng/mg hair [14-16, 60, 61], and IR-MALDESI response to all six ARVs is highly linear (R2 > 0.984) over this range (Figure 6-4) [49].

124

Table 6-1. Physicochemical properties of selected ARVs for hair strand incubation and IR-MALDESI calibration.

Antiretroviral Physiochemical Properties

pKa logP

Nonnucleoside RT inhibitor

Efavirenz EFV 10.2 4.5

Integrase Inhibitor

Dolutegravir DTG 8.2 1.7

Raltegravir RAL 6.7 -0.4

Protease Inhibitor

Strongest basic Strongest acidic

Darunavir DRV 2.4 13.6 2.8

Entry Inhibitor

Maraviroc MRV 7.3 3.6

Nucleoside Analogue RT Ihibitor

Lamivudine 3TC 4.3 -1.1

125

Figure 6-3. Calibration of IR-MALDESI MSI response to ARV-incubated hair strands. (A) IR- MALDESI ion maps of EFV response, representative of other investigated ARVs, from strands incubated in stock solution concentrations of 1 μg/mL, 10 μg/mL, and 100 μg/mL as noted in each panel. (B) Log–log plot summarizing +ESI IR-MALDESI response to all six ARVs relative to LC– MS/MS measurements of ARV accumulation from incubated hair strands. Linear plots comparing IR- MALDESI and LC–MS/MS results for each individual analyte are shown in the Supporting Information, Figure S-1. Table 6-2. Summary of ARV response in incubated hair strands from LC-MS/MS and IR-MALDESI analysis.

LC-MS/MS (ng/mg hair) IR-MALDESI (signal abundance) HIV Blank Stock concentration (ng/mL) Blank ARVs Low-1000 Med – 10000 High - 100000 휒̅ σ Low-1000 Med – 10000 High - 100000 3TC 0.00 1.00±0.21 6.78±0.21 61.76±1.56 3.66×102 9.78×102 4.87±0.54×103 1.75±0.20×104 1.25±0.060×105 MRV 0.00 1.23±0.35 6.20±1.19 69.77±3.11 5.42×102 2.66×103 2.14±0.28×104 2.70±0.66×105 2.64±1.44×105 DRV 0.00 2.45±0.31 15.29±0.56 136.24±22.45 3.48×101 2.35×102 5.44±3.20×103 2.88±0.33×104 1.19±0.71×105 RAL 0.00 1.38±0.09 24.25±2.9 337.86±2.22 0.00 0.00 7.88±1.08×103 2.32±0.75×105 9.94±0.54×105 DTG 0.00 8.60±0.76 264.34±22.19 900.43±51.11 6.63×101 4.45×102 1.03±0.34×105 3.73±1.17×106 3.85±0.95×106 EFV 0.00 24.34±1.39 165.02±20.41 1772.58±44.45 4.12×101 1.73×102 4.49±0.93×104 3.79±0.46×105 1.32±0.56×106

126

Figure 6-4. Calibration of IR-MALDESI signal abundance to LC-MS/MS response for ARVs in incubated hair strands. For ARVs in which the highest incubated concentrations exceed 150 ng/mg hair, linear estimates are shown using all data (black lines) and excluding the highest data points (red lines).

127

Table 6-3. Linear regression estimates of IR-MALDESI calibration data. All Calibration Data Excluding High Calibration Data > 150 ng/mg hair HIV Linear Regression of Calibration Estimated Linear Regression of Calibration Estimated LOD* ARVs LOD* Slope Intercept R2 (ng/mg Slope Intercept R2 (ng/mg hair) hair) EFV 6.99×102 9.22×104** 0.96226 0.81 2.33×103 -5.35×103 0.99919 0.24 DRV 8.38×102 6.11×103 0.98448 0.92 8.38×102 6.11×103 0.98448 0.92 3TC 1.99×103 2.38×103 0.99931 1.63 1.99×103 2.38×103 0.99931 1.63 RAL 2.81×103 5.35×104 0.97330 N/A 9.69×103 -2.68×103 0.99956 N/A DTG 4.10×103 7.18×105 0.64508 0.36 1.41×104 -8.98×103 0.99998 0.10 MRV 3.78×104 3.12×103 0.99961 0.23 3.78×104 3.12×103 0.99961 0.23

IR-MALDESI signal abundance for a given analyte concentration varies according to the ionization efficiency of the analyte in the electrospray [33], and the current MSI conditions indicate an increase in sensitivity from MRV > DTG > RAL > 3TC > DRV > EFV based on calibration slope. Limits of detection (LOD) were not reached experimentally for any of the six ARVs over the 3-log range in drug concentrations used for incubating hair strands but have been estimated to be less than or equal to 1.6 ng/mg hair based on the calibration and response to blank hair strands (Table 6-3). A systematic determination of instrumental LOD will be the focus of future work targeting ARV analysis in hair.

6.3.3 IR-MALDESI Response to ARVs in Dosed Hair

Brown hair strands cut from the scalp of three HIV infected, virologically suppressed subjects receiving fixed-dose Atripla (a combination of EFV, FTC, and prodrug form of TFV) for ≥1 year, were evaluated to determine interpatient and intrapatient variability in IR- MALDESI response. MSI analysis was simultaneously performed on 5–10 single hair strands from each HIV positive donor on ARV therapy to account for natural variability in hair growth cycles that can influence the accumulation of drug, with up to 20% of scalp hair in the dormant

128

telogen phase as well as incubated strands used as positive controls. The resulting ion maps, Figure 6-5A, showed continuous distribution of EFV response across the entirety of each strand of hair without detectable gaps, indicating uninterrupted dosing in accordance with the condition of virologic suppression measured by PCR by which these patients were classified as adherent. MS2I of dosed and incubated hair strands by IR-MALDESI confirmed the unique assignment of EFV based on known transition patterns [62]. The MSiSlicer feature of MSiReader allows visualization and interpretation of ion abundance along the curvilinear path of a hair strand. Figure 6-5B shows the temporal profile of EFV over 3 mm of a hair strand from Patient 1 starting at the proximal end, corresponding to approximately 9 days of hair growth. Per-voxel MSI response was evaluated using MSiReader software and average response for each patient is summarized in Figure 6-5A. Average IR-MALDESI response to multiple strands from individual donors indicated high repeatability (<12% relative standard deviation, % RSD) of response and a ∼1.4-fold intrapatient variability in IR-MALDESI EFV ion abundance. A ∼4-fold interpatient variability was observed. These results were in good agreement with LC–MS/MS analysis of extracts from matching single strands, Figure 6-6, which indicated a ∼2.3-fold interpatient variability in EFV (mean =1.720 ng/mg hair, range = 0.968–2.200 ng/mg hair). LC–MS/MS extraction efficiency for EFV from dosed hair was estimated to be 91% based on serial extraction of hair strands. No detectable response to TFV or FTC was observed by IR-MALDESI MSI from dosed patient samples. LC–MS/MS analysis of extracts from matching single strands indicated that FTC (mean = 0.333 ng/mg hair, range = 0.213–0.535 ng/mg hair) was present in the hair strands at concentrations below EFV and that TFV was below the limit of quantitation. TFV (pKa = 3.75) and FTC (pKa = 2.7) accumulate at low levels in hair and typically require large thatches of hair for quantitative analysis [14]

129

Figure 6-5. Analysis of EFV in hair strands from three dosed patients. (A) IR-MALDESI MSI shows uniform longitudinal distribution of EFV in hair strands collected from virally suppressed dosed patients, with high intrapatient repeatability of response. (B) Longitudinal profiling of EFV response from proximal end of dosed hair strands using MSiReader.

Figure 6-6. Calibration of IR-MALDESI signal abundance to LC-MS/MS response for EFV in dosed hair strands.

130

Given that IR-MALDESI sensitivity toward EFV in hair was lower than other relevant ARVs calibrated in this work and that the method was capable of evaluating the longitudinal distribution of EFV at concentrations typically observed for ARVs in hair [14-16, 60, 61], we anticipate that this approach has broad applicability for monitoring the disposition of ARVs in hair samples collected from patients. While MSI sample preparation has been minimized to reduce processing time and avoid axial and transverse analyte delocalization [25] and compromise the temporal record of drug use, solvent extraction [25] and longitudinal sectioning [27] steps will likely be explored in future analysis to improve limits of quantitation when targeting lower accumulating compounds. Since all patients examined were on the same fixed dose regimen of EFV, differences in the response observed in hair between patients may arise from metabolism or physiological incorporation into the hair strands. While explicating the contributing sources of these differences was beyond the scope of the current work, an approach to normalizing analyte response using a marker for melanin content is discussed in the next section.

6.3.4 Evaluating Hair Melanin Content by MSI

Drug response in hair can vary greatly depending on hair pigmentation, treatment, and texture. Cosmetic and ultrastructural alterations to hair through treatments such as bleaching and dyeing have been shown to reduce the presence of drug in hair [63, 64]. The amount of drug incorporated in hair can correlate strongly to melanin content based on drug basicity [35, 37], and is likely one of the factors contributing to differences in EFV response observed between patients in the previous section. While Poetzsch et al. have recently demonstrated that melanin does not influence the ionization process of UV MALDI [25], development of an MSI method to evaluate hair melanin content is needed to facilitate comparison in drug accumulation between different subjects, a necessity for an assessment of adherence based strictly on dose–response. Eumelanin, the predominate polymeric form of melanin, is composed mainly of the monomer unit 5,6-dihydroxyindole-2-carboxylic acid (DHICA). A

131

common strategy for quantification of eumelanin involves the chemical oxidation of DHICA to PTCA [39-41], as shown in Figure 6-7A. On the basis of the method of Szekely-Klepser et al. [40], hair strands of different colors from multiple donors were adhered to a glass slide and

15% H2O2 was uniformly sprayed onto the surface of the hair strands. Once wetted, samples were allowed to incubate overnight prior to analysis. Targeted negative ion mode MS2I monitoring of characteristic PTCA transitions, verified by direct infusion of a PTCA reference standard, was conducted on both pigmented and unpigmented hair strands as an initial validation step. Ion maps of [PTCA – H]− and fragments associated with the loss of one (m/z 154.0133) and two (m/z 110.0232) carboxylic acid units are shown in Figure 6-7B. These ion maps indicate that PTCA and its characteristic fragments were positively identified in the four pigmented hair strands analyzed but do not show any presence of the ions from gray, unpigmented hair strands (denoted by white asterisks in the overlaid images of Figure 6-7B) and served to confirm the PTCA peak assignment. PTCA response from hair strands with a range of pigmentation (blonde, brown, black, dyed) was then evaluated in a full MS polarity switching MSI mode. PTCA ion abundance was observed to increase with darkening hair color (Figure 6-7C), and there is some indication that delocalization of PTCA away from the hair strands can occur during H2O2 incubation. Average PTCA response across each color hair strand is summarized in Figure 6-7D, with a range of almost an order of magnitude differentiating highly pigmented black strands from blonde. Dyed brown, chemically altered hair showed very little PTCA ion abundance. Additional assay optimization is expected to reduce the intrastrand variability in response (% RSD = 39–75%).

132

Figure 6-7. (A) Scheme of melanin degradation to analytical target PTCA based on hair oxidation by 2 − 15% H2O2 and 1 M NH4OH. (B) −ESI IR-MALDESI MS I ion maps of [PTCA – H] (m/z 198.0039) and fragments associated with the loss of one (m/z 154.01330) and two (m/z 110.02320) carboxylic acid units from oxidized pigmented and unpigmented hair strands. Unpigmented, gray strands (denoted by white asterisks) show no response to PTCA. (C) Ion map of PTCA response from hair strands varying in color (dyed, blonde, brown, and black). D) Average [PTCA – H]− ion abundance from colored hair strands indicating increased abundance as hair color darkens. On the basis of the method for determining hair PTCA response, we evaluated the applicability of a normalization scheme for EFV in hair based on melanin content to facilitate direct interpatient comparison of drug accumulation. Strands (n = 4) from each of the three patients on therapy were first analyzed by IR-MALDESI for response to EFV before the strands were sprayed, incubated, and reanalyzed for EFV and PTCA. Optical images of the brown hair strands in Figure 6-8A show the depletion of pigmentation after the exposure to

H2O2. Also seen in Figure 6-8A are ion maps of EFV distribution in hair strands, which do not indicate a significant difference in EFV ion abundance measured before and after exposure to

133

H2O2. While exposure to peroxide-based bleaching creams has recently been shown to rapidly degrade the detectability of cocaine in hair during MSI analysis [26], our results suggest that analysis of PTCA may be possible while still preserving the integrity of a drug bound in hair. Ion abundances of both EFV and PTCA were averaged based on longitudinal distance from the proximal end across all strands from each patient. The ion map of EFV response and average profiles of EFV distribution from each patient are shown in the top panel of Figure 6-8B, again indicating a 4-fold difference in response between Patient 1 and Patient 3 that is in good agreement with the LC–MS/MS response (Figure 6-6). PTCA response between the brown hair strands from the patients varied (Figure 6-9) and normalizing the per-voxel ion abundance of EFV ([EFV + H]+) by that of PTCA ([PTCA – H]−), shown in the middle panel of Figure 6-8B, causes all profiles of patient response to fixed dose therapy to overlap within a standard deviation of measurement. EFV ionizes well in both positive and negative ionization modes, allowing us to also investigate normalizing the ion abundance of EFV ([EFV – H]−) by PTCA([PTCA – H]−) since we have shown recently that the % RSD for a normalization scheme can be sensitive to the ionization mechanism of the two ions compared based on potential differences in ionization efficiency [33]. The bottom panel of Figure 6-8B shows profiles of [EFV – H]−/PTCA – H]−, yielding similar results to [EFV + H]+/PTCA – H]−. This is the first evidence for a normalization approach to MSI drug response in hair based on melanin content. Determining the applicability of the technique for establishing adherence based on threshold drug response will require a much greater sample size of patients in future work, but the approach may hold great promise.

134

Figure 6-8. (A) IR-MALDESI response to [EFV + H]+ from strands (n = 4) of three dosed patients before (left) and after (right) oxidation of melanin by H2O2 indicating no significant degradation in response to EFV. (B) Top panel: ion map of [EFV + H]+ (left) and average longitudinal profile (right) for each of three patients, indicating a 4-fold difference in response to accumulated EFV in hair. Middle panel: Normalization of [EFV + H]+ response by [PTCA – H]− results in similar longitudinal profiles for each of three patients to fixed-dose intake of EFV. Bottom panel: Comparative normalization approach for EFV, matching ionization mechanisms ([EFV – H]−/PTCA – H]−).

135

Figure 6-9. IR-MALDESI ion map of PTCA response from strands (n=4) of three dosed patients after oxidation of hair melanin by H2O2. Additionally, imaging dosed hair strands over the range m/z 150–600 in both positive mode ESI and negative mode ESI yielded over 1000 peaks uniquely associated with hair. As shown in Figure 6-10, the mass excess plot of these peaks, representing the difference between the nominal and monoisotopic ion mass, indicate a high degree of overlap with the mass excess distribution for lipid classes, particularly fatty acyls [65]. There is potential for discovery within these data of biomarkers for hair pigmentation without the requirement for an oxidation sample step, in a similar manner to recent identification of biomarker candidates for aging by MSI [30].

136

Figure 6-10. Mass excess of all hair-specific peaks (+ESI, black dots; -ESI, white dots) overlaid on mass excess distribution of lipids generated from LIPIDMAPS Structure Database. The plots indicate that all of the hair-specific peaks correspond to lipids across different classes, particularly fatty acyls.

137

6.4 Conclusions

The application of IR-MALDESI to the analysis of ARVs in hair for the purpose of characterizing drug adherence has been successfully demonstrated. Infrared laser desorption of hair samples penetrates through the cuticle layer of hair strands, allowing imaging of analytes in hair without any sample extraction and very minimal sample preparation for rapid analysis. Linear response of IR-MALDESI MSI relative to LC–MS/MS methods was observed for a range of HIV ARV agents over a 3-log concentration range. The high spatial resolution of IR-MALDESI MSI allows profiling of drug response along individual hair strands, offering a retrospective picture of a patient’s drug adherence over a long temporal window. An approach to normalizing MSI response to drugs based on native melanin content of hair has also been demonstrated, facilitating the development of broadly applicable adherence evaluation criteria based solely on drug response.

6.5 Acknowledgements

The authors gratefully acknowledge financial support received from National Institutes of Health (Grants U19AI096113, R01AI111891, R01GM087964, and P30 AI50410).

