Mass Spectrometric Analysis of Neurologically-Relevant Molecules

Item Type text; Electronic Dissertation

Authors Smith, Catherine L.

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 03/10/2021 20:43:26

Link to Item http://hdl.handle.net/10150/626689 MASS SPECTROMETRIC ANALYSIS OF NEUROLOGICALLY-RELEVANT MOLECULES by Catherine L. Smith

A Dissertation Submitted to the Faculty of the DEPARTMENT OF CHEMISTRY AND BIOCHEMISTRY

In Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY WITH A MAJOR IN CHEMISTRY

In the Graduate College

THE UNIVERSITY OF ARIZONA

2018 2 STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of the requirements for an advanced degree at the University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that an accurate acknowledgement of the source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.

SIGNED: Catherine L. Smith

3 DEDICATION

For Brandon

I couldn’t have done this without you.

4 TABLE OF CONTENTS

TITLE PAGE

LIST OF FIGURES 13

LIST OF TABLES 17

LIST OF ABBREVIATIONS 18

LIST OF CONTRIBUTORS 21

ABSTRACT 22

CHAPTER 1. Introduction: Quantifying neurochemicals with mass 23

spectrometry

1.1 Introduction 23

1.2 Sampling and analysis of neurochemicals 25

1.2.1 Ex situ 26

1.2.1.1 Affinity assays 26

1.2.1.2 Separation and detection techniques 27

1.2.2 In vitro 30

1.2.2.1 Biological fluids 30

1.2.2.2 Cell models 31

1.2.3 In vivo 31

1.2.3.1 Direct measurement techniques 32

1.2.3.2 Microdialysis 32

1.3 Tandem mass spectrometry 36

1.4 Neurotransmission in insects 42

1.4.1 Classical neurotransmission 43

1.4.2 Quantitation of neurotransmitters 46

5 1.5 Drug penetration of the blood-brain barrier 47

1.5.1 Mechanisms of BBB penetration 48

1.5.2 Peptide-based drug development 50

1.5.3 Current measurement techniques and challenges 51

1.6 Overview 53

CHAPTER 2. Separation-free quantitation of biogenic amines in 55

brain tissue with mass spectrometry

2.1 Introduction 56

2.2 Materials and methods 59

2.2.1 Chemicals and reagents 59

2.2.2 Rats 59

2.2.3 Unilateral 6-hydroxydopamine (6-OHDA) lesion 59

method

2.2.4 Amphetamine-induced rotation 60

2.2.5 Rat tissue collection and preparation 60

2.2.6 Drosophila melanogaster 60

2.2.7 Apis mellifera 61

2.2.8 Derivatization reaction 62

2.2.9 Mass spectrometric detection and quantitation 63

2.3 Results and Discussion 63

2.3.1 Mass spectrometric detection of labeled biogenic 63

amines

2.3.2 Comparison to “gold standard” LC-EC 75

2.3.3 Quantification of biogenic amines in insect brain 80

homogenate

6 2.3.4 Sexual dimorphism was not detected across three 85

Drosophila strains

2.3.5 Deletion of ABC transporter does not affect 87

whole-head histamine

2.3.6 Genetic inactivation of DAT in Drosophila 89

melanogaster does not systematically alter whole-

head biogenic amine content

2.4 Conclusions 90

2.5 Author contributions 90

2.6 Acknowledgments 91

CHAPTER 3. Nosema ceranae parasitism in honey bees (Apis 92

mellifera) increases biogenic amines associated with

foraging behavior and alters olfactory learning and

memory

3.1 Introduction 93

3.2 Materials and Methods 95

3.2.1 Animals 95

3.2.2 Feeding 95

3.2.3 Nosema inoculum 96

3.2.4 Spore counts 97

3.2.5 Learning and memory 97

3.2.6 Amino acid analysis of brain tissue and pollen 98

3.2.7 Biogenic amine analysis of brain tissue 101

3.2.8 Statistics 102

3.3 Results and Discussion 102

7 3.3.1 Odor-associative learning and memory in nurse- 102

and forager-aged bees

3.3.2 Amino acid concentrations in the whole brain of 106

nurse- and forager-aged bees

3.3.3 Biogenic amine levels in the whole brain of nurse- 112

and forager-aged bees

3.3.4 Comparing whole-brain content of compounds 118

along biological synthesis pathways can elucidate

mechanisms of change in infected bees

3.4 Conclusions 122

3.5 Author Contributions 123

3.6 Acknowledgments 124

CHAPTER 4. Glycosylation of peptide-based drug candidates 125

improves in vivo stability and penetration of the

blood-brain barrier

4.1 Introduction 126

4.2 Materials and Methods 131

4.2.1 Chemicals and reagents 131

4.2.2 Peptide synthesis 131

4.2.3 MS identification and quantification of peptides 131

4.2.4 In vitro stability of peptides 133

4.2.5 Animals 133

4.2.6 Carotid artery catheterization 133

4.2.7 Microdialysis surgery 134

4.2.8 Microdialysis and blood draws 135

8 4.3 Results and Discussion 135

4.3.1 High ion throughput of glycosylated peptides 135

during tandem MS yields lower limits of detection

4.3.2 Glycosylated compounds have improved in vitro 144

stability

4.3.3 “Shotgun microdialysis” for direct comparison of 149

in vivo lifetime and BBB penetration of peptide

derivatives

4.4 Conclusions 152

4.5 Author Contributions 154

4.6 Acknowledgements 155

CHAPTER 5. Identification of optimal structural modifications for 156

the improvement of peptide-based drug delivery

properties: an Angiotensin 1-7 study of in vivo

stability and BBB penetration

5.1 Introduction 157

5.2 Materials and Methods 161

5.2.1 Chemicals and reagents 161

5.2.2 Animals 161

5.2.3 Peptide synthesis 162

5.2.4 Mass spectrometric identification and 162

quantification of compounds

5.2.5 In vitro stability of Angiotensin derivatives. 162

5.2.6 Carotid artery catheterization 164

5.2.7 Microdialysis surgery 165

9 5.2.8 Microdialysis and blood draws 165

5.3 Results and Discussion 166

5.3.1 Mass spectrometric identification and 166

quantification of compounds

5.3.2 Modified compounds have increased in vitro 175

lifetime

5.3.3 Non-selective protein binding may decrease free 179

fraction of peptides

5.3.4 “Shotgun microdialysis” allows for the direct 181

comparison of in vivo lifetime and BBB

penetration of Ang 1-7 derivatives

5.4 Conclusions 192

5.5 Author Contributions 193

5.6 Acknowledgments 193

CHAPTER 6. Conclusions and future directions 194

6.1 Expanding the portfolio of BzCl-DI-MS 195

6.1.1 Increasing the number of amine compounds 195

being investigated

6.1.2 Application to additional factors in honey bee hive 196

health

6.2 Glycosylation and other modifications for improved 200

pharmaceutical properties of peptide-based drugs

6.2.1 Investigation of apparent PACAP degradation 202

10 6.2.2 Positive and negative controls for blood-brain 205

barrier penetration will strengthen our

methodology

6.3 Potential routes for the elucidation of enkephalin 207

dynamics in vivo

6.3.1 Quantitation of enkephalins in chronic pain 209

models

6.3.2 Pharmaceutical probing of enkephalin dynamics 209

6.4 Concluding remarks 211

APPENDIX A Procedure: Benzoyl Chloride Labeling Reaction 214

APPENDIX B Quantification of endogenous opioid peptide 216

dynamics in the anterior cingulate cortex by online-

preservation microdialysis

B.1 Introduction 217

B.2 Materials and Methods 219

B.2.1 Chemicals 219

B.2.2 Statistical analysis & validation criteria 219

B.2.3 Animals 220

B.2.4 Intracranial (ACC) cannula implantation for 221

microdialysis

B.2.5 CSF collection from cisterna magna for in vitro 221

experiments

B.2.6 In vivo microdialysis procedure 222

B.2.7 Online-preservation system 223

B.2.8 Sample preparation for nano LC-MS3 analysis 224

11 B.2.9 Nano LC-MS3 analysis of opioids 224

B.2.10 Flow-injection MS study of opioid degradation 225

B.3 Results and Discussion 226

B.3.1 Chip-based ESI for improved nano LC-MS3 226

quantification of endogenous opioids

B.3.2 Initial in vivo measurements in the ACC and 231

degradation of opioid peptides

B.3.3 Online-preservation of endogenous peptides 237

B.3.4 Online-preservation microdialysis allowed 239

quantifiable measurement of EOPs in the ACC

B.3.5 Identification and quantification of methionine 239

enkephalin sulphoxide

B.3.6 Endomorphin II is present in dialysate from the 241

ACC

B.3.7 Stimulated release of enkephalins in the ACC 241

B.4 Conclusions 244

B.5 Author Contributions 246

B.6 Acknowledgements 246

APPENDIX C Compound structures, mass spectra, and 247

chromatograms

REFERENCES 261

12 LIST OF FIGURES FIGURE TITLE PAGE

1.1 Principles of microdialysis 34

1.2 Principles of electrospray ionization 38

1.3 Neural transmission 45

1.4 Mechanisms of penetration of the blood-brain barrier 49

2.1 Schotten-Baumann reaction mechanism 64

2.2 Benzoyl chloride derivitization 65

2.3 Workflow for sample preparation 67

2.4 MS2 spectra for target compounds 68

2.5 Representative calibration curves 69

2.6 Labeled compounds show stability against oxidation 74

2.7 Comparison to existing LC-EC method for validation of 77 quantitation

2.8 Dopamine and serotonin correlations between methods 79

2.9 Biogenic amine quantification in honey bees matches previously 81 reported results

2.10 Biogenic amine content (ng/head) in Drosophila melanogaster 86 whole heads, by sex

2.11 Biogenic amine content (ng/head) in Drosophila melanogaster 88 whole heads, by strain

3.1 Honey bee learning and memory trials 104

3.2 N. ceranae affects amino acid levels in Day 7 and Day 15 honey 108 bees

3.3 Effects of amino acids in whole brain tissue in Day 7 and Day 15 109 honey bees with and without N. ceranae infection

3.4 Biogenic amine concentrations in whole brain tissue 114

13 3.5 Changes in whole-head content of compounds by biosynthetic 119 pathway

4.1 Glycosylated and non-glycosylated compounds 129

4.2 Fragmentation spectra and CID breakdowns 137

4.3 Added selectivity with glycosylation 138

4.4 LC-MS3 quantitation adds selectivity and sensitivity 140

4.5 Investigation of glycosylated PACAP’s mass deficiency 143

4.6 In vitro degradation of Ang peptides in rSer at 37 °C 145

4.7 In vitro degradation of SAM-995 and MMP-2200 in rSer at 37 °C 146

4.8 In vitro degradation of PACAP derivatives 148

4.9 Workflow for collection and analysis of in vivo samples 151

4.10 In vivo quantitation of SAM-995 and MMP-2200 153

5.1 Ang 1-7 modifications for improved in vivo lifetime and blood- 160 brain barrier penetration

5.2 Annotated MS1-3 spectra for Angiotensin 1-7 167

1-3 5.3 Annotated MS spectra for Ang 1-6-Ser(O-Glc)-NH2 169

5.4 Improved chromatography with addition of octyl sulfonate pairing 171 agent

5.5 Selected ion chromatograms and MS3 spectra 173

5.6 In vitro stability of Ang 1-7 derivatives 176

5.7 Apparent degradation in the first minutes may be due to protein 180 binding

5.8 Effect of flow rate on probe recovery 184

5.9 In vivo results following injection of six Ang 1-7 derivatives 188

5.10 Calculated area under the curve (AUC) for each of the 190 Angiotensin derivatives in vivo

6.1 Biosynthetic pathways for biogenic amines 197

14 6.2 Biogenic amine content in whole honey bee brains shows 198 differences based on season

6.3 Varroa destructor may affect biogenic amine content in honey 201 bee whole brains

17 6.4 No degradation products of PACAP 1-27-S-(O-Glc)-NH2 [ Leu] 203 have been identified

6.5 An interfering species prevents the use of loratadine as a BBB- 208 control in animal experiments

6.6 SNL surgery alters baseline concentration of EOPs 210

6.7 Injection of gabapentin does not affect enkephalin release 212

B.1 Analysis of endogenous opioid peptides 227

B.2 Degradation studies of enkephalins in biological fluids by DI-MSn 234

B.3 Online-preservation apparatus scheme allows for improved 238 microdialysis recoveries from the ACC

B.4 Representative nano LC-MS3 data from in vivo microdialysis 240 sampling of the rat ACC

B.5 Stimulated release of LE and ME 243

C.1 SAM-995 spectra 247

C.2 MMP-2200 spectra 248

C.3 LC gradient for SAM-995, MMP2200, and six Angiotensin 1-7 249 derivatives

C.4 Chromatograms and MS3 spectra for SAM-995 and MMP2200 250

C.5 Ang 1-7 spectra 251

C.6 Ang 1-7-NH2 spectra 252

C.7 Ang 1-6-Ser-NH2 spectra 253

C.8 Ang 1-6-Ser(O-Glc)-NH2 spectra 254

C.9 Ang 1-6-Ser(O-Cb)-NH2 spectra 255

15 C.10 Ang 1-6-Ser(O-Lac)-NH2 spectra 256

17 C.11 [Leu ] PACAP 1-27-NH2 spectra 257

17 C.12 [Leu ] PACAP 1-27-Ser(O-Glc)-NH2 spectra 258

C.13 LC gradient for PACAP derivatives 259

C.14 LC-MS3 chromatograms and spectra for PACAP derivatives 260

16 LIST OF TABLES TABLE TITLE PAGE

1.1 Common mass analyzers and their resolution 40

2.1 Sensitivity and range of biogenic amine quantification 71

2.2 Variability in measurements 72

2.3 Comparison to LC-EC method 78

2.4 Previously published Drosophila values 83

2.5 Quantitation of biogenic amines in Drosophila melanogaster 84 whole heads

3.1 Summary of compounds that vary significantly with N. ceranae 110 infection

3.2 Examining the effects of age and infection on compounds in the 111 brains of honey bees

4.1 Fragment masses and identification for quantification of 132 compounds of interest

4.2 Sensitivity and range of peptide quantification 141

5.1 Fragment masses and identification for quantification of Ang 1-7 163 derivatives

5.2 Addition of octyl sulfonate pairing agent improves recovery 170 during ZipTip® purification

5.3 Sensitivity and range of peptide quantification 174

5.4 Lifetime of Angiotensin 1-7 derivatives in vitro 177

5.5 Microdialysis probe selection 183

5.6 Post-experiment probe recoveries for probes used in vivo 187

B.1 Descriptive summary of specific ion transitions involved in MS3 230 quantification of enkephalins

B.2 Relevant amount & concentration values for each of the opioid 242 peptides monitored in this study

17 LIST OF ABBREVIATIONS

Abbreviation Description 3-MT 3-methoxytryptamine 6-OHDA 6-hydroxydopamine ACC Anterior cingulate cortex ACN Acetonitrile aCSF Artificial cerebrospinal fluid ACTH Adrenocorticotropic hormone ANOVA Analysis of variance AUC Area under the curve BBB Blood-brain barrier BOLD MRI Blood-oxygen level dependent magnetic resonance imaging BzCl Benzoyl chloride CFMEs Carbon fiber microelectrodes CI Chemical ionization CID Collision induced dissociation CNS Central nervous system CS Canton S CSF Cerebrospinal fluid Cupr Cuprophane DA Dopamine dAdLE d-Alanine d-Leucine enkephalin DAT Dopamine transporter DCM Dichloromethane DI Direct injection DOP Delta opioid receptor DOPAC 3,4-dihydroxyphenylacetic acid EC Electrochemical detection EI Electron impact ELISA Enzyme-linked immunosorbent assay EM Endomorphin EOPs Endogenous opioid peptides ESI Electrospray ionization FA Formic acid Fmn Fumin FSCV Fast-scan cyclic voltammetry GABA Gamma-aminobutyric acid GC Gas chromatography GPCR G-protein coupled receptor HAc Acetic acid

18 HPLC High performance liquid chromatography ID Inner diameter IS Internal standard KOP Kappa opioid receptor LC Liquid chromatography LDR Linear dynamic range LE Leucine enkephalin LIT Linear ion trap l-l Liquid-liquid LOD Limit of detection MALDI Matrix-assisted laser desorption ionization ME Methionine enkephalin MEKC Micellar electrokinetic chromatography MFB Medial forebrain bundle MOP Mu opioid receptor MPTP 1-methyl-1,2,3,6-tetrahydropyridine MS Mass spectrometry MS/MS, MSn Tandem MS MSE Methionine sulfoxide enkephalin MW Molecular weight nkd Naked cuticle gene NOP Nociceptin receptor NT Neurotransmitter OA Octopamine OD Outer diameter PACAP Pituitary adenylate-cyclase activating polypeptide PAES Polyarylethersulfone PDA Photo diode array detector PER Proboscis extension reflex PES Polyethersulfone PET Positron emission tomography PITC Phenylisothiocyanate Q Quadrupole RAS Renin-Angiotensin system RIA Radio immunoassay RSD Relative standard deviation SEM Standard error of the mean SNL Spinal nerve ligation TEA Triethylamine TFA Trifluoroacetic acid TOF Time of flight VAP Vascular access port

19 VIP Vasoactive intestinal peptide VLS Ventral lateral striatum ZT ZipTip®

20 LIST OF CONTRIBUTORS INITIALS NAME ARM Adam R. Meier CL Chenxi Liu CLK Catherine L. (Kramer) Smith CS Chris Stagg DCF Drew C. Farrell DSK Diana S. Meske EL Eric Lemister EMJ Evan M. Jones EN Edita Navratilova KLP Kate L. Parent LS Lajos Szabo MJB Mitchell J. Bartlett NDL Nicholas D. Laude SAL Sara A. Lewis SC Samantha Calle SLG Stephanie L. Gage

21 Abstract The analysis and quantitation of neurologically-relevant molecules requires detection methods that are sensitive, selective, and applicable to a wide range of molecules.

Targeted analysis using tandem mass spectrometry allows for the detection of molecules from complex matrices with an added level of selectivity. Mass spectrometry is on the leading edge of technological advances and improvements in our understanding of the intricate workings of the brain, allowing us to develop better models and better therapeutic approaches.

In this thesis, I use tandem mass spectrometry to investigate two classes of neurochemicals: classical neurotransmitters, and potential therapeutic drugs based on endogenous neuropeptides. Chapter 1 will introduce existing sampling techniques and detection schemes for small molecule neurotransmitters and small peptides. We will also introduce two key concepts: insect models for understanding human neurotransmission, and the role of the blood-brain barrier in developing CNS-active pharmaceuticals. In

Chapter 2 we develop a method to quantify small molecule neurotransmitters in tissue homogenate for the purpose of understanding how the bulk content of an insect brain can change under differing circumstances. Our approach allows for the analysis of a wider range of compounds with improved throughput compared to existing methods. Chapter 3 expands this method for the quantitation of five biogenic amines in Apis mellifera, to investigate the effect of infection by the microsporidian Nosema ceranae. Chapter 4 explores the role of glycosylation on the stability and blood-brain barrier permeability of peptide-based drugs. Chapter 5 expands this work to a series of Angiotensin 1-7 derivatives, for a study of the effect of different structural modifications to peptide-based drugs, with the goal of driving drug development toward more effective pharmaceuticals.

Chapter 6 concludes this work and outlines the future directions of the research.

22 Chapter 1 Introduction: Quantifying neurochemicals with mass spectrometry

1.1 Introduction Analytical science is problem-driven. The roots of chemistry lay in the discoveries of the alchemists, as they searched for ways to change matter to gold. For centuries, the human body has served as an inspiration to artists and scientists alike, from

Leonardo da Vinci’s Vitruvian Man, to Galvani and Volta’s experiments with electricity that ultimately inspired Mary Shelley’s description of a monster brought to life by the power of electricity. These early experiments paved the way for our modern knowledge of the anatomy, physiology, and chemistry of the human body. From the discovery of the potential for blood transfusion1 to seemingly science fiction-based advances in gene editing using CRISPR that bring to mind shades of Gattaca,2 advances in science and medicine are driven by the technological advances that allow us to investigate ever more complex aspects of the human body.

One system that scientists have been trying to understand for centuries is the brain. This highly complex system is responsible for a vast number of biological processes, from conscious thought and emotion to regulation of heart rate. However, understanding the brain requires probing every level of function, from the molecular to the whole system.

Each level of complexity involves different approaches and analytical techniques. As chemists, we are mainly concerned with measuring perturbations on the molecular scale.

This information can be used to understand how chemical changes affect downstream processes such as behavior, and how we can develop more effective treatments for diseases.

23 Here, we use advances in mass spectrometry and liquid chromatography to investigate two classes of neurochemicals: classical neurotransmitters, and potential therapeutic drugs based on endogenous neuropeptides.

Neurotransmitters play a pivotal role in biology, as they are responsible for the chemical communication that drives many biological processes, from motor behavior to emotional responses. Altered neurotransmission is implicated in many diseases and disorders, including Parkinson’s Disease,3–6 dementia,7 addiction,6,8 schizophrenia,9,10 and mood disorders such as depression.9,11–13 Quantitation of neurotransmitters has been complicated by many factors, such as low biological concentrations, limitations of sampling techniques, and certain inherent challenges to some analytical methods (as will be discussed in section 1.2). In this work, we quantify small molecule neurotransmitters in insect tissue homogenate for the purpose of understanding how the bulk content of an insect brain can change under various conditions, such as genetic mutation or infection with a parasite. Our approach allows for the analysis of a wider range of compounds with improved throughput compared to existing methods.

However, chemical perturbations in the brain do not occur in a vacuum. Local concentrations are constantly changing as a function of biological stimuli, synthesis or degradation, translocation to different parts of the body, and other causes. It is naïve to assume that we can measure fluctuations in chemicals without taking into account the systems at work and how they affect these fluctuations. With this in mind, we began to investigate neurochemicals from a systems perspective- in this case, that of the blood- brain barrier. The blood-brain barrier serves to protect the brain from potentially harmful compounds in the blood, and is of particular relevance to the development of CNS-active pharmaceuticals that must be able to penetrate the BBB to reach their target sites. In this work, we investigate peptide-based drugs and how structural modifications can affect their

24 in vivo lifetime and ability to penetrate the blood-brain barrier. This is accomplished using microdialysis sampling with nanoflow LC-MSn analysis, which allows for quantification of small peptides in a volume-limited complex matrix with high sensitivity and selectivity.