6.6 References

1. Haubrich, R.H., Little, S.J., Currier, J.S., Forthal, D.N., Kemper, C.A., Beall, G.N., Johnson, D., Dubé, M.P., Hwanga, J.Y., McCutchan, J.A.: The value of patient- reported adherence to antiretroviral therapy in predicting virologic and immunologic response. Aids. 13, 1099-1107 (1999)

2. Nieuwkerk, P.T., Oort, F.J.: Self-reported adherence to antiretroviral therapy for HIV- 1 infection and virologic treatment response: a meta-analysis. JAIDS Journal of Acquired Immune Deficiency Syndromes. 38, 445-448 (2005)

3. Gross, R., Bilker, W.B., Friedman, H.M., Strom, B.L.: Effect of adherence to newly initiated antiretroviral therapy on plasma viral load. Aids. 15, 2109-2117 (2001)

138

4. Haberer, J.E., Baeten, J.M., Campbell, J., Wangisi, J., Katabira, E., Ronald, A., Tumwesigye, E., Psaros, C., Safren, S.A., Ware, N.C., Thomas, K.K., Donnell, D., Krows, M., Kidoguchi, L., Celum, C., Bangsberg, D.R.: Adherence to Antiretroviral Prophylaxis for HIV Prevention: A Substudy Cohort within a Clinical Trial of Serodiscordant Couples in East Africa. PLOS Medicine. 10, e1001511 (2013)

5. van der Straten, A., Stadler, J., Montgomery, E., Hartmann, M., Magazi, B., Mathebula, F., Schwartz, K., Laborde, N., Soto-Torres, L.: Women’s Experiences with Oral and Vaginal Pre-Exposure Prophylaxis: The VOICE-C Qualitative Study in Johannesburg, South Africa. PLOS ONE. 9, e89118 (2014)

6. Van Damme , L., Corneli , A., Ahmed , K., Agot , K., Lombaard , J., Kapiga , S., Malahleha , M., Owino , F., Manongi , R., Onyango , J., Temu , L., Monedi , M.C., Mak'Oketch , P., Makanda , M., Reblin , I., Makatu , S.E., Saylor , L., Kiernan , H., Kirkendale , S., Wong , C., Grant , R., Kashuba , A., Nanda , K., Mandala , J., Fransen , K., Deese , J., Crucitti , T., Mastro , T.D., Taylor , D.: Preexposure Prophylaxis for HIV Infection among African Women. New England Journal of Medicine. 367, 411- 422 (2012)

7. Van der Straten, A., Van Damme, L., Haberer, J.E., Bangsberg, D.R.: Unraveling the divergent results of pre-exposure prophylaxis trials for HIV prevention. Aids. 26, F13- F19 (2012)

8. Baumgartner, W., Hill, V., Blahd, W.: Hair Analysis for Drugs of Abuse. (1989)

9. Pragst, F., Balikova, M.A.: State of the art in hair analysis for detection of drug and alcohol abuse. Clinica Chimica Acta. 370, 17-49 (2006)

10. Uematsu, T.: Therapeutic Drug Monitoring in Hair Samples. Clinical Pharmacokinetics. 25, 83-87 (1993)

11. Beumer, J., Bosman, I., Maes, R.: Hair as a biological specimen for therapeutic drug monitoring. International journal of clinical practice. 55, 353-357 (2001)

12. Huang, Y., Gandhi, M., Greenblatt, R.M., Gee, W., Lin, E.T., Messenkoff, N.: Sensitive analysis of anti-HIV drugs, efavirenz, lopinavir and ritonavir, in human hair by liquid chromatography coupled with tandem mass spectrometry. Rapid Communications in Mass Spectrometry. 22, 3401-3409 (2008)

13. Huang, Y., Yang, Q., Yoon, K., Lei, Y., Shi, R., Gee, W., Lin, E.T., Greenblatt, R.M., Gandhi, M.: Microanalysis of the antiretroviral nevirapine in human hair from HIV-

139

infected patients by liquid chromatography-tandem mass spectrometry. Analytical and Bioanalytical Chemistry. 401, 1923 (2011)

14. Liu, A.Y., Yang, Q., Huang, Y., Bacchetti, P., Anderson, P.L., Jin, C., Goggin, K., Stojanovski, K., Grant, R., Buchbinder, S.P., Greenblatt, R.M., Gandhi, M.: Strong Relationship between Oral Dose and Tenofovir Hair Levels in a Randomized Trial: Hair as a Potential Adherence Measure for Pre-Exposure Prophylaxis (PrEP). PLOS ONE. 9, e83736 (2014)

15. Hickey, M.D., Salmen, C.R., Tessler, R.A., Omollo, D., Bacchetti, P., Magerenge, R., Mattah, B., Salmen, M.R., Zoughbie, D., Fiorella, K.J.: Antiretroviral concentrations in small hair samples as a feasible marker of adherence in rural Kenya. Journal of acquired immune deficiency syndromes (1999). 66, 311 (2014)

16. Gandhi, M., Ameli, N., Bacchetti, P., Anastos, K., Gange, S.J., Minkoff, H., Young, M., Milam, J., Cohen, M.H., Sharp, G.B., Huang, Y., Greenblatt, R.M.: Atazanavir Concentration in Hair Is the Strongest Predictor of Outcomes on Antiretroviral Therapy. Clinical Infectious Diseases. 52, 1267-1275 (2011)

17. Kleeberger, C.A., Phair, J.P., Strathdee, S.A., Detels, R., Kingsley, L., Jacobson, L.P.: Determinants of heterogeneous adherence to HIV-antiretroviral therapies in the Multicenter AIDS Cohort Study. JAIDS Journal of Acquired Immune Deficiency Syndromes. 26, 82-92 (2001)

18. Parienti, J.-J., Massari, V., Descamps, D., Vabret, A., Bouvet, E., Larouzé, B., Verdon, R.: Predictors of Virologic Failure and Resistance in HIV-Infected Patients Treated with Nevirapine- or Efavirenz-Based Antiretroviral Therapy. Clinical Infectious Diseases. 38, 1311-1316 (2004)

19. Prideaux, B., Stoeckli, M.: Mass spectrometry imaging for drug distribution studies. Journal of Proteomics. 75, 4999-5013 (2012)

20. Murphy, R.C., Hankin, J.A., Barkley, R.M.: Imaging of lipid species by MALDI mass spectrometry. Journal of Lipid Research. 50, S317-S322 (2009)

21. Gode, D., Volmer, D.A.: Lipid imaging by mass spectrometry–a review. Analyst. 138, 1289-1315 (2013)

22. Burnum, K.E., Frappier, S.L., Caprioli, R.M.: Matrix-Assisted Laser Desorption/Ionization Imaging Mass Spectrometry for the Investigation of Proteins and Peptides. Annual Review of Analytical Chemistry. 1, 689-705 (2008)

140

23. Greer, T., Sturm, R., Li, L.: Mass spectrometry imaging for drugs and metabolites. Journal of Proteomics. 74, 2617-2631 (2011)

24. Schwamborn, K.: Imaging mass spectrometry in biomarker discovery and validation. Journal of proteomics. 75, 4990-4998 (2012)

25. Poetzsch, M., Steuer, A.E., Roemmelt, A.T., Baumgartner, M.R., Kraemer, T.: Single Hair Analysis of Small Molecules Using MALDI-Triple Quadrupole MS Imaging and LC-MS/MS: Investigations on Opportunities and Pitfalls. Analytical Chemistry. 86, 11758-11765 (2014)

26. Cuypers, E., Flinders, B., Bosman, I.J., Lusthof, K.J., Van Asten, A.C., Tytgat, J., Heeren, R.M.A.: Hydrogen peroxide reactions on cocaine in hair using imaging mass spectrometry. Forensic Science International. 242, 103-110 (2014)

27. Nakanishi, T., Nirasawa, T., Takubo, T.: Quantitative Mass Barcode-Like Image of Nicotine in Single Longitudinally Sliced Hair Sections from Long-Term Smokers by Matrix-Assisted Laser Desorption Time-of-Flight Mass Spectrometry Imaging. Journal of Analytical Toxicology. 38, 349-353 (2014)

28. Shen, M., Xiang, P., Shi, Y., Pu, H., Yan, H., Shen, B.: Mass imaging of ketamine in a single scalp hair by MALDI-FTMS. Analytical and Bioanalytical Chemistry. 406, 4611-4616 (2014)

29. Porta, T., Grivet, C., Kraemer, T., Varesio, E., Hopfgartner, G.: Single Hair Cocaine Consumption Monitoring by Mass Spectrometric Imaging. Analytical Chemistry. 83, 4266-4272 (2011)

30. Waki, M.L., Onoue, K., Takahashi, T., Goto, K., Saito, Y., Inami, K., Makita, I., Angata, Y., Suzuki, T., Yamashita, M., Sato, N., Nakamura, S., Yuki, D., Sugiura, Y., Zaima, N., Goto-Inoue, N., Hayasaka, T., Shimomura, Y., Setou, M.: Investigation by Imaging Mass Spectrometry of Biomarker Candidates for Aging in the Hair Cortex. PLOS ONE. 6, e26721 (2011)

31. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared Matrix-Assisted Laser Desorption Electrospray Ionization (IR-MALDESI) Imaging Source Coupled to a FT-ICR Mass Spectrometer. Journal of The American Society for Mass Spectrometry. 24, 92-100 (2013)

32. Barry, J.A., Robichaud, G., Bokhart, M.T., Thompson, C., Sykes, C., Kashuba, A.D., Muddiman, D.C.: Mapping antiretroviral drugs in tissue by IR-MALDESI MSI

141

coupled to the Q Exactive and comparison with LC-MS/MS SRM assay. Journal of The American Society for Mass Spectrometry. 25, 2038-2047 (2014)

33. Bokhart, M.T., Rosen, E., Thompson, C., Sykes, C., Kashuba, A.D.M., Muddiman, D.C.: Quantitative mass spectrometry imaging of emtricitabine in cervical tissue model using infrared matrix-assisted laser desorption electrospray ionization. Analytical and Bioanalytical Chemistry. 407, 2073-2084 (2015)

34. Thompson, C.G., Bokhart, M.T., Sykes, C., Adamson, L., Fedoriw, Y., Luciw, P.A., Muddiman, D.C., Kashuba, A.D.M., Rosen, E.P.: Mass Spectrometry Imaging Reveals Heterogeneous Efavirenz Distribution within Putative HIV Reservoirs. Antimicrobial Agents and Chemotherapy. 59, 2944-2948 (2015)

35. Rollins, D.E., Wilkins, D.G., Krueger, G.G., Augsburger, M.P., Mizuno, A., O'Neal, C., Borges, C.R., Slawson, M.H.: The Effect of Hair Color on the Incorporation of Codeine into Human Hair. Journal of Analytical Toxicology. 27, 545-551 (2003)

36. Slawson, M.H., Wilkins, D.G., Rollins, D.E.: The Incorporation of Drugs into Hair: Relationship of Hair Color and Melanin Concentration to Phencyclidine Incorporation. Journal of Analytical Toxicology. 22, 406-413 (1998)

37. Kronstrand, R., Förstberg-Peterson, S., KÅgedal, B., Ahlner, J., Larson, G.: Codeine Concentration in Hair after Oral Administration Is Dependent on Melanin Content. Clinical Chemistry. 45, 1485-1494 (1999)

38. Henderson, G.L., Harkey, M.R., Zhou, C., Jones, R.T., Jacob, I.I.I.P.: Incorporation of Isotopically Labeled Cocaine into Human Hair: Race as a Factor. Journal of Analytical Toxicology. 22, 156-165 (1998)

39. Ito, S., Wakamatsu, K.: Quantitative Analysis of Eumelanin and Pheomelanin in Humans, Mice, and Other Animals: a Comparative Review. Pigment Cell Research. 16, 523-531 (2003)

40. Szekely-Klepser, G., Wade, K., Woolson, D., Brown, R., Fountain, S., Kindt, E.: A validated LC/MS/MS method for the quantification of pyrrole-2,3,5-tricarboxylic acid (PTCA), a eumelanin specific biomarker, in human skin punch biopsies. Journal of Chromatography B. 826, 31-40 (2005)

41. Panzella, L., Manini, P., Monfrecola, G., D'Ischia, M., Napolitano, A.: An easy-to-run method for routine analysis of eumelanin and pheomelanin in pigmented tissues. Pigment Cell Research. 20, 128-133 (2007)

142

42. Rosen, E.P., Bokhart, M.T., Nazari, M., Muddiman, D.C.: Influence of C-Trap Ion Accumulation Time on the Detectability of Analytes in IR-MALDESI MSI. Analytical Chemistry. 87, 10483-10490 (2015)

43. Robichaud, G., Barry, J.A., Muddiman, D.C.: IR-MALDESI mass spectrometry imaging of biological tissue sections using ice as a matrix. Journal of the American Society for Mass Spectrometry. 25, 319-328 (2014)

44. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Generation and detection of multiply-charged peptides and proteins by matrix-assisted laser desorption electrospray ionization (MALDESI) fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry. 17, 1712-1716 (2006)

45. Rosen, E.P., Bokhart, M.T., Ghashghaei, H.T., Muddiman, D.C.: Influence of Desorption Conditions on Analyte Sensitivity and Internal Energy in Discrete Tissue or Whole Body Imaging by IR-MALDESI. Journal of The American Society for Mass Spectrometry. 26, 899-910 (2015)

46. Cochran, K.H., Barry, J.A., Robichaud, G., Muddiman, D.C.: Analysis of trace fibers by IR-MALDESI imaging coupled with high resolving power MS. Analytical and Bioanalytical Chemistry. 407, 813-820 (2015)

47. Nazari, M., Muddiman, D.C.: Polarity switching mass spectrometry imaging of healthy and cancerous hen ovarian tissue sections by infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). Analyst. 141, 595-605 (2016)

48. Jurchen, J.C., Rubakhin, S.S., Sweedler, J.V.: MALDI-MS imaging of features smaller than the size of the laser beam. Journal of the American Society for Mass Spectrometry. 16, 1654-1659 (2005)

49. Olsen, J.V., de Godoy, L.M.F., Li, G., Macek, B., Mortensen, P., Pesch, R., Makarov, A., Lange, O., Horning, S., Mann, M.: Parts per Million Mass Accuracy on an Orbitrap Mass Spectrometer via Lock Mass Injection into a C-trap. Molecular & Cellular Proteomics. 4, 2010-2021 (2005)

50. Kessner, D., Chambers, M., Burke, R., Agus, D., Mallick, P.: ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics. 24, 2534- 2536 (2008)

51. Race, A.M., Styles, I.B., Bunch, J.: Inclusive sharing of mass spectrometry imaging data requires a converter for all. Journal of Proteomics. 75, 5111-5112 (2012)

143

52. Robichaud, G., Garrard, K.P., Barry, J.A., Muddiman, D.C.: MSiReader: an open- source interface to view and analyze high resolving power MS imaging files on Matlab platform. Journal of the American Society for Mass Spectrometry. 24, 718-721 (2013)

53. Dreisewerd, K.: The Desorption Process in MALDI. Chemical Reviews. 103, 395-426 (2003)

54. ChemAxon. http://www.chemicalize.org, accessed October 8, 2015

55. Kashuba, A.D.M., Dyer, J.R., Kramer, L.M., Raasch, R.H., Eron, J.J., Cohen, M.S.: Antiretroviral-Drug Concentrations in Semen: Implications for Sexual Transmission of Human Immunodeficiency Virus Type 1. Antimicrobial Agents and Chemotherapy. 43, 1817-1826 (1999)

56. ViiV Healthcare Group: Retrieved from http://www.medicines.org.au/files/viptivic.pdf, 2005

57. Moss, D.M., Siccardi, M., Murphy, M., Piperakis, M.M., Khoo, S.H., Back, D.J., Owen, A.: Divalent Metals and pH Alter Raltegravir Disposition In Vitro. Antimicrobial Agents and Chemotherapy. 56, 3020-3026 (2012)

58. Abel, S., Davis, J.D., Ridgway, C.E., Hamlin, J.C., Vourvahis, M.: Pharmacokinetics, safety and tolerability of a single oral dose of maraviroc in HIV-negative subjects with mild and moderate hepatic impairment. Antivir Ther. 14, 831-837 (2009)

59. Shalaeva, M., Kenseth, J., Lombardo, F., Bastin, A.: Measurement of Dissociation Constants (pKa Values) of Organic Compounds by Multiplexed Capillary Electrophoresis Using Aqueous and Cosolvent Buffers**Advanced Analytical Technologies, Inc. (Formerly CombiSep), 2711 South Loop Drive, Suite 4200, Ames, IA 50010. Journal of Pharmaceutical Sciences. 97, 2581-2606 (2008)

60. Koss, C.A., Natureeba, P., Mwesigwa, J., Cohan, D., Nzarubara, B., Bacchetti, P., Horng, H., Clark, T.D., Plenty, A., Ruel, T.D.: Hair concentrations of antiretrovirals predict viral suppression in HIV-infected pregnant and breastfeeding Ugandan women. AIDS (London, England). 29, 825 (2015)

61. Gandhi, M., Mwesigwa, J., Aweeka, F., Plenty, A., Charlebois, E., Ruel, T.D., Huang, Y., Clark, T., Ades, V., Natureeba, P.: Hair and plasma data show that lopinavir, ritonavir and efavirenz all transfer from mother to infant in utero, but only efavirenz transfers via breastfeeding. Journal of acquired immune deficiency syndromes (1999). 63, 578 (2013)

144

62. Nirogi, R., Bhyrapuneni, G., Kandikere, V., Mudigonda, K., Komarneni, P., Aleti, R., Mukkanti, K.: Simultaneous quantification of a non-nucleoside reverse transcriptase inhibitor efavirenz, a nucleoside reverse transcriptase inhibitor emtricitabine and a nucleotide reverse transcriptase inhibitor tenofovir in plasma by liquid chromatography positive ion electrospray tandem mass spectrometry. Biomedical Chromatography. 23, 371-381 (2009)

63. Pötsch, L., Skopp, G., Becker, J.: Ultrastructural alterations and environmental exposure influence the opiate concentrations in hair of drug addicts. International Journal of Legal Medicine. 107, 301-305 (1995)

64. Jurado, C., Kintz, P., Menéndez, M., Repetto, M.: Influence of the cosmetic treatment of hair on drug testing. International Journal of Legal Medicine. 110, 159-163 (1997)

65. Sud, M., Fahy, E., Cotter, D., Brown, A., Dennis, E.A., Glass, C.K., Merrill Jr, A.H., Murphy, R.C., Raetz, C.R., Russell, D.W.: Lmsd: lipid maps structure database. Nucleic acids research. 35, D527-D532 (2006)

145

7 IR-MALDESI Mass Spectrometry Imaging at 50 Micron Spatial Resolution

The following work was reprinted with permission from: Bokhart, M. T., Manni, J., Garrard, K. P., Ekelöf, M., Nazari, M., & Muddiman, D. C.: IR-MALDESI Mass Spectrometry Imaging at 50 Micron Spatial Resolution. Journal of The American Society for Mass Spectrometry, 28, 10, 2099-2107 (2017). DOI: 10.1007/s13361-017-1740-x. Copyright © 2017 American Society for Mass Spectrometry. The original publication may be accessed via the Internet at https://link.springer.com/article/10.1007/s13361-017-1740-x

7.1 Introduction

Mass spectrometry imaging (MSI) combines the molecular specificity of mass spectrometry with the spatially resolved analysis of an imaging technique [1]. This can be performed in a label-free manner over a range of m/z values, allowing thousands of analytes to be analyzed simultaneously. The resulting data may be visualized using a variety of software [2] to depict the localization and abundance of each molecule. MSI has been routinely applied to the study of plants [3], proteins [4], lipids [5], and drug distributions [6]. A common goal of nearly all applications is to achieve high spatial resolution without a loss in analyte detectability. Laser-based MSI methods are the most commonly used methods to date, and an article reviewing the broad field of high resolution laser-based MSI across different technologies was recently published [7]. The spatial resolution of all laser-based MSI methods is inherently tied to the laser spot size. Infrared (IR) laser desorption methods for improving spot size follow the same general principles as those in the UV regime [8]; however, IR lasers, which are the focus of this manuscript, are not as commonly used in the MSI field. For a Gaussian profile laser, the diffraction limited spot size is given by 4Mf2 (1)  D

146

, where M2 is the beam propagation quality factor, λ is the wavelength, f is the lens focal length, and D is the Gaussian input beam diameter at the lens. To decrease the spot size of the focused laser, it is possible to use a shorter focal length lens (f), increase the beam diameter (D), and/or improve the laser beam quality (M2). Spherical aberration can cause significant enlargement of the minimum spot size if not corrected [9]. The spot size determined by spherical aberration is given by kD3 (2) f 2 , where D and f are as described above and k is a material constant. While these can be easily changed in custom sources, adjustments to commercial MSI instruments are more difficult [10]. A short focal length lens presents a unique challenge in MSI because the ablated material must be proximal to and accessible for sampling by the mass spectrometer, often preventing the use of very short focal length lenses. To circumvent this issue, researchers may use transmission geometry IR laser ablation, where the laser is focused on the opposite side of the sample being analyzed [11]. Alternatively, Römpp and coworkers used an optic with a central drilled hole to allow the ablated material to pass through [12]. The laser beam propagation factor M2 is a defining parameter detailing the beam quality, with a perfect Gaussian beam having an M2 of 1. To reduce the detrimental effect of lasers with M2>1, a spatial filter may be used to remove the non-Gaussian and low laser energy edge [13]. As seen from equation 1, using the maximum diameter of the lens is beneficial to achieving a diffraction-limited spot size, however, the larger diameter increases the contribution of spherical aberration (equation 2) if a plano-convex or meniscus lens is used, which do not correct for spherical aberration. Incorporating beam expanding lenses reduces beam divergence and the focal spot size of the IR laser. Alternatively, Hieta and coworkers have used the inherent beam divergence of their laser with a very long path length to completely fill the focusing lens and remove the beam edge [13].

147

Several methods have been used to achieve spatial resolutions lower than the IR laser spot size. Oversampling involves moving the sample by less than the laser ablation diameter with complete ablation at each position, effectively sampling only a small portion of the total laser spot size [14, 15]. This method results in irregular shapes of tissue ablated from a Gaussian profile laser beam. The coupling of IR laser ablation and a spatially resolved detector has been reported [16, 17] to achieve MSI resolution below the diffraction limit of the IR laser. Infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) is a technique combining infrared laser ablation and electrospray ionization [18, 19]. IR- MALDESI MSI has been used to analyze a variety of biospecimens [20]. The laser ablation diameter on tissue has been previously reported as approximately 150 µm using a single focusing element [19, 21]. This method has been used and optimized for tissue imaging experiments of lipids, metabolites and small molecule drugs [21]. This work details the utility of a multi-element optical system for IR-MALDESI MSI to simultaneously reduce spot size and improve spatial resolution. The multi-element optical system incorporates an adjustable iris, 3.75x beam expander, and aspheric focusing lens to improve the spatial resolution of IR-MALDESI MSI to 50 µm.

7.2 Experimental

7.2.1 Materials

LC/MS grade methanol, ethanol, water, and formic acid were purchased from Fisher Scientific (Pittsburgh, PA, USA). Laser burn paper was purchased from ZAP-IT (ZFC-23, ZAP-IT, Concord, NH, USA). Pre-cleaned microscope slides were purchased from VWR (Radnor, PA, USA).

148

7.2.2 Tissues

Mouse liver tissues were obtained from the College of Veterinary Medicine at North Carolina State University. Hen ovarian tissue was obtained from a C-Strain white leghorn commercial egg laying hen. Animals were managed in accordance with the Institute for Laboratory Animal Research Guide, and all husbandry practices were approved by North Carolina State University Institutional Animal Care and Use Committee. Tissues were harvested and immediately flash-frozen with a dry ice in isopentane bath and stored at -80 ºC until the time of the experiment. A Leica CM1950 cryomicrotome (Buffalo Grove, IL, USA) operated at -20ºC was used to prepare cryosections for analysis. Optimum cutting temperature (OCT) embedding medium (Fisher Scientific, Waltham, MA, USA) was used to adhere the tissue to a 40 mm specimen disc. The tissues were sectioned into 10-µm thick sections and thaw-mounted onto standard glass microscope slides immediately prior to analysis.

7.2.3 IR-MALDESI Source

A custom acrylic enclosure housed the ionization source consisting of an X-Y translation stage, electrospray emitter and a portion of the laser optics. Custom software developed in-house was used to drive the X-Y stage with the sample on a Peltier-cooled plate. The plate was cooled to -9 ºC to form a thin ice matrix layer on the animal tissue sections to enhance energy absorption and improve ablation dynamics. The ice matrix layer was maintained by purging the enclosure with dry nitrogen (Arc3 Gases, Raleigh, NC, USA) to a relative humidity of approximately 10% throughout the imaging experiments. Two laser pulses at 20 Hz from a mid-IR laser (IR opolette, Opotek, Carlsbad, CA, USA) tuned to 2940 nm were focused on the sample surface at each stage position. The ablated material was ejected normal to the target surface where it overlapped with an orthogonal electrospray plume. Stable electrospray was maintained using a solvent composition of 1:1 methanol:water with 0.2% formic acid, a 2.0 µl/min flow rate, 4.0 kV potential, and a 75 µm i.d. silica taper

149

tip emitter (New Objective, Woburn, MA, USA). Geometric parameters of the IR-MALDESI source remained constant through all experiments presented herein. The emitter tip was placed 5 mm above the sample, the laser ablation spot was 1 mm from the emitter tip and 5 mm from the customized extended MS inlet [21, 22]. The IR-MALDESI source was coupled to a Q Exactive™ Plus mass spectrometer (ThermoScientific, Bremen, Germany). Data were collected in positive ion mode with a scan range of 250-1000 m/z and resolving power was set to 140,000 (FWHM at 200 m/z). Automatic gain control (AGC) was turned off and the injection time was fixed based on the number of laser pulses. The total injection time (IT) was calculated as t=25+50(n-1) ms, where n is the number of laser pulses (i.e. IT = 25 ms for 1 laser pulse, 75 ms for 2 laser pulses, etc.). The MS .RAW data files were converted to the mzML format using MSConvert [23], and the mzML files were subsequently converted to the imzML format using the imzMLconverter software [24, 25]. The imzML files were then loaded into MSiReader for visualization and analysis of MSI data [26]. Averaged mass spectra and S/N were obtained from Xcalibur version 2.2 (Thermo Fisher, San Jose, CA, USA).

7.2.4 Safety

IR laser safety glasses (pn 100-38-200, Laser Safety Industries, Minneapolis, MN, USA) were worn while the laser was on. Curved laser safety shields (TPS8, Thorlabs, Newton, NJ, USA) were installed on the optical breadboard around the mirrors used to steer the beam. All lenses were enclosed in a 1” diameter lens tube. A laser curtain (Laz-R-Shroud, Rockwell Laser Industries, Cincinnatti, OH, USA) was erected around the IR-MALDESI laboratory space to isolate the IR laser from the rest of the laboratory.

150

7.2.5 Laser and Optics

An IR opolette tunable (2700-3100 nm) laser system was purchased from Opotek Inc. (Carlsbad, CA, USA). Protected gold mirrors (PF10-03-M01, Thorlabs, Newton, NJ, USA) were mounted on kinematic mounts (KM100, Thorlabs, Newton, NJ, USA) and were used to align the laser beam. The laser head and two aligning mirrors were mounted on an optical breadboard above the mass spectrometer. A third mirror directed the beam path down through the focusing lenses and in front of the mass spectrometer inlet. Two laser focusing designs were tested sequentially to compare the laser focusing ability of each. The first design consisted of a single, uncoated CaF2 planoconvex lens (focal length (f) = +62.5 mm, Edmund Optics, Irvine, CA, USA) (Figure 7-1A) and was previously used for IR-MALDESI MSI analyses [19]. The second design incorporated an adjustable iris (SM1D12C, Thorlabs, Newton, NJ, USA), a beam expander, and an aspherical focusing lens. The Galilean beam expander was constructed using a 1” CaF2 concave lens (f = -40 mm, antireflective (AR)-coated 3-5 μm)

(Thorlabs, Newton, NJ, USA) and a 1” CaF2 convex lens (f = +150 mm, AR-coated 3-5 μm) (Thorlabs, Newton, NJ, USA). The expanded beam was then focused on the target using a 1” ZnSe aspheric lens (f = 50.0 mm, AR-coated 3-5 μm) (Thorlabs, Newton, NJ, USA) (Figure 7-1B). A 25.4 mm diameter quartz coverslip (Alfa Aesar, Haverhill, MA, USA) was placed between the final lens and the target to protect the ZnSe lens. Laser energy was measured using a Nova II laser power meter with a 3A-P laser thermal power sensor (Ophir Optronics, Jerusalem, Israel) for both optical systems.

151

Figure 7-1. Schematic of IR-MALDESI using (a) a single spherical focusing lens, and (b) a beam expander with an iris and an aspherical focusing lens.

7.2.6 Laser Beam Caustic Visualization

A zap-it film card (ZFC-23, Zap-IT, Concord, NH, USA) was mounted on a microscope slide. After the lens system was focused onto the sample plane, the stage height was adjusted to defined positions located above, at, and below the focal point. A single laser ablation event was recorded on the laser film card at these defined z height positions for both optical systems. Images of each laser ablation spot were obtained using a 10x objective on an LMD7000 (Leica, Buffalo Grove, IL, USA). The burn paper ablation images were then loaded into MATLAB 2016b (Mathworks, Natick, MA, USA) where the ablation spots were manually fitted with an ellipse. The major and minor axes of each ellipse were recorded, with their corresponding z height, and are provided in Figure 7-2 and Figure 7-3. The fitted ellipses were plotted in 3- dimensional space with each corresponding projection (XY, XZ, and YZ) as shown in Figure 7-4. The average fluence at each z position was calculated by taking the average energy divided

152

by the area of each ellipse, which is reported as J/cm2. The ellipses are color coded based on the fluence using the “hot” heatmap available in MATLAB. The laser caustics for both systems were then visualized as a 3D surface through interpolation between z heights and the surfaces were colored according to the average fluence.

Figure 7-2. A laser burn spot was recorded on laser burn paper at the specified z positions for the single focusing lens. A 10x objective was used to take an optical image of each burn spot, and fit with an ellipse in MATLAB. The ellipse major and minor axes were converted into µm using a microscope micrometer.

153

Figure 7-3. A laser burn spot was recorded on laser burn paper at the specified z positions for the multi- element optical design. A 10x objective was used to take an optical image of each burn spot, and fit with an ellipse in MATLAB. The ellipse major and minor axes were converted into µm using a microscope micrometer.

Figure 7-4. Visualization of laser beam caustic for (a) single focusing optic, and (b) beam expander. The data is represented as projections of ellipses on three axes showing the laser focus profile in the XY, XZ, and YZ planes. The color of each ellipse is representative of the average laser fluence at each Z position. These ellipses were interpolated and represented as a 3D surface for both laser optic designs.

154

7.2.7 Fitting Real Data to Theoretical Real Laser Beam Focus

The equation for a real laser beam propagation is given by

2 2 M  2 WzWzz()1 2   (3) W

, where W(z) is the real beam radius at position z along the beam optical axis, Wo is the real 2 beam waist radius, M is the beam quality factor, and zo is the position of the position of the beam minimum radius with respect to a reference position [27]. Parameters defining the laser beam were calculated by fitting the theoretical curve of a real laser beam in the XZ and YZ planes to the measured values. The fit was performed in Excel using generalized nonlinear regression. The residuals were calculated for each data point and plotted as a function of z height. The half angle divergence was calculated using a linear regression on the last three points in the curve.

7.2.8 Laser Ablation Diameter on Tissue

Mouse liver tissue was cryosectioned and mounted onto microscope slides. The number of laser pulses was varied from 1-5 and the diameter of the beam shaping iris was set to 2, 4, or 12 mm. For each combination tested, 20 positions on tissue were ablated. After laser ablation, the slides were removed from the stage, excess water was removed, and the tissues were stained with 100 µL of histogene staining solution (Applied Biosystems, Foster City, CA, USA) pipetted directly on top of the tissue and allowed to sit for approximately 2 minutes. The slide was then washed with 70% ethanol for 30 seconds followed by 100% ethanol for 30 seconds. Excess ethanol was removed and approximately 50 μL of Permount mounting medium (Fisher Scientific, Waltham, MA, USA) was pipetted onto the tissue and a coverslip was mounted on the slide. Digital images of the stained tissues with ablation spots were taken using an LMD7000 (Leica Microsystems, Buffalo Grove, IL, USA) with a 5x objective. The JPEG images were loaded into MATLAB 2016b (Mathworks, Natick, MA, USA) and circles

155

were fitted to the ablation spots using the “imfindcircles” function in the Image Processing Toolbox.

7.2.9 MSI with High Spatial Resolution

Hen ovary tissue was analyzed using the beam expander optic design to perform high spatial resolution IR-MALDESI MSI. A region on the edge of a hen ovarian tissue section was analyzed using a 2 mm iris diameter and two laser pulses, corresponding to the smallest ablation diameter reliably achievable with this optical system. Five regions of interest (ROIs) 41 x 11 voxels in size were analyzed using decreasing step size of the stage in an oversampling mode with 100, 75, 50, 40, and 30 µm step sizes. After IR-MALDESI MSI, the tissue was stained using the tissue staining protocol described above. Additionally, a hen ovary section was analyzed with three paired conditions: 2 mm iris diameter with 50 µm raster step size, 4 mm iris diameter with 75 µm raster step size, and 12 mm iris diameter with 100 µm raster step size, resulting in minimal undersampling of the tissue. A custom birefringent optical attenuator was used to maintain a constant 1.2 J/cm2 fluence for each iris setting.

7.3 Results and Discussion

7.3.1 Multi-element Optical System

The optical systems presented in Figure 7-1 were easily installed and aligned using a 1” tube lens system (Thorlabs, Newton, NJ, USA) to secure all lens components. The inclusion of two mirrors on kinematic mounts on the optical breadboard allowed precise adjustment of the laser path, which was needed to keep the laser beam centered throughout the multi-lens system. The total path length for both optical systems was approximately 1 meter. The output beam profile of the OPO laser was measured using a Pyrocam III HR (Ophir-Spiricon, North Logan, UT, USA) beam profiling camera as seen in Figure 7-5. The beam profile of the laser

156

output was not a perfect Gaussian, which effects the minimum spot size achievable due to the dependence on M2. To this end, a variable iris was incorporated into the multi-lens system to block the low-energy fringe of the laser output and thereby improve the minimum focal point. A 3.75x magnification Galilean beam expander was constructed from plano-concave (f = -40 mm) and plano-convex (f = +150 mm) lenses placed such that their focal points were coincident. The expanded beam (approx. 20 mm diameter) was incident on a 1” ZnSe aspherical lens (f = +50 mm), whose shape corrects spherical aberration. A quartz coverslip was placed immediately after the last focusing lens to protect the lens from ablated sample debris.

Figure 7-5. Beam profile of the opolette OPO IR laser at 2940 nm measured using a Pyrocam IIIHR profiling camera. Circles represent the energy profile transmitted at each iris setting; 2 mm (black), 4 mm (white), and 12 mm (gold). A 12 mm iris setting is effectively 6.8 mm diameter due to beam clipping of the 1” optics of the 3.75x beam expander and absorption of laser energy in the lens tube wall.