Quantifying the effect of structural modifications on BBB penetration will allow for the development of more effective CNS-active pharmaceuticals in the future.

1.2 Sampling and analysis of neurochemicals Measurement of chemicals in the brain has been approached many different ways, depending on the problem at hand. A scientist has to take into consideration many aspects before embarking on a scientific quest. Which chemicals are being targeted? How can they be measured? What role do they play?

What information is the scientist trying to gain? The answers to these questions will affect the best approach to solve a problem.

In the case of many neurochemical analyses, the measurement technique falls under one of three categories: 1) ex situ, 2) in vitro, or 3) in vivo. Traditionally, ex situ experiments are defined as those made away from a native site. For our purposes, we will define ex situ as tissues that have been removed from their native site, and are limited to a single point in time. Ex situ experiments by this definition occur in Chapters 2 and 3, where the heads or brains of insects are removed, homogenized, and analyzed for biogenic amine content. These experiments are limited to a single time-point.

In vitro experiments are defined as taking place outside a living organism. Herein, we will define in vitro experiments as those that are performed under conditions that mimic a biological system but are not occurring in an organism, and that can be measured over a period of time. Chapters 4 and 5 use rat serum to mimic the biological degradation of peptide-based drugs with time.

In vivo experiments occur in a living system. Classically, the measurements considered in vivo are made within the animal. However, for our purposes, we will classify

25 experiments wherein we are sampling from a living system over a period of time as in vivo.

Chapters 4 and 5 use microdialysis sampling followed by nano LC-MSn analysis for the quantification of blood-brain barrier penetration of peptide-based drugs over a period of time.

1.2.1 Ex situ Long before scientific analysis techniques were mature enough for in vivo sampling, scientists were quantifying chemicals from biological samples. This was mainly done by removing tissues, homogenizing them, and quantifying the components of the sample. For example, adrenocorticotrophic hormone (ACTH), a polypeptide hormone produced by the pituitary gland,14 was quantified in pituitary tissue as early as 1944.15

However, this was done by freezing and lyophilizing the tissue, injecting it into the adrenal gland of another animal, and comparing the concentration of ascorbic acid in the control adrenal gland versus the injected gland, as ascorbic acid concentrations were seen to decrease proportionally to ACTH concentration, using a “photo-electric method”. 15–18 This was followed by assaying its potency against LA-1-A, a purified ACTH standard.16 This method was complicated, introduced a large amount of error, and was not a direct measurement of the target of interest.

In order to circumvent these limitations, scientists began to apply separations. Samples can be purified based on the inherent characteristics of the chemical components, such as solubility, size, or binding ability. Early separations often took advantage of small molecules that could easily be extracted or proteins with specific binding that could be used to produce an assay.

1.2.1.1 Affinity assays Affinity assays, also known as binding assays, measure the interaction between two molecules.19 These interactions can be measured in and of themselves (for example pharmacokinetic/pharmacodynamic studies), or can be used to extract target compounds from a solution. Two of the most common type of the latter are

26 radioimmunoassays (RIA) and enzyme-linked immunosorbent assays (ELISA). RIA and

ELISA are used to measure the concentration of antigens by use of antibodies.20,21 These methods are highly selective, and can be used for any antigen for which a specific antibody can be found or developed (though this is easier said than done with structurally analogous molecules).

Radioimmunoassay, first described for the measurement of insulin,21 is based on the competitive binding of radiolabeled versus unlabeled antigens and antibodies.20,22 The amount of antibody and radiolabeled antigen is known, and by measuring the bound/free radiolabeled antigen ratio at different concentrations of added unknown, is it possible to calculate the unlabeled antigen in the sample. While RIA can be highly sensitive, with detection limits down to 10 amol for some species,23 it is limited to species with specific antibodies, otherwise cross-reactivity will occur. Additionally, radioisotopes can be expensive and inherently have short lifetimes if they are decaying rapidly enough to be useful.

Enzyme-linked immunosorbent assay (ELISA) is similar to RIA, but relies upon the binding of antigens to enzyme-linked antibodies. The substrates for the enzyme are used for quantification.21 ELISA avoids the safety concerns of radioactivity, but still requires highly specific antibodies. Commercially-available ELISAs are reported to have sensitivities in the low femtomole regime.24

RIA and ELISA are sensitive and selective quantitation methods, but their selectivity is also a limitation. RIA and ELISA are restricted in the number of analytes that can be quantified in a sample. In many cases, researchers wish to quantify more compounds in a sample. This is where separations come into play.

1.2.1.2 Separation and detection techniques One of the earliest separation techniques is liquid-liquid (l-l) extraction. This technique takes advantage of the relative solubility of

27 molecules in immiscible liquids. By adjusting the characteristics of the solutions (identity, pH, salt content, etc.) it is possible to optimize for a target compound, but in a complex matrix l-l extraction will still yield a complex sample. Liquid-liquid extraction can be used to slightly purify a sample, but cannot be used for identification; it must be coupled to an analytical detection method. However, many detection methods do not work well with mixtures, and samples require further separation.

What we now consider to be instrumental liquid chromatography has been a staple of separation science since the early 1970s, though the first instrument was reported by

Spackman, et al. in 1958.25 Separation can be accomplished using many different chemical distinctions between molecules, including size, shape, hydrophobicity, and charge. High-performance liquid chromatography (HPLC) is the only technique used for the research presented herein.

HPLC is characterized by flowing a liquid mobile phase containing the analytes of interest through a column packed with a stationary phase material under pressures in the thousands of PSI regime.26 The separation is accomplished due to differences in the ratio of time that analytes spend in the stationary phase versus the mobile phase. Separations can be optimized based on the composition of the mobile phase, the composition of the stationary phase, flow rate, and several other parameters.

With the advent of column separations, it was possible for scientists to separate many of the components of a homogenized sample, and the ability to couple different detectors to

HPLC (such as UV/Vis, refractive index, fluorescence, electrochemical, mass spectrometry, and more) meant more information could be gleaned from complex samples. Spackman’s original instrument was developed to separated amino acids and used a column that was 150 cm long x 0.9 cm in diameter, a flow rate of 30 mL/h, and took over 1000 minutes (more than 16 hours) to run. He was able to quantify micromoles

28 of amino acids in 2-mL samples.25 Liquid chromatography has made great strides since

Spackman’s work. Of particular relevance to this work, miniaturized LC with nanospray

ESI-MS allows for the identification and quantitation of thousands of peptides in pmol/μL samples at flow rates less than 1 μL/min on columns that are 10-cm long with 75-μm inner diameters.27 This is due to advances in both chromatography and mass spectrometry.

Reproducibility in flow due to improvements in pump design, reproducibility in column packing and packing material, and fabrication of zero dead volume fittings capable of withstanding higher pressures all play a role in improved chromatography. Recent advances in nanospray technology using electrospray nozzles less than 500 μm in diameter stabilize low-flow spray (< 1 μL/min) and make it possible to analyze volume- limited samples.28,29

While there are many detection methods that are coupled to separations, most have inherent limitations. Liquid chromatography with electrochemical detection (LC-EC) is a well-characterized and sensitive technique, with femtogram limits of detection reported in

1980.30–34 However, techniques utilizing electrochemical detection require electrochemically active species, limiting the compounds that can be probed. Additionally, electrochemical detection can be non-specific, so the separation must be robust. UV/Vis detection is limited to species that absorb within a certain wavelength range, and is not as sensitive with LODs in the single μM – hundreds of nM regime.35–37 Fluorescence detection limits have been reported in the sub-pM regime,38 but require fluorescent analytes or derivitization. By comparison, mass spectrometry can be used to detect virtually any ionizable molecule, and using LC-MS, sensitivities in the pM regime are reported in the literature and in this work.39–41

With the advances in separation science, ex situ testing remains a useful tool for understanding biological systems. The downfall to ex situ sampling is that it is inherently

29 quantifying a moment frozen in time. In order to study systems in depth, scientists want to gauge the effect of perturbations upon the system, which is where in vitro and in vivo techniques come into play.

1.2.2 In vitro In vitro techniques attempt to mimic biological systems without the complexity or expense of in vivo research. For example, cell culture models can be used to investigate neurotoxicity or neuroprotection,42–44 to evaluate therapeutic penetration into a tissue,45,46 or even to develop models for structures such as the blood-brain barrier and evaluate new pharmaceuticals.47,48 However, in vitro techniques are inherently less complex than the systems under investigation. While intentional, and clearly useful, this is a drawback for understanding systems as a whole.

1.2.2.1 Biological fluids One way to mimic a biological system is to take advantage of components of that system, such as blood, serum, or cerebrospinal fluid. Particularly in drug development, determining the functional lifetime of a drug under biological conditions is important. In Chapters 4 and 5 we apply an in vitro method to study the degradation of peptide-based drugs in rat serum as a mimic for their stability in an experimental animal.

Blood serum contains most of the components of whole blood, (including peptides, proteins, antibodies, small molecules, enzymes, and hormones) but does not contain clotting factors or blood cells. While less complex than whole blood, serum does allow for the measurement of degradation due to factors such as peptidases, proteases, and oxidation.49 It can also be used to account for specific or non-specific binding to proteins in solution (as will be touched upon in Chapter 5). Additionally, it is less expensive than in vivo experiments.

Cerebrospinal fluid (CSF) is also a useful biological fluid for in vitro methods. CSF is found in the brain and spinal cord, and serves as a medium for nutrient and waster transfer, and as a shock absorber for the brain.50,51 CSF contains fewer compounds than blood, due to

30 the regulation of the blood-brain barrier, which will be discussed later. CSF can also be used to mimic biological behavior in terms of binding or degradation,52 or to model diffusion and fluid flow processes in the brain.53 Admittedly, the main research interest for CSF analysis is quantitation of the unbound or “free” brain concentration of CSN-active drugs, which falls under in vivo methods, by our definitions.53,54

Another in vitro technique for mimicking biological systems is the use of cell models.

1.2.2.2 Cell models Cell models are arguably one of the most common model systems for studying biology outside of in vivo animal experiments. HeLa cells, an immortalized cervical cancer cell strain, are widely used in cancer drug development and delivery research.55–57 PC12 cells are undifferentiated cells derived from a rat adrenal pheochromocytoma.58 PC12 cells can be used to model the behavior of neuronal cells, from peptide uptake,59 to simulate neuronal growth, investigate causes of and protection against neuronal death,60 or measure neurochemical release.61,62

It is also common to co-culture several types of cells to mimic tissues. The most relevant example for our purposes is the recent successes in modelling of the blood-brain barrier, using a combination of endothelial cells and glial cells.63,64 This model will be discussed in greater detail later in Chapter 1.

In vitro methods can be useful for modeling biological systems, and are less expensive than in vivo experiments. However, they are inherently less complex than the actual system under investigation, and thus may not be perfectly representative. For this reason, it is sometimes necessary to directly study systems in vivo.

1.2.3 In vivo While ex situ testing is useful for comparing averages, it cannot be used for tracking changes with time in individuals. For example, Chapters 2 and 3 use ex situ sampling to compare whole-head biogenic amine content across different genetic strains,

31 but this method could not be used to monitor changes in dopamine signaling in a single animal over the course of a day. In vitro techniques mimic living systems but tend to be more static and less complex. In vivo measurements are made in a living system and can provide useful information about changes occurring as a function of certain stimuli, with some sort of temporal resolution.

1.2.3.1 Direct measurement techniques There are many techniques for monitoring chemicals in the brains of living animals, from classic techniques such as electrochemistry, to more modern imaging methods such as positron emission tomography (PET). One of the most common in vivo chemical measurement techniques is electrochemistry.

Electrochemical measurements in vivo involve the implantation of a microelectrode. At the surface of the electrode, electro-active molecules can be oxidized or reduced by a small voltage change, resulting in an electron flux that creates a measureable current through the electrode. The current created can be used to calculate concentration.

Electrochemical measurements can have temporal resolution in the millisecond regime,65 often being used to quantify vesicular release from single cells,62,66 but have the downfall of a lack of specificity. Currently, specificity can only be gained in a few ways: 1) modification of electrodes for selectivity toward particular molecules,67,68 or 2) tailoring of applied waveforms to target particular molecules.69–71

1.2.3.2 Microdialysis Another approach to in vivo analysis involves sampling from the system, rather than measuring chemicals in the system. Two common sampling methods are push-pull perfusion and microdialysis. The two methods are similar on the surface: a probe is implanted that allows for sampling from the matrix. The difference lies in how the sample is collected. In push-pull perfusion, first reported for sampling from the CNS in

1962,72 two cannulae, either side-by-side or concentric, are implanted into the region of interest. Through one cannula a fluid is perfused into the brain, through the other,

32 removed. This allows for direct sampling of the compounds in the region of interest.73,74

However, push-pull perfusion mainly fell out of favor for sampling from the brain due to its tendency to create lesions caused by flow through the cannulae.73 It is now used mainly to sample from liquid-heavy regions, such as the extracellular fluid of the spinal cord,75 although there have been studies focused on limiting this damage using low-flow push- pull perfusion.76,77

In microdialysis sampling, the probe is comprised of two concentric cannulae. The outer cannula ends in a semi-permeable membrane with a set molecular weight cut-off. Solution is flowed into the probe through the inner cannula, and as it flows through the second cannula, molecules can permeate across the membrane as a function of the concentration gradient.73,78 Figure 1.1 shows an illustration of a microdialysis probe with a blowup of the permeation of compounds in the brain across the probe membrane.

There are several benefits to microdialysis sampling. 1) Decoupling sampling from detection allows for a wider range of detection schemes. With electrochemical detection only electroactive species can be monitored, but microdialysis sampling can be coupled to mass spectrometry, allowing for the detection of any ionizable species. 2) Microdialysis can be performed in awake and behaving animals, allowing for detection of more biologically-relevant responses. Compared to direct sampling, microdialysis does not change the bulk volume. Small volume changes in cerebrospinal fluid can have extreme effects on the health and behavior of animals and should be avoided if possible. 3) The microdialysis probe membranes can be used as a first level of filtration, preventing proteases and enzymes from entering the dialysate and degrading target compounds during collection.73 4) Studies have shown that the blood-brain barrier regains its integrity after probe implantation, which will be relevant in Chapters 4 and 5.79

Figure 1.1 Principles of microdialysis. Modern microdialysis probes are designed with a concentric inlet and outlet. Perfusate flows in via the internal inlet probe, and target molecules permeated across the membrane as the solution is flowing. The membrane has a molecular weight cut-off that prevents the permeation of large molecules. Typical lengths for the probe are 0.5 - 5 mm with 0.25 – 0.5 mm diameters.

Adapted from Ungerstedt 1991, with permission.78

34 In the low- to no-flow atmosphere of the brain, permeation across the probe membrane is diffusion-limited. Diffusion rates in solution can be described by the Stokes-Einstein equation (Eqn 1.1), which states that the diffusion coefficient D of a molecule is directly related to the temperature and inversely related to the viscosity of the solution η, number of molecules N and size (radius) of the molecule r.

푅푇 퐷 = Eqn 1.1 6휋휂푟푁

Thus, diffusion is inversely related to the size of the molecule, leading to improved recovery for smaller molecules due to larger diffusion coefficients. However, in the case of implanted probes, diffusion is further affected by the tortuosity of the matrix, i.e. “the extent to which the cellular obstructions, such as the bodies or processes of cells, hinder the movement of substances”.80 Microdialysis samples from the interstitial volume, which is theorized to be about 80% of the volume of the brain, but diffusion is hindered by the tortuosity which effectively increases the path length of travel for molecules.80

Probe membranes require biocompatible materials for implantation. Some common commercially available membrane materials are cellulose (cuprophane), polyethersulfone

(PES), polyarylethersulfone (PAES). Selection of membrane material is typically dependent upon target molecular weight (MW) cut-off. For example, CMA Microdialysis, a subsidiary of Harvard Apparatus, sells cuprophane membranes with a cut-off of 6 kDa,

PAES with 20 kDa, and PES with 100 kDa

(http://www.microdialysis.se/us/products/probes). Smaller MW cut-offs will serve as filters and prevent larger molecules such as proteases from reaching the dialysate. MW cut-offs are determined based upon the target mass of the molecules under investigation.

Microdialysis flow rates are typically 0.5 – 10 μL/min.81,82 As microdialysis sampling is diffusion limited, lower flow rates lead to increased recovery.83 However, they also have

35 lower temporal resolution; for example, a flow rate of 1 μL/min will have a ten-minute temporal resolution to collect 10 μL. Microdialysis is limited in recovery, due to the inherent decrease in concentration as a function of slow diffusion in the brain and through the membrane compared to direct sampling.

Microdialysis has a few limitations, namely that it is 1) invasive, 2) has poor spatial and temporal resolution, 3) requires a sensitive detection scheme due to low volume and low concentration samples, and 4) probes can be expensive. Compared to direct measurement techniques such as implanted carbon fiber microelectrodes for electrochemical detection, microdialysis sampling results in a loss of spatial and temporal resolution. Microdialysis probes tend to be 0.5 – 5 mm in length, and 0.25 – 0.5 mm in diameter, compared to carbon fiber microelectrodes (CFMEs) which are on the scale of

30 μm in length and single microns in diameter.52 Additionally, CFMEs can have subsecond temporal resolution, whereas microdialysis is typically 5 – 20 min temporal resolution.

However, what is lost in spatial and temporal resolution with microdialysis sampling, is gained in chemical resolution. Microdialysis sampling opens the door to alternative detection methods, such a mass spectrometry, which can be used to monitor a wider range of species. The advent of HPLC, especially nano-HPLC, has led to microdialysis becoming a quite popular technique for monitoring biochemical events due to increased sensitivity.73,79,83–85 Coupling microdialysis sampling to nano-LC-MSn has resulted in limits on detection (LODs) on the single pM scale. Li et al. reported 0.5 – 50 pM LODs for neuropeptides in 4 μL samples.81

1.3 Tandem mass spectrometry Mass spectrometry, meaning quite literally “the measurement of mass” in its earliest form dates all the way back to J.J. Thomson’s experiments with cathode rays.86 Mass spectrometry as we understand it today is defined

36 by the generation of ions and the measurement of the mass to charge ratio of these ions in the gas phase. Instruments are typically comprised of three components: 1) the ionization source, 2) the mass analyzer, and 3) the detector, though in some cases the mass analyzer and the detector are the same, such as FT-ICR and Orbitrap instruments.

For decades mass spectrometry could only be performed on molecules already in the gas phase; in fact, many of the mass analyzers still used today predate what we now consider the most common ionization source: electrospray ionization (ESI), used to ionize liquid samples.87 Prior to ESI, the most common method of ionization was electron impact (EI), a hard ionization method that causes molecules to fragment during ionization. However, many biomolecules were not volatile enough to reach the gas phase, and could not be analyzed with this method.87 With the advent of chemical ionization (CI) and ESI in the late 1960’s,87 it became possible to not only ionize biomolecules, but to detect them in their whole state, rather than inherently fragmented, as with EI. In fact, ESI has become so important and pervasive in the world of analytical chemistry that its inventor, John Fenn, was awarded a quarter share of the Nobel Prize in Chemistry in 2002.

The most common ionization methods in use today are EI, CI, ESI, and MALDI (matrix- assisted laser desorption ionization), though new ionization techniques are constantly being developed to approach new technical challenges.87–89 However, we will limit the discussion to electrospray ionization, as it is the only technique applied herein.

In positive-mode ESI, a liquid (typically a miscible aqueous:organic mixture with some acidic content) is flowed into the ESI source. The ESI needle is heated, sheathed in an inert desolvation gas, and a voltage difference is applied between the ESI needle and the inlet to the instrument. Figure 1.2 illustrates the principles of electrospray ionization, from

Staniforth and Stavros, 2013.90 For an in-depth review of ESI, see Banerjee and

Mazumdar, 2012.91

Figure 1.2 Principles of electrospray ionization. Samples are solubilized in a slightly acidic aqueous/organic mixture (for positive mode). This solution is sprayed through the nozzle, to which a voltage is applied, typically 2 – 5 kV. The spraying ESI solution forms a Taylor cone and breaks into charged droplets. A sheath gas, typically N2 or Ar, desolvates the droplets, which undergo coulombic explosion as they reach their

Rayleigh limit. By the time the sample reaches the inlet, it is composed of desolvated charged analyte molecules. From Staniforth and Stavros 2013, with permission.90

38 Mass analyzers come in many shapes and sizes, and are typically selected based on two parameters: 1) which ionization source is used, and 2) the target resolution of the experiment. Ionization sources are typically either pulsed or continuous, and traditionally the mass analyzer was selected based on which type of ionization source was more compatible. For example, ESI is a continuous source, and works well with quadrupole mass analyzers, whereas MALDI, and other pulsed sources, work well with time-of-flight

(TOF) mass analyzers. However, with improvements in ion optics and traps, these challenges have mainly been overcome.

Resolution in mass spectrometry is defined as

푚 푅 = Eqn. 1.2 훥푚 where m is the mass to charge ratio of an ion and Δm is the peak width. Some common mass analyzers and their resolutions are listed in Table 1.1. Many of the mass analyzers can be coupled to accomplish additional goals. For example, a TOF-TOF, also called a reflectron-TOF, increases the resolution of a typical TOF instrument by more than 100x.

Ion traps are commonly placed before pulsed mass analyzers so they can be coupled with continuous ionization sources like ESI. It is also common to couple mass analyzers to allow for the collection of both parent and fragmentation spectra for a wider range of useful information. One common example of this is a “triple-quad”, or QQQ, that has three subsequent quadrupoles. In full MS, only one is used. However, for the collection of fragmentation spectra, the first quadrupole is used for selection, the second for fragmentation, and the third to scan through the fragment masses to produce a spectrum.

In this work we apply tandem MS for added sensitivity and selectivity. Tandem MS, also referred to as MS/MS or MSn, is a technique where ions are subjected to an additional level or levels of mass spectrometric analysis by inducing fragmentation.92 Fragmentation

39 Table 1.1 Common mass analyzers and their resolution. Mass analyzers are typically selected based on the ionization source and the required resolution for a particular experiment.

Mass Analyzer Pulsed or continuous Resolution

Quadrupole (Q, quad) Continuous unit

Time-of-flight (TOF) Pulsed ~1000 - 20,000

Ion Trap Pulsed 103 - 104

FT-ICR Pulsed >106

Orbitrap Pulsed >106

Sector Continuous ~105

40 occurs at the most labile bonds in a molecule, yielding highly specific and reproducible structural information. Fragmentation is most commonly accomplished using collision- induced dissociation (CID), which was introduced in the early 1960’s. In CID, gaseous ions enter a collision cell where they collide with an inert collision gas, such as helium, nitrogen, or argon. Kinetic energy is transferred during the collision, resulting in dissociation of the activated species, as shown in Eqn 1.3

퐴퐵+ + 푁 → 퐴퐵+∗ + 푁 → 퐴+ + 퐵 + 푁 Eqn 1.3

In some instruments, this process can be repeated to the nth degree- limited only by detection of the ions, as some ions are lost in each fragmentation step. This fragmentation efficiency, or how many of the ions are transmitted through each fragmentation step, plays an important role, as will be discussed later. The two main instruments used in this research are an ESI-LTQ-Orbitrap and an ESI-Q-TOF, (LT = linear trap, Q = quadrupole).

QTOF instruments can traditionally only perform one level of fragmentation (MS2), whereas trap or triple quad instruments can perform several levels of fragmentation (MSn).

We can apply these techniques for highly selective and specific detection of molecules for quantitation in complex matrices.

Additionally, mass spectrometry can be coupled to many biological sampling techniques.

The main analytical challenges to using MS for detection of biological species are 1) sample complexity, and 2) salt content. The sample complexity is a challenge for most analytical techniques, and is typically handled with sample preparation procedures and/or chromatographic separations, which can usually be coupled with a mass spectrometer.

The salt content of biological samples is a challenge in mass spectrometry because high salt tends to have a negative effect on the stability of ESI spray and the ionization of compounds within the sample. A common desalting procedure for biological samples in preparation for MS analysis is the use of a ZipTip®, a small micropipette tip packed with

41 stationary phase. ZipTips® act as small columns, binding the target molecules in a sample, allowing them to be washed and then eluted into a more compatible solution.

With these approaches, it is possible to quantify a wide range of compounds in biological samples with high sensitivity using mass spectrometry. An additional benefit to using mass spectrometric detection is that it can act as an orthogonal separation, in that it separates species by m/z in addition to the chemical characteristics by which they are separated in a chromatographic separation (such as hydrophobicity). This allows for the separation and identification of a larger number of species, inherently yielding more information.

Tandem mass spectrometry provides structural information about the species in a sample that could not otherwise be probed. In this work, we focus on targeted analyses, where rather than identifying or quantifying as many of the compounds in a sample as possible, we focus on quantifying specific compounds. While tandem MS is often used to glean structural information, in this case we use it to add another level of selectivity to our analyses. Sample preparation and analysis of complex biological mixtures can require in- depth procedures,93,94 and by using tandem MS for selectivity we can circumvent some of the steps involved. The selectivity of our measurements is then determined by the selection and fragmentation of target masses, followed by quantitation of the previously identified fragment masses. Particularly when coupled to a chromatographic separation, high sensitivity and selectivity can be achieved while adding the identification capabilities of tandem MS.95 In this work, we apply tandem MS for the quantification of small molecules and peptides in animal models.

1.4 Neurotransmission in insects Animal models are currently our best method for studying disease states and treatment approaches under controlled conditions.96–98

However, mammalian models of disease have several drawbacks, including cost, time

42 investment, and ethical concerns.99–104 Insect models circumvent many of these concerns, and models for human disease have been discovered or genetically induced in several insect species.105–109

A common model for neurological disorders is Drosophila melanogaster, the fruit fly.

Drosophila are a powerful tool for studying disorders due to their phylogenetic homology with humans, which means that many fundamental mechanisms are remarkably well conserved.110,111 About 75% of human disease genes have fly homologs.111 Models of

Parkinson’s Disease, Huntington’s Disease, and Alzheimer’s Disease have all been developed.110,112 Our ability to probe these systems directly translates to our ability to understand function and malfunction in humans. Another benefit of Drosophila models is that Drosophila are easy to rear, inexpensive, have short lifespans, and have several external features that can be manipulated for easy phenotyping of mutants.113,114

Insect research is also important for industries such as agriculture. Honey bees (Apis mellifera) play an important role in crop pollination, and diseases that affect honey bee populations or behaviors could markedly affect agricultural output.115–119

Neurotransmitters play a role in behavior and many diseases are characterized by changes in neurotransmission. For example, Parkinson’s disease results from selective neurodegeneration of dopaminergic neurons and its main pharmacological treatment is with the dopamine precursor L-DOPA.34 The ability to quantify neurotransmitters in insects can help elucidate chemical changes that occur as a function of behavior or disease states.

1.4.1 Classical neurotransmission The classical definition of neurotransmission began with Santiago Ramón y Cajal’s discovery of neurons and the subsequent acceptance of the Neuron Doctrine, which asserts that the nervous system is comprised of discrete cells.120 We now know that neurons communicate by sending chemical signals across the

43 synaptic cleft, or the space between the axon terminal of the pre-synaptic neurons and the dendrites of the post-synaptic neuron. Neurons are stimulated by depolarization, which occurs when the difference in the ion concentration outside of the cell and inside of the cell cause the cell membrane to reach a threshold potential. At that point, the action potential is generated by a sudden flux of sodium ions entering the cell as sodium channels open. The action potential fires down the axon until it reaches the axon terminals, where it stimulates the release of neurotransmitters from vesicles into the synaptic cleft to stimulate the next neuron.3 An illustration of the chemical neurotransmission between a presynaptic and postsynaptic neuron is shown in Figure 1.3. Neurotransmitters released into the synaptic cleft typically undergo one of three routes of clearance, 1) re-uptake by the presynaptic neuron, 2) uptake by receptors on the postsynaptic neuron, or 3) diffusion away from the active site and subsequent degradation.

The criteria for a compound to be classified as a neurotransmitter (NT) are as follows: 1)

An NT must be synthesized by and released from neurons. 2) The NT must be released in a chemically or pharmacologically recognizable form. 3) The NT should stimulate the same effects in the post-synaptic neuron that stimulated the pre-synaptic neuron. 4) The effects should be blocked by competitive antagonists in a dose-dependent manner. 5)

There should be active mechanisms to terminate action.3 Classical neurotransmitters are those such as acetylcholine, the biogenic amines, and amino acid neurotransmitters, versus other compounds that fit the classification of an NT, such as some peptides.

Dopamine is a catecholamine neurotransmitter, i.e. a molecule that contains both a catechol and an amine. Dopamine is associated with the reward/motivation pathway and controls aspects of locomotion, sleep, arousal, and addiction.121–124 L-DOPA is produced from the amino acid tyrosine. Epinephrine and norepinephrine are catecholamines and downstream products of dopamine. More colloquially known as adrenaline and

Figure 1.3 Neural transmission. Depolarization of a neuron causes the cell membrane to reach a threshold potential. An action potential is generated by a sudden flux of sodium ions entering the cell as sodium channels open. The action potential fires down the axon until it reaches the axon terminal, where it stimulates the release of neurotransmitters from synaptic vesicles into the synaptic cleft. Neurotransmitters can be taken up by the receptors on the post-synaptic neuron to cause further stimulation, be reclaimed by re-uptake pumps on the pre-synaptic neuron, or diffuse away. From

Tawfik 2016, with permission.189

45 noradrenaline, they play a role in the fight-or-flight response. In clinical and laboratory research on normal physiology, as well as disease pathogenesis and treatment efficacy, catecholamines are measured as indicators of acute or chronic stress responses.125,126

Another monoamine neurotransmitter is serotonin, which plays a role in neural development, aggression, and depression.127,11,128 Serotonin is produced from the amino acid tryptophan..

1.4.2 Quantitation of neurotransmitters Neurotransmitter measurement has been occurring for decades, starting with Henry Hallett Dale’s identification of acetylcholine in

1914 and Otto Loewi’s subsequent frog heart experiment, for which the two men shared the 1936 Nobel Prize in Physiology or Medicine.

Quantitation of neurotransmitters is often performed with liquid chromatography-coupled techniques such as electrochemical detection (LC-EC) or mass spectrometry (LC-MS).

Amperometric detection requires a front-end separation due to its lack of specificity. LC-

EC is a well-characterized and sensitive technique (picogram limits of detection were reported in the early 1970s and femtogram limits of detection were reported in 1980).30–34

Additionally, electrochemical detection has a linear dynamic range over several orders of magnitude, which can be of great benefit in some experiments. However, techniques utilizing electrochemical detection require electrochemically active species, limiting the compounds that can be probed. For example, dopamine and serotonin are electroactive, whereas gamma-aminobutyric acid (GABA) and glutamate are not. In contrast, mass spectrometry can be used for any molecule that can be ionized.

Some researchers have approached the problem with more resourceful methods, such as fast-scan cyclic voltammetry (FSCV) and micellar electrokinetic chromatography (MEKC).

These specialized approaches also have certain inherent limitations. Background- subtracted FSCV adds selectivity over typical electrochemical detection methods, has

46 subsecond temporal resolution, and is sensitive (into the pmol regime), but is still limited to electrochemically active analytes.129 MEKC-EC techniques were developed due to their high sensitivity in low-volume samples (femtomoles of material in nanoliter volumes), which is necessary for mass-limited biological sources such as the fruit fly Drosophila melanogaster.130 However, the inherent limitations of electrochemical detection are still present, and MEKC cannot be directly coupled to a mass spectrometer because of the surfactants used in the separation.

Determination of the optimal approach for quantitation is determined by several factors:

1) what is the origin of the sample? 2) how is the sample collected? 3) which compounds are being quantified? These questions alter the approach. For example, microdialysis sampling works well in a rat, but not so well in fruit flies, due to their small size.

Compounds that are not electroactive cannot be quantified by electrochemistry.

Researchers must decide the approach based on the problem. In 2009, Perry et al. published a review of the analytical techniques for the determination of neurotransmitters and came to much the same conclusion.131

Chapters 2 and 3 present our work in the quantitation of five biogenic amines in insect tissue homogenate. Most existing methods use chromatographic separation and electrochemical detection, leading to a lower throughput and limitations on the types of compounds that can be analyzed. Our method uses a rapid l-l extraction followed by direct injection tandem MS, which improves the throughput to less than 10 minutes per samples for preparation and analysis but has limits of detection on par with existing methods.

1.5 Drug penetration of the blood-brain barrier Originally observed by Ehrlich in the early 1900s during studies with dye, the blood-brain barrier (BBB) is still not fully understood.132,133 The blood brain barrier is comprised of the cerebral capillary endothelial cells that form tight junctions formed mainly of the protein occludin.134,135 It serves to

47 protect the central nervous system (CNS) from harmful compounds like neurotoxins, bacteria, and some macromolecules, while simultaneously allowing necessary compounds like sugars, salts, and small-molecule precursors to neurotransmitters to penetrate.134,136 For drugs meant to target diseases of the central nervous system, the ability to penetrate the blood-brain barrier is paramount. Few small molecule drugs can cross the BBB- greater than 98% of small-molecule drugs do not penetrate the BBB.137

Even fewer peptide-based drugs are capable of crossing the BBB.137 The BBB provides a challenge to development of CNS-active pharmaceuticals.138

1.5.1 Mechanisms of BBB penetration The blood-brain barrier separates the blood from the cerebrospinal fluid, which is contained in the ventricles and subarachnoid space, and secreted by the choroid plexus.139 The tight junctions of the endothelial cells prevent most molecules from crossing the BBB, except by specific mechanisms. These mechanisms are illustrated in Figure 1.4.139 Small hydrophilic molecules can be transported via a paracellular aqueous pathway (a). However, this pathway is restricted by the tight junctions, and few water-soluble compounds (such as polar drugs) penetrate. Gaseous molecules like O2 and CO2, and lipophilic compounds can enter through the lipid membranes (b) via diffusion.140 This mechanism is non-saturable.141 Transport proteins

(c) can transport certain necessary compounds across the BBB, such as glucose and amino acids.142 A few examples are the GLUT1 and Pgp transporters, which transport glucose and P-glycoprotein, respectively. Some of these transporters also act as efflux transporters. Any other species that penetrate must do so by transcytosis, either receptor- mediated (d) or adsorptive (e).139,143 Transport can be increased in some cases, such as cationization, which improves the BBB penetration of albumin. In some cases, poly- arginine tails have been added to peptides to increase their basicity and thus their BBB penetration.143,144 Most drugs target pathways b-e. Pathway b, which depends on

Figure 1.4 Mechanisms of penetration of the blood-brain barrier. There are five main pathways by which compounds can cross the blood brain barrier. a) Water-soluble agents can pass through the tight junctions. Greater than 98% of water-soluble small molecules are excluded. b) Lipophilic compounds can diffuse across the lipid membranes. c) Transport proteins transport specific target molecules, such as glucose.

The remaining compounds that penetrate must do so via transcytosis, either d) receptor- mediated, or e) adsorptive. Most drugs target b-e. From Abbott 2010, with permission.134

49 lipophilicity, is the easiest to target with structural modifications.137,139,144 However, this does pose the risk of altering the drug’s CNS activity or making it a target for the Pgp efflux transporter.142,144,145 Another option is to target receptors with endogenous ligands, taking advantage of the “Trojan horse” approach by linking a drug to a ligand with a known transporter.138,141 In one study, the cerebral vasodilatory vasoactive intestinal peptide (VIP) was conjugated with a monoclonal antibody, resulting in increased cerebral blood flow, which served as evidence of BBB penetration.137,146 However, modification of endogenous ligands can lead to decreased transport of these ligands via their typical routes.147

1.5.2 Peptide-based drug development While more small molecules cross the BBB than peptides, small molecule drugs often have unanticipated side effects due to non-specific binding to receptors, whereas drugs derived from endogenous peptides have high affinity for pre-existing receptors.148–150 Peptide-based pharmaceuticals show great promise due to their high specificity and biocompatibility. Thus, recent focus has been on development of endogenous peptide-based drugs that will bind with high specificity, resulting in more efficient drugs with fewer side effects from non-specific binding.149,151,152 A mechanism of action based on natural processes of the body is less likely to have deleterious side effects. Additionally, peptide-based drugs can typically be broken down into non- hazardous degradation products, as they are comprised of naturally-occurring amino acids.

Peptide-based drugs do have one major downfall that has limited their development; compounds based on endogenous species tend to be rapidly degraded in the body, often too quickly to be of use as a therapeutic treatment.150 In order to develop highly effective peptide-based drugs, many researchers have explored ways to increase the lifetime of peptide-based drugs in vivo as well as their ability to permeate the BBB.137,153–158 Some approaches include enzyme inhibition or masking of enzyme target sites for increased

50 lifetime through cyclization via thioether or disulfide bridges, addition of a sugar moiety, halogenation, enzyme inhibition, and replacement of native residues with D-amino acids.141,155,157,159–162 Glycosylation can improve in vivo lifetime by altering the conformation of the peptide or hindering peptidases, and there is evidence that it may also improve BBB penetration, though the mechanism is unknown.49,161,163

This approach is now being taken to develop peptide-based drug candidates for other disorders. For example, the endogenous opioid leucine enkephalin (YAGFL) was modified to develop the drug MMP-2200, AKA lactomorphin, a glycosylated opioid peptide and a mu and delta receptor agonist164 with potential for alleviating the effects of dopaminergic hyper-stimulation.165 Derivatives of PACAP, or pituitary adenylate-cyclase activating polypeptide, are under investigation for neuroprotective effects.43,44,166–172

Glycosylated derivatives of these compounds are investigated in Chapter 4. Angiotensin

1-7, a Mas receptor activator, has been shown to provide cardioprotective effects, protective effects in hepatic fibrosis, and neuroprotective effects in cases of ischemic stroke.173–176 Structural modifications of Ang 1-7 and their effect on the in vivo stability and

BBB penetration are investigated in Chapter 5.

1.5.3 Current measurement techniques and challenges Due to the importance of the

BBB on pharmaceutical development, researchers have been trying to find ways to study

BBB permeation. Early studies were performed with injection of visible dyes and subsequent histology, but this has the obvious downfall of being a postmortem procedure.

Another approach to studying the permeability of compounds across the BBB is the use of radioactive compounds and positron emission tomography (PET) for detection.177

However, these studies require the synthesis of radioactive compounds, the assumption that these compounds behave the same as non-labeled compounds, and additionally has

51 relatively poor spatial resolution (several millimeters).177 In fact, this method is used more often to study disruption of the BBB than specific compounds and their ability to cross.178

More recently, some researchers have been developing in vitro cell culture models using endothelial cells, often with a combination of endothelial cells and astrocytes to mimic the cellular structure of the BBB as closely as possible.47,64,179,180 In general, the gauges of a

BBB cell model are: 1) reproducibility of results, 2) saves time and money compared to in vivo experiments, 3) cell types and transporters are characteristically represented, and 4) permeability of compounds matches observed in vivo permeability. If a cell model fits these requirements, it can be used to model the BBB penetration of potential therapeutics.64 Due to the complexity of co-cultures, and the challenge of optimizing the cell-type ratios, some researchers use endothelial cell cultures only.171 However, the most successful in vitro BBB models to date use the co-culture of endothelial cells with glial cells, as there is evidence that the glial cells strongly affect the microenvironment of the

BBB.64 (Though there is some disagreement as to the importance of astrocytes on BBB penetration.47) In a 2007 study, brain endothelial cells from an immortalized rat cell line were co-cultured with glial cells, and showed representative morphology in the endothelial cell monolayer as well as tight junctions formed by occludin, serving as a basic BBB model.64 Recently researchers have developed even more complex cell models that co- culture up to four cell types (brain endothelial cells, pericytes, neurons, and astrocytes), resulting in good correlation between penetration data in vitro and in vivo.48,179 However, despite the increasing complexity of these models, it is difficult to develop in vitro cell models that perfectly mimic the BBB in terms of types and quantities of receptors, cell- type ratio, microvasculature, and transport mechanisms.63,133 Additionally, endothelial cells in the BBB have been shown to have different properties than peripheral endothelial cells, such as an increase in mitochondria, which means that the origin of the cells used

52 in culture models is important.147 Finally, at some point the time and expense of these elaborate co-cultures negates the purpose of circumventing in vivo experiments.

In vivo analysis of BBB penetration is mainly restricted to imagining techniques such as positron emission tomography (PET), and microdialysis. PET requires radiotracers and expensive instrumentation, has spatial resolution on the order of millimeters at best, and requires an animal be under anesthesia. Intracerebral microdialysis has been used for decades to investigate the in vivo BBB permeability of drugs.73,82,182–184 While the spatial resolution is no better than PET, microdialysis can be used in awake and freely moving animals, does not require radioactive compounds, and can easily be coupled to offline detection methods such as mass spectrometry. Mass spectrometric quantitation of peptides in microdialysate has seen much development in the past decade due to the promise of unsurpassed chemical resolution and structural information.40,185–188 In

Chapters 4 and 5 we will present a method for quantitation of BBB penetration of peptide- based drugs using microdialysis coupled to LC-MSn with limits of detection in the picomolar regime.

1.6 Overview In this work we will present uses of tandem mass spectrometry for the quantitation of small-molecule neurotransmitters and peptide-based drugs in complex biological samples. In Chapter 2 we will present a method for the separation-free quantitation of five biogenic amines in tissue homogenate with improved throughput and chemical specificity compared to existing methods. Chapter 3 applies this method to the quantitation of neurotransmitters in whole honey bee brains, for the purpose of investigating the effect of a gut-dwelling parasite on whole-brain biogenic amine content.

In Chapter 4 we will investigate the effect of glycosylation of peptides on in vitro stability.

We also investigate the blood-brain barrier penetration of these peptide-based drugs using microdialysis sampling with nanoflow LC-MSn analysis, which allows for quantification of

53 small peptides in a volume-limited complex matrix with high sensitivity and selectivity.

Chapter 5 investigates the effect of structural modifications on the stability and penetration properties of six Angiotensin 1-7 derivatives. We introduce the concept of shotgun microdialysis for the direct comparison of the in vivo properties of multiple related drugs in a single animal. Finally, Chapter 6 will outline the future directions of this research.

54 Chapter 2

Separation-free quantitation of biogenic amines in brain tissue with mass

spectrometry

Abstract

Biogenic amines such as dopamine are of interest for many reasons, including their roles in normal behavior and neuropsychiatric disorders. Here, we present a high-throughput method for the simultaneous quantitation of five biogenic amines. This method allows for the quantitation of dopamine, L-DOPA, serotonin, octopamine, and histamine in less than

10 minutes per sample and with limits of quantitation in the low-nM range (6.4 ± 0.7, 40 ±

11, 7.7 ± 0.8, 12 ± 4, and 15 ± 5 nM in 20 µL injections, respectively). The speed and sensitivity were achieved using a benzoyl chloride derivatization reaction, with mass spectrometric detection and no chromatographic separation. This simple method is appropriate for application to complex matrices and is sufficiently sensitive for mass- limited samples, with limits of detection in the pmol regime. Dopamine and serotonin were quantified in brain punch biopsies from the striatum of rats with unilateral 6- hydroxydopamine (6-OHDA) lesions. The striatal tissue samples from the two hemispheres were found to have significantly different amounts of dopamine (p < 0.01) but not serotonin (p > 0.05). Quantitation with this method was not significantly different than quantitation with an accepted LC-EC method (p > 0.05) and has increased throughput. The five biogenic amines were quantified in whole heads from each sex of three Drosophila laboratory strains, one wild-type and two mutant. Mutations that eliminate the eye-pigment ABC transporter or the dopamine transporter did not cause variations in whole-head content of the five biogenic amines under investigation.

55 2.1 Introduction

Determination of biogenic amine levels is important as they play a large role in several branches of chemistry including food chemistry, medicinal chemistry, and neuroscience.1–

7 Biogenic amine indicators of food spoilage or fermentation such as putrescine and cadaverine are quantified for quality control purposes.2,8,9 Additionally, histamine and tyramine have been used to validate the origin and fermentation process of wines.1,2,8 In clinical medicine, epinephrine and norepinephrine are measured in the diagnosis and management of neuroendocrine tumors.10–14 The wide range of sources of biogenic amines is reflected in the variety of matrices in which they can be found, from solid foods such as fish or cheese, to biological samples such as urine or tissue. These matrix differences add a layer of complexity that leads to challenges when developing quantitation and sample preparation methods. The large chemical space of the source materials requires that we develop rapid analysis techniques for different matrices that improve throughput, as well as being accurate and robust.