157

7.3.2 Visualization of Laser Beam Caustic

The laser beam caustic was visualized by placing laser burn paper at various heights relative to the focal point. Although this limited the visualization to regions where there was sufficient fluence to ablate the laser burn paper, it was found to be acceptable to obtain a profile of the laser focus. It is likely that the ablation spots on the burn paper do not exactly reflect the laser beam width where the intensity drops to 1/e2, the common definition of laser beam width; however, the method is sufficient for comparison between two optical designs, as presented here. The laser ablation for the single spherical focusing lens was recorded over a 10.16 mm distance shown in Figure 7-2. MATLAB was used to manually fit ellipses (blue) to the laser ablation spots and major and minor axes were recorded with their corresponding z height. The laser ablation spots for the multi-lens system are shown in Figure 7-3. These ellipses were then plotted to give the 3 projections, XY, XZ, and YZ, seen in Figure 7-4. The two XY projections show the observed laser ablation spots. The beam expander design gave much more circular ablation spots than the single lens design, a desirable quality of a laser used for MSI. A circular laser ablation allows MS images to be acquired with the same spatial resolution in both the X and Y dimensions, even in the undersampling mode. The reduced ablation area of the beam expander design resulted in a much higher fluence, as indicated by the heatmap. Laser fluence has a direct impact on IR-MALDESI MSI data quality. First, the laser system must have sufficiently high fluence to ablate or desorb the sample and matrix. Second, at energies above the ablation threshold, the fluence will have an impact on the ablation dynamics and overlap with the electrospray [22]. Third, the ablation fluence determines the number of laser pulses required to completely ablate through the sample [21], which is a requirement for quantitative IR-MALDESI analyses [28]. By the above points, the higher fluence of the multi-element design indicated it would successfully ablate tissue material, possibly with fewer pulses, than the single optic design. Optimizing the laser fluence to a wide variety of sample types for

158

imaging or direct analysis could be accomplished by attenuation, using a polarization based attenuation and an iris to balance the total energy and desired spot size. To better visualize the laser beam caustic, a 3D surface was constructed from the stacked ellipses. The 3D surface is representative of the true laser focus as it is not composed of discrete sampling points, rather a continuum of ablation spots based on interpolation of the data collected. To aid in comparing the two beam caustics, the axes and color maps are presented on the same scale. This emphasizes the reduction in the ablation diameter along with reduction in the depth of focus (Figure 7-4). The equation for a real laser beam propagation (equation 2) was fit to the beam profile in the XZ and YZ projections. Figure 7-6 displays the laser profile for both projections along with the equation fit with parameters solved. The residuals were plotted for each fit to estimate the goodness of fit of the model to the data. The fit residuals for the single optic (Figure 7-6A) indicated that the data deviated from the theoretical fit moderately when the laser spot size was far from the focal point, whereas this was not the case when the beam expander was used (Figure 7-6B). One explanation for this discrepancy may be that the single optic was situated further away from the focal point to fully cover the laser beam caustic. As the laser ablation spots at the furthest points (XZ/YZ distance) approached the ablation threshold of the laser burn paper and were larger than the image size of the 5x objective used to view it, it could lead to an inaccurate estimation of the laser diameter at these points. Linear regression on the last three points of each projection was used to estimate the far field divergence of the laser beam. The model was used to estimate the half angle divergence of both laser optic designs.

159

Figure 7-6. The equation for a real laser beam was fit to the beam profile in the XZ and YZ planes using a generalized reduced gradient nonlinear regression for both (a) single optic, and (b) beam expander. Residuals of each fit are presented to the right. Linear regression was used to estimate the far field beam divergence from the last three points of each curve. Using the model of the propagation of a real laser beam, important laser parameters describing the laser focus were defined to quantitatively compare the two optical systems. Table 7-1 summarizes some of the most important parameters for laser ablation in IR- MALDESI MSI using the single focusing element or optical design with an iris set to 12 mm, 3.75x beam expander and aspherical lens. M2 is the beam propagation factor and is a commonly used metric of the beam quality. A diffraction limited, perfectly Gaussian beam would have an M2 = 1. The comparison of the multi-lens beam expander designs shows an improvement in M2 for both X and Y dimensions over the single optic, from 27.0 to 15.4 and 16.9 to 14.7, respectively. The minimum beam waist, Wo, is the beam radius in micrometers at its minimum size. The value is a representation of the laser focus minimum spot size. The multi-element optical system improved Wo on laser burn paper by a factor 2.6 and 1.9 for the X and Y dimensions. This reduction in both dimensions is particularly important for microprobe MSI, where the sample is rastered in the XY plane, to give greater spatial resolution. The Rayleigh length of the laser focus is defined as the distance along the propagation direction where the

160

cross-sectional area is doubled. This corresponds to where the beam waist is 2 Wo. The depth of focus, or confocal parameter (b), is twice the Rayleigh length centered about the minimum. IR-MALDESI MSI with the single focusing lens has an average depth of focus of 2.78 mm, which made the technique quite impervious to minor changes in sample height on the Peltier stage. The beam expander design has a significantly smaller depth of focus (0.78 mm). While this may require more careful focusing of the optic for the smallest ablation diameters in MSI use, it is, importantly, sufficiently greater than the sample thickness (10 µm). Related to the depth of focus is the half angle divergence. This value describes the angle of light being focused and is a direct result of the size of the incoming diameter and the distance to the focusing lens. Like the minimum beam waist, the half angle divergence is influenced by the beam quality propagation factor, M2. The calculated half angle divergence for the beam expander lens system is 3-4º larger than the single focusing lens and is most likely the result of a large input diameter at the final focusing lens.

Table 7-1. Defining beam characteristics.

Single lens Beam expander

Parameter X Y X Y

M2, beam quality 27.0 16.9 15.4 14.7 W beam waist o, 197.0 140.3 74.7 73.1 (µm) b, depth of focus (mm) 3.07 2.49 0.78 0.75 θ, Half angle 7.9º 6.8º 11.3º 11.1º divergence

161

7.3.3 Ablation Diameter on Mouse Liver Tissue

To assess the efficacy of the multi-element optical system for MSI applications, it is necessary to move to tissue, which is a more challenging substrate to ablate and introduces the use of an ice matrix. A combination of iris diameters and number of laser pulses (n = 1-5) was used to define the tissue ablation diameter of the beam expander optical design. The 2.0 and 4.0 mm iris only let the most intense laser energy through the optical system, while the 12 mm iris (max) would let nearly all of the laser beam into the system. With the 3.75x beam expander in use, only the center 6.8 mm of the laser output beam would travel through the 1” lens tube system housing the optics without being absorbed by the tubing wall. All combinations of iris and number of pulses were able to completely ablate through the 10 µm thick mouse liver tissue to give a clear circular ablation spot, except for the 2.0 mm iris and 1 laser pulse (Figure 7-7A). The laser ablation areas were calculated by fitting circles to the ablation spots on tissue using the “imfindcircles” algorithm in MATLAB. The reproducibility of ablation spots is represented by the 95% confidence interval in the mean ablation diameter shown in Figure 7-7B, indicating that the ablation diameters are highly reproducible. Most of the tissue was ablated in the first 2 laser pulses, as shown in the marginal increase of ablation diameter in pulses 3-5. In a balance of minimum spot size with complete ablation and shortest ion injection time, the 2 laser pulse setting was selected for evaluation of the multi-element optical system on IR-MALDESI MSI.

162

Figure 7-7. Measurement of ablation diameters on mouse liver tissue using ice as a matrix while varying iris diameter and number of pulses. (A) Stained tissue images after ablation by an IR laser with ablation spots fitted with circles. (B) Average tissue ablation diameters for the iris diameters and number of pulses.

163

7.3.4 High Spatial Resolution IR-MALDESI MSI

The lowest ablation diameter setting, 2.0 mm iris and 2 laser pulses, was chosen to investigate the smallest raster step size that could be used while still generating useful MSI data. Using this configuration, data were collected from the edge of a structurally complex hen ovary tissue at 100, 75, 50, 40, and 30 µm step sizes. The step sizes less than 50 µm represent analysis in an oversampling mode, where the area of tissue being ablated is smaller than the laser ablation diameter. However, as seen in Figure 7-8, the ion abundance of cholesterol (m/z 369.3516 ± 2.5 ppm) at the 40 and 30 µm step sizes neared the limit of detection of the IR- MALDESI source coupled to a Q Exactive™ Plus mass spectrometer, and the detection frequency was not optimal. The MS peak of cholesterol for each raster step size is displayed on the right of Figure 7-8. The ion abundance and signal-to-noise ratio (S/N) is significantly reduced for 40 and 30 µm step sizes compared to the 100, 75, and 50 µm step size. Cholesterol was chosen as it was the most abundant ion detected by IR-MALDESI and, therefore, would require the least volume of ablated tissue to be detected. Methods to improve ion abundance including reoptimization of the IR-MALDESI source geometry to improve ablation and electrospray plume overlap [21], the use of a higher repetition rate laser to reduce ion loss in the MS C-trap , and ionization agent optimization [28, 30, 31] will be the focus of future work.

164

+ Figure 7-8. MSI of a lipid (cholesterol [M-H2O+H] , m/z 369.3516 ± 2.5 ppm) overlaid on the optical image of the tissue. Regions of interest were analyzed with a range of raster step sizes to demonstrate + MSI at high spatial resolution. The average cholesterol [M-H2O+H] peak for each raster step size is displayed with corresponding S/N values. The desired ablation diameter was achieved through adjustment of the iris and the rotation of a polarization-based optical attenuator placed in the beam path. A hen ovary tissue was analyzed at 50, 75, and 100 μm spatial resolution using 2.0 mm, 4.0 mm and 12.0 mm iris diameters, respectively. Because more energy enters the optical system with larger iris diameters, a custom polarization-based attenuator was placed in the beam path to maintain a constant fluence (1.2 J/cm2) between the spatial resolutions of the experiment. Figure 7-9A shows the ion distribution map of cholesterol for 3 regions of the same hen ovary tissue section analyzed at different spatial resolutions. A serial tissue section was stained and an optical image was included in Figure 7-9B to visually compare tissue morphology to the MS image. The distribution of cholesterol showed concentric rings of high abundance in the developing follicles on the left. A full scan mass spectrum is displayed in Figure 7-9C showing numerous

165

endogenous lipids and metabolites in the hen ovary tissue analyzed at 50 μm resolution. Tissue specific peaks were generated by averaging an on-tissue region and subtraction of an off-tissue region to minimize peak contribution by ambient contaminants. Figure 7-10 demonstrates that mass measurement accuracy (MMA) was better than ± 2.5 ppm for all pixels in the imaging experiment.

Figure 7-9. (A) IR-MALDESI MSI of cholesterol [M-H2O+H]+, m/z 369.3516 ± 2.5 ppm in hen ovary tissue at 50, 75, and 100 μm spatial resolution. (B) An optical image of a stained serial section showing the complex morphology of the ovary tissue. (C) An averaged mass spectrum for the 50 μm spatial resolution experiment.

166

Figure 7-10. Mass measurement accuracy (MMA) of cholesterol MSI presented in Figure 7-9. RAW files were loaded into RawMeat (Vast Scientific) and exported using MZ monitor. MMA was calculated for every cholesterol peak and a histogram was created using 0.1 ppm bin size.

7.4 Conclusion

A multi-element optical design composed of an adjustable iris, beam expander, and aspherical focusing lens was implemented for high resolution IR-MALDESI MSI. The design was compared to the previously used single lens design, and demonstrated improved beam 2 quality factor (M ) and a smaller minimum waist, Wo, two of the most important laser ablation

167

parameters in the context of MSI. A hen ovarian tissue was analyzed using the multi-element system, yielding high quality MS images with 50 micron spatial resolution.

7.5 Acknowledgments

The authors gratefully acknowledge financial support from the National Institutes of Health (R01GM087964), the W.M. Keck Foundation, and North Carolina State University.

7.6 References

1. Amstalden van Hove, E.R., Smith, D.F., Heeren, R.M.: A concise review of mass spectrometry imaging. Journal of chromatography. A. 1217, 3946-3954 (2010)

2. Ràfols, P., Vilalta, D., Brezmes, J., Cañellas, N., del Castillo, E., Yanes, O., Ramírez, N., Correig, X.: Signal preprocessing, multivariate analysis and software tools for MA(LDI)-TOF mass spectrometry imaging for biological applications. Mass spectrometry reviews. (2016). doi:10.1002/mas.21527

3. Boughton, B.A., Thinagaran, D., Sarabia, D., Bacic, A., Roessner, U.: Mass spectrometry imaging for plant biology: a review. Phytochemistry Reviews. 15, 445- 488 (2016)

4. Gessel, M.M., Norris, J.L., Caprioli, R.M.: MALDI imaging mass spectrometry: Spatial molecular analysis to enable a new age of discovery. Journal of proteomics. 107C, 71-82 (2014)

5. Gode, D., Volmer, D.A.: Lipid imaging by mass spectrometry - a review. The Analyst. 138, 1289-1315 (2013)

6. Nilsson, A., Goodwin, R.J., Shariatgorji, M., Vallianatou, T., Webborn, P.J., Andrén, P.E.: Mass spectrometry imaging in drug development. Analytical chemistry. 87, 1437- 1455 (2015)

7. Murray, K.K., Seneviratne, C.A., Ghorai, S.: High resolution laser mass spectrometry bioimaging. Methods. 104, 118-126 (2016)

8. Riedl, M.J. Optical design fundamentals for infrared systems. SPIE press, (2001)

168

9. Alkelly, A.A.: Spot size and radial intensity distribution of focused Gaussian beams in spherical and non-spherical aberration lenses. Optics communications. 277, 397-405 (2007)

10. Zavalin, A., Yang, J., Caprioli, R.: Laser Beam Filtration for High Spatial Resolution MALDI Imaging Mass Spectrometry. Journal of the American Society for Mass Spectrometry. 24, 1153-1156 (2013)

11. Jacobson, R.S., Thurston, R.L., Shrestha, B., Vertes, A.: In Situ Analysis of Small Populations of Adherent Mammalian Cells Using Laser Ablation Electrospray Ionization Mass Spectrometry in Transmission Geometry. Analytical chemistry. 87, 12130-12136 (2015)

12. Römpp, A., Schäfer, K.C., Guenther, S., Wang, Z., Köstler, M., Leisner, A., Paschke, C., Schramm, T., Spengler, B.: High-resolution atmospheric pressure infrared laser desorption/ionization mass spectrometry imaging of biological tissue. Analytical and bioanalytical chemistry. 405, 6959-6968 (2013)

13. Hieta, J.-P., Vaikkinen, A., Auno, S., Räikkönen, H., Haapala, M., Scotti, G., Kopra, J., Piepponen, P., Kauppila, T.J.: A Simple Method for Improving the Spatial Resolution in Infrared Laser Ablation Mass Spectrometry Imaging. Journal of the American Society for Mass Spectrometry. (2017). doi:10.1007/s13361-016-1578-7

14. Jurchen, J.C., Rubakhin, S.S., Sweedler, J.V.: MALDI-MS imaging of features smaller than the size of the laser beam. Journal of the American Society for Mass Spectrometry. 16, 1654-1659 (2005)

15. Nazari, M., Muddiman, D.: Cellular-level mass spectrometry imaging using infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) by oversampling. Analytical and bioanalytical chemistry. 407, 2265-2271 (2015)

16. Luxembourg, S.L., McDonnell, L.A., Mize, T.H., Heeren, R.M.A.: Infrared Mass Spectrometric Imaging below the Diffraction Limit. Journal of proteome research. 4, 671-673 (2005)

17. Soltwisch, J., Göritz, G., Jungmann, J.H., Kiss, A., Smith, D.F., Ellis, S.R., Heeren, R.M.A.: MALDI Mass Spectrometry Imaging in Microscope Mode with Infrared Lasers: Bypassing the Diffraction Limits. Analytical chemistry. 86, 321-325 (2014)

18. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Generation and detection of multiply-charged peptides and proteins by matrix-assisted laser desorption electrospray

169

ionization (MALDESI) Fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry. 17, 1712-1716 (2006)

19. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) imaging source coupled to a FT-ICR mass spectrometer. Journal of the American Society for Mass Spectrometry. 24, 92-100 (2013)

20. Bokhart, M.T., Muddiman, D.C.: Infrared matrix-assisted laser desorption electrospray ionization mass spectrometry imaging analysis of biospecimens. The Analyst. 141, 5236-5245 (2016)

22. Rosen, E., Bokhart, M., Ghashghaei, H.T., Muddiman, D.: Influence of Desorption Conditions on Analyte Sensitivity and Internal Energy in Discrete Tissue or Whole Body Imaging by IR-MALDESI. Journal of the American Society for Mass Spectrometry. 26, 899-910 (2015)

23. Kessner, D., Chambers, M., Burke, R., Agus, D., Mallick, P.: ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics. 24, 2534- 2536 (2008)

24. Schramm, T., Hester, A., Klinkert, I., Both, J.P., Heeren, R.M., Brunelle, A., Laprevote, O., Desbenoit, N., Robbe, M.F., Stoeckli, M., Spengler, B., Rompp, A.: imzML--a common data format for the flexible exchange and processing of mass spectrometry imaging data. Journal of proteomics. 75, 5106-5110 (2012)

25. Race, A.M., Styles, I.B., Bunch, J.: Inclusive sharing of mass spectrometry imaging data requires a converter for all. Journal of proteomics. 75, 5111-5112 (2012)

26. Robichaud, G., Garrard, K.P., Barry, J.A., Muddiman, D.C.: MSiReader: an open- source interface to view and analyze high resolving power MS imaging files on Matlab platform. Journal of the American Society for Mass Spectrometry. 24, 718-721 (2013)

27. Keicher, D.M.: Laser beam characterization results for a high-power cw Nd: YAG laser. Photonics West'95. 162-171 (1995)

170

28. Bokhart, M., Rosen, E., Thompson, C., Sykes, C., Kashuba, A.M., Muddiman, D.: Quantitative mass spectrometry imaging of emtricitabine in cervical tissue model using infrared matrix-assisted laser desorption electrospray ionization. Analytical and bioanalytical chemistry. 407, 2073-2084 (2015)

29. Rosen, E.P., Bokhart, M.T., Nazari, M., Muddiman, D.C.: Influence of C-Trap Ion Accumulation Time on the Detectability of Analytes in IR-MALDESI MSI. Analytical chemistry. 87, 10483-10490 (2015)

30. Meier, F., Garrard, K.P., Muddiman, D.C.: Silver dopants for targeted and untargeted direct analysis of unsaturated lipids via infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). Rapid Communications in Mass Spectrometry. 28, 2461-2470 (2014)

31. Nazari, M., Muddiman, D.C.: Polarity switching mass spectrometry imaging of healthy and cancerous hen ovarian tissue sections by infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). The Analyst. 141, 595-605 (2016)

171

8 MSiReader v1.0: Evolving Open-Source Mass Spectrometry Imaging Software for Targeted and Untargeted Analyses

The following work was reprinted from the recently accepted: Bokhart, M. T.†, Nazari, M. †, Garrard, K. P., & Muddiman, D. C. (2017). MSiReader v1.0: Evolving Open-Source Mass Spectrometry Imaging Software for Targeted and Untargeted Analyses. Journal of The American Society for Mass Spectrometry. †Authors contributed equally.