Quantitation of biogenic amines in biological samples is common, as biogenic amine neurotransmitters such as dopamine and serotonin are involved in many neurobiological processes. Specifically, dopamine signaling controls aspects of locomotion, sleep, arousal, and addiction.15–18 Parkinson’s disease results from selective neurodegeneration of dopaminergic neurons and its main pharmacological treatment is with the dopamine precursor L-DOPA.19 Serotonin plays a role in neural development, aggression, and depression.20–22 In clinical and laboratory research on normal physiology, as well as disease pathogenesis and treatment efficacy, catecholamines are measured as indicators of acute or chronic stress responses.23,24 However, quantitation of these compounds in biological samples can be difficult, due to sample complexity and variability between types of samples (urine, blood, tissue, etc). As analysis techniques can be limited by sensitivity,

56 specificity, and throughput, it is necessary to develop methods specific to the analytes and sample type of interest.

Quantitation of biogenic amines has classically been performed with liquid chromatography-coupled techniques such as electrochemical detection (LC-EC) or mass spectrometry (LC-MS). Amperometric detection requires a front-end separation due to its lack of specificity. Column separations typically require around 20 minutes (commonly ranging from 5-60 minutes) per sample.25–27 LC-EC is a well-characterized and sensitive technique (picogram limits of detection were reported in the early 1970s and femtogram limits of detection were reported in 1980).27–30,19 Additionally, electrochemical detection has a linear dynamic range over several orders of magnitude, which can be of great benefit in some experiments. However, techniques utilizing electrochemical detection require electrochemically active species, limiting the compounds that can be probed. For example, dopamine and serotonin are electroactive, whereas gamma-aminobutyric acid

(GABA) and glutamate, two neurotransmitters that control neural circuit activity, are not.

In contrast, mass spectrometry can be used for any molecule that can be ionized.

Some researchers have approached the problem with more resourceful methods, such as fast-scan cyclic voltammetry (FSCV) and micellar electrokinetic chromatography (MEKC).

These specialized approaches also have certain inherent limitations. Background- subtracted FSCV adds selectivity over typical electrochemical detection methods, has subsecond temporal resolution, and is sensitive (into the pmol regime), but is still limited to electrochemically active analytes.31 MEKC-EC techniques were developed due to their high sensitivity in low-volume samples (femtomoles of material in nanoliter volumes), which is necessary for mass-limited biological sources such as the fruit fly Drosophila melanogaster.32 However, the inherent limitations of electrochemical detection are still

57 present, and MEKC cannot be directly coupled to a mass spectrometer because of the surfactants used in the separation.

For an efficient and versatile quantitation method for neurologically relevant biogenic amines, a technique is required that couples a rapid column-free sample preparation with an unrestricted detection method for increased throughput and the ability to quantify a wide range of species. An optimal method would be capable of being applied to complex matrices of common biological tissues and fluids, use widely accessible reagents and instrumentation, and require basic, rather than expert, technical skill.

To achieve these goals, we have developed an approach based on derivatization with benzoyl chloride33,34 followed by liquid-liquid extraction coupled to direct-injection MS2 targeted analysis. This sample preparation circumvents a column separation while still providing the necessary sample workup, such as desalting. Derivatization was used to increase the mass and ionization efficiency of the target compounds, which yields improved limits of detection, down to 100 femtomoles of material. This approach allows for simultaneous detection of five biogenic amines without upstream column separation.

In selecting biological samples for methods development, we focused on two premier model organisms of high value in neuroscience research, rats and fruit flies. Because biogenic amine receptors and transporters are targets of therapeutic drugs for many indications, rodents are essential tools for preclinical testing in the drug-development pipeline. The same targets are bound by drugs of abuse with high addictive potential, adding to the need for rodent studies of their neuropharmacology and pathophysiological mechanisms. Rodents also provide several Parkinson’s disease models based on administration of neurotoxins selective for dopaminergic neurons, such as rotenone, paraquat, 6-hydroxydopamine (6-OHDA), and 1-methyl-1,2,3,6-tetrahydropiridine

(MPTP), or on transgenic introduction of dominant disease-causing mutations.35,36 The 6-

58 OHDA model has the experimentally appealing feature of unilateral toxin injection, with the contralateral brain hemisphere available as a within-animal control.35,37

The Drosophila genetic system has been used for decades to study biogenic amine biochemistry, pharmacology, neuroanatomy, and behavioral effects.38 The phylogenetic conservation of molecular mediators of biogenic amine action are sufficient to allow

Drosophila to serve as a model of mammalian biology.39

2.2 Materials and Methods

2.2.1 Chemicals and reagents All chemicals and reagents were purchased from Sigma

Aldrich (St. Louis, MO, USA). Standards were prepared fresh at a concentration of 1 mM in 0.1 M perchloric acid and diluted to target concentrations in 0.1 M perchloric acid prior to the derivatization reaction, which was adapted from Song et al.33 Carbonate buffer (200 mM, pH 11) was made in a large batch (1 L) and stored at room temperature for up to six months. The internal standard was dopamine derivatized with deuterated benzoyl chloride, which was spiked into each prepared sample for a final concentration of 50 nM.

2.2.2 Rats Male Sprague-Dawley rats (375-425 grams; Envigo, Indianapolis, IN) were housed in a temperature and humidity controlled room with 12 hour reversed light/dark cycles with food and water available ad libitum. All animal handling procedures were approved by the Institutional Animal Care and Use Committee, University of Arizona and in accordance with the National Institutes of Health Guidelines for the Care and Use of

Laboratory Animals.

2.2.3 Unilateral 6-hydroxydopamine (6-OHDA) lesion method Animals used in this study were combined from two different experiments in which a unilateral lesion was induced by injecting 20 g of 6-OHDA (5.0 g/L in 0.9% sterile saline with 0.02% ascorbic acid). All surgical procedures were followed as previously published.40,41 The first group

59 (n = 3) received injections at two stereotactic coordinates (10 g/coordinate) in the medial forebrain bundle (MFB), with the following modifications: AP -1.8, ML +2.0, DV -8.2 and

AP -2.8, ML +1.8, DV -8.2.41 The second group (n = 4) received a single injection (20

g/coordinate) in the ventral lateral striatum (VLS) at the following coordinate: AP +0.8,

ML +2.5, DV -5.2.40

2.2.4 Amphetamine-induced rotation In both groups, amphetamine-induced rotations were assessed 3-weeks post-lesion, following D-amphetamine (5 mg/kg, i.p.). The total number of contralateral and ipsilateral rotations (i.e., toward the side of the lesion, an indicator of dopamine depletion36,37,42,43) were recorded during one-minute intervals, every five minutes, over the course of one hundred minutes. Rats selected for this study demonstrated 4.56 ± 1.07 (mean ± s.e.m.) net ipsilateral rotations per minute.

2.2.5 Rat tissue collection and preparation Rats were euthanized using carbon dioxide followed by cervical dislocation. Coronal brain slices were collected and 2-mm biopsy punches were used to sample striatal tissue. Two samples each from left and right hemispheres were collected and immediately flash frozen on an aluminum pan at -70 °C.

Samples from the same hemisphere were combined. Samples massed at approximately

5 mg and were homogenized in 0.1 M perchloric acid. Samples were stored at -80 °C until analysis was performed. At time of analysis, samples were centrifuged and the supernatant was divided into two aliquots. One aliquot was analyzed using LC-EC for quantitation of dopamine and serotonin content. The other aliquot was analyzed using benzoylation and mass spectrometric detection for the quantitation of dopamine, L-DOPA, histamine, and serotonin. Analyses were time-locked.

2.2.6 Drosophila melanogaster Fly stocks were maintained at room temperature on a nutrient medium44 comprised of corn flour, yeast, sugars, and agar, obtained from the

Restifo Lab (Department of Neurology, University of Arizona). For experimental cultures,

60 rearing was done at optimal density in an incubator at 25 °C, with 12-hour light-dark cycles.

Control strains included a wild-type laboratory strain, Canton S (maintained in-house, originally obtained from the Han Lab, University of Texas, El Paso), and a white-eyed strain, w1118, that had been “Cantonized” by repeated backcrosses to Canton S (obtained from the Jackson Lab, Tufts University). This allows direct comparison of Canton S and w1118. The w1118 strain also serves as the genetic control for fumin.45 The fumin strain

(w1118; fmn recombinant 19, also from the Jackson Lab) has a homozygous insertion mutation in the gene encoding the dopamine transporter (DAT; FBgn0034136); this mutation prevents reuptake of dopamine.31,46

Animals were isolated by sex within 2 hours of eclosion (adult emergence) and aged three days prior to sacrifice. Thus, the flies used for analysis were young adults and both sexes were virgins. Flies were anesthetized under humidified carbon dioxide and decapitated using iridectomy scissors. Heads were placed in 0.1 M perchloric acid on ice immediately after dissection. Once ten heads were collected, they were immediately homogenized using a polypropylene pestle, then stored at -80 °C until reaction and analysis (no more than one month). The mean number of samples analyzed per genotype/sex group was

14, with a range of 7 to 22.

2.2.7 Apis mellifera Brood frames were collected from colonies at the Carl Hayden Bee

Research Center locations in Tucson, Arizona between March and July, 2016. Frames were taken from European colonies, Apis mellifera ligustica, headed by queens from

Pendell Apiaries (Stonyford, CA). Bees were kept in a Binder BD (E2) incubator at 31.7

°C with 50% relative humidity, under constant dark conditions. Bee-collected pollen was removed from pollen traps in the surrounding Tucson, AZ area on February 26, 2016 and kept at -20 °C until use. Each cage was given an insert with three grams of pollen, 50% sucrose solution, and water ad libitum. Pollen, sucrose, and water were changed every

61 seven days. Cages were checked daily and dead bees were removed. Whole brains were dissected from 7- and 15-day-old bees. A bee was placed into a scintillation vial and chilled on ice until immobile (between 2 - 5 minutes). The brain was rapidly dissected, weighed, and placed into 50 µL of chilled 0.1 M perchloric acid. The brain was ground manually, frozen in liquid nitrogen, and transferred into -80 °C storage until analysis.

Because levels of biogenic amines are known to fluctuate throughout the day,47–49 all brains were dissected between 2 - 5 p.m.; and each dissection alternated between treatments to control for any time of day variances between 2 - 5 p.m.

2.2.8 Derivatization reaction Benzoyl chloride reacts with primary and secondary amines and phenols. Derivatization allows for differentiation of isobaric species such as dopamine and octopamine, as dopamine labels three times but octopamine labels only twice.

Histamine labels once, serotonin labels twice, and L-DOPA labels three times. All samples were stored at -80 °C in 100 µL 0.1 M perchloric acid between sample collection and reaction. As the labeling reaction occurs at pH greater than 9, first carbonate buffer was added to the sample (200 µL of 200 mM carbonate buffer, pH 11.0). Then benzoyl chloride was added (100 µL of 2% v/v in acetonitrile), and was in excess by several orders of magnitude. The mixture was vortexed for five seconds. Two extraction steps with 200

µL dichloromethane (DCM) were performed, followed by two washes of the organic layer with 200 µL each of basified water (adjusted to pH 8 with ammonium hydroxide) to minimize the carryover of unwanted species such as benzoic acid and salt products. The organic layer was vacuum-concentrated to dryness and the final product was dissolved in an electrospray-compatible solvent (50:50 acetonitrile:H2O, with 0.1% formic acid as a proton source) as necessary to bring signals into the linear dynamic range of the MS instrument.

62 2.2.9 Mass spectrometric detection and quantitation Detection and quantitation were performed with an AB Sciex QStar Elite using Analyst® QS 2.0. Samples were injected using a 20-µL injection loop, with 50:50:0.1% ACN:H2O:formic acid spray solvent flowing at 8 µL/min. The most abundant fragment peak for benzoylated compounds is the 105 m/z benzoyl fragment.33 The area of the 105 m/z peak was quantified over the course of the 4-min collection. Reaction and run time combined are less than 10 minutes per sample.

Samples were run in duplicate and values were averaged for each sample. Values were normalized to the signal of the internal standard (50 nM deuterated labeled dopamine) and concentrations were calculated based upon the resulting normalized signal and calibration curves. Statistical analyses were performed using GraphPad Prism 5 (La Jolla,

CA). Data sets were tested for normality using a D’Agostino and Pearson omnibus normality test. If the data sets passed, they were compared using one-way ANOVA with

Bonferroni post-hoc. Data sets that did not pass the normality test were compared using a Kruskal-Wallis test, which does not assume a Gaussian distribution, with Dunn’s Multiple

Comparison test. In all cases, α = 0.05 and significance required p < 0.05.

2.3 Results and Discussion

2.3.1 Mass spectrometric detection of labeled biogenic amines Five biogenic amine compounds (dopamine, histamine, L-DOPA, serotonin, and octopamine) were labeled with benzoyl chloride (BzCl) using protocols modified from Song et al.33 Though referred to herein as benzoyl chloride derivatization, the recognized name for this reaction is the

Schotten-Baumann reaction. Originally published in the mid 1880’s (1884 by Schotten and 1886 by Baumann),50,51 the Schotten-Baumann reaction is defined as “the synthesis of esters from alcohols and amides from amines with acyl halides and anhydrides in the presence of aqueous base”.52 The mechanism for the reaction of an amine is shown in

Figure 2.1 Schotten-Baumann reaction mechanism. In the presence of aqueous base, amides and esters are produced from amines and alcohols, respectively, via reaction with acyl halides. In this example, the lone pair of electrons on the deprotonated amine can attack the acyl halide, producing a quaternary intermediate.

The ketone reforms and the halide is kicked off. In the basic solution, the final product is deprotonated.

Figure 2.2 Benzoyl chloride derivitization. a) Adapted from Song et al.33 Benzoyl chloride labels primary and secondary amines and phenols at reaction pH greater than

9.0. b) Benzoylation of dopamine (top) and octopamine (bottom). Dopamine and octopamine are isobaric, but dopamine labels three times, whereas octopamine labels only twice, due to the location of the second hydroxyl group. This allows for differentiation of the two species by mass spectrometry.

65 Figure 2.1. BzCl labels primary and secondary amines and phenols, as shown in Figure

2.2a. This leads to increased molecular weight and hydrophobicity of compounds for improved extraction and quantitation. Additionally, it differentially labels the isobaric compounds dopamine and octopamine, due to the location of one of the alcohol groups

(Figure 2.2b). Octopamine is a biogenic amine that plays the role of norepinephrine in invertebrates, such as Drosophila melanogaster.53–56 Differential labeling allows for mass spectrometric quantitation where it would typically be impossible to distinguish between the two species.

Figure 2.3 shows the work flow for sample preparation and analysis. Samples were homogenized in 0.1 M perchloric acid (HClO4) for storage. When preparing for analysis, samples were added to a solution buffered to pH 11 and immediately reacted with 2% v/v benzoyl chloride. After labeling, samples were extracted with dichloromethane, dried, and reconstituted in ACN:H2O:FA (1:1:0.1%) for electrospray compatibility. There are liquid- liquid extraction steps (two organic and two aqueous) to provide the necessary desalting and sample cleanup. As the reaction presented herein is largely aqueous (75% aqueous,

25% acetonitrile), benzoyl chloride also undergoes a competitive hydrolysis reaction, producing benzoic acid.57 Samples were injected onto the mass spectrometer and quantified by using the main MS2 fragment ion. Labeling with BzCl has the added benefit of yielding an easily identifiable fragment peak in the MS2 as the benzoyl group (m/z 105) is cleaved. For example, Figure 2.4 shows the MS2 spectra for fragmentation of five labeled biogenic amines into their benzoyl fragment (m/z 105). Area under the curve

(AUC) for each compound is normalized to the benzoyl fragment of the internal standard

(dopamine labeled with deuterated benzoyl chloride, m/z 110). The m/z 105 fragment peak for each compound is quantified over the course of the 4-minute injection. Figure

2.5 shows a representative calibration curve for each compound.

Figure 2.3 Workflow for sample preparation. Samples are homogenized and stored in 100 µL 0.1 M perchloric acid at -80°C until reaction and analysis. The reaction requires introduction of buffer (200 µL carbonate, pH 11.0) to increase the pH, at which time 100 µL of 2% v/v benzoyl chloride in acetonitrile is added and reaction proceeds.

Sample cleanup involves a series of liquid-liquid extractions (200 µL each), first with

DCM to extract the organic compounds, then a washing of the DCM phase with water adjusted to pH 8 to remove excess benzoic acid. The resulting DCM phase is dried and reconstituted in 100 µL ESI-compatible solvent (ACN:H2O:FA 1:1:0.1%). Samples are injected onto the instrument using a 20 µL injection loop and the MS2 product peak is quantified by area.

Figure 2.4 MS2 spectra for target compounds. For benzoylated compounds, the main fragment peak is m/z 105. This peak is used to quantify each compound. For the heavy-labeled internal standard (dopamine labeled with deuterated benzoyl chloride, DA-d15), the fragment peak has a mass of 110, due to the five deuterium atoms on the benzoyl ring. [DA, Hist, OA, L-DOPA] = 750 nM, [5HT] = 500 nM, [DA-d15] = 50 nM.

Figure 2.5 Representative calibration curves. a) Dopamine, b) histamine, c) serotonin, d) octopamine, and e) L-DOPA. Signals are normalized to the area under the curve (AUC) for the internal standard (dopamine labeled with deuterated benzoyl chloride). [IS] = 50 nM, n = 4 - 6, error bars are s.e.m.

69 Due to the nature of liquid-liquid extraction, higher extraction efficiency was obtained for compounds that labeled at more sites, and therefore were less hydrophilic (i.e. dopamine

> octopamine > histamine). This led to variability in limit of detection (LOD) and linear dynamic range (LDR) for each of the target compounds, as shown in Table 2.1. Limit of detection here is defined as the lowest concentration at which signal can be detected (as the level of selectivity of MS2 detection yields a null baseline), and LDR is determined by the r2 value of a normalized calibration curve, where r2 > 0.99. LODs for histamine, serotonin, octopamine, dopamine, and L-DOPA were (15 ± 5 nM, 7.7 ± 0.8 nM, 12 ± 4 nM,

6.4 ± 0.7 nM, and 40 ± 11 nM, respectively.

Measurement error due to reaction variability and quantitation variability were investigated. A bulk solution of all five compounds at 1 µM was prepared. This solution was split into seven vials and each was derivatized then run in triplicate. For analysis of the measurement error, see Table 2.2. Here, “technical replicate” refers to the reproducibility of the labeling process for the same original sample split and labeled in different vials. “Pseudo-replicates” are the same derivatized sample run in triplicate on the instrument, and account for the variability of the instrument itself. For these samples, n = 21 (7 trials of 3). For the technical replicates, the triplicate values were averaged for n = 7. The deviation from linearity was determined by building a calibration curve and retroactively calculating the difference between the detected signal for a particular concentration and the expected signal based on the calibration curve without that sample.

For a calibration curve with six concentrations, only the middle four concentrations were used to calculate deviation, and these deviations were averaged (n = 4).

In general, the calibration curves had an average percent relative standard deviation (%

RSD) of about 10.5%, which is higher than we would like, but acceptable. The largest variation in our method seemed to arise from the variability in labeling efficiency, which

70 Table 2.1 Sensitivity and range of biogenic amine quantification.

Limits of detection and linear dynamic range (in nM and pmol, using a

20 µL injection loop) for MS2 quantitation (average ± s.e.m., n = 11 samples).

Compound LOD (nM) LOD (pmol) LDR (nM) r2

Dopamine 6.4 ± 0.7 0.13 ± 0.01 10 - 500 0.9964 Histamine 15 ± 5 0.3 ± 0.1 75 - 500 0.9997 Serotonin 7.7 ± 0.8 0.15 ± 0.02 25 - 750 0.9991 Octopamine 12 ± 4 0.24 ± 0.08 100 - 750 0.9932 L-DOPA 40 ± 11 0.8 ± 0.2 100 - 750 0.9933

71 Table 2.2 Variability in measurements. Average percent relative standard deviation (%

RSD) was calculated to determine variability in the method.

Dopamine Histamine Serotonin Octopamine L-DOPA

Technical 12.2 58.1 14.8 21.1 23.7 replicates

Pseudo-replicates 11.8 20.4 11.3 18.6 21.1

Deviation from 9.0 11.2 12.2 10.4 9.78 linearity

72 averaged around 18% (discounting histamine, as there is some optimizing that needs to occur to measure this compound with higher accuracy). The variability in the instrument averages around 16.6%, again within acceptable limits, especially considering the high number of samples analyzed in the Drosophila study.

An added benefit to this method is that samples labeled with benzoyl chloride are stable against oxidation, as previously reported.33 Figure 2.6a shows the stability of labeled dopamine over the course of 9 days. Seven samples were prepared in replicate on day 1 and stored as dry samples at 4 °C. On each day, one sample was reconstituted in ESI spray solvent (ACN:H2O:FA 1:1:0.1%) and run in triplicate. No measurable degradation occurs on this time scale. Figure 2.6b shows the stability of labeled dopamine against voltage-induced oxidation. Labeled dopamine was subjected to slow-scan voltammetry to determine its oxidation properties. For comparison, voltammograms for the buffer and for non-labeled dopamine are shown. Labeled dopamine did not appear to oxidize under these conditions. Cyclic voltammograms were obtained at a glassy carbon electrode

(black = buffer, red = dopamine, blue = benzoylated dopamine). Geometric electrode area

= 4.9 x 10-2 cm2, scan rate = 0.050 V/s, buffer = 1:1 ACN:25 mM phosphate 100 mM NaCl pH = 7.40, and all analyte concentrations = 500 µM.

Due to the nature of the analysis (sans separation), the samples are injected using a 20-

µL injection loop at a flow rate of 8 µL/min, resulting in a 2.5-minute bolus. Each scan cycle (which contains a full MS scan and MS2 scans of the internal standard and the five compounds) takes seven seconds, one second per scan. Therefore, 21 scans are collected for each compound during the 2.5-minute bolus. By comparison, if liquid chromatography was coupled to the mass spectrometer, the 30-second chromatographic peaks would yield only 4 scans per compound, based on the separation used for comparison later in this paper. Using direct injection, we could quantify up to 35

Figure 2.6 Labeled compounds show stability against oxidation. a) Labeled dopamine can be stored at 4 °C at least nine days without apparent degradation or loss of signal. n = 3 – 5 b) Labeled dopamine was subjected to slow-scan voltammetry to determine its oxidation properties. For comparison, voltammograms for the buffer and for non-labeled dopamine are shown. Labeled dopamine did not appear to oxidize under these conditions. Cyclic voltammograms were obtained at a glassy carbon electrode (black = buffer, red = dopamine, blue = benzoylated dopamine). Geometric electrode area = 4.9 x 10-2 cm2, scan rate = 0.050 V/s, buffer = 1:1 ACN:25 mM phosphate 100 mM NaCl pH = 7.40, all analyte concentrations = 500 µM.

74 compounds and have the same number of scans per compound as we would with a chromatographic separation.

Our separation-free approach increases throughput, such that each sample can be labeled, extracted, and run in less than 10 minutes on average. This compares highly favorably to liquid chromatography methods wherein the gradient itself is often 30 minutes long. Additionally, the short analysis time compared to a chromatographic method means that more standards can be run for a more representative calibration curve in a shorter period of time. For example, with this method, six standard concentrations can be run, in duplicate, in less than an hour, the time it would take to run two standards with an LC method.

2.3.2 Comparison to “gold standard” LC-EC Liquid chromatography coupled to electrochemical detection has become the gold standard for the analysis of electroactive biogenic amines, due to its high sensitivity and reproducibility.27,28,58 LC-EC sensitivities for biogenic amines on the picogram scale have been reported since the early seventies.30

However, this technique is limited by its time resolution and the inherent requirement that target analytes be electroactive.

We chose to validate the methodology discussed herein by comparison with a previously published LC-EC method.29,41 We compared quantitation of dopamine and serotonin content in striatum samples from rats with unilateral 6-hydroxydopamine (6-OHDA) lesions of moderate severity, based on behavioral criteria.36 Lesioning with 6-OHDA is used as a

Parkinson’s Disease model in rats, as it is a neurotoxin that specifically targets dopaminergic neurons, producing Parkinson’s-like symptoms.42 Unilateral lesioning allows for direct between-hemisphere comparison, reducing biological variability. Paired samples from the lesioned and unlesioned hemispheres of seven animals were collected, homogenized in perchloric acid, split, and run in parallel by LC-EC and our method. Figure

75 2.7 shows dopamine and serotonin quantification in samples from both lesioned and unlesioned brain hemispheres obtained by the two methods. As expected, with both methods there was a significant difference in dopamine content between the lesioned and unlesioned hemispheres (one-way ANOVA with post-hoc Bonferroni, p < 0.01) and no significant difference (p > 0.05) between hemispheres for serotonin, which is unaffected by 6-OHDA. For dopamine and serotonin content values in these samples, see Table 2.3.

Paired LC-EC and BzCl-DI-MS values for each biogenic amine were compared graphically and tested for correlations (Figure 2.8). Dopamine values were well correlated (slope =

0.5515, F = 30.47, and p < 0.0001). Our method reports slightly lower values than LC-

EC. Interestingly, the serotonin values, despite similar means, were not well correlated

(slope = 0.1256, F = 0.7116, and p = 0.4154). Using this model, our LC-EC method has been detecting a species that co-elutes with serotonin in some animals. This may affect the correlations. The co-eluting species is currently under investigation.

This serves as validation that our method is a suitable alternative for the quantitation of biogenic amines in brain homogenate. While LC-EC has a wider LDR, our method is more than sufficient to quantify the relatively small range of concentrations seen in these samples (from 3.4 - 14.7 ng/mg for dopamine and 0.3 – 1.1 ng/ mg for serotonin, in both cases a full order of magnitude greater than our limit of detection).

Additionally, while LC-EC methods often use a cooled autosampler, there is a limitation on the number of samples that can be loaded, due to the sample instability. Our method does not have this weakness, as our samples are stable against oxidation for over a week.

As a result, once samples are labeled, they can be stored for longer periods of time before analysis, and analysis time is shorter. The method presented herein is sufficiently sensitive to quantify a wider range of biogenic amines than is possible with LC-EC, with a longer shelf-life of samples and improved throughput.

Figure 2.7 Comparison to existing LC-EC method for validation of quantitation.

Comparison to LC-EC quantitation for dopamine a) and serotonin b) in matched rat striatal samples (n = 7 rats). Lesioning with 6-OHDA produced a significant drop in dopamine levels measured by both methods (One-Way ANOVA with Bonferroni post- hoc, p < 0.01), but had no effect on serotonin levels (p > 0.05). LC-EC and BzCl-DI-MS quantitation were not statistically different (p > 0.05, One-way ANOVA with Bonferroni post hoc).

77 Table 2.3 Comparison to LC-EC method. Numerical results of the LC-EC vs BzCl-DI-

MS validation experiment (n = 7) are shown. The LC-EC and BzCl-DI-MS methods were not statistically different within hemispheres (one-way ANOVA with post-hoc Bonferroni, p > 0.05).

Dopamine (ng/µg, mean ± SEM) Serotonin (ng/µg, mean ± SEM)

Hemisphere LC-EC BzCl-DI-MS LC-EC BzCl-DI-MS

Lesioned 4.3 ± 2.0 7.8 ± 1.7 0.9 ± 0.1 0.6 ± 0.1 (right) Unlesioned 22.3 ± 1.3 18.1 ± 1.5 1.0 ± 0.2 0.6 ± 0.1 (left)

Figure 2.8 Dopamine and serotonin correlations between methods. Correlation of values for dopamine and serotonin in each tissue sample were plotted. Dopamine values were well-correlated (slope = 0.5515, F = 30.47, and p < 0.0001). Our method reports slightly lower values than LC-EC. The serotonin values, despite similar means, were not well correlated (slope = 0.1256, F = 0.7116, and p = 0.4154), perhaps due to a co-eluting species that is currently under investigation.

79 2.3.3 Quantification of biogenic amines in insect brain homogenate The comparison to the LC-EC method validated our separation-free method for the quantitation of biogenic amines in relatively large, soft biological matrices from mammalian sources. The next step was to move into the realm of insects. Apis mellifera, or honey bees, have brains with a wet mass of about 1 mg. Analysis of honey bee brains allowed for the validation of our method with the brains of insects, without being as mass-limited as studies with

Drosophila, but with the same chemicals present- namely octopamine, which acts as a neurotransmitter in invertebrates. Additionally, removal of the brain from the chitinous exoskeleton circumvents complications that might arise during sample preparation, as will be discussed later. Five biogenic amines were quantified in the whole brains of honey bees. Figure 2.9 shows the results for dopamine, octopamine, and serotonin for the samples analyzed herein compared to previously published literature values.59–64 Only three compounds are shown, as there are fewer published studies that report whole-head content of L-DOPA and histamine in honey bees. In most cases, the published values were provided in graphical form, therefore values were estimated. Previously published values also span a large range of honey bee ages and circumstances (e.g. role in the hive, and presence of queen). The data shown herein presents values for bees under conditions similar to ours to the best of our abilities.

The results for quantification of three biogenic amines in whole honey bee brains does not differ significantly from previously published data (Mann-Whitney test, p > 0.05). Our octopamine values do trend slightly higher in general, but not significantly. Comparison to previously published data for the whole-brain content of serotonin, octopamine, and dopamine in honey bees again validated our quantification method, this time for quantification in insect tissue.

Figure 2.9 Biogenic amine quantification in honey bees matches previously reported results. Five biogenic amines were quantified in 7- and 15-day old honey bees. The results for dopamine, octopamine, and serotonin are shown compared to published literature values. In most cases, published values were eyeballed from bar graphs. Data collected with our method does not vary statistically from published values

(Mann-Whitney test, p > 0.05).

81 To investigate the feasibility of this approach in smaller, more complex tissues, we applied our method to the heads of adult fruit flies, which are quite small (~0.1 mg), and have an exoskeleton of sclerotized cuticle. Dopamine, L-DOPA, histamine, serotonin, and octopamine content were quantified in Drosophila melanogaster across three different strains and two sexes. The issues we addressed were overall feasibility, sexual dimorphism, and the impact of mutations in each of two genes that encode specific transporters. The white gene encodes an ABC transporter that imports eye pigment precursors tryptophan and guanine.65 This gene was named a century ago based on the eye-color phenotype of what was later shown to be a complete loss-of-function mutant.66

For historical reasons related to transgenic technology, w-mutant flies (in particular, w1118) are very commonly the control strain for mutants of interest. Prior studies of biogenic amine content of w mutants have reported conflicting results.67–70 Drosophila DAT was named for its encoded dopamine transporter, which binds cocaine and has been proposed to represent an ancestral catecholamine-carrier gene.71 The fumin (fmn) mutation in DAT was identified due to a behavioral phenotype, excessive arousal and motor activity; “fumin” is Japanese for “sleepless”. The reduced uptake of dopamine and hyperactivity of fmn mutants31,46 raise the question of whether total dopamine content of the brain might be elevated.

Table 2.4 breaks down previously reported biogenic amine content in the heads and brains of Drosophila melanogaster. Table 2.5 reports the whole-head biogenic amine content of both sexes and across the three genotypes, as measured with our method.

Previously reported values for dopamine in adult Drosophila vary greatly, from 0.002 to

0.7 ng/ head, and 0.01-0.08 ng/brain.4,32,45,47–51 Our results, from 0.15 ± 0.03 to 0.25 ±

0.07 ng of dopamine per head, fall within the range of previous whole-head values, and on the high end of the previous order-of-magnitude range. For histamine, the levels

82 Table 2.4 Previously published Drosophila values. Reported values for biogenic amine content in Drosophila melanogaster whole heads and brains. Values are reported as ng/head or ng/brain with mean ± s.e.m where provided. Group Year Method Histamine Octopamine Serotonin Dopamine L-DOPA Comments (author) Kaiser Heads 1993 GC-MS Canton S 0.0005 0.009 ± 0.001 0.137 ± 0.037 Freeze and sieve (Watson) Meinertzhagen 2000 HPLC-EC Oregon R 1.98 ± 0.15 Freeze and sieve (Borycz) Ewing, Han 2003 MEKC-EC Canton S 0.064 ± 0.007 0.002 ± 0.0004 0.147 ± 0.012 Males; Freeze and sieve (Ream) Adults 3-5 days old; prep Ewing (Powell) 2005 MEKC-EC Canton S 0.260 ± 0.06 1.87 ± 0.32 0.398 ± 0.14 unspecified Canton S 0.029 ± 0.01 0.010 ± 0.002 Ewing (Paxon) 2005 MEKC-EC 3-5 days old; freeze and sieve iav (inactive) 0.005 ± 0.0004 0.008 ± 0.001 Hirsh 2006 HPLC-EC w1118 0.129 ± 0.014 0.038 ± 0.001 0.203 ± 0.005 Males (Hardie)

Meinertzhagen Oregon R 2.08 0.203 0.678 2008 HPLC-EC Adults: day 7; freeze and sieve (Borycz) w1118 1.06 0.138 0.271

HPLC-EC, Canton S 0.264 ± 0.061 0.169 ± 0.036 0.599 ± 0.14 Zars (Sitaraman) 2008 Prep unspecified immunoassay w1118 0.233 ± 0.091 0.033 ± 0.009 0.148 ± 0.003 White Brains 1996 HPLC-EC Canton S 0.263 0.066 0.081 Males; dissected in HClO (Monastirioti) 4

Hirsh Oregon R 0.074 ± 0.002 1999 HPLC-EC Males; dissected on cold block (McClung) iav (inactive) 0.086 ± 0.001 Hirsh Males, Dissected into citrate 2006 HPLC-EC w1118 0.083 ± 0.003 0.039 ± 0.001 0.052 ± 0.002 (Hardie) acetate on ice Canton S 0.060-0.085 0.025-0.032 0.012-0.022 5F + 5M; 2-3 days old; dissected in Ringer’s sln; values Gerber (Yarali) 2009 HPLC-MS/MS 1118 w1118 0.045-0.080 0.025-0.032 0.015-0.025 eyeballed from box plot; w backcrossed to CS

Ewing Canton S 0.119 ± 0.012 0.014 ± 0.0007 0.315 ± 0.045 2010 MEKC-EC Males (Kuklinski) “white” 0.107 ± 0.021 0.010 ± 0.001 0.177 ± 0.051 Canton S 0.038 ± 0.003 0.023 ± 0.002 0.019 ± 0.002 Larval brains, 3rd instar; CNS Venton 2011 CE-FSCV dissected into Schneider's buffer (Fang) w118 0.025 ± 0.004 0.016 ± 0.002 0.011 ± 0.001 then into HClO4 Venton 2015 CE-FSCV Canton S 0.045 ± 0.007 0.022 ± 0.005 0.085 ± 0.01 72h males; dissections in PBS (Denno)

Table 2.5 Quantitation of biogenic amines in Drosophila melanogaster whole heads.

Five biogenic amines where quantified across two sexes and three genetic strains. Values are reported as mean ± s.e.m, ng/head. *Denotes statistical significance between strains, within sex. ǂDenotes statistical significance between sexes, within strain. Statistical significance was only found for male versus female whole-head histamine content in w1118 flies

(one-way ANOVA with Bonferroni post-hoc, p < 0.05).

Histamine Octopamine Serotonin Dopamine L-DOPA Canton S 0.75 ± 0.08 1.6 ± 0.3 0.42 ± 0.05 0.20 ± 0.03 1.7 ± 0.3 w1118 Male 0.78 ± 0.09ǂ 0.79 ± 0.2 0.32 ± 0.07 0.21 ± 0.03 2.8 ± 0.3 fumin 1.2 ± 0.2 1.0 ± 0.4 0.42 ± 0.1 0.19 ± 0.04 3.5 ± 0.7 Canton S 1.2 ± 0.2 2.2 ± 0.6 0.48 ± 0.1 0.21 ± 0.05 6.0 ± 2 w1118 Female 1.4 ± 0.2ǂ 1.4 ± 0.5 0.42 ± 0.1 0.25 ± 0.07 4.9 ± 0.9 fumin 1.5 ± 0.2 1.0 ± 0.4 0.30 ± 0.09 0.15 ± 0.03 4.4 ± 0.8

84 presented herein, from 0.75 ± 0.08 to 1.5 ± 0.2 ng/head are also similar to previously reported values of ~1-2 ng of histamine per whole head.75,76

However, our results for whole-head serotonin, octopamine, and L-DOPA content are much higher than in previous reports. Serotonin levels in this study ranged from 0.30 ±

0.09 to 0.5 ± 0.1 ng/head. In contrast, previously reported serotonin levels in Drosophila were quite low, from 0.009 to 0.038 ng/head, and 0.016-0.039 ng/brain.56,69,72,74 For example, based on 25-head samples analyzed by negative ion GC-MS with chemical ionization, Watson et al. obtained values of ~8.8 pg of serotonin per Canton S head.72

Two papers by the Venton group, using CE-FSCV, reported about 0.022 ng per Canton S brain, (in one case the reported value is for the larval brain, i.e. before the onset of metamorphosis).69,74 Hardie et al. measured 37.7 ± 1.0 pg and 39.3 ± 1.3 pg of serotonin in the heads and brains of w1118 flies, respectively.56

Reported octopamine levels vary from 0.0005 ng/head to 0.203 ng/head,56,72,76 whereas our values average 0.8 – 2.2 ng/head. Ream et al. reported the whole head concentration of L-DOPA in Canton S flies as 747 ± 59.8 fmol/head (approximately 0.15 ng/head), based on MEKC-EC, whereas our values were much higher, in the 1.7 - 6.0 ng range.32

Considering the validation of quantification methods with both comparison to LC-EC and published honey bee brain values, the likely cause of this discrepancy is the additional complication of the Drosophila cuticle. Whole Drosophila heads were used rather than brains to increase the throughput of sample collection. Homogenization is performed manually with plastic pestles and breaking up the cuticle entirely and reproducibly is challenging. For the sake of reproducibility, a different homogenization method should be used, or future samples should be dissected out of the cuticle.

2.3.4 Sexual dimorphism was not detected across three Drosophila strains As females are considerably larger than males, one might expect female heads to show

Figure 2.10 Biogenic amine content (ng/head) in Drosophila melanogaster whole heads, by sex. Sexual dimorphism was detected in only one instance. w1118 females had higher whole-head histamine content than males. While whole-head content often trended slightly higher in females (particularly for histamine, octopamine, and L-DOPA), it was only significant for histamine in w1118 flies. Numbers in white are the n for each sample type, where each n is a preparation for 10 whole heads. Error bars are s.e.m. *Denotes statistical significance (one- way ANOVA with Bonferroni post-hoc, p > 0.05).

86 across-the-board increases in whole-head contents of many biologically relevant compounds. Results are shown in Figure 2.10. While whole-head content seems slightly higher in females for some biogenic amines (namely histamine, octopamine, and L-

DOPA), we found statistically significant elevations in female heads compared to males in only one instance. Histamine content was elevated in female w1118 flies compared to male w1118 flies, but not female Canton S or fmn flies (one-way ANOVA, with Bonferroni’s

Multiple Comparison Test p < 0.05) compared to males of the same strain. We did not see differences in whole-head content for any other biogenic amines between sexes.

Histamine, octopamine, and L-DOPA appear higher in females, but the difference is not statistically significant in most cases (p > 0.05). The published literature on dopamine in the Canton S strain includes conflicting reports on sexual dimorphism, but these studies are difficult to compare with our data because of biological sample differences. Denno et al. dissected brain tissue from adult heads and quantified biogenic amines by CE-FSCV, finding a significant sex difference only for dopamine, which was higher in female tissue.74

Neckameyer et al. used whole flies and quantified dopamine by HPLC, observing higher concentrations in females at some ages, but not at 3 days (the age of our fly samples).77

Ream et al. used heads and MEKC-EC and did not see a sex difference.32 We saw no significant difference in dopamine content in females versus males of any strain. In general, we did not detect sexual dimorphism across the three Drosophila strains.

2.3.5 Deletion of ABC transporter does not affect whole-head histamine The results of our quantification of whole-head biogenic amines is presented by strain in Figure 2.11.

The white gene encodes an ABC transporter that imports eye pigment precursors tryptophan and guanine.65 Comparison of the Canton S strain and the w1118 strain (which has a Canton S background) allows for the determination of changes in whole-head biogenic amine content with this mutation. In Drosophila, histamine is the major

Figure 2.11 Biogenic amine content (ng/head) in Drosophila melanogaster whole heads, by strain. Statistically significant changes in whole-head biogenic amine content were not detected for any of the five compounds tested across three strains of Drosophila, leading to the conclusion that these genetic mutations affect biogenic amine signaling but not whole-head content. Numbers in white are the n for each sample type, where each n is a preparation for

10 whole heads. Error bars are s.e.m.

88 neurotransmitter used by photoreceptors at synapses in the brain.78,79 Most of this histamine exists in the eyes.76,80,81 Pigmentation mutants (such as the w1118 used herein) are believed to have variations in their histamine content; in fact, Borycz et al. reported three pigmentation mutants of Oregon S that had about half the histamine content of the wild type flies using LC-EC.67 Previously published studies have also reported a decrease in the histamine content, as well as the dopamine and serotonin levels in wild type versus white mutants.67–69 However, a conflicting report states that there is no statistical difference between Canton S and w1118 mutants, quantified with HPLC-MS.70 For histamine in males of Canton S, w1118, and fumin strains, we measured 0.75 ± 0.08, 0.78

± 0.09, and 1.2 ± 0.2 ng/head, respectively. For histamine in females of Canton S, w1118, and fumin strains, we measured 1.2 ± 0.3, 1.4 ± 0.2, and 1.5 ± 0.2 ng/head, respectively.

Our study does not detect statistical significance across strains for histamine (one-way

ANOVA with Bonferroni post-hoc, p < 0.05). Interestingly enough, there was actually an increase in histamine in the fumin strain, though non-significant. The fumin strain is hyperactive, as the lack of DAT means that stimuli cause increased responses.46 There is some evidence that histamine may mimic light-evoked neurotransmitter release and plays a role in sleep/wake cycles75,82; perhaps an increase in histamine content plays a role in the reported hyperactivity and reduced sleep45,46 in fumin strains, though a genetic cause and why this would be restricted to males is unclear. A larger sample set would allow us to determine if this is a real effect or simply due to sampling. In general, we did not see differences in whole-head biogenic amine content with an ABC transporter mutation.

2.3.6 Genetic inactivation of DAT in Drosophila melanogaster does not systematically alter whole-head biogenic amine content In Drosophila, dopamine is a regulator of arousal.16,46 The reduced uptake of dopamine and hyperactivity of fmn

89 mutants31,46 raise the question of whether total dopamine content of the brain might be elevated. We did not see differences between strains, despite the mutation that eliminates the dopamine transporter. This leads us to believe that while dopamine signaling is affected, synthesis is not. In general, however, we are led to conclude that genetic inactivation of DAT in Drosophila melanogaster does not systematically alter whole-head biogenic amine content.

2.4 Conclusions

In this work, we have developed an approach for the quantitation of biogenic amines using benzoyl chloride derivatization followed by liquid-liquid extraction coupled to direct- injection MS2 analysis. This method couples a rapid column-free sample preparation with a more widely applicable detection method for high throughput and the ability to quantify a wide range of biogenic amines. This method is simple and reproducible, is capable of being applied to complex matrices, and is sufficiently sensitive for mass-limited samples such as Drosophila melanogaster. Due to the robustness of this technique, it can be applied to many systems. This method is inexpensive, sensitive, and high-throughput.

2.5 Author Contributions

LC-EC experiment was performed by DCF and KLP. MJB performed lesions, and both

MJB and KLP collected rat brain samples. SAL provided advice and training on Drosophila strains and rearing. SLG reared honey bees and collected brain tissue. CLK performed all Drosophila sample collection, and all sample preparation, and mass spectrometric analysis.

90 2.6 Acknowledgements

The authors would like to thank the Han Lab at UTEP and the Jackson Lab at Tufts

University for providing fly strains. Thanks also to Dr. Adam R. Meier for performing the oxidation study.