8.1 Introduction

Mass spectrometry imaging (MSI) is a rapidly growing research area where mass spectrometric analysis is performed in a spatially resolved manner. MSI data is most commonly represented as heat maps based on ion abundance mapped on the Cartesian coordinates of the analyzed sample. This allows for the generation of images for each spectrally resolved m/z value across the analyzed region. A color scale bar is used to represent the ion abundance. A wide variety of ionization sources and mass analyzers have been used to create MSI datasets. Most commercial software available for analysis of MSI data are vendor-specific and proprietary. However, in recent years there has been increased support for data sharing through a universal data format imzML [1] and the introduction of open-source and/or free data analysis software such as MSiReader [2], Cardinal [3], BioMAP [4], msIQuant [5], SpectralAnalysis [6], and METASPACE [7]. Each software has unique advantages and disadvantages. Recently, a new version of msIQuant [8] was introduced with the ability to analyze imzML data with software tools to handle MSI quantification data and multimodal imaging capabilities. Cardinal and SpectralAnalysis are software packages that allow advanced statistical analysis of MSI data sets in an untargeted fashion [3, 6]. The recent developments of software packages available for analyzing and processing MSI data were outlined in a recent review article [9]. MSiReader was introduced in 2013 as an open-source, vendor neutral MSI data analysis software written in MATLAB® [2]. At the time of publication, it was the only software capable of analyzing high resolving power mass spectrometry imaging data without data

172

compression. The maximum file size was and still is only limited by the amount of RAM available. MSiReader was initially designed to process and analyze imaging data generated in our laboratory using the infrared matrix-assisted laser desorption electrospray ionization (IR- MALDESI) imaging source [10, 11]; however, since then, it has become an essential tool for research in our laboratory as well as others. At the time of the submission of this manuscript, the original manuscript has been cited over 119 times for diverse applications in proteomics [12] , plant and animal metabolomics [13-15], and forensics [16]. Community feedback has led to the incorporation of numerous improvements in the user interface that enhance workflow, as well as tools for spectral export, image export, data binning, colocalization, normalization methods, polarity switching and filtering, as well as quantification. The evolution of MSI in the past several years has required the constant development and improvement of software packages to facilitate the analysis of a wide variety of MSI data. MSiReader has evolved to include tools that allow complex and customized data analysis workflows to be incorporated into the software, allowing unprecedented analysis of MSI data. In this manuscript, the authors present v1.0 MSiReader, with improved stability, faster data loading and processing, and newly incorporated features for the analysis of MSI datasets.

8.2 Experimental

All experiments mentioned in this manuscript were performed in accordance with local ethical guidelines. Human cervical tissues were obtained from the University of North Carolina Tissue Procurement Facility through UNC IRB #09-0921, and written informed consent was obtained from all patients. Two-day-old whole neonatal mouse pups were obtained from NCSU Department of Molecular Biomedical Sciences. Hen ovarian tissues were acquired from commercial egg laying hens in the NCSU Department of Poultry Science. All husbandry practices were approved by North Carolina State University Institutional Animal Care and Use Committee (IACUC).

173

All imaging experiments were performed in our laboratory using infrared matrix- assisted laser desorption electrospray ionization (IR-MALDESI) coupled to high resolving power mass spectrometers. The details of IR-MALDESI source design and steps involved in imaging experiments are described elsewhere in detail [10, 17, 18]. Quantitative MSI and whole-body MSI were performed using a Q Exactive mass spectrometer (Thermo Scientific, Bremen, Germany) as described by Bokhart et al. [19] and Rosen et al. [20], respectively. Polarity switching IR-MALDESI MSI was performed using a Q Exactive Plus mass spectrometer (Thermo Scientific, Bremen, Germany) as outlined by Nazari and Muddiman [21]. The .RAW files generated by the Thermo instruments were first converted to mzML format using the msConvert tool from ProteoWizard [22], and then converted to imzML using the imzML converter [23]. The imzML files were subsequently loaded into MSiReader v1.00 for visualization and analysis.

8.3 Results and Discussions

8.3.1 Overview of MSiReader Features

MSiReader was developed because of the need to process high resolving power MSI data sets without compromising the integrity of the data with software-imposed limitations on dynamic range or resolution. Initial efforts were directed at providing an efficient workflow for untargeted analysis using new instrumentation. The first version could simply load an mzXML data set and display a heatmap image of ion abundance for any m/z value and window size chosen by the user. It also provided a tool to plot the spectrum for a single pixel and another to find putative peaks based on a parabolic fitting and thresholding algorithm. Since then, many additional capabilities have been implemented and performance has been greatly improved. Some examples of new capabilities of the software include: saving and reloading an entire session as a MATLAB binary (.MAT) file; peak picking based on differences between

174

reference and interrogated regions of interest; four channel colocalization plots; m/z abundance rank plots; several normalization options; numerous ways to export, save and reuse data; exporting binned abundance data into a continuous m/z range in preparation for multivariate analysis; batch generation of a images for each m/z value in a list; extensive customization options via a preferences file; and an application programming interface (API) that eases the task of prototyping new ideas and implementing custom scripts. The focus of this paper is on the performance improvements made since the initial release and five new and useful tools for: loading multiple data sets at once, absolute quantification, polarity switching, image overlay, and mass measurement accuracy QA plots.

8.3.2 Loading and Processing of Imaging Data Sets

In addition to the preferred imzML format, MSiReader can load data stored as mzXML, IMG (Analyze 7.5), and Bruker ASCII files. All features of MSiReader are supported for each of the formats and internally the spectra are stored as sparse, processed data. That is, the spectra for each pixel are stored as an independent sequence of m/z, abundance pairs using double precision floating-point values following the IEEE 754 standard. As each scan is loaded, it can be filtered by any combination of m/z range, abundance threshold, spatial location or polarity. This can have a significant impact on performance, especially removing zero and very low abundance values from the data cube from a high resolving power instrument. Additionally, a smart abundance filter can be selected whereby any consecutive sequence with more than two values below the threshold is replaced by threshold (or zero) values at each end, thereby preserving the appearance of the spectral plot for that pixel. Table 8-1 shows the dramatic improvement in data set loading time for the supported formats that have been made since MSiReader release 0.03 in July of 2013. Even though load time is dominated by the speed of the file storage medium and the amount of RAM available, significant improvements have been made. All tests were done on the same computer, a Dell Precision Tower 5810 (Xeon ES-1607 v3, 3.1 GHz Processor, 64 GB RAM, 256 GB SSD and

175

4 TB HD) in a newly started MATLAB 2017a session. All code was stored on the SSD and the data sets were on the hard drive. The measured time required for various operations in MSiReader are obtained using the built-in MATLAB tic and toc functions and displayed in the MATLAB command window.

Table 8-1. MSiReader dataset loading time improvement Data Set File Size Number Memory Release Release % Decrease Format (GB) of Scans (GB) 0.03 (sec) 1.00Q (sec) imzML(1) 5.87 7,426 3.97 140 67 52 imzML(2) 3.93 10,608 2.55 140 80 43 imzML(3) 36.40 32,060 25.17 936 706 25 mzXML 1.87 5,551 0.931 109 77 29 ASCII 0.227 4,048 0.064 316 8 97 IMG 0.530 7,275 3.55 73 30 59 IMG 3.23 69,680 27.84 333 80 76

The performance of other tasks in MSiReader has also been improved even though new features (hence new code) have been added. Table 8-2 compares several common operations: updating the heatmap display after changing the m/z value the normalization option or the window tolerance; exporting raw spectra for the scans in a region of interest (ROI); peak picking based on differences between a reference and an interrogated ROI; and batch export of heatmap images for a list of m/z values into a folder (png file format). The heatmap update and batch export tasks were done using the imzML(2) file from Table 1 and the pixel export and peak picking tasks were done using the imzML(3) file. The 42 scan ROI was the same for both export operations and that ROI was also the interrogated region for the peak picking operation. As shown in the table, significant improvements have been made for these tasks, greatly enhancing the user experience.

176

Table 8-2. MSiReader operation time improvement. Release Release Operation % Decrease 0.03 (sec) 1.00Q (sec) Heatmap Update 13 1 92 10,608 scans; 3.93 GB data set Spectrum Export to Excel 365 21 94 42 scan ROI; 50 MB output file Spectrum Export to Text 83 7 92 42 scan ROI; 51 MB output file Peak Picking 91 14 85 ROIs: 42 scan reference vs 42 scan interrogated Batch Image Export 681 59 91 10608 scans; 3.93 GB data set; 57 m/z values

8.3.3 Loading Multiple Data Sets

A commonly requested feature that is now available in MSiReader is the ability to load and process more than one data set simultaneously. Previously, the only way to accomplish this was to merge all the files into one file, which could be a tedious and time-consuming process, especially for files with different spatial dimensions. Now MSiReader can load all the imzML or mzXML files in a folder on-the-fly and distribute the images into a tiled mosaic defined by the user. The files need not have the same dimensions and smaller files are padded with empty scans such that all images in a row have the same number of scan rows and all images in a column have the same number of scan columns. Each image is centered in its tile and surrounded by at least one empty row and column. After selecting to load a folder, the user is prompted to define the number of tile rows and columns (Figure 8-1). The files are then loaded in alphanumeric order by name or they may be given a sequence number suffix. The source of each pixel (file and scan number) is retained throughout all subsequent processing. All ROI, peak picking analysis, export and visualization tools are supported across the image mosaic. For example, the interrogated and reference ROIs for peak picking may be in different image tiles and thus from different data sets.

177

Figure 8-1. Loading multiple imaging data sets.

8.3.4 Absolute Quantification in MSiReader for Mass Spectrometry Imaging

Quantification in mass spectrometry imaging is a challenging experiment requiring consideration to many factors that may influence ion abundance during the experiment. This requires the incorporation of a calibration curve and may involve the incorporation of a normalization compound. Visualization and analysis software must support common and advanced data analysis methods. Most MSI software packages allow normalization (ratio) of images to a specific compound or total ion current (TIC), thereby accounting for per-pixel ionization efficiencies. This normalization step has proven itself essential for quantitative mass spectrometry imaging [19]. Incorporation of a spatial calibration curve is essential to any quantitative MSI experiment to relate ion abundance to an absolute amount of a compound. Integration of

178

standards with tissues has been performed in various ways [24] but all require the selection of an ROI for each calibration point in addition to the tissue area to be quantified. The newly introduced MSiQuantification tool (Figure 8-2) in MSiReader allows the user to select up to 10 calibration ROIs and define the concentration of each. Once the tissue ROI and at least 3 calibration points have been selected, the user can generate a linear least squares regression for the calibration curve and calculate the concentration of analyte across the selected tissue ROI. After the calibration curve has been defined, the tissue image may be represented as absolute concentration based on the calibration curve. To preserve data analyses and allow additional analyses, the quantification process parameters and ROIs can be saved into a binary file and reloaded later to repeat the analysis using the same exact parameters.

Figure 8-2. MSiQuantification tool for absolute quantification MSI experiments. The tool allows ROIs to be drawn for up to 10 calibration spots (green lines) and an area of the tissue to be quantified (light blue line).

179

8.3.5 Polarity Switching Mass Spectrometry Imaging Data Processing

Lipids and metabolites are some of the most commonly analyzed biomolecules using MSI. Alterations in lipidome and metabolome have been linked to several diseases spanning from hypertension, diabetes, Alzheimer’s diseases, and cancer [25, 26]; therefore, assessing the distribution and abundance of lipids and metabolites in biological tissue specimens is of paramount importance in order to provide insight into the onset and progression of disease, as well as discovery of potential biomarkers. Due to their remarkable structural diversity, lipids and metabolites exhibit preferential ionization efficiencies. Generally, metabolites that are detected in high ion abundance as positive ions are poorly detected (or not detected at all) as negative ions, and vice versa. Therefore, to comprehensively characterize the metabolome, analysis of tissue specimens in both polarities is required. One could perform two MSI experiments, one in each polarity, to obtain extensive metabolome coverage; however, this approach would be costly in terms of time, sample, and materials used. Alternatively, polarity switching MSI can be used to obtain the same biochemical information, while shortening the analysis time and reducing sample consumption by almost half. Indeed, polarity switching MSI has become a valuable analysis mode in recent years [21, 27-30]. The recent surge in reports of polarity switching MSI calls for new software options for analyzing these datasets. One of the new features introduced in MSiReader is the ability to parse data using the polarity in which the spectra were acquired. There are different approaches to acquiring polarity switching spectra such as alternating polarities from pixel to pixel [21, 30], from one line to the next[29], or even using a “spiral” step[28] to acquire spectra in positive and negative modes. MSiReader can analyze these patterns while parsing the data and load the desired polarity. Furthermore, it can be used to filter out “equilibration spectra” that might have been collected in between polarity switching events during data acquisition. MSiReader implements four polarity switching patterns: [+--+], [-++-], [+-] and [-+]. For the 4-tuple patterns, either the odd (1,3) or the even (2,4) scans are equilibrium scans used to allow the electrospray to

180

stabilize when the polarity is changed from one pixel to the next with no advancement of the sample raster stage. An example of a polarity switching MSI method and equilibration spectra was recently demonstrated by our group [21] and a representative image showing the loading process in MSiReader is shown in Figure 8-3. Additional patterns for data analysis can be readily incorporated into the software using the existing algorithm for reading the file header and scan polarities. In addition, an optional filter parameter allows the user to select the positive image, the negative image, or both images for subsequent processing. If both polarities are retained, polarity is selected when the MSiPeakfinder or MSiSpectrum tools are launched. The filter and polarity selection can also be used without switching on a file with an arbitrary distribution of polarities – for example, a tiled composite imzML image of multiple tissue samples.

Figure 8-3. Polarity switching and polarity filtering are implemented for the mzXML and imzML file formats as the data is loaded. In this case, the polarity pattern selected is [+--+, odd] and only the (+) polarity scans are retained.

181

8.3.6 Image Overlay Tool

Mass spectrometry imaging allows the spatial distribution of ions to be depicted in a two-dimensional array. The overlap of two complimentary imaging techniques allows researchers to gain a deeper insight into the significance of spatial localizations. In combining MSI images with other imaging modalities, the molecular distributions gain meaning in their context to the tissue morphology. MSiImage tool has been implemented in MSiReader to import, scale, and align an image (e.g. optical, fluorescence) on an MSiReader heatmap. This includes an adjustable transparency of the image along with resizing, cropping, and rotation. Any file format supported by MATLAB can be imported allowing connection between two analysis methods to be compared. Alignment of classical biology techniques such as staining by haemotoxylin and eosin (H&E) or immunohistochemistry with a MSI data set allows for histology-directed molecular analysis of tissues [31]. An example of an image overlay with the MSI data is demonstrated in Figure 8-4, where the spatial distribution of m/z 367.3363, putatively assigned + + to desmosterol [M+H -H2O] , is displayed on a semi-transparent optical image of the whole- body section. In addition, the MSiImage function can be used in combination with the colocalization tool in MSiReader to show the distribution of 3 distinct ions overlaid with another image.

182

Figure 8-4. MSiImage tool for overlaying an optical image with the ion image of putatively assigned + desmosterol (m/z 367.3363, [M+H-H2O] ).

8.3.7 Mass Measurement Accuracy (MMA) Heatmap and Histograms

The importance of maintaining high mass measurement accuracy (MMA) throughout MS-based experiments is well-known. This point is more vital in MSI analyses since there are no chromatographic separations and deviations in MMA could lead to generation of incorrect

183

ion images and/or identification of incorrect analytes in untargeted studies. In our laboratory, the MMA for an analyte observed in MSI analyses were calculated by exporting the centroid m/z value for the peak at each scan over the user-defined ROI (often the tissue section) to an Excel sheet using MSiReader and then calculating the MMA for each scan. Another option could be to use a separate program to export the centroid m/z from the raw file. These processes are laborious and time consuming, and do not provide information about any change in the MMA as a function of spatial location. To circumvent these issues and speed up the analysis process, we have introduced new quality assurance features in MSiReader that allow users to generate heatmaps of MMA for the scans within the user-identified ROI or all scans in the file. This feature uses the m/z value inputted by the user as the “true” value and picks the most abundant peak in the m/z window as the “experimental” value to calculate the MMA at each scan. The MMA for each scan is then plotted in a heatmap, where possible deviations at specific locations within the ROI can be easily observed (Figure 8-5). In addition, a histogram of the MMAs, as a function of number of scans or the abundance at each scan, can also be exported to a separate figure (along with the MMA heatmap or separately). This tool can serve as a facile method to check the MMA of the analyte(s) of interest throughout the imaging experiment. If needed, the MMA values, along with the scan number from which they were calculated, can be exported into an Excel sheet for further investigation.