91 Chapter 3 Nosema ceranae parasitism in honey bees (Apis mellifera) increases biogenic amines associated with foraging behavior and alters olfactory learning and memory Abstract

Nosema sp. is an internal parasite of the honey bee, Apis mellifera, and one of the leading contributors to colony losses worldwide. This parasite is found in the honey bee midgut, and has profound consequences on the host’s physiology. Nosema sp. impairs foraging performance in honey bees, yet it is unclear how this parasite affects the bee’s neurobiology. In this study, we examined whether Nosema sp. affects odor learning and memory and whether the brains of parasitized bees show differences in amino acids and biogenic amines. We took newly emerged bees and fed them with Nosema ceranae. At approximate nurse and forager ages, we employed an odor-associative conditioning assay and two bioanalytical techniques to measure changes in brain chemistry. We found that nurse-aged bees infected with N. ceranae significantly outperform controls in odor learning and memory—suggestive of precocious foraging; but by forager age, infected bees were slower to learn and showed memory impairment. We detected significant differences in amino acid concentrations, some of which were age-specific; as well as altered serotonin, octopamine, dopamine, and L-DOPA concentrations in the brain of parasitized bees. These findings suggest the effects of N. ceranae parasitism extend to the brain and behavioral tasks may be compromised. These results yield new insight into the host-parasite dynamic of honey bees and N. ceranae, as well the neurochemistry of odor learning and memory under normal and parasitic conditions.

92 3.1 Introduction

Nosema sp. is an internal parasite of the honey bee Apis mellifera, and one of the most significant factors contributing to colony losses.1 Given the global importance of honey bee pollination to the reproduction of floral species and to agricultural productivity, it is important to understand how Nosema sp. parasitism affects honey bee health. Nosema sp. is an example of a microsporidian, which constitute a group of spore-forming unicellular parasites classified as fungi. Bees typically become infected with Nosema sp. through the ingestion of spores found in contaminated food and water. Once infected, spores begin to thrive in the epithelial cells of the midgut.2 Over time, cell walls will rupture and spores will be excreted through the fecal matter. At very high levels of infection,

Nosema sp. resembles dysentery. Bees, who are naturally hygienic and excrete outside of the hive, will defecate in and around the hive, spreading the infection to other workers and to the queen.3,4

Nosema ceranae, the Nosema species used in this study, affects several areas of honey bee physiology. The rRNA for N. ceranae was first coded in 1996, and was believed to be restricted to Asian honey bees, Apis cerana.5 However, N. ceranae was isolated from

Apis mellifera in 2006.6 This led researchers to screen older stored samples, and evidence of N. ceranae was found in American bee colonies dating back to at least 1995.7

N. ceranae induces gene expression changes in nutritional, metabolic, and hormonal pathways in the midgut and fat body, and alters gene expression in the brain.8–10 N. ceranae obtains energy for replication from the honey bee midgut, harming the honey bee’s epithelial cells and development as the infection grows.8,11 This effect increases oxidative stress in the bee12 and depletes nutrient reserves, affecting the hypopharyngeal glands in both structure and function.13–15 Portions of the midgut proteome responsible for energy production, protein regulation, and antioxidant defense are also altered.16 This

93 evidence is part of a growing body of knowledge that N. ceranae disrupts nutrient digestion and metabolism in its host. As a result, infected honey bees show increased hunger. They demonstrate increased sucrose sensitivity12 and are less likely to share food with nestmates via trophallaxis.17

At the behavioral systems level, less is known about Nosema sp. infection, but there are intriguing observations related to foraging behavior. Bees infected with Nosema sp. are more likely to forage at a younger age than uninfected bees.18–22 Precocious foraging alters population age structure in the colony and can lead to colony losses.3,18,20,23–25

There are reports of infected bees being unable to return to the hive, and generally exhibiting poor foraging performance.19,26–28 Infected bees are also more likely to engage in riskier behavior, such as increased foraging trips during adverse weather conditions29 and robbing other hives for resources.30 Nosema sp. affects the number of flights taken, and the average duration.19,31,32

These observations give rise to our hypothesis that Nosema sp. may affect foraging behavior through a decline in cognitive ability or an alteration of neurochemical signaling.

Several neurological tasks are required for successful communication of foraging sites- processes that include spatial navigation, visual processing, and odor learning and memory.33 Given that energy stores are depleted under Nosema sp. infection, it is possible that these foraging tasks, which are metabolically expensive, are impaired under infection. It becomes necessary then to determine whether Nosema sp., a gut-dwelling pathogen, exerts specific effects on the brain. We sought to address this question by asking the following: (1) Does Nosema sp. interfere with the ability to associate an odor with a reward? If so, (2) does Nosema sp. impair memory of the odor-association? And,

(3) if there are changes in odor learning and memory, how might Nosema sp. alter neurochemistry, such that amino acids and key behavior-regulating biogenic amines are

94 affected? In this study, we use a forward-paired, associative odor-conditioning assay using the proboscis extension reflex (PER), to test the learning and memory of caged bees at approximate nurse and forager ages (days 7 and 15, respectively).

Nosema infection clearly affects behavior, therefore it is reasonable to hypothesize that neurochemicals related to behavior are being altered. To understand the effects of

Nosema ceranae infection, we quantified several biogenic amines and amino acids, looking for links between changes in the whole-brain content of these compounds and the behavioral effects seen in infected bees. For in-depth method development for biogenic amine quantification, see Chapter 2.

3.2 Materials and Methods

3.2.1 Animals Brood frames were collected from colonies at the Carl Hayden Bee

Research Center locations in Tucson, Arizona between March and July, 2016. Frames were taken from European colonies, Apis mellifera ligustica, headed by queens from

Pendell Apiaries (Stonyford, CA). Bees, less than 12 hours old, were inoculated with

100,000 spores of Nosema sp. After inoculation, bees were separated into cages according to treatment with 50 bees per cage. Bees were kept in a Binder BD (E2) incubator at 31.7 °C with 50% relative humidity, under constant dark conditions.

3.2.2 Feeding Bee-collected pollen was removed from pollen traps in the surrounding

Tucson, AZ area on February 26, 2016 and kept at -20 °C until use. Each cage was given an insert with three grams of pollen, 50% sucrose solution, and water ad libitum. Pollen, sucrose, and water were changed every seven days. Cages were checked daily and dead bees were removed.

95 3.2.3 Nosema inoculum Spores were collected from bees found at the entrance of an infected hive the day before, or the day of, inoculation. Over the course of the experiment, multiple hives were used as the source of Nosema spores.

A single infected bee abdomen was crushed with a mortar and pestle in 1 mL of water.

Ten µL was transferred to a hemocytometer for a spore count. Five squares were counted and the following equation was applied to yield 50,000 spores per 1 µL:

Number of Spores Counted / Number of Squares Counted = # of spores in 4 nL

# of spores in 4 nL * 250,000 = total # of spores in 1 mL of water

Total # of spores in 1 mL of water / 1000 µL = Number of spores per µL in sample

Once the Nosema suspension was determined, the sample was spun down, and the supernatant was removed and reconstituted in a 50% sucrose solution for the desired volume. Each bee was placed into an Eppendorf tube cut with a hole, which was large enough for a proboscis to extend through. Each bee was hand fed with 2 µL of Nosema inoculum (for a delivery of 100,000 spores per bee) or 2 µL of 50% sucrose solution to the proboscis for controls. Each bee was observed feeding either through the extension of the proboscis through the hole in the tube upon treatment, or through the opening of the tube itself and stimulating the antennae to elicit proboscis extension and feeding. Pollen, sucrose and water were kept from cages for one hour after feeding to ensure infection. A sample of the inoculated bees was sent to the Bee Research Center in Beltsville, MD

(ARS-USDA) for identification of Nosema species. DNA was extracted from infected and control honey bee abdomens and amplified using primers specific to Nosema apis and N. ceranae.34 N. ceranae was confirmed as the source of infection according to size.34

Nosema apis was not detected.

96 3.2.4 Spore counts Spore counts from Day 7 and Day 15 bees was determined using the abdomen from a bee whose brain was analyzed for biogenic amines. Spores were counted using a hemocytometer and calculated as reported in Fries et al., 2013.34 On day

7, Nosema-infected bees averaged 14,575,000 spores (±13,203,254 stdev, n = 20). Two of 20 control bees revealed Nosema spores (800,000 and 30,900,000) on Day 7.

Chemical analysis for these two bees was placed into the Nosema-infected category. On

Day 15, spore counts of Nosema-infected bees increased to an average of 109,390,104 spores (± 39,526,537 stdev, n = 16). Zero spores were counted in Day 15 controls (n =

16).

3.2.5 Learning and memory Learning and memory experiments took place between

March and May 2016. The night before associative-learning tests, sucrose was removed from the cage between 5 – 6 p.m. The next morning, bees were restrained in a 1 mL pipette tip cut such that the body was restrained and the neck was free to rotate. Wax was used around the opening for further restraint. Bees were tested for the PER by applying a wooden applicator soaked in 50% sucrose to the tip of the antenna. Bees were not allowed to lick. If the bee did not exhibit a strong PER (rapid full extension) it did not proceed to the study.

Seven- and 15-day-old bees were assessed for associative odor learning in a forward- paired conditioning paradigm. Clove oil (diluted 1:1000 in mineral oil; Sigma) was placed on filter paper in a 10 µL volume and inserted into a 0.5 mL glass syringe. The syringe was placed one inch from the bee, and connected to a solenoid-controlled air stream. The solenoid was powered by an Interval Generator 1830 (W.P. Instruments, Sarasota, FL) to deliver a five-second odor pulse (7 km/h). Three seconds into the pulse, a wooden applicator soaked with 50% sucrose was presented to the antenna. The bee was allowed to lick for one second. This sequence was repeated for three trials spaced ten minutes

97 apart. Three odor conditioning trials was found to be the least number of trials needed for long-term memory.35 Three trials were chosen to assess a potentially subtle difference in learning and memory with Nosema infection, which may be masked with a stronger conditioning paradigm of more spaced trials. All experiments were performed between 10 a.m. - 12 p.m. under red light.

Animals were tested for odor learning and memory at three time points after conditioning: one hour, four hours, and twenty-four hours. These are approximate periods of time when memory traces occur and are indicative of late short-term, mid-term, and early long-term memory.35 At each time point, the bee was presented with the five-second odor pulse, and scored on proboscis extension immediately following the odor. At each time point, the bee was tested twice.

At the end of the experiment day, all animals were fed until satiation, typically between 12

– 16 µL of 50% sucrose between 5 – 6 p.m. Feeding was performed away from the odor delivery area to ensure place-conditioning did not occur. Bees were kept restrained overnight at room temperature in a covered box with 1 – 2 inches of water to maintain humidity. Memory tests at 24 hours were performed the following day.

3.2.6 Amino acid analysis of brain tissue and pollen Whole brains were dissected from

7- and 15-day-old bees between July and August 2016. Bees were flash frozen in liquid nitrogen between 2 – 5 p.m. and stored at -80 °C until dissection. Each brain was rapidly dissected, weighed using a Sartorius CP2P microscale, and frozen in liquid nitrogen before transferring to -80 °C until analysis.

Each brain was homogenized using a bead beater for 30 seconds (100 mg of 1.0 mm beads, 500 μL of deionized water). Two-hundred µL aliquots of brain homogenate was subjected to one of three analysis procedures to control for losses with digestion:

98 (1) Conventional Acid Hydrolysis: 500 μL of 6 M HCl with 4% thioglycolic acid was added to the sample, sealed in an inert atmosphere and digested at 70 °C for 24 hours.

Fifty μL aliquots were filtered and dried down for derivatization.

(2) Base Hydrolysis: 600 μL of 4 M NaOH was added to the sample, sealed in an inert atmosphere, and digested at 90 °C for 4 hours. Two hundred μL aliquots were filtered, neutralized with 6 M HCl, and dried down before derivatization.

(3) Sodium Azide Acid Hydrolysis: 780 μL of 6 M HCl, 20 μL of 1% phenol (in 6 M

HCl), 100 μL of 12 M HCl, and 100 μL of 8% sodium azide were added to the sample, sealed and then digested at 70 °C for 24 hours.36 Twenty-five μL aliquots were transferred to 2 mL amber glass vials for derivatization.

Conventional acid hydrolysis with chloroformate derivatization was used to quantify all amino acids, with the exception of tryptophan, cysteine and arginine. Tryptophan and cysteine are destroyed under acidic conditions, and arginine cannot be derivatized using chloroform. Tryptophan was recovered using base hydrolysis with chloroformate derivatization. Cysteine and arginine were quantified with sodium azide acid hydrolysis followed by phenylisothiocyanate (PITC) derivatization. The latter method enables cysteine to be quantified in its oxidative form, cysteic acid, and arginine into a phenylthiocarbamyl derivative. Asparagine and glutamine are hydrolyzed to their acidic forms and are reported here as asparagine/aspartic acid and glutamine/glutamic acid, respectively.37

Conventional acid hydrolyzed and base hydrolyzed samples were analyzed using the EZ: faast Amino Acid Analysis Kit for Protein Hydrolysates by Gas Chromatography – Mass

Spectrometry (Phenomenex, Torrence, CA, USA). The re-dissolved chloroformate derivatives were analyzed by EI GC-MS on an Agilent 7890A gas chromatography system coupled with a 5975C EI mass spectrometer detector. One µL extracts were injected on

99 a Zebron ZB-50 capillary column (30 m x 0.25 mm I.D. x 0.25 µm film) in a 1:15 split mode with an injector temperature of 250 °C and helium as the carrier gas (1.1 mL/min).

Separation was achieved using an oven program with an initial temperature of 110 °C that increased to 320 °C at a rate of 30 °C/min recommended by the EZ: faast protocol for amino acid analysis. Chloroformate derivatives were identified by comparison of retention times and mass fragmentation patterns with derivatized standards and quantified using major ions (SIM).

Sodium azide hydrolyzed samples were analyzed using a modified method from Elkin and

Wasynczuk.38 The method consists of a neutralization step using a 2:2:1 mixture of methanol:water:triethylamine (TEA) (v/v) followed by a 20 min derivatization using a

7:1:1:1 mixture of methanol:TEA:water:PITC (v/v). Methanol washes were applied to remove interfering compounds before samples were dried down. Due to time- and light- sensitivity, the phenylthiocarbamyl amino acids were re-dissolved with a 5 mM solution of disodium hydrogen phosphate containing 5% acetonitrile (pH 7.4) directly before being analyzed.

The re-dissolved phenylthiocarbamyl derivatives were analyzed by reverse-phase HPLC-

PDA on a Thermo Scientific Spectra System coupled with a Finnigan Surveyor PDA Plus

Detector. Twenty µL of the extract was injected and separated using a Pico-Tag column

(3.9 × 150 mm) with a linear gradient pattern adopted from Kwanyuen and Burton.39 The gradient started with 100% of solvent A, a mixture of 150 mM CH3COONa·3H2O, 0.05%

TEA, and 6% acetonitrile (pH 6.1), and finished with 100% solvent B, a 6:4 acetonitrile:water (v/v) mixture. The flow rate was set at 1 mL/min with a column temperature of 38 °C, and the detection wavelength was set to 254 nm.

Phenylisothiocyanate derivatives were quantified and identified by comparison of retention times of derivatized standards.

100 3.2.7 Biogenic amine analysis of brain tissue Whole brains were dissected from 7- and

15-day-old bees taken from the same cage as those bees used for behavioral experiments. A bee was placed into a scintillation vial and chilled on ice until immobile

(between 2 - 5 minutes). The brain was rapidly dissected, weighed, and placed into 50

µL of chilled 0.1 M perchloric acid. The brain was ground manually, frozen in liquid nitrogen, and transferred into -80 °C storage until analysis. Because levels of biogenic amines are known to fluctuate throughout the day,40–42 all brains were dissected between

2 - 5 p.m.; and each dissection alternated between treatments to control for any time of day variances between 2 - 5 p.m. The corresponding abdomen was saved at -20 °C degrees for Nosema spore counts.

Sample volume was brought to 100 µL in 0.1 M perchloric acid. Benzoylation was performed by adding 200 µL of 200 mM carbonate buffer (pH 11.0) followed by 100 µL of

2% benzoyl chloride in acetonitrile (v/v). The mixture was vortexed for five seconds. Two liquid-liquid extraction steps were performed using 200 µL each of dichloromethane

(DCM), followed by two washes of the organic layer with 200 µL each of basified water

(nanopure water adjusted to pH 8 with ammonium hydroxide) to minimize the carryover of unwanted species such as benzoic acid and salt products. The solution was evaporated to dryness using a Speedvac Concentrator (Thermo Savant). The resulting product was reconstituted in 50:50 H2O:acetonitrile with 0.1% formic acid for electrospray compatibility.

Solutions were diluted in 50:50 H2O:acetonitrile with 0.1% formic acid as necessary to bring signals into the linear dynamic range of the instrument.

Detection and quantitation were performed with an Applied Biosystems QStar Elite mass spectrometer. Samples were injected using a 20 µL injection loop, with 50:50

H2O:acetonitrile with 0.1% formic acid spray solvent flowing at 8 µL/min. The most abundant fragment peak for benzoylated compounds is the 105 m/z benzoyl fragment.43

101 The area of the 105 m/z peak was quantified over the course of the 2.5-min bolus.

Samples were run in duplicate, values were averaged for each sample, and concentrations were calculated from a standard calibration curve.

3.2.8 Statistics JMP 12.0.1 was used for all statistics. The behavioral results were analyzed for effect of Nosema using a Wilcoxon (Rank Sums) test with a Chi-Square approximation. This test was applied separately for learning trials 2 and 3, and memory testing at 1, 4, and 24 hours. Amino acids and biogenic amines were individually analyzed using a two-way, full factorial ANOVA with an LS Means Differences Student’s T-test. All error bars are reported as standard error of the mean (s.e.m.). All tests employ α = 0.05 and a 95% confidence interval.

3.3 Results and Discussion

3.3.1 Odor-associative learning and memory in nurse- and forager-aged bees

Reports of Nosema-infected bees foraging differently prompted us to consider whether N. ceranae, a pathogen that resides in the midgut, affects the brain of the honey bee. We focused this question by examining odor learning and memory performance in the laboratory. Odor learning and memory is a task necessary for successful foraging and involves the coordination of several areas of the insect brain, including the antennal lobe, mushroom bodies, and the subesophageal zone. For review see Gauthier and

Grunewald.44 To test whether N. ceranae might affect the neurobiology of honey bees, we tested olfactory learning and memory in nurse- and forager-aged bees. We used a forward-paired, odor associative conditioning assay using the proboscis extension reflex

(PER). Bees at Day 7 and Day 15 were trained to associate an odor with a sucrose reward using three pairings spaced ten minutes apart.

We found that N. ceranae does affect odor learning and memory, and that the effects are age-specific. At 7 days old, bees infected with Nosema learned to associate an odor with

102 a reward similar to control bees (Figure 3.1a). Memory performance, however, differed with infection. Nosema-infected bees showed increased PER when tested for memory of the conditioned odor (Figure 3.1c). Significant increases in PER occur with Nosema at 1 hr [X2 (1, n = 122) = 25.98, p < 0.0001] and 4 hr [X2 (1, n = 106) = 5.14, p = 0.02]; and continued slightly at 24 hr [X2 (1, n = 94) = 3.28, p = 0.07] (Figure 3.1c), indicating that

Nosema-infected bees have enhanced odor learning and memory performance at nurse age. This effect may be an indicator of accelerated maturation in response to N. ceranae infection that may lead to precocious foraging, which has been reported elsewhere in

Nosema-infected honey bees.18–22 We can add that infected bees demonstrate heightened odor learning and memory performance at nurse age, suggesting an increased physiological capacity for mechanisms of memory.

At 15 days old, differences in trial learning emerged (Figure 3.1b). PER was evaluated during trial learning at the onset of the odor stimulus prior to sucrose reward. Control bees had higher PER than Nosema-infected bees at trial two [X2 (1, n = 67) = 3.058, p = 0.08] and significantly higher PER at trial three [X2 (1, n = 67) = 3.99, p = 0.04]. When PER performance was compared between Day 7 and Day 15, Nosema-infected bees did not improve in trial learning. Infected bees showed an average of 61.2% PER (Trial 3) to the conditioned odor on Day 7, and 56.7% PER (Trial 3) on Day 15. Control bees, in contrast, showed a significant increase in trial learning from 50% on Day 7 to 80% on Day 15 [X2

(1, n = 60) = 3.28, p = 0.02] (Figure 3.1a & b).

Day 15 also revealed differences in memory performance between treatment groups.

Nosema-infected bees had a significant deficit in memory at the 1-hour time point [X2 (1, n = 148) = 6.41, p = 0.01] (Figure 3.1d). Odor memory tested at 4 hours and 24 hours was reduced in comparison to control bees, but was not significant at either time.

103

Figure 3.1 Honey bee learning and memory trials. Honey bee olfactory learning and memory in 7- and 15-day old bees with and without Nosema ceranae. a & c) Day 7 infected bees (n = 31) and controls (n = 30) were tested for learning and memory using the proboscis extension reflex (PER). b & d) Day 15 infected bees (n = 37) and controls

(n = 30) were tested for learning and memory of the conditioned odor. Asterisks denote significance, α = 0.05, and error bars denote s.e.m.

104 The impact of N. ceranae in odor learning and memory changes with the age of the bee.

Average spore counts rose from approximately 14 million at nurse age to over 100 million per bee at forager age. In trial learning, an indication of learning acquisition, forager-aged

Nosema-infected bees showed reduced proboscis extension responses at trials two and three. This suggests that at this age and level of infection, Nosema-infected bees are slower to learn. When memory is tested, Nosema-infected bees show a significant memory deficit one hour after training, although we are reluctant to suggest N. ceranae affects memory formation specific to this time window. Our paradigm for testing memory

(two odor puffs at three different time periods) appears to induce extinction of the odor association which may mask a potential role for N. ceranae in specific forms of memory, i.e. long-term vs. short-term. These results suggest that forager bees with Nosema- infection may be compromised due to deficits in odor learning and memory. These results, however, contrast with those found in the Charbonneau et al. 2016 study, which found limited effect of Nosema sp. in odor learning and memory using PER.45 We speculate these variant results may be due to the strength of the conditioning paradigm used. We employed three spaced odor/sucrose pairings as the minimum conditioning required for long-term odor memory35 and tested this association at three time points within twenty- four hours. Charbonneau et al., in contrast, used eight pairings and tested at 24 hours.

This latter, robust paradigm is interesting in that given extensive training, odor memory is similar to controls. In the field, this may translate to Nosema-infected bees needing to make more flower visits than controls to learn and remember odors at a similar rate. N. ceranae infection and the act of foraging is energetically demanding, and it may be unlikely infected bees could compensate in this manner. A direct comparison of the two training paradigms, perhaps even including intermediate-level training would be an interesting study into the effect of Nosema on memory formation.

105 3.3.2 Amino acid concentrations in the whole brain of nurse-and forager-aged bees

If odor learning and memory is disrupted under N. ceranae infection, it would suggest N. ceranae imposes brain-specific effects that may cause the altered foraging behavior that has been observed. We may also expect to find dysregulation in amino acids and biogenic amines in the brain that regulate signaling pathways. To further test the impact of N. ceranae infection on brain physiology, whole bee brains were analyzed for 18 amino acids.

Each amino acid was analyzed for infection and age using a two-way ANOVA, with a post- hoc LS Means Differences Student’s T-test.

Amino acid analysis in biological solutions dates back to at least the 1950’s,46–48 though method development continued for decades to optimize for different amino acids.

Conventional acid hydrolysis using hydrochloric acid (HCl) and thioglycolic acid was reported as far back as 1962,49,50 though not all amino acids are stable under acidic conditions. For example, tryptophan is not easily quantified under these conditions,51 nor is cysteine.36,52 For that reason, the following amino acids were analyzed using conventional acid hydrolysis with chloroformate derivatization: Ala, Asp/Asn, Glu/Gln, Gly,

Ile, Leu, Lys, Met, Phe, Pro, Ser, Thr, Tyr, His, Val. Trp was quantified using base hydrolysis, which uses sodium hydroxide to break down the proteins, rather than HCl.53

Cysteine and arginine were quantified using sodium azide acid hydrolysis, followed by phenylisothiocyanate (PITC) derivitization. NaN3 oxidizes cysteine to cysteic acid for reproducible quantitation.36,54 Chloroformate derivitization does not work on arginine,55 though PITC derivitization works for both arginine and cysteine.39,56

At 7 days old, bees infected with Nosema show significant concentration differences in eight amino acids as compared to control bees: (1) alanine, (2) asparagine/aspartic acid,

(3) cysteine, (4) glutamine/glutamic acid, (5) isoleucine, (6) lysine, (7) methionine, and (8) proline (Figure 3.2, Table 3.1). Nosema-infected bees had statistically lower levels in each

106 of these amino acids with the exception of cysteine. Cysteine levels in infected bees were notably higher.

At 15 days old, bees infected with N. ceranae show significant concentration differences in ten amino acids: (1) alanine, (2) asparagine/aspartic acid, (3) glutamine/glutamic acid,

(4) isoleucine, (5) lysine, (6) methionine, (7) proline, (8) arginine, (9) histidine, and (10) tyrosine (Figure 3.2, Table 3.1). The seven amino acids found to be lower on Day 7 were also found to be lower on Day 15. In addition, arginine, histidine, and tyrosine are also significantly lower with infection (Tables 3.1 & 3.2, and Figures 3.2 & 3.3). At day 15, cysteine values in infected bees were no longer higher than in control bees. Altogether, these ANOVA results showed that ten amino acids in the brain were significantly affected by N. ceranae, most often decreased.

Eleven of 18 amino acids were significantly affected by age (Figure 3.3, Table 3.2), typically showing an increase in whole-head amino acid content with age, with the exception of valine. Only five amino acids- glycine, leucine, serine, tryptophan and threonine- were unaffected by age or Nosema infection. Glycine showed a borderline p- value with age, p = 0.05, and tryptophan showed a borderline p -value with infection, p =

0.05. Three amino acids—arginine, cysteine, and methionine—showed a significant interaction with both infection and age (Table 3.2).

Cysteine is the only amino acid that is significantly increased with N. ceranae, but only in nurse-aged bees. This anomaly merits further study into the physiological role of cysteine under parasitic infection. It is possible that cysteine, which is important for immune response, plays a role in precocious foraging in response to parasitic infection. By forager age, three additional amino acids are lower- arginine, tyrosine, and histidine, while cysteine is no longer increased. These amino acids either directly or indirectly affect immune response57 and lower levels might suggest reduced immunity in Nosema-infected

107

Figure 3.2 N. ceranae affects amino acid levels in Day 7 and Day 15 honey bees.

Amino acid concentrations were measured in 79 whole brain homogenates in N. ceranae-infected bees and controls. Amino acid levels in control and N. ceranae- infected bees were compared for Day 7 (top) and Day 15 (bottom). Asterisks denote significance using a Two-Way ANOVA, with Post-hoc Student’s T-test, α = 0.05, and error bars denote s.e.m.

108

Figure 3.3 Effects of amino acids in whole brain tissue in Day 7 and Day 15 honey bees with and without N. ceranae infection. Amino acid levels on Day 7 and Day 15 were compared for control bees (top) and N. ceranae-infected bees (bottom). Asterisks denote significance using a Two-Way ANOVA, with Post-hoc Student’s T-test, α = 0.05.

Error bars denote s.e.m.

109 Table 3.1 Summary of compounds that vary significantly with N. ceranae infection.

Significance determined using Two-Way ANOVA with Post-hoc Student’s t-test, α = 0.05.

n Compound P-Value Effect of N. ceranae (Control, Infected) Day 7 Ala < 0.0001 Decrease 20, 19 Asp/Asn < 0.0001 Decrease 20, 19 Glu/Gln 0.0007 Decrease 20, 19 Ile 0.01 Decrease 20, 19 Lys 0.04 Decrease 20, 19 Met < 0.0001 Decrease 20, 18 Pro 0.007 Decrease 20, 19

Cys 0.0006 Increase 20, 18 Serotonin 0.001 Increase 8,11 L-DOPA 0.01 Increase 7,11 Octopamine 0.05 Increase 7, 11

Day 15 Ala < 0.0001 Decrease 20, 19 Asp/Asn 0.007 Decrease 20, 19 Glu/Gln < 0.0001 Decrease 19, 19 Ile 0.02 Decrease 20, 19 Lys 0.0003 Decrease 20, 19 Met 0.001 Decrease 20, 16 Pro < 0.0001 Decrease 20, 19 Arg 0.001 Decrease 16, 18 His 0.001 Decrease 15, 18 Tyr 0.003 Decrease 20, 19

110 Table 3.2 Examining the effects of age and infection on compounds in the brains of honey bees. Results from a two-way ANOVA examining the effects of age (Day 7 or Day 15) and N. ceranae infection on compounds in the honey bee brain. Asterisks denote P-values less than 0.05.

Effect of Nosema, Effect of Age, Compound Nosema*Age P-Value P-Value Ala < 0.0001* < 0.0001* 0.8984 Arg 0.0394* < 0.0001* 0.0117* Asp/Asn < 0.0001* < 0.0001* 0.2329 Cys 0.386 0.2682 0.0019* Glu/Gln < 0.0001* 0.0051* 0.3122 Gly 0.1107 0.0504 0.5657 His 0.0073* 0.0022* 0.1634 Ile 0.0019* < 0.0001* 0.8237 Leu 0.1116 0.9893 0.3058 Lys < 0.0001* 0.0003* 0.2208 Met < 0.0001* < 0.0001* 0.0003* Phe 0.9589 < 0.0001* 0.6966 Pro < 0.0001* < 0.0001* 0.0831 Ser 0.2623 0.6546 0.2963 Thr 0.1003 0.4857 0.4782 Trp 0.0572 0.1071 0.7528 Tyr 0.0039* 0.0288* 0.2685 Val 0.7552 0.0049* 0.6434 Histamine 0.1656 0.2009 0.5209 Octopamine 0.0797 0.0163* 0.2805 Serotonin 0.0011* 0.302 0.1251 Dopamine 0.0472 0.071 0.8789 L-DOPA 0.069 0.2266 0.0411*

111 bees at this age or level of infection. Arginine, tyrosine, and histidine are also in the biosynthesis pathways of known neurotransmitters involved in odor learning and memory:

(1) nitric oxide, (2) dopamine and octopamine, and (3) histamine, respectively. These amino acids may be candidate compounds to better understand normal learning and memory in the honey bee, as well as under parasitic challenge. These results also raise the possibility of diet supplementation as a means to rescue cognitive impairment and improve overall health. This strategy could include a diet supplement that beekeepers use to feed their colonies. It needs to be determined whether an enhanced diet can overcome the nutrient deficiency N. ceranae imposes. A goal moving forward might be to supplement honey bee nutrition in such a way that Nosema sp. becomes asymptomatic.

3.3.3 Biogenic amine levels in nurse- and forager-aged bees Biogenic amines were also quantified in honey bee brains. BAs can function as neurotransmitters, neuromodulators, and neurohormones. For honey bee reviews regarding biogenic amines, see Bicker, 1999; Scheiner et al., 2006; Gauthier and Grunewald, 2012.44,58,59

As discussed previously, measurement of biogenic amines dates back several decades, but is most often accomplished used LC-EC, which requires a separation (decreasing sample throughput), and can only be used to quantify electrochemically-active compounds. The method we introduced in Chapter 2 can be used to quantify biogenic amines in complex matrices without a chromatographic separation. In the previous chapter we showed that this method can be applied to brain homogenate with higher throughput than existing methods.

The brain of a honey bee is approximately 1 mg wet mass, and published literature values for biogenic amine content in the brain of a honey bee are in the single nanogram regime.60–66 Therefore, any method for quantifying these molecules must be capable of sensitive detection in mass-limited samples. While it is possible to pool samples to

112 increase the sample mass and therefore concentration, this will result in the loss of information about individual insects. We would be able to broadly compare the effects of infection, but would have no individual information. With a method sensitive enough for single brain analysis, we can collect and correlate data on multiple facets of infection, such as behavior, number of Nosema spores found on the body, and biogenic amine content.

Additionally, any anomalous insects will be easily identified, rather than skewing the data for an average of several insects. The method we use here has detection limits on the 5-

50 nM scale for a 20 μL injection volume, correlating to the ability to detect tens of picograms of material, easily sensitive enough for quantitation of biogenic amines in single honey bee brains. As shown in Chapter 2, the presence of exoskeleton fragments can cause additional complications in the analysis, so these samples have been dissected out of the exoskeleton and are only brain homogenate.

We hypothesized that if N. ceranae infection affected the neurochemistry of a honey bee brain, we would see differences in biogenic amine levels correlating to behavioral changes. Additionally, since changes in the amino acid content of bee brains were seen, tracking compounds downstream in the biosynthetic pathways could allow us to better understand how Nosema infection causes altered behavior. Five biogenic amines were measured to understand how these key behavioral-regulating chemical messengers might be affected by N. ceranae infection. Whole brains were analyzed for histamine, octopamine, serotonin, dopamine, and L-DOPA to test for effects caused by infection and age (Table 3.2, Figure 3.4). Data was normalized to nurse-age control bees for comparison and two-way ANOVA was used to determine significance.

We found that serotonin and dopamine levels appeared higher in Nosema-infected bees at both ages (Serotonin: F = 12.43, p = 0.001; Dopamine: F = 4.21, p = 0.04. In the case of serotonin, a post-hoc analysis showed significantly higher levels of serotonin with N.

113

Figure 3.4 Biogenic amine concentrations in whole brain tissue. Five biogenic amines were quantified in whole brain tissue in Day 7 and Day 15 honey bees with and without N. ceranae infection. Biogenic amine content is normalized to control bees, day

7, for comparison. For Control, Day 7, n = 8. For Control, Day 15, n = 12. For infected,

Day 7, n = 11. For infected, Day 15, n = 11. Groups connected by the same letter are not significantly different. α = 0.05 and error bars denote s.e.m.

114 ceranae at Day 7 than age-matched controls (t = 2.02, p = 0.001). Dopamine was not found to be significantly higher than age-matched controls at Day 7 or Day 15.

Histamine was not significant for age or infection, though Nosema-infected bees may have a slight decrease in histamine content (a larger sample set would be required to tease out this possibility).

Octopamine and L-DOPA are higher in Nosema-infected bees, though their p-values are borderline respectively, p = 0.07, and p = 0.06. When we dissected these results further, there are significant changes in biogenic amine levels that were specific to age. L-DOPA is the only compound of this group to show a significant interaction effect with Nosema and age (Infection*Age F = 4.47, p = 0.04). Post-hoc analysis showed that L-DOPA is significantly higher in Nosema-infected bees at Day 7 (t = 2.02, p = 0.01), but no difference was found at Day 15. It is worth noting that while L-DOPA was significantly higher in Day

7 bees, dopamine was not, despite being the biological product of L-DOPA.

Octopamine is the only biogenic amine significant with age. Octopamine is elevated from

Day 7 to Day 15 for both Nosema (18.3 ng of OA/mg of brain tissue ± 2.03 s.e.m.) and control bees (16.3 ng of OA/mg of brain tissue ± 2.34 s.e.m.). Nosema-infected bees at

Day 7 had elevated octopamine levels (14.2 ± 3.17) in comparison to age-mated controls

(5.77 ± 3.46) [t = 0.05]. Nosema-infected bees at Day 7 showed comparable octopamine levels to control bees at Day 15 (t = 0.56).

From the literature, we know high levels of serotonin and octopamine are associated with foraging.63,67–69 In fact, octopamine treatment is sufficient to induce precocious foraging70; and improves learning and memory in newly emerged bees.71 Moreover, bees induced to forage precociously show higher, forager-like levels of serotonin and octopamine in their antennal lobes, and bees that revert back to nursing have lower levels.72 Our results in the whole brain show similar trends with N. ceranae infection. Levels of serotonin and

115 octopamine are elevated in Nosema-infected nurses and do not differ from control foragers. We suspect that heightened serotonin and octopamine in infected nurses underscore our findings of enhanced odor learning and memory and could be an additional indication that Nosema-infected bees at nurse age are precocious foragers. We would also like to point out that we measured neurotransmitters between March and August of

2016 and found wide variation in response to N. ceranae, which needs further study. We matched the neurotransmitter data, the bulk of which was collected in the spring, to the time behavior was performed between March and May. In effect, our studies report the results of spring bees fed spring pollen and it is possible that fall bees respond to N. ceranae differently.

It is necessary to consider our results in light of the host-parasite relationship. Precocious foraging may be the bee’s effort to replace lost nutrients to N. ceranae, but it could also be advantageous to the parasite. For instance, the behavioral change to foraging for pollen may be aiding parasite replication within the honey bee. Two studies15,73 and unpublished results in our collaborator’s lab found that Nosema-infected bees fed pollen had significantly higher spore loads than bees without pollen, suggesting spore replication is greater when the host has consumed pollen. A hive that is rich in pollen would therefore fuel N. ceranae replication within the hive. Precocious foraging could provide a means of dispersal for the parasite as well, as infected bees are known to drift, or rob other hives, potentially spreading infection.74 As bees age and spore counts rise, the probability of drifting increases. Based on these observations, it is possible that behavioral changes in

Nosema-infected bees may be the host’s response to fight off infection (which in some cases end up being beneficial to the parasite, as discussed by Campbell et al.), or they could be parasitic “manipulation” of honey bee behavior.75

116 The biogenic amine changes we observed in infected bees resemble those described in other examples of parasitic manipulation of the host. Serotonin, octopamine, and dopamine are neuromodulators commonly affected by parasitism.76,77 These neuromodulators can modify neural circuits to accommodate for behavioral changes to meet the animal’s immediate survival needs and to adapt to a changing environment. This behavioral plasticity can also come at a price, because it can open the animal to manipulation by another organism.76 Intra- and extra-CNS parasites of gammarids, for example, alter serotonin signaling in the host. Serotonin modulates escape behaviors in crustaceans, and it is suggested that parasites of gammarids manipulate serotonin to make the host more susceptible to predation, an effect suggested to enhance parasite transmission.78 The parasitic wasp Cotesia congregata, through an unknown mechanism, elevates octopamine levels in the brain, thoracic and abdominal ganglia in its host larvae,

Manduca sexta.79 Normally hungry caterpillars show reduced feeding, which is a behavior that increases parasite survival.79 The parasitic wasp Ampulex compressa secretes a variety of substances to zombify the cockroach, Periplaneta americana.80 These substances act upon multiple neurotransmitter systems including the cholinergic,

GABAergic, dopaminergic, and octopaminergic systems of the host.

In each of these examples, multiple mechanisms occur to affect the behavior of the host, such as changes in the neuromodulatory system, the neuroendocrine system, and the immune system.76 We can only speculate about the honey bee-N. ceranae dynamic, but if it is like other host-parasite examples, N. ceranae could be affecting multiple mechanisms to ensure its survival. One mechanistic possibility for how a gut-dwelling parasite could affect the brain is described by the neuro-immune hypothesis. This theory proposes that parasite-induced behavioral change may be the result of the parasite’s attempt at circumventing/defeating host immune responses.81 If the parasite can

117 manipulate the immune system, it may not need to reside in the brain to affect behavior.

Immune-derived molecules have privileged routes of transmission to the brain82 to affect neural cells; and biogenic amines function as neurohormones circulating throughout the insect body, and are affected by the immune response through interactions with cytokines.83 Octopamine, for example, is released during both stress and immune responses in insects and is thought to be one aspect of the immune-neural connection manipulated by the parasitic wasp, C. congregata in M. sexta.84 The schistosome parasite

Tricholbilharzia ocellata is perhaps the clearest example of a parasite manipulating the immune system. T. ocellata secretes schistosomin, a molluscan cytokine-like molecule85 into its snail host, Lymnaea stagnalis, which suppresses the snail’s neuroendocrine cells leading to a reduction in egg laying. The energy from the snail is redirected to support parasitic growth. Something similar may be occurring in honey bees with Nosema infection. N. ceranae has been found to suppress the immune system of the honey bee.8,86,87 N. ceranae upregulates the naked cuticle gene, nkd, a negative regulator of host immune function.87 This has the effect of suppressing the host’s immune response and when nkd is knocked down, several immune genes are upregulated and N. ceranae spore loads are reduced.87 The oxidative stress and metabolic disruption N. ceranae imposes could also be viewed as a parasite extracting energy from the host to aid its own survival.

3.3.4 Comparing whole-brain content of compounds along biological synthesis pathways can elucidate mechanisms of change in infected bees The quantification of both amino acids and biogenic amines presented herein has the benefit of allowing us to compare whole-head content of compounds along several biosynthetic pathways. For example, histamine is directly synthesized from histidine by the action of histidine decarboxylase. Figure 3.5a shows the pathway and the data for both compounds across

118

Figure 3.5 Changes in whole-head content of compounds by biosynthetic pathway. By comparing the changes in whole-head content of the biogenic amines and their precursors, including amino acids, it may be possible to determine where in the pathway changes are occurring. a) Histamine is directly synthesized from histidine, and both compounds show similar trends. b) Serotonin is synthesized from tryptophan.

The large increases seen in serotonin but not tryptophan require further study. c)

Tyrosine in the amino acid precursor to L-DOPA, dopamine, and octopamine. An increase in whole-head content of L-DOPA in infected nurse-age bees indicates that tyrosine hydroxylase may be affected. Octopamine is higher than control nurse-aged bees for control foragers and both ages of infected bees.

119 age and infection, normalized to nurse-aged control bees for each compound. Both histidine and histamine show the following trends: 1) a slight increase with age in both control and infected bees, 2) a slight decrease in nurse-age infected bees, and 3) forager- aged infected bees have whole-head content similar to control nurse-age bees, but lower than control bees of the same age. The trends in histidine and histamine across age and infection are similar, which indicates that the pathway itself is not necessarily affected by infection. Therefore, the decrease in both histidine and histamine seen in infected bees at day 15 compared to control bees at day 15 is likely due to changes in the metabolism of histidine.

Figure 3.5b shows the pathway for the synthesis of serotonin from tryptophan.

Tryptophan is hydroxylated by the action of tryptophan hydroxylase, producing 5- hydroxytryptophan (5-HTP), which is decarboxylated to produce serotonin (5- hydroxytryptamine, 5HT). In this study, only tryptophan and serotonin were quantified.

Tryptophan shows non-significant decreases in whole-head content with both age and infection, whereas serotonin shows increases in whole-head content with both age and infection. Therefore, tryptophan metabolism appears unaffected by infection, though the biosynthetic pathway that produces serotonin is altered. Without data for 5-HTP it is difficult to tease out where in the pathway this alteration is occurring. In the future, it would be beneficial to quantify 5-HTP as well. Structurally, it would be a good candidate for benzoyl chloride derivatization.

Figure 3.5c shows the biosynthetic pathway for L-DOPA, dopamine, and octopamine, all of which are synthesized from tyrosine. Tyrosine is hydroxylated by tyrosine hydroxylase to form L-DOPA, which is decarboxylated by dopa-decarboxylase (DDC) to form dopamine. All three compounds were quantified in this study. Tyrosine shows a slight increase with age and a slight decrease with infection. L-DOPA, however, shows a

120 significant increase with infection in nurse-aged bees, indicating that tyrosine hydroxylase may be upregulated at this young age. Dopamine shows increased whole-head content with both age and infection, though the trend is slightly different than that of L-DOPA.

Dopamine is only slightly increased in infected nurse-age bees compared to nurse-age control bees, whereas L-DOPA is significantly higher. Despite a larger whole-head content of L-DOPA, the dopamine content is not significantly increased, which may indicate an upregulation of L-DOPA metabolism, or a downregulation of DDC activity.

Additionally, dopamine whole-head content is highest in infected forager-aged bees. L-

DOPA decreases from nurse- to forager-age in infected bees, indicating that DDC activity may be upregulated during this stage in the bee lifespan. It would be interesting to quantify these compounds at a few more stages to track these changes with time.

Tyrosine can also be decarboxylated to form tyramine, which is hydroxylated to form octopamine. Tyramine was not quantified. Octopamine shows the same general trends as tyrosine (increases with both age and infection), though the increases are more significant, which may indicate that slight changes in tyrosine content are magnified as compounds progress down the pathway. It would be useful to add tyramine to the biogenic amine quantification to determine where this is occurring.

Changes in whole-head content as a function of age or infection can be compared between the compounds along their biosynthetic pathways. These comparisons can be used to tease out how Nosema infection may cause changes in neurochemicals and, subsequently, behavior. Our results indicate that in some cases (namely histidine) the metabolism of amino acids appears to be altered in infected bees, whereas in other cases

(such as tryptophan) the pathway itself seems to undergo differential regulation in infected bees. Adding further biosynthetic products would provide useful information on where dysregulation is occurring in infected bees. Additionally, quantifying both amino acids and

121 biogenic amines at a wider range of ages would provide a smoother overview of neurochemical changes as a function of both age and infection.