184

Figure 8-5. Top: Screenshot of MSiReader interface showing the ion image of glutathione (m/z 306.0766, [M-H]-) in hen ovary analyzed with negative mode IR-MALDESI. The mass measurement accuracy function can be accessed by right-clicking on the image axes. Bottom: Heatmap of MMA in ppm of glutathione over the tissue region (denoted by magenta line on the ion image). Calculated MMA values for all interrogated voxels are presented in histogram form with an overlaid Gaussian fit, demonstrating normality of the MMA distribution. The dashed lines on the histogram demonstrate the ±2.5 ppm tolerance that the ion image was generated with.

185

8.4 Conclusions

Since its introduction to the MSI community in 2013, MSiReader has been downloaded and used by over 800 researchers worldwide and has been cited in more than 119 publications. Over the past four years, new algorithms have been implemented into the software, new features have been added to enhance the analysis workflows, and significant performance improvements have been made based on the feedback from the MSI community. The influence of the user community and our own research objectives continue to provide an opportunity for further development of this useful and free software tool.

8.5 Acknowledgements

The authors thank Drs. Guillaume Robichaud, Jeremy Barry, and Eli Rosen for their discussions regarding MSiReader development. Additionally, feedback from MSiReader users across the world is greatly appreciated. The authors gratefully acknowledge financial assistance received from the National Institutes of Health (R01GM087964), the W. M. Keck foundation, and North Carolina State University.

8.6 References

1. Schramm, T., Hester, A., Klinkert, I., Both, J.P., Heeren, R.M., Brunelle, A., Laprevote, O., Desbenoit, N., Robbe, M.F., Stoeckli, M., Spengler, B., Rompp, A.: imzML--a common data format for the flexible exchange and processing of mass spectrometry imaging data. Journal of proteomics. 75, 5106-5110 (2012)

2. Robichaud, G., Garrard, K.P., Barry, J.A., Muddiman, D.C.: MSiReader: an open- source interface to view and analyze high resolving power MS imaging files on Matlab platform. Journal of the American Society for Mass Spectrometry. 24, 718-721 (2013)

3. Bemis, K.D., Harry, A., Eberlin, L.S., Ferreira, C., van de Ven, S.M., Mallick, P., Stolowitz, M., Vitek, O.: Cardinal: an R package for statistical analysis of mass spectrometry-based imaging experiments. Bioinformatics. (2015)

186

4. Stoeckli, M., Staab, D., Staufenbiel, M., Wiederhold, K.-H., Signor, L.: Molecular imaging of amyloid β peptides in mouse brain sections using mass spectrometry. Analytical Biochemistry. 311, 33-39 (2002)

5. Kallback, P., Shariatgorji, M., Nilsson, A., Andren, P.E.: Novel mass spectrometry imaging software assisting labeled normalization and quantitation of drugs and neuropeptides directly in tissue sections. Journal of proteomics. 75, 4941-4951 (2012)

6. Race, A.M., Palmer, A.D., Dexter, A., Steven, R.T., Styles, I.B., Bunch, J.: SpectralAnalysis: Software for the Masses. Analytical chemistry. 88, 9451-9458 (2016)

7. Palmer, A., Phapale, P., Chernyavsky, I., Lavigne, R., Fay, D., Tarasov, A., Kovalev, V., Fuchser, J., Nikolenko, S., Pineau, C., Becker, M., Alexandrov, T.: FDR-controlled metabolite annotation for high-resolution imaging mass spectrometry. Nat Meth. 14, 57-60 (2017)

8. Källback, P., Nilsson, A., Shariatgorji, M., Andrén, P.E.: msIQuant – Quantitation Software for Mass Spectrometry Imaging Enabling Fast Access, Visualization, and Analysis of Large Data Sets. Analytical chemistry. 88, 4346-4353 (2016)

9. Ràfols, P., Vilalta, D., Brezmes, J., Cañellas, N., del Castillo, E., Yanes, O., Ramírez, N., Correig, X.: Signal preprocessing, multivariate analysis and software tools for MA(LDI)-TOF mass spectrometry imaging for biological applications. Mass spectrometry reviews. n/a-n/a (2016)

10. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) imaging source coupled to a FT-ICR mass spectrometer. Journal of the American Society for Mass Spectrometry. 24, 92-100 (2013)

11. Bokhart, M.T., Muddiman, D.C.: Infrared matrix-assisted laser desorption electrospray ionization mass spectrometry imaging analysis of biospecimens. The Analyst. 141, 5236-5245 (2016)

12. Gemperline, E., Keller, C., Jayaraman, D., Maeda, J., Sussman, M.R., Ané, J.-M., Li, L.: Examination of Endogenous Peptides in Medicago truncatula Using Mass Spectrometry Imaging. Journal of proteome research. 15, 4403-4411 (2016)

187

13. Bhandari, D.R., Wang, Q., Friedt, W., Spengler, B., Gottwald, S., Römpp, A.: High resolution mass spectrometry imaging of plant tissues: towards a plant metabolite atlas. The Analyst. 140, 7696-7709 (2015)

14. Gemperline, E., Jayaraman, D., Maeda, J., Ané, J.-M., Li, L.: Multifaceted Investigation of Metabolites During Nitrogen Fixation in Medicago via High Resolution MALDI-MS Imaging and ESI-MS. Journal of the American Society for Mass Spectrometry. 26, 149-158 (2015)

15. Liebeke, M., Strittmatter, N., Fearn, S., Morgan, A.J., Kille, P., Fuchser, J., Wallis, D., Palchykov, V., Robertson, J., Lahive, E.: Unique metabolites protect earthworms against plant polyphenols. Nature communications. 6, (2015)

16. Forbes, T.P., Sisco, E.: Chemical imaging of artificial fingerprints by desorption electro-flow focusing ionization mass spectrometry. The Analyst. 139, 2982-2985 (2014)

17. Robichaud, G., Barry, J.A., Muddiman, D.C.: IR-MALDESI mass spectrometry imaging of biological tissue sections using ice as a matrix. Journal of the American Society for Mass Spectrometry. 25, 319-328 (2014)

18. Nazari, M., Bokhart, M.T., Muddiman, D.C.: Whole-body Mass Spectrometry Imaging by Infrared Matrix-assisted Laser Desorption Electrospray Ionization (IR-MALDESI). e53942 (2016)

19. Bokhart, M., Rosen, E., Thompson, C., Sykes, C., Kashuba, A.M., Muddiman, D.: Quantitative mass spectrometry imaging of emtricitabine in cervical tissue model using infrared matrix-assisted laser desorption electrospray ionization. Analytical and bioanalytical chemistry. 407, 2073-2084 (2015)

20. Rosen, E., Bokhart, M., Ghashghaei, H.T., Muddiman, D.: Influence of Desorption Conditions on Analyte Sensitivity and Internal Energy in Discrete Tissue or Whole Body Imaging by IR-MALDESI. Journal of the American Society for Mass Spectrometry. 26, 899-910 (2015)

21. Nazari, M., Muddiman, D.C.: Polarity switching mass spectrometry imaging of healthy and cancerous hen ovarian tissue sections by infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). The Analyst. 141, 595-605 (2016)

188

22. Kessner, D., Chambers, M., Burke, R., Agus, D., Mallick, P.: ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics. 24, 2534- 2536 (2008)

23. Race, A.M., Styles, I.B., Bunch, J.: Inclusive sharing of mass spectrometry imaging data requires a converter for all. Journal of proteomics. 75, 5111-5112 (2012)

24. Porta, T., Lesur, A., Varesio, E., Hopfgartner, G.: Quantification in MALDI-MS imaging: what can we learn from MALDI-selected reaction monitoring and what can we expect for imaging? Analytical and bioanalytical chemistry. 407, 2177-2187 (2015)

25. Gowda, G.N., Zhang, S., Gu, H., Asiago, V., Shanaiah, N., Raftery, D.: Metabolomics- based methods for early disease diagnostics. Expert review of molecular diagnostics. 8, 617-633 (2008)

26. Johnson, C.H., Ivanisevic, J., Siuzdak, G.: Metabolomics: beyond biomarkers and towards mechanisms. Nature reviews Molecular cell biology. 17, 451-459 (2016)

27. Thomas, A.l., Charbonneau, J.L., Fournaise, E., Chaurand, P.: Sublimation of new matrix candidates for high spatial resolution imaging mass spectrometry of lipids: enhanced information in both positive and negative polarities after 1, 5- diaminonapthalene deposition. Analytical chemistry. 84, 2048-2054 (2012)

28. Korte, A.R., Lee, Y.J.: Multiplex mass spectrometric imaging with polarity switching for concurrent acquisition of positive and negative ion images. Journal of the American Society for Mass Spectrometry. 24, 949-955 (2013)

29. Janfelt, C., Wellner, N., Hansen, H.S., Hansen, S.H.: Displaced dual-mode imaging with desorption electrospray ionization for simultaneous mass spectrometry imaging in both polarities and with several scan modes. Journal of mass spectrometry : JMS. 48, 361-366 (2013)

30. Ellis, S.R., Cappell, J., Potočnik, N.O., Balluff, B., Hamaide, J., Van der Linden, A., Heeren, R.M.: More from less: high-throughput dual polarity lipid imaging of biological tissues. The Analyst. 141, 3832-3841 (2016)

31. Chaurand, P., Schwartz, S.A., Billheimer, D., Xu, B.J., Crecelius, A., Caprioli, R.M.: Integrating Histology and Imaging Mass Spectrometry. Analytical chemistry. 76, 1145- 1155 (2004)

189

APPENDIX

190

Appendix A

A. Infrared Matrix-Assisted Laser Desorption Electrospray Ionization Mass Spectrometry Imaging of Biospecimens

The following work was reprinted with permission from: Bokhart, M. T. & Muddiman, D. C. (2016). Infrared Matrix-Assisted Laser Desorption Electrospray Ionization Mass Spectrometry Imaging of Biospecimens. Analyst, 141, 5236-5245. DOI: 10.1039/C6AN01189F. Copyright © 2016 The Royal Society of Chemistry. The original publication may be accessed via the Internet at http://dx.doi.org/10.1039/c6an01189f

A.1 Introduction

Mass spectrometry imaging (MSI) is a rapidly evolving technique applied to a wide variety of applications from proteins [1], pharmaceuticals [2], endogenous metabolites [3], and -omic level analysis [4]. In MSI, there are two general modes of operation – microscope and microprobe [5]. MSI data collected in a point-by-point manner where the analysis is performed sequentially across the sample is referred to as microprobe mode. MSI instruments that acquire data in microscope mode desorb ions from the sample and use a spatially resolved detector. Microprobe mode analysis is the most common and is the method by which IR-MALDESI collects MSI data. By mapping m/z abundances and the positions from which they were acquired, unique ion maps of analytes can be generated, showing the spatial distribution within the sample. Heat maps depicting ion abundance generated with careful consideration of factors affecting analyte detection provide important visual representations of analyte distribution and relative concentration. The concept of spatially resolved MSI using secondary ion mass spectrometry was originally proposed by Castaing and Slodzian in the 1960s [6], but its utility for biological mapping was not realized until the mid-1990s with Spengler's report of spatial distribution of biomolecules [7] and subsequent application of matrix assisted laser desorption ionization

191

(MALDI) MSI for biomolecules by Caprioli et al. in 1997 [8]. Since then, MALDI has become the most commonly used MSI technique [9] as it provides sensitive detection for a wide variety of compounds and commercial MSI instruments are now available. However, MALDI requires the careful selection of matrix/analyte pair, matrix application protocol and typically requires analysis under vacuum. Ambient ionization techniques are desirable for MSI as they do not require the sample to be placed under high vacuum and represent analysis of biological samples closer to in vivo conditions. A number of ambient ionization techniques have been successfully applied to MSI and were the subject of a recent review [10]. Each technique has its advantages and disadvantages, allowing researcher to choose the technique best suited for the analysis. Hybrid ionization methods offer unique advantages of two or more ionization methods. Laser-based desorption methods with post ionization allow the spatial information of analytes to be retained. Techniques such as electrospray laser desorption ionization (ELDI) [11], matrix assisted laser desorption electrospray ionization (MALDESI) [12], laser ablation electrospray ionization (LAESI) [13], laser electrospray mass spectrometry (LEMS) [14], or infrared laser ablation desorption electrospray ionization (IR-LADESI) [15] combine resonant or nonresonant laser desorption of the substrate with postionization by electrospray ionization (ESI) [16]. Resonant desorption requires absorption of laser energy by a matrix. The term matrix referring to any substance that can resonantly absorb energy from a laser radiation source, be it an organic matrix for UV excitation or water, which is commonly used in IR laser excitation. MALDESI was introduced by Muddiman and co-workers in 2006 as the first hybrid ionization technique combining resonant laser desorption with electrospray postionization [12] The constant evolution of the MALDESI source has involved numerous optimizations of geometry [17–19], source design [12,20–22], and chemical compositions [23] based on the target analyte. MALDESI employing an IR laser with endogenous and exogenous ice matrix is particularly advantageous for fresh-frozen biological tissue analysis [24]. The higher fluence

192

of an IR laser compared to a UV laser allows the complete ablation of material within the focal volume of the laser. MSI microprobe analysis requires MS analysis at each rastered position, which needs to be performed sequentially and rapidly; therefore, it precludes the use of chromatographic separation prior to MS acquisition. The complex chemical composition of a biological tissue in conjunction with tissue heterogeneity places high demand on the instruments spectral resolving power to separate all compounds based on m/z. Additional structural information may be obtained from MSI using MSn methods in the case of isobaric species. IR-MALDESI has been developed using Fourier transform (FT) based MS instruments to obtain high resolving power MSI data. This review summarizes the latest developments of IR-MALDESI imaging of biological tissues, highlighting its capabilities while identifying areas of improvement and future work. Parameters affecting imaging quality are optimized and applications in drug distribution studies and untargeted metabolomics are presented. The development of absolute quantification MSI demonstrates the analytical capabilities of IR-MALDESI for biological samples.

A.2 Methods

The main steps to an IR-MALDESI MSI experiment are the same as any MSI experiment [25] which can include collection, storage, and sectioning of the specimen along with instrument or ionization specific variables. Each of these components can be and tailored to the analyzed material for optimal results. Collection, storage and sectioning are highly dependent on the sample and analyte in the investigation. Analysis of animal tissue, plant tissue, materials research substrates and neat or buffered solutions all require sample specific collection and storage procedures to ensure sample integrity throughout the analysis process. Sectioning of a sample is required for some specimens to produce a flat 2 dimensional plane

193

for laser ablation and to control the amount of material being sampled, which is the case for tissue analysis. Aqueous solutions can be analyzed by placing the liquid surface at the focal point of the laser or creating a dried droplet of the solution and deposition of an ice matrix on the surface prior to analysis. Other samples that are roughly planar can be analysed without the need for sectioning, such as fabric [26], fibers [27], and hair [28]. Tissues are the most commonly analyzed biomaterial to date and tissue sample preparation steps will be discussed in detail.

A.2.1 IR-MALDESI MSI Tissue Preparation

Biological tissues are harvested and immediately immersed in an isopentane/dry ice bath to prevent tissue fracturing during freezing as experienced with liquid nitrogen submersion [25]. Fresh frozen tissues are stored at −80 °C until time of analysis. Tissues are mounted on a cryostat specimen disc using a thin layer of embedding media. The embedding media is used to only adhere the tissue to the cryostat specimen disc, such that the tissue is not completely embedded in the media. The tissue is faced down to the desired plane for analysis using a cryomicrotome (Leica CM1950, Buffalo Grove, IL, USA) with section thicknesses of ≤50 μm prior to thaw mounting onto precleaned glass microscope slides. A comparison of serial tissue sections with thicknesses of 50, 25, and 10 μm showed complete ablation of the tissue and similar ion abundance for 3 pharmaceutical compounds and endogenous analytes for each thickness [29].

A.2.2 Quantitative IR-MALDESI Imaging Preparation

In experiments where quantitative information is desired, additional steps are required for IR-MALDESI MSI analysis. A normalization compound is incorporated at a spatially uniform concentration using an automated pneumatic sprayer (TM Sprayer, LEAP technologies, Carrboro, NC) to account for variability. Prior optimization of the spray

194

conditions are needed to ensure the uniformity of the coating. The normalization compound may be a stable isotope labelled (SIL) version of the analyte or carefully selected structural analogue of the quantified analyte given that it has a similar ionization efficiency. A precleaned microscope slide is uniformly coated with this normalization compound prior to sectioning of the tissue at a concentration sufficiently above the limit of detection for the tissue imaging experiment. The 10 μm tissue section is then thaw mounted onto the coated slide. A calibration curve may be constructed using SIL version of the analyte which allows the calibration curve to be placed directly on the analysed tissue [30]. Alternatively, natural isotope abundance standards may be placed on a blank tissue mounted and analysed in parallel to the tissue to be quantified [31]. Both calibration curve methods require the generation of a calibration series in solution prior to spotting on tissue. A constant volume of each calibration solution is pipetted on top of the thaw mounted tissue without spatial overlap to convert the liquid concentrations to a spatial concentration. Both methods have proven effective for calibration in MSI in good agreement with LC-MS/MS analysis of serial sections [30,31] Others have proposed spiked tissue homogenate cores as calibration points for quantitative MALDI MSI, but we have demonstrated this is not essential in IR-MALDESI because there is complete ablation of the tissue at each rastered position and IR-MALDESI does not require the extraction of analytes into an organic matrix like MALDI MSI does [32].

A.2.3 IR-MALDESI Tissue Imaging

The IR-MALDESI source and operation have been described in more detail elsewhere [18,24,33,34]. Briefly, the IR-MALDESI source is enclosed within an acrylic box, effectively isolating the sampled environment from the laboratory. The slide-mounted tissue section is placed on a Peltier cooled sample plate within the source enclosure. The sample plate is located on a computer controlled XY translational stage with manual Z-axis height adjustment which allows the tissue section to be moved in an array of discrete positions under the fixed position of the laser. By translating the stage by less than the desorption diameter in an oversampling

195

method [35], in combination with complete desorption of the tissue, voxels smaller than the focal volume of the laser can be sampled [18,36]. A voxel refers to the three-dimensional tissue volume desorbed by the laser. This ensures the complete ablation of the entire tissue during the analysis, providing reproducible volumes of tissue sampled.

A.2.4 Formation of Ice Matrix Layer

The enclosure is purged with dry nitrogen to a relative humidity (RH) less than 3% prior to bringing the Peltier stage to −9 °C. This is done in order to prevent condensation of water on the tissue, which could potentially delocalize analytes in or on the tissue if droplets form prior to freezing. The tissue is maintained at −9 °C for approximately 10 minutes prior to exposure to the ambient RH in the lab (>10% RH), causing the rapid formation of an ice layer on the sample plate. This exogenous ice layer serves as an energy absorbing matrix in addition to endogenous water present within the biological tissue. Resonant excitation of OH stretching mode of water by a tunable mid-IR (2700–3100 nm) laser (Opolette, Opotek, Carlsbad, CA, USA) is used to desorb tissue material at each rastered position with a user-defined number of laser pulses. Recently, it was found that two pulses at 2940 nm provides complete, reproducible ablation of a 10 μm thick tissue section within the focal volume of the laser [29] The desorbed tissue volume is ejected normal to the surface where the ablation plume overlaps with an orthogonal electrospray plume as depicted in Figure A-1. Analytes from the desorbed tissue material partition into the electrospray droplets where they undergo desolvation and charge transfer in a manner similar to ESI. The ions are then sampled by a high resolving power mass analyser (FT-ICR [24], Q Exactive [29], Q Exactive Plus [33]).

196

Figure A-1. Schematic of IR-MALDESI source. The sample is placed on a Peltier cooled X–Y translational stage where an exogenous ice matrix layer may be formed. A 2.94 μm laser is used to desorb material from the sample. The desorption plume interacts with an orthogonal electrospray plume where analytes partition into the charged droplets and ionization occurs in an ESI-like manner. The geometry of the source will have a direct effect on the laser desorption and electrospray plume overlap, and therefore on ion formation. Statistical design of experiments (DoE) was used in several studies to explore the experimental space with distinct optimized parameters for direct analysis of liquid droplets, solid state sample and tissue imaging [17–19]. All tissue imaging experiments discussed here were performed with the same geometric configuration as determined by DoE for tissue imaging [18].

A.2.5 Data Analysis and Visualization in MSiReader

Thermo .RAW data files were converted to mzML format using MSConvert tool from ProteoWizard [37]. These mzML data files were subsequently converted to imzML file format [38] using imzML converter [39]. Individual imzML files may be then concatenated into a single MSI file for MSI analysis in MSiReader [40]. All IR-MALDESI images presented were analysed in MSiReader. MSiReader is an open-source, vendor-neutral software created for visualization analysis of high resolving power MSI data [40]. Since its initial publication and public release in 2013, more than 20 new

197

features have been added to the software package. Five subsequent public updates include normalization strategies, peak export, customizable heat maps, colocalization, batch processing, optical image overlay, and a standalone version that does not require a MATLAB license. These features have been developed to better visualize MSI data and understand the complex data sets, such as those presented here.

A.3 Discussion

A.3.1 Effects of Ice Matrix on IR-MALDESI Analysis

The positive effects of the formation of an exogenous ice matrix layer were discovered when an ion abundance increase was observed when ice from ambient RH formed on the Peltier sample stage [18]. An investigation of the effects on desorption plume dynamics was subsequently undertaken using a novel shadowgraphy imaging system [18]. This system allowed images of the desorption plume to be taken using UV-laser excitation of a fluorophore behind the desorption event. The short (∼10 ns) fluorescence emission of the fluorophore allowed images to be captured of the desorption plume at the nanosecond timescale. Comparison of no ice and ice matrix desorption plumes showed distinct difference in the tissue desorption processes. The mass spectra measured from each laser pulse shows nearly all material is desorbed in the first pulse without ice matrix while ion abundance actually increases with increasing number of laser pulses with an ice matrix. Note this analysis is performed on a 50 μm thick fresh frozen tissue section. Subsequent evaluation of the ice layer effects in a targeted manner is presented in Figure A-2(B–D). The average abundance (휒̅) and detection frequency (ƒ) of three antiretroviral drugs incubated in cervical tissue were used as a metric to evaluate ice layer effects. As shown in Figure A-2, ice matrix results in higher abundance and more frequent detection for all three compounds [29].

198

Figure A-2. Influence of ice as a matrix for a 2940 nm IR laser. Targeted analysis of three antiretroviral (ARV) drugs in tissue from tissue with and without ice matrix. Ice matrix increased the frequency of detection and the average ion abundance for all three analytes of interest. Reproduced from ref. 29 with permission from Springer.

A.3.2 Comparison of IR-MALDESI to MALDI MSI

MSI allows for the multiplexed analysis of analytes in a single experiment, having the potential to show the distribution of drug, metabolite, and endogenous species in a single experiment. The capabilities of IR-MALDESI for drugs and metabolites in an imaging experiment were compared to that of MALDI MSI [41]. The spatial distribution of lapatinib and metabolites were used as a comparison of serial sections of a dosed liver between the two MSI ionization methods. Figure A-3 shows (top to bottom) optical image of the tissue, matrix application, distribution of parent drug, heme b indicating vasculature, and a colocalization plot of drug and heme b. The main differences between the two MSI techniques stem from matrix preparation and the laser used. MALDI requires the uniform deposition of an organic acid matrix while IR-MALDESI uses an ice matrix. The difference in matrix has major effects

199

on the resulting mass spectra as organic acid matrices create many matrix related peaks in the low m/z range potentially obscuring an analyte ion whereas an ice matrix does not. Additionally, the IR laser used in IR-MALDESI completely ablates the ice matrix and tissue at each rastered position compared to the UV laser used in MALDI, which excites and ablates a much smaller volume of matrix and cocrystallized analytes. Nonetheless, the targeted analyte(s) can be detected with either method after some amount of method development. Lapatinib and metabolites were detected using both ionization methods, with 24 and 11 metabolites being imaged using MALDI and IR-MALDESI, respectfully. This work showcases the similarity and complementary data that can be obtained from using two different ionization methods that rely on separate mechanisms for the ionization of analytes. It should be noted that both analyses used an FT-ICR detector, though these were from different manufacturers. For a more accurate comparison of ionization methods the data should be collected on the same MS platform, although, this is often not experimentally feasible.

Figure A-3. Comparison of MALDI and IR-MALDESI MSI analysis of serial sections from a lapatinib dosed liver. (A, F) optical images of the analysed tissue, (B, G) image of tissue with matrix applied (C, H) lapatinib (D, I) heme b (E, J) colocalization of lapatinib (blue) and heme b (red) in a single MS image. Reproduced from ref. 41 with permission from Elsevier.

200

A.3.3 Electrospray Ionization Agent

MALDI uses the careful pairing of organic matrix and analyte for sensitive and selective detection. Additionally, metal salts may be used to preferentially form adducts to the analyte. In IR-MALDESI, the matrix only facilitates desorption of material and ionization is achieved by electrospray. This allows for easy adjustment of ionization characteristics by changing ESI conditions. This permits selection of aqueous and organic solvent system to be optimized for particular class of compounds [41] although to date, only binary solvent systems have been investigated. Additionally, the ionization agent can be adjusted for operation in positive or negative ionization mode [42] using typical ionization agents such as formic acid or ammonium hydroxide. Some instruments, like the Q Exactive, allow rapid change of polarity such that a polarity switching imaging experiment may be performed on the same sample [42–44]. In this case, the electrospray solvent system must be selected as a compromise in performance between the polarities. As previously demonstrated for LC ESI analysis, weak acids can actually increase negative ion formation [45] and are often most suitable for polarity switching experiments. Because the laser completely desorbs the material at each rastered position in IR-MALDESI, an image consists of spectra of alternating polarity and does not allow the collection of spectra in both polarities from the same spatial location.

A.3.4 Metal Adduct Formation

Additionally, changing the traditional ionization agents (acid, base) by doping the electrospray with a constant concentration of metal adduct has proven to be useful. Meier et al. doped the electrospray solvent with silver nitrate to selectively form Ag+ adducts, which have an affinity toward double bonds [46]. Silver is not found biologically, so there is complete control over adduct formation. Using its unique isotopic distribution, roughly 1:1 for 107Ag and 109Ag, multivariate analysis of spectra and selection of mass spectra peak pairs with this ratio allow facile identification of unsaturated compounds. This work is exemplified by the

201

confident identification of 43 olefinic lipids in human serum. Also in this work, the high affinity of Ag+ for double bonds was used to lower limits of detection of cholecalciferol by 2 orders of magnitude. This, however, is still above physiological serum levels. Increases in ion abundance and reproducibility may be obtained using cationization agents in a targeted manner depending on the analyte. Bokhart et al. doped electrospray solvent with sodium chloride to preferentially form a sodium adduct of antiretroviral drugs [30]. This afforded differences in ionization efficiency per-spectrum to be accounted for as demonstrated by the reproducibility of normalization to the protonated and sodiated forms of a normalization compound. The increased reproducibility allowed quantitative MSI data to be obtained from emtricitabine in a human cervical tissue model [30].

A.3.5 Drug Distribution Studies

A.3.5.1 Quantitative IR-MALDESI imaging A major research area in MSI is obtaining reproducible and accurate quantitative information from a MSI experiment. This is not trivial due to tissue heterogeneity, careful incorporation of a calibration curve and the minute volume of tissue sampled at each voxel. IR-MALDESI is inherently quantitative with the complete ablation of tissue within the focal volume of the laser and employ of oversampling. MALDI, on the other hand, relies on extraction and cocrystallization of analytes with the matrix and ablation depths of approximately 100 nm. Ionization in MALDI is dependent on the organic acid matrix chosen, crystal size, and sample preparation steps. Ionization in IR-MALDESI is independent of the matrix as it relies on ionization through an electrospray like mechanism. In a proof of concept quantitative experiment, 5 sequential sections of a single human cervical tissue incubated in 100 μg/mL antiretroviral (ARV) drug solution were quantified using IR-MALDESI and serial sections to each were quantified using a validated LC-MS/MS method on the homogenate to evaluate the accuracy of quantitative MSI. The incorporation of a structural analogue normalization compound allowed for the ratio of analyte/normalization

202

compound response to be taken at each voxel, corresponding to one orbitrap mass spectrum, to help account for tissue specific ablation, ionization efficiency, and ion loss through the MS. The incorporation of a calibration curve consisting of SIL analogue of the drug allows the calibration curve to be placed directly on top of the analysed tissue. The quantification of ARV in a tissue section in a MSI experiment is summarized in Figure A-4. The SIL-calibration curve has a +3 m/z shift, where the natural abundance of the A + 3 isotope peak will have negligible contribution to the abundance. Data extraction is easily performed in MSiReader software with automated quantification procedure being available within the software in MSiReader v0.8+. Briefly, the analyte and calibration curve abundances are normalized to the normalization compound response at a per-voxel level (Figure A-4A and B). Data is extracted for regions of interest – specifically for calibration spots and tissue areas to be quantified. The tissue concentration for each ROI is calculated using the average normalized response and the volume from which it has been detected to give a spatial concentration. The resulting calibration curve (Figure A-4C) can then be used to calculate the concentration of analyte in tissue (Figure A-4D). IR-MALDESI analysis is typically performed on 10 μm thick sections and the IR laser completely ablates the material within the focal volume of the laser, truly sampling a volume of tissue and therefore yielding accurate tissue concentrations of analyte. Using this method, IR-MALDESI gave a similar result to a validated LC-MS/MS method for the 5 replicates of a model tissue system designed to define the quantitative analytical capabilities of IR-MALDESI with a concentration of 17.2 ± 1.8 μg/gtissue compared to 28.4 ± 2.8 μg/gtissue determined by LC-MS/MS homogenate. These values represent the average concentration for the five tissue sections with a 95% confidence interval. The similar absolute concentration by the two methods indicates great potential for IR- MALDESI to be applied to quantify drug concentrations within tissue microenvironments where concentration may reach therapeutic or potentially toxic levels – information that is lost in the homogenization process required for LC-MS/MS analysis. The IR-MALDESI MSI quantification was compared to LC-MS/MS of an adjacent tissue section, since both IR-

203

MALDESI and LC-MS/MS are destructive techniques. By comparing MSI to LC-MS/MS, it is assumed that the LC-MS/MS data is the true concentration of analyte in tissue based on LC- MS/MS acceptance in the literature and by regulatory agencies. Differences between the two methods could be attributed to the sample preparation steps for the two quantification methods as they are necessarily different at nearly every step, with an amount of error associated with each deviation. LC-MS/MS requires homogenization of the tissue to aid extraction of analytes into the solvent and removal of tissue debris. LC-MS/MS quantification is based on area under a chromatographic peak where ionization suppressing compounds such as lipids are chromatographically resolved from the antiretroviral drug. The extracted analyte in LC- MS/MS of a tissue homogenate represents the average concentration of the drug in the tissue section with the loss of all spatial information in the homogenization process. In IR-MALDESI MSI quantification, the tissue is carefully prepared with a normalization compound present uniformly across the analyzed area and a calibration curve spatially integrated with the analyzed tissue. In MSI, the spatial concentration of the analyte is preserved by analyzing the tissue at many locations across the tissue, with each location essentially a separate MS analysis but without chromatographic separation. Additionally, the calibration curve is based on MS peak height for each voxel of a calibration spot instead of the area under the chromatographic peak for individual calibration standards in LC-MS/MS.

204

Figure A-4. Summary of a quantitative MSI experiment using IR-MALDESI. (A) Emtricitabine (FTC) was quantified in a cervical tissue model system using a (B) stable isotope labeled (SIL) calibration curve on the same tissue. (C) MSI calibration curve showed good linearity and (D) summary of calculated FTC in the model tissue. Reproduced from ref. 30 with permission from Springer. IR-MALDESI MSI is ideally suited for drug distribution studies, specifically ARVs because they are often administered in a multidrug regimen referred to as highly active antiretroviral therapy (HAART) requiring the analysis of numerous drugs and metabolites in a single analysis. This is in comparison to the industry standard quantitative whole body autoradiography which requires synthesis and administration of radiolabeled compounds, which has significant cost and safety concerns associated with it. Additionally, this method only detects the radiolabel and is unable to differentiate parent and metabolites. A.3.5.2 Multi-organ Pharmaceutical Distribution The developed quantitative IR-MALDESI was utilized for a multi-organ drug distribution study of HIV integrase inhibitor efavirenz (EFV) within a rhesus macaque dosed to steady state [31]. The quantitative MSI method was applied to a tissue set identified as putative HIV reservoirs where latent virus may be dormant and current ARV regimens may

205

not penetrate. MSI allows vital inter- and intra-organ assessment of drug penetration needed to eradicate HIV from an infected individual. Eleven putative HIV reservoirs were analysed by IR-MALDESI MSI and distributions of EFV were quantified using a calibration curve on organ-matched, non-dosed tissue. The quantification process was performed in parallel to absolute concentration determination by LC-MS/MS, hematoxylin and eosin (H & E) staining, and immunohistochemistry (IHC) staining of CD3+ cells. Representative MS and IHC images for 5 tissues are shown in Figure A-5. Interpretation of the three imaging modalities allowed for comprehensive determination of drug distribution into specific regions of an organ and its corresponding immune response by CD3. Heterogeneous EFV distribution was observed in the tissues, with the most heterogeneity observed in the gastrointestinal tract, specifically the colon (Figure A-5A). Combining information gained from MSI, H&E, and IHC, it was determined that EFV had the highest concentration in the colon mucosa and lamina propria.

Figure A-5. MSI of efavirenz and CD3+ IHC imaging of (A) colon (B) ileum (C) inguinal lymph node (D) cerebellum (E) spleen. Reproduced from ref. 31 with permission from American Society for Microbiology. Copyright 2015.

206

A.3.5.3 MSI Imaging of Antiretroviral Drugs in Hair Hair has long been used as a temporal record for determining use of drugs of abuse. However, this is typically performed using a thatch of hair which will represent the average incorporation over a long period of time. IR-MALDESI MSI was used to analyze single hair strands at a spatial resolution of 100 μm, which based on an average growth rate of 1 cm per month, gives a temporal resolution of approximately 7 hours [28]. This analysis is of particular interest to HIV research in patients receiving pre-exposure prophylaxis where consistent dosing is vital to the programs efficacy. IR-MALDESI proved to be a useful technique for the spatial analysis of hair without the need for sample preparation steps required for MALDI MSI, such as analyte extraction and/or longitudinal slicing of the individual hair strands [47]. Microscopy images of a hair after IR-MALDESI clearly showed ablation of the cuticle and into the cortex of the hair strands. Many factors influence the incorporation of drug into hair including the drug basicity, lipophilicity, and melanin content of the hair. The authors present a method of normalization based on the oxidation of 5,6-dihydroxyindole-2-carboxylic acid, the monomeric unit of eumelanin, to pyrrole-2,3,5-tricarboxylic acid. This was performed by spraying the individual hair strands with 15% hydrogen peroxide and 1 M ammonium hydroxide using a thermally- assisted pneumatic sprayer. Figure A-6A shows the distribution of EFV before and after treating the hair with H2O2. The sample preparation step does not delocalize or extract EFV significantly. Normalizing the response of incorporated EFV to the oxidized eumelanin product reduced variability between patients with different hair color indicating IR-MALDESI could be broadly applicable method for drug adherence evaluation (Figure A-6B).

207

Figure A-6. IR-MALDESI analysis of ARVs in hair. (A) EFV distribution before and after H2O2 sample preparation for the normalization of drug incorporation to melanin content. MSI images show that incorporated EFV does not delocalize during sample preparation. (B) Normalization of EFV to monomeric eumelanin unit potential for broad application of IR-MALDESI MSI analysis to hair analysis of drug incorporation. Reprinted with permission from ref. 28. Copyright 2016 American Chemical Society.

A.3.6 Untargeted Metabolomic MSI Analyses

A.3.6.1 Polarity Switching MSI of Cancerous vs. Healthy Ovarian Tissue IR-MALDESI is well suited for untargeted metabolite and lipid analysis of biospecimens, allowing researchers to better understand underlying biological factors in health and disease [46,48]. Ionization by electrospray allows facile acquisition in either positive or negative mode. A strategy for repeated acquisition of both polarities in a spatially resolved manner was originally applied to the forensic analysis of fibers and dyes. In this polarity

208

switching MSI method, positive and negative ionization was performed on alternating voxels of the image [27]. Interpretation of the data requires parsing of the data file by polarity prior to MSI interpretation. This allows for analysis by both polarities on the same tissue with minimal loss in spatial resolution. This acquisition method was optimized and applied to sections of cancer and control ovarian tissue from a chicken model of ovarian cancer (OVC) [42]. Prior to analysis, the authors investigated the influence of several electrospray compositions and their effects on lipid response in both polarities. The optimized conditions were applied to the untargeted analysis of healthy and cancerous chicken ovary tissue, with fold differences in ion abundance being compared in MSiReader. Spatial interrogation of OVC is important in characterizing alterations in lipid metabolism, which has been linked to the disease [49]. Figure A-7 shows MS images of selected lipids in a healthy (left) and cancerous (right) chicken ovary tissue. The dashed line indicates maturing follicles in the healthy tissue while the solid white line indicates the cancerous regions in an ovary tissue. The untargeted method revealed compounds up and down regulated in the two tissues, indicating potential for IR-MALDESI to discover potential biomarkers of disease (Figure A-7).

209

Figure A-7. Control (left) and cancer (right) ovarian tissue was analysed using IR-MALDESI MSI. The two tissue types showed significant changes in lipid and metabolite profiles especially in maturing follicles (dashed line) and within cancerous regions (solid line). Reproduced from ref. 42 with permission from The Royal Society of Chemistry.

Fundamental mass spectrometry principles of spectral accuracy and sulfur counting were used in the identification of lipid and metabolites from full scan MS1 data. For example, glutathione was identified using accurate mass (<1 ppm) and the concept of sulfur counting to eliminate other possible isobaric species from MS1 data.

210

A.3.7 Limitations

The utility of IR-MALDESI has been demonstrated but there are several inherent limitations of the technique. Nearly every MSI technique requires cryosectioning of tissues, which does not allow analysis at multiple time points from a single biological experiment. Instead, analysis of several time points must be carried out using separate subjects, increasing time and cost while introducing biological variability. MSI is primarily used as a destructive ex vivo imaging technique and is not typically applied to in vivo imaging. However, recent publication of in vivo analysis using laser desorption with secondary ionization was recently reported [50] describing a new way of performing MSI for in vivo imaging. Quantification using IR-MALDESI is method development intensive, requiring careful selection of a normalization compound and creation of a calibration curve for each compound. Typically, quantitative MSI experiments are run in parallel to a validated LC/MS assay for comparison of total amount of analyte in the tissue. Analysis of tissue in microprobe mode requires the sequential analysis of thousands of tissue desorption events, thereby effectively diluting the amount of tissue sampled by several orders of magnitude. In addition to the minute amount of tissue sampled per spectrum, no pre-concentration or chromatographic separation steps may be performed due to the repetitive, sequential acquisition in MSI. This repeated acquisition of MSI results in analysis times of several hours for a modest-sized tissue. Additionally, IR- MALDESI MSI shows preference for compounds that are highly abundant and easily ionisable and most applications are towards small molecules and lipids. To date, large biomolecules such as proteins have not been observed from tissue, however, they are easily detected from purified protein solutions for concentrations as low as 10 pM [17]. Detection of intact or tryptic peptides from tissue will require extensive method development that differs significantly from our current protocols presented here and are an ongoing research focus.

211

A.4 Conclusions

IR-MALDESI has proven to be a robust MSI technique for the analysis of biospecimens. Significant improvements have been made in analysis from a breadth of sample types and analytes while retaining the spatial location. There is still opportunity in the MSI field for improvement, in comprehensiveness of analysis, exhaustively defining quantification abilities, reducing spatial resolution, and validation of and application to real biological questions. The application of IR-MALDESI MSI to targeted, quantitative drug distribution studies and untargeted metabolomics studies demonstrates the versatility of the IR-MALDESI imaging platform for a variety of analytes and applications.

A.5 Acknowledgements

The authors thank Milad Nazari and Brian Cartiff for their assistance in preparing this manuscript. The authors gratefully acknowledge financial support from the National Institutes of Health (R01GM087964), the W. M. Keck Foundation, and North Carolina State University.

A.6 References

1. Groseclose, M.R., Andersson, M., Hardesty, W.M., Caprioli, R.M.: Identification of proteins directly from tissue: in situ tryptic digestions coupled with imaging mass spectrometry. Journal of Mass Spectrometry. 42, 254-262 (2007)

2. Prideaux, B., Stoeckli, M.: Mass spectrometry imaging for drug distribution studies. Journal of Proteomics. 75, 4999-5013 (2012)

3. Miura, D., Fujimura, Y., Wariishi, H.: In situ metabolomic mass spectrometry imaging: Recent advances and difficulties. Journal of Proteomics. 75, 5052-5060 (2012)

4. Crecelius, A.C., Schubert, U.S., von Eggeling, F.: MALDI mass spectrometric imaging meets "omics": recent advances in the fruitful marriage. Analyst. 140, 5806-5820 (2015)

212

5. Rubakhin, S.S., Jurchen, J.C., Monroe, E.B., Sweedler, J.V.: Imaging mass spectrometry: fundamentals and applications to drug discovery. Drug Discovery Today. 10, 823-837 (2005)

6. Castaing, R., Slodzian, G.: Microanalyse par émission ionique secondaire. J. Microscopie. 1, 395-410 (1962)

7. Spengler, B., Hubert, M., Kaufman, R.: In Proceedings of the 42nd Annual Conference on Mass Spectrometry and Allied Topics., Chicago, Illinois, p. 1041 (1994)

8. Caprioli, R.M., Farmer, T.B., Gile, J.: Molecular Imaging of Biological Samples: Localization of Peptides and Proteins Using MALDI-TOF MS. Analytical Chemistry. 69, 4751-4760 (1997)

9. Palmer, A., Trede, D., Alexandrov, T.: Where imaging mass spectrometry stands: here are the numbers. Metabolomics. 12, 107 (2016)

10. Robichaud, G., Barry, J.A., Muddiman, D.C.: Atmospheric Pressure Mass Spectrometry Imaging. John Wiley & Sons, Ltd, (2006)

11. Shiea, J., Huang, M.-Z., Hsu, H.-J., Lee, C.-Y., Yuan, C.-H., Beech, I., Sunner, J.: Electrospray-assisted laser desorption/ionization mass spectrometry for direct ambient analysis of solids. Rapid Communications in Mass Spectrometry. 19, 3701-3704 (2005)

12. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Generation and detection of multiply-charged peptides and proteins by matrix-assisted laser desorption electrospray ionization (MALDESI) fourier transform ion cyclotron resonance mass spectrometry. Journal of the American Society for Mass Spectrometry. 17, 1712-1716 (2006)

13. Nemes, P., Vertes, A.: Laser ablation electrospray ionization for atmospheric pressure, in vivo, and imaging mass spectrometry. Analytical chemistry. 79, 8098-8106 (2007)

14. Brady, J.J., Judge, E.J., Levis, R.J.: Mass spectrometry of intact neutral macromolecules using intense non-resonant femtosecond laser vaporization with electrospray post-ionization. Rapid Communications in Mass Spectrometry. 23, 3151- 3157 (2009)

15. Rezenom, Y.H., Dong, J., Murray, K.K.: Infrared laser-assisted desorption electrospray ionization mass spectrometry. Analyst. 133, 226-232 (2008)

213

16. Fenn, J., Mann, M., Meng, C., Wong, S., Whitehouse, C.: Electrospray ionization for mass spectrometry of large biomolecules. Science. 246, 64-71 (1989)

17. Barry, J.A., Muddiman, D.C.: Global optimization of the infrared matrix-assisted laser desorption electrospray ionization (IR MALDESI) source for mass spectrometry using statistical design of experiments. Rapid Communications in Mass Spectrometry. 25, 3527-3536 (2011)

18. Robichaud, G., Barry, J.A., Muddiman, D.C.: IR-MALDESI Mass Spectrometry Imaging of Biological Tissue Sections Using Ice as a Matrix. Journal of The American Society for Mass Spectrometry. 25, 319-328 (2014)

19. Rosen, E.P., Bokhart, M.T., Ghashghaei, H.T., Muddiman, D.C.: Influence of Desorption Conditions on Analyte Sensitivity and Internal Energy in Discrete Tissue or Whole Body Imaging by IR-MALDESI. Journal of The American Society for Mass Spectrometry. 26, 899-910 (2015)

20. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Construction of a versatile high precision ambient ionization source for direct analysis and imaging. Journal of the American Society for Mass Spectrometry. 19, 1527-1534 (2008)

21. Sampson, J.S., Muddiman, D.C.: Atmospheric pressure infrared (10.6 µm) laser desorption electrospray ionization (IR-LDESI) coupled to a LTQ Fourier transform ion cyclotron resonance mass spectrometer. Rapid Communications in Mass Spectrometry. 23, 1989-1992 (2009)

22. Sampson, J.S., Murray, K.K., Muddiman, D.C.: Intact and top-down characterization of biomolecules and direct analysis using infrared matrix-assisted laser desorption electrospray ionization coupled to FT-ICR mass spectrometry. Journal of the American Society for Mass Spectrometry. 20, 667-673 (2009)

23. Sampson, J.S., Hawkridge, A.M., Muddiman, D.C.: Development and Characterization of an Ionization Technique for Analysis of Biological Macromolecules: Liquid Matrix- Assisted Laser Desorption Electrospray Ionization. Analytical Chemistry. 80, 6773- 6778 (2008)

24. Robichaud, G., Barry, J.A., Garrard, K.P., Muddiman, D.C.: Infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) imaging source coupled to a FT-ICR mass spectrometer. Journal of the American Society for Mass Spectrometry. 24, 92-100 (2013)

214

25. Goodwin, R.J.A.: Sample preparation for mass spectrometry imaging: Small mistakes can lead to big consequences. Journal of Proteomics. 75, 4893-4911 (2012)

26. Cochran, K.H., Barry, J.A., Muddiman, D.C., Hinks, D.: Direct Analysis of Textile Fabrics and Dyes Using Infrared Matrix-Assisted Laser Desorption Electrospray Ionization Mass Spectrometry. Analytical Chemistry. 85, 831-836 (2013)

27. Cochran, K.H., Barry, J.A., Robichaud, G., Muddiman, D.C.: Analysis of trace fibers by IR-MALDESI imaging coupled with high resolving power MS. Analytical and Bioanalytical Chemistry. 407, 813-820 (2015)

28. Rosen, E.P., Thompson, C.G., Bokhart, M.T., Prince, H.M.A., Sykes, C., Muddiman, D.C., Kashuba, A.D.M.: Analysis of Antiretrovirals in Single Hair Strands for Evaluation of Drug Adherence with Infrared-Matrix-Assisted Laser Desorption Electrospray Ionization Mass Spectrometry Imaging. Analytical Chemistry. 88, 1336- 1344 (2016)

29. Barry, J.A., Robichaud, G., Bokhart, M.T., Thompson, C., Sykes, C., Kashuba, A.D., Muddiman, D.C.: Mapping antiretroviral drugs in tissue by IR-MALDESI MSI coupled to the Q Exactive and comparison with LC-MS/MS SRM assay. Journal of The American Society for Mass Spectrometry. 25, 2038-2047 (2014)

32. Groseclose, M.R., Castellino, S.: A Mimetic Tissue Model for the Quantification of Drug Distributions by MALDI Imaging Mass Spectrometry. Analytical Chemistry. 85, 10099-10106 (2013)

33. Rosen, E.P., Bokhart, M.T., Nazari, M., Muddiman, D.C.: Influence of C-Trap Ion Accumulation Time on the Detectability of Analytes in IR-MALDESI MSI. Analytical Chemistry. 87, 10483-10490 (2015)

215

34. Nazari, M., Bokhart, M.T., Muddiman, D.C.: Whole-body Mass Spectrometry Imaging by Infrared Matrix-assisted Laser Desorption Electrospray Ionization (IR-MALDESI). e53942 (2016)

35. Jurchen, J.C., Rubakhin, S.S., Sweedler, J.V.: MALDI-MS imaging of features smaller than the size of the laser beam. Journal of the American Society for Mass Spectrometry. 16, 1654-1659 (2005)

36. Nazari, M., Muddiman, D.C.: Cellular-level mass spectrometry imaging using infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI) by oversampling. Analytical and Bioanalytical Chemistry. 407, 2265-2271 (2015)

37. Kessner, D., Chambers, M., Burke, R., Agus, D., Mallick, P.: ProteoWizard: open source software for rapid proteomics tools development. Bioinformatics. 24, 2534- 2536 (2008)

38. Schramm, T., Hester, A., Klinkert, I., Both, J.-P., Heeren, R.M.A., Brunelle, A., Laprévote, O., Desbenoit, N., Robbe, M.-F., Stoeckli, M., Spengler, B., Römpp, A.: imzML — A common data format for the flexible exchange and processing of mass spectrometry imaging data. Journal of Proteomics. 75, 5106-5110 (2012)

39. Race, A.M., Styles, I.B., Bunch, J.: Inclusive sharing of mass spectrometry imaging data requires a converter for all. Journal of proteomics. 75, 5111-5112 (2012)

41. Barry, J.A., Groseclose, M.R., Robichaud, G., Castellino, S., Muddiman, D.C.: Assessing drug and metabolite detection in liver tissue by UV-MALDI and IR- MALDESI mass spectrometry imaging coupled to FT-ICR MS. International journal of mass spectrometry. 377, 448-455 (2015)

42. Nazari, M., Muddiman, D.C.: Polarity switching mass spectrometry imaging of healthy and cancerous hen ovarian tissue sections by infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). Analyst. 141, 595-605 (2016)

43. Schuhmann, K., Almeida, R., Baumert, M., Herzog, R., Bornstein, S.R., Shevchenko, A.: Shotgun lipidomics on a LTQ Orbitrap mass spectrometer by successive switching between acquisition polarity modes. Journal of Mass Spectrometry. 47, 96-104 (2012)

216

44. Korte, A.R., Lee, Y.J.: Multiplex Mass Spectrometric Imaging with Polarity Switching for Concurrent Acquisition of Positive and Negative Ion Images. Journal of The American Society for Mass Spectrometry. 24, 949-955 (2013)

45. Wu, Z., Gao, W., Phelps, M.A., Wu, D., Miller, D.D., Dalton, J.T.: Favorable Effects of Weak Acids on Negative-Ion Electrospray Ionization Mass Spectrometry. Analytical Chemistry. 76, 839-847 (2004)

46. Meier, F., Garrard, K.P., Muddiman, D.C.: Silver dopants for targeted and untargeted direct analysis of unsaturated lipids via infrared matrix-assisted laser desorption electrospray ionization (IR-MALDESI). Rapid Communications in Mass Spectrometry. 28, 2461-2470 (2014)

47. Nakanishi, T., Nirasawa, T., Takubo, T.: Quantitative Mass Barcode-Like Image of Nicotine in Single Longitudinally Sliced Hair Sections from Long-Term Smokers by Matrix-Assisted Laser Desorption Time-of-Flight Mass Spectrometry Imaging. Journal of Analytical Toxicology. 38, 349-353 (2014)

48. Loziuk, P., Meier, F., Johnson, C., Ghashghaei, H.T., Muddiman, D.C.: TransOmic analysis of forebrain sections in Sp2 conditional knockout embryonic mice using IR- MALDESI imaging of lipids and LC-MS/MS label-free proteomics. Analytical and Bioanalytical Chemistry. 408, 3453-3474 (2016)

49. Menendez, J.A., Lupu, R.: Fatty acid synthase and the lipogenic phenotype in cancer pathogenesis. Nat Rev Cancer. 7, 763-777 (2007)

50. Lee, J.K., Jansson, E.T., Nam, H.G., Zare, R.N.: High-Resolution Live-Cell Imaging and Analysis by Laser Desorption/Ionization Droplet Delivery Mass Spectrometry. Analytical Chemistry. 88, 5453-5461 (2016)

217