3.4 Conclusions

Nosema ceranae infection is known to cause behavioral changes such as precocious foraging and altered hygiene. We show here that N. ceranae infection also affects memory and olfactory learning, and neurochemical content. We observed two distinct behaviors to N. ceranae infection that occurs with age, or length of parasite incubation. We saw evidence of precocious foraging occurring at nurse age, which could be viewed as a novel behavior for the host, and a deleterious behavior with reduced learning and memory performance in forager ages. We found that infected bees exhibit improved memory at nurse-age, which may indicate accelerated maturation and lead to early foraging.

However, we also found that infected bees exhibit decreased memory and learning at forager-age, which may explain why infected bees are more likely to forage under dangerous conditions, or forage more often. In order to elucidate the underlying causes of these behavioral changes, we analyzed eighteen amino acids and five biogenic amines in whole honey bee brains. We found that seven amino acids were decreased in infected bees at both ages, and also that cysteine was increased in infected bees, but only at nurse-age. Three amino acids that affect immune response are decreased in forager- aged bees, perhaps signaling a decreased ability to fight the infection at this later stage.

We also found that both serotonin and octopamine are increased in Nosema-infected bees at day 7 and in both control and infected bees at day 15. As both serotonin and octopamine are implicated in foraging, the increase at an early age is likely related to precocious foraging, and the increase in Day 15 bees may be the natural evolution of foraging. This certainly warrants further investigation into precocious foraging to determine if foraging is caused by increased serotonin and octopamine, or if there is

122 another variable that affects both behavior and these biogenic amines. We also saw that in some cases amino acid metabolism appears to be altered while the downstream biosynthesis is not. In other cases this is reversed, leading to the conclusion that Nosema parasitism has many modes of action that alter the neurochemistry and thus the behavior of the host.

Combining our behavioral and chemical data and examples of other host - parasite relationships, we theorize that N. ceranae may be manipulating honey bee behavior to aid its proliferation and dispersal. These results may have special importance for managed apiaries where colonies are close together. A better understanding of how N. ceranae is affecting the honey bee brain could provide better strategies to curb infection. To better understand the mechanisms of altered behavior in Nosema infection, we need to continue studies into the neurochemistry of infected bees, particularly targeting compounds we have shown are affected. As further studies into the effect of N. ceranae infection on Apis mellifera, we intend to expand our repertoire of biogenic amines, to increase the age range under investigation, and to examine the effects of season and diet. We will begin studies into the correlation between extent of infection (spores found per bee) and changes in the biogenic amine content of the brain. We will also be looking into the effects of other parasites, such as the mite Varroa destructor.

3.5 Author Contributions

SLG designed experiments, reared animals, performed behavioral experiments, collected specimen samples, and provided the data for Figure 3.1 and Tables 3.1 and 3.2. SC performed the amino acid quantitation shown in Figure 3.2 and 3.3. CLK performed biogenic amine method development and quantitation, and prepared all figures. Writing was done by SLG and CLK.

123 3.6 Acknowledgements

We would like to thank Mona Chambers, Geoff Hildago and Henry Graham for their assistance with colony management at the CHBRC. We would like to thank Michele

Hamilton and Dr. Judy Chen for help in identifying the Nosema species. Als thank you to

Dr. Hong Lei for advice in setting up the odor delivery system. This work was funded through an APHIS grant to GDH.

124 Chapter 4

Glycosylation of peptide-based drug candidates improves in vivo stability

and penetration of the blood-brain barrier

Abstract

Peptide-based drugs show great promise for the treatment of many diseases, due to their inherent selectivity and biocompatibility. However, they are rapidly degraded in vivo and tend to be unable to penetration the blood-brain barrier, which is necessary for drugs targeting the CNS. Glycosylation can improve the in vivo properties of a peptide, such as stability against peptidases, and penetration across the blood-brain barrier. We present herein the effects of glycosylation on three classes of peptide-based drugs, including changes in the fragmentation patterns, increased lifetime both in vitro and in vivo, and improved penetration properties. We have developed an in vitro model for the degradation of peptides in rat serum and show that glycosylation of peptides increases their stability against peptidases in serum. We have also developed an in vivo method for the quantitation of peptides in serum and CSF. With this method, we are able to co-inject related drugs of interest and directly compare the effect of glycosylation while circumventing the challenge of inter-animal variability. We show that glycosylation can improve the in vivo lifetime and blood-brain barrier penetration of a peptide-based drug, providing support for further development of peptide-based drugs.

125 4.1 Introduction

Peptide-based pharmaceuticals show great promise due to their high specificity and biocompatibility. Small molecule drugs often have unanticipated side effects due to non- specific binding to receptors, whereas drugs derived from endogenous peptides have high affinity for pre-existing receptors.1–3 A mechanism of action based on natural processes of the body is less likely to have deleterious side effects. Additionally, peptide-based drugs can typically be broken down into non-hazardous degradation products, as they are comprised of naturally-occurring amino acids. This natural degradation process does have one major downfall that has limited the use of peptide-based drugs; compounds based on endogenous species tend to be rapidly degraded in the body, often too quickly to be of use as a therapeutic treatment.3 For this reason, methods for screening the stability of peptides are necessary as a part of drug development.

Additionally, for drugs meant to target diseases of the central nervous system, the ability to penetrate the blood-brain barrier (BBB) is paramount. Originally observed by Ehrlich in the early 1900s during studies with dye, the BBB is still not fully understood.4 The blood brain barrier is comprised of the cerebral capillary endothelial cells that form tight junctions, preventing compounds from crossing into the cerebral spinal fluid from the blood.5 The BBB serves to protect the CNS from harmful compounds like neurotoxins, bacteria, and some macromolecules, while simultaneously allowing necessary compounds like sugars, salts, and small-molecule precursors to neurotransmitters to penetrate.5 Few small molecule drugs can cross the BBB, and even fewer peptide drugs.6

Unfortunately, it is rare for peptides to cross the BBB. In order to develop highly effective peptide-based drugs, many researchers have explored ways to increase the lifetime of peptide-based drugs in vivo as well as their ability to permeate the BBB.6–12 Some approaches include enzyme inhibition or masking of enzyme target sites for increased

126 lifetime through cyclization, addition of a sugar moiety, and halogenation.9,11,13,14 In this chapter we will investigate the effect of glycosylation.

In order to determine the permeability of a compound across the BBB, it is necessary to develop screening methods. Early studies were performed with injection of visible dyes and subsequent histology, but this has the obvious downfall of being a postmortem procedure. Another approach to studying the permeability of compounds across the BBB is the use of radioactive compounds and positron emission tomography for detection.15

However, these studies require the synthesis of radioactive compounds, the assumption that labeled compounds behave the same as non-labeled compounds, and additionally have relatively poor spatial resolution (several millimeters).15 In fact, this method is used more often to study disruption of the BBB than specific compounds and their ability to cross.16

More recently, some researchers have been developing in vitro cell culture models using endothelial cells, often with a combination of neurons and astrocytes to mimic the cellular structure of the BBB as closely as possible.17–20 However, it is difficult to develop in vitro cell models that perfectly mimic the BBB in terms of types and quantities of receptors, microvasculature, and transport mechanisms.

Another way to measure BBB permeability is using microdialysis. Intracerebral microdialysis is an in vivo procedure that samples from the cerebrospinal fluid (CSF) of the brain, opposite the BBB from the blood. Microdialysis samples from solution based on permeability across a membrane with a set molecular weight cut-off. The concentration gradient across the membrane causes compounds in the solution to cross the membrane into the perfusate, which is then collected as dialysate and analyzed. For a more in-depth explanation of microdialysis, see Chapter 1. Measuring the presence of target compounds in vivo accounts for more parameters than model systems, such as degradation and blood

127 clearance. Intracerebral microdialysis has been used for decades to investigate the BBB permeability of drugs.21–25

Previous studies have shown that glycosylation of peptides can retard degradation by hindering peptidases, and can lead to improved BBB penetration.1,14,26–29 In this work, we address the behavioral alterations that occur when a peptide is modified with a sugar moiety, from changes in the fragmentation patterns to in vivo stability and BBB penetration. We have developed an in vitro method for screening the stability of peptides in a biological solution. We have also developed an in vivo method using dual microdialysis and blood draws to measure the penetration of compounds across the BBB and to monitor the lifetime of compounds in the blood and in the cerebrospinal fluid. We present herein three classes of compounds, with and without glycosylation, and compare their behavior under these conditions. The structures are shown in Figure 4.1.

The first class of compounds (referred to as the Ang derivatives) are based on Angiotensin

1-7, an endogenous peptide that is of interest in several fields of medicine, as it has been shown to have provide cardioprotective effects, protective effects in hepatic fibrosis, and neuroprotective effects in cases of ischemic stroke.30–32 However, Ang 1-7 has the same downfalls as many other peptide-based drugs, in that it has rapid degradation in vivo and poor BBB penetration. Thus, synthesis of a derivative that can overcome these challenges in necessary in order to progress in drug development studies. Here, we investigate two derivatives of Ang 1-7. The first compound is Ang 1-6-Ser-NH2 (DRVYIHS-NH2, Figure

4.1a), which is amidated Ang 1-7 with the seventh residue replaced with a serine. The second compound is a sugar-modified derivative (DRVYIHS[O-Glc]-NH2, Figure 4.1b).

The second class of compounds contains the compound MMP-2200 (Figure 4.1c), also known as lactomorphin. Synthesized and patented by our collaborators in the Polt Lab at the University of Arizona, MMP-2200 is a glycosylated opioid peptide and a mu and delta

128

Figure 4.1 Glycosylated and non-glycosylated compounds. a) Ang 1-6-Ser-NH2,

DRVYIHS-NH2, is currently of interest in the medical field for its potential for neuroprotective effects. b) A glycosylated modification of Ang 1-7, with the C-terminus amidated and the seventh residue replaced with a serine with a glucose moiety attached. c) MMP-2200, AKA lactomorphin, is an endogenous opioid-based peptide synthesized by our collaborators in the Polt Lab. d) SAM-995 is the non-glycosylated version of MMP-2200. e) PACAP may provide neuroprotection in cases of

17 neurodegenerative diseases, so we are investigating [Leu ] PACAP 1-27-NH2, which is PACAP 1-27 but with the methionine residue replaced with leucine for oxidative stability, and amidation of the C-terminus for improved stability against peptidases. f) A

17 glycosylated PACAP, [Leu ] PACAP 1-27-Ser(O-Glc)-NH2 has an added serine residue with a glucose moiety attached.

129 receptor agonist28 with potential for alleviating the effects of dopaminergic hyper- stimulation.33 Lactomorphin is based on the endogenous opioid leucine enkephalin

(YAGFL), and has the sequence YTGFLS, with an amidated C-terminus and a lactose moiety on the serine residue. BBB penetration was reported for MMP-2200 in 2012 using a microdialysis-based approach.34 For comparison, we also analyze the “non-modified” version of this compound, i.e. the same peptide sequence but without the sugar moiety on the serine residue, referred to herein as SAM-995 (Figure 4.1d).

The third class of compounds are PACAP derivatives. PACAP, or pituitary adenylate- cyclase activating polypeptide, is a member of the glucagon family, and is known to play a role in neuromodulation and neuroprotection.35–43 Two active isoforms have been identified, PACAP 1-38, and PACAP 1-27. Here we investigate two derivatives of PACAP

1-27. The first derivative (Figure 4.1e) has the seventeenth residue (a methionine) replaced with a leucine for stability against oxidation, and is amidated on the C-terminus for protection against peptidases. This peptide is referred to henceforth as [Leu17] PACAP

1-27-NH2 and has the sequence HSDGIFTDSYSRYRKQLAVKKYLAAVL-NH2. The glycosylated variant under investigation has an added serine with a glucose moiety

17 attached (Figure 4.1f, [Leu ] PACAP 1-27-Ser(O-Glc)-NH2).

In this work, we investigate the effect of glycosylation on three classes of peptide-based drugs, from fragmentation properties in the instrument during method development and quantitation, to in vivo behavior. We have developed methods for comparing the effect of glycosylation on peptide-based drugs in terms of in vivo stability and BBB penetration across different classes and sizes of peptides. We show that glycosylation of peptides increases their stability against peptidases in vitro and in vivo, and improves blood-brain barrier penetration in vivo. This work will open to door for pharmaceutical development of

130 more biocompatible drugs with higher specificity and fewer side effects, while avoiding the inherent limitations of current peptide-based drugs.

4.2 Materials and Methods

4.2.1 Chemicals and reagents All chemicals were purchased through Sigma Aldrich (St.

Louis, MO, USA), unless otherwise stated. Internal standard (d-Alanine d-Leucine enkephalin, dAdLE) was purchased through American Peptide Company, Inc. (Sunnyvale,

CA).

4.2.2 Peptide synthesis Peptides and glycosylated peptides were synthesized by the Polt

Lab (University of Arizona) from commercial material via solid-phase synthesis, as previously reported.1,44,45

4.2.3 MS identification and quantification of peptides Compounds were initially analyzed on a Thermo LTQ Velos-Orbitrap instrument in flow infusion mode for determination of exact mass and identification of fragmentation patterns. Experiments for the determination of in vitro degradation rates were performed on an Applied Biosystems

QStar Elite mass spectrometer, using direct injection (20 μL sample loop, 8 μL/min

2 ACN:H2O:FA) and MS fragments for quantification. The area under the curve for the main MS2 fragments was used for quantification over the course of the 2.5-minute bolus.

Values were normalized to the 0-minute fraction as 100% signal. LC-MS3 methods were developed for in vivo samples and a Proxeon nano-LC coupled to a Thermo LTQ Velos-

Orbitrap instrument using Advion TriVersa NanoMate chip-based nanoelectrospray technology. The compounds were fragmented twice, yielding an MS3 spectrum, and the identifiable fragments were used for quantification. Table 4.1 shows the fragment masses and identifications used for quantification with both methods.

131

Table 4.1 Fragment masses and identification for quantification of compounds of interest.

The main peptides fragments are identified for each compound. In vitro stability studies quantified

MS2 fragments, whereas quantification in BBB penetration studies was performed using LC-MS3 quantification for an additional level of selectivity.

Compound Parent Main MS2 fragment MS3 fragments

+2 +2 +2 Ang 1-6-Ser-NH2 444.74 [M+2H] 392.71 [b6+2H] 370.20 [a6-NH3+2H] +2 378.71 [a6+2H] + 647.35 [b5+H]

+2 +2 +2 Ang 1-6-Ser(O-Glc)-NH2 525.77 [M+2H] 444.74 [-Glc+2H] 378.71 [a6+2H] +2 392.71 [b6+2H] +2 435.73 [-Glc-H2O+2H] + 647.35 [b5+H]

+ + + SAM-995 686.35 [M+H] 582.29 [b5+H] 554.30 [a5+H] + 564.28 [b5-H2O+H]

+ + + MMP-2200 1010.46 [M+H] 686.35 [M-Lac+H] 441.21 [a4+H] + 469.21 [b4+H] + 538.23 [a5-NH2+H] + 582.29 [b5+H] + 651.31 [-Lac-NH3-H2O+H] + 669.32 [-Lac-NH3+H]

+3 17 +5 +4 709.75 [b18+3H] [Leu ] PACAP 1-27-NH2 626.75 [M+5H] 750.65 [b26+4H] + 855.36 [b8-H2O+H] + 873.37 [b8 +H] + 887.50 [y8-NH3+H] +3 925.84 [c23+3H] 954.85 [YSRYRKQ+H]+ [Leu17] PACAP 1-27- +5 +5 +4 Ser(O-Glc)-NH2 675.95 [M+5H] 643.54 [M-Glc+5H] 778.16 [b26+4H]

132 4.2.4 In vitro stability of peptides Degradation rate studies were performed in rat serum

(rSer) in a 37 °C water bath. Peptides were spiked in for a final concentration of ~100 µM and time points were determined by apparent degradation rate after n = 2 trials. At each time point, 10 µL rSer + peptide was removed and spiked with 1 µL glacial acetic acid to quench peptidase activity and 1 µL of 100 µM d-alanine d-leucine enkephalin (dAdLE) as an internal standard. Samples were desalted using µ-C18 Zip Tips® (EMD Millipore) according to the following protocol, on a 10 µL scale: three conditioning steps with acetonitrile (ACN), followed by three conditioning steps with 0.1% trifluroacetic acid (TFA) in nanopure water (v/v), then 15 binding steps. A single wash with 0.1% TFA desalted the column, and the procedure was completed with 10 elution steps in 60:40 ACN:H2O:0.1%

TFA (v/v). Samples were brought to 100 µL in 50:50 ACN:H2O with 0.1% formic acid for electrospray compatibility. Samples were analyzed using tandem MS on an AB Sciex

QStar Elite mass spectrometer with a 20 µL injection loop. Samples were run in duplicate, averaged for each time point, and normalized to the 0-minute time point as 100% signal.

Analyses were performed in GraphPad Prism 5 (La Jolla, CA), and degradation trials with exponential fits with r2 < 0.7 were discarded.

4.2.5 Animals Male Sprague-Dawley rats (275-325 grams) were used for in vivo experiments. Animals were purchased from Harlan Laboratories (Indianapolis, IN) and were housed in a temperature and humidity controlled room with 12 h reversed light/dark cycles with food and water available ad libitum. All animals were treated as approved by the Institutional Animal Care and Use Committee, University of Arizona and in accordance with the NIH Guidelines for the Care and Use of Laboratory Animals. Both the number of animals used and their suffering were minimized.

4.2.6 Carotid artery catheterization Carotid artery catheterizations were performed by staff at Envigo Laboratories. Animals were anesthetized with 1.5-2% isoflurane mixed in

133 medical grade oxygen (1.5 L/minute) and two incisions were made: (1) a 0.5 cm midline incision between the scapulae; (2) a 2.0 cm incision to the right of the midline of the ventral neck. The right carotid artery was then isolated from surrounding tissues. Using 4-0 silk suture, two loose ties were placed at the caudal and cranial ends of the vessel. The cranial suture was then tied off and a bulldog clamp was placed above the caudal tie to temporarily occlude blood flow. A small incision was made in the vessel and a piece of

PE-10 catheter tubing, pre-flushed with a heparin:glycerol (500 IU Heparin/1 mL; 50:50 ratio) solution, was passed through the vessel towards the heart. The bulldog clamp was then released and the caudal suture was tied off to secure the catheter in place. A small subcutaneous tunnel was then formed behind the ear connecting with the incision at the scapulae. The catheter was then connected to a vascular access port (VAP). The ventral neck incision was closed with stainless steel wound clips, while the VAP was secured with and its incision closed with 4-0 silk sutures. Animals were monitored during recovery for

3-4 days before being shipped University of Arizona facilities. Catheter patency was evaluated within first 24 hours of arrival and subsequently every 2-3 days to ensure patency prior to microdialysis.

4.2.7 Microdialysis surgery Upon arrival at our facility, animals were allowed to acclimate for 5-7 days. Animals were anesthetized with 1.5-2% isoflurane mixed in medical grade oxygen (1.5 L/minute) and positioned inside a stereotaxic frame (David Kopf Instruments,

Tujunga, CA). A midsaggital incision was made along the skull and a 26 AWG guide cannula (AZ-08; Eicom, San Diego, CA) was implanted in the striatum at the following coordinates AP: +1.0 mm and ML: +3.2 mm relative to bregma and DV -3.4 mm from the surface of the brain. The guide was then replaced with an Eicom AZ-08-04 probe containing a 4 mm cuprophane (Cupr) membrane with a molecular weight cut-off of 50

134 kDa. After the probe was positioned, the animals remained under anesthesia for the remainder of the experiment.

4.2.8 Microdialysis and blood draws The microdialysis set-up included the use of a micro-tee to couple in an acidic preservation solution, as reported in Laude et al.

(Appendix B), which prevents peptidase activity during collection. For microdialysis, a dual syringe pump (PHD 2000, Harvard Apparatus, Holliston, MA) was set to a flow rate of 0.5 L/minute. As samples were collected, the micro-tee coupled in preservation solution, comprised of 10% v/v acetic acid, 2% v/v acetonitrile, and 140 pM d-alanine d- leucine enkephalin (dAdLE). dAdLE was used as an internal standard. Baseline blood

(~100 L) and dialysate samples were collected. Subsequent time-locked samples (with blood being drawn at the median of the dialysate time range) were then collected every

10 minutes for 90 minutes, following a tail vein injection at t = 0 min. In addition, blood samples were also collected at both 1- and 5- minutes post-injection. CSF samples were immediately frozen on dry ice. Blood samples were spun down to separate serum from the red blood cells. Serum was drawn off, and 10 L of serum was spiked with glacial acetic acid to quench peptidase activity and internal standard to a final concentration of

10% HAc and 140 pM IS before being frozen on dry ice. Samples not run immediately were stored in a -80 °C freezer and thawed immediately before analysis. Sample preparation for analysis required ZipTip® cleanup, as discussed previously, drying of samples, and reconstitution in 5 μL of 0.1% TFA for injection onto the column.

4.3 Results and Discussion

4.3.1 High ion throughput of glycosylated peptides during tandem MS yields lower limits of detection Peptides were initially analyzed via flow-injection on a high resolution

Orbitrap instrument for determination of exact mass and identification of peptide fragmentation patterns. With the intention of developing sensitive and selective LC-MS3

135 quantitation methods for in vivo studies, CID breakdown curves were built by increasing the CID Energy (an arbitrary value) and recording the subsequent ion intensity. By building CID curves, we can optimize the applied CID Energy in our quantitation methods for the maximum ion intensity of target fragment peaks. During this process, we came across the fortuitous discovery that glycosylation yields simple and highly efficient collision-induced dissociation (CID) fragmentation patterns. Figure 4.2a shows the full

+2 mass spectrum for Ang 1-6-Ser-NH2, where the most abundant peak is the [M+2H] ion.

Figure 4.2b shows the MS2 spectrum, with the inset CID breakdown curve. Fragmentation of peptides has been studied in great depth, and most peptides have been shown to fragment along the peptide backbone in CID, providing b and y ions.46 This leads to several identifiable fragments in the MSn spectra, as shown in Figure 4.2b. With multiple stable fragmentation pathways, the gas phase concentration of any fragment is lower. In

+2 the case of 1-6-Ser-NH2, the main fragment, [b6+2H] , has only about 40% of the ion intensity of the parent. Figure 4.2c shows the full mass spectrum for Ang 1-6-Ser(O-Glc)-

+2 NH2, where the most abundant peak is again the [M+2H] ion. We have seen that with glycosylated compounds, the main fragment in MS2 is cleavage of the sugar moiety with high efficiency. As shown in Figure 4.2d, the [M–Glc+2H]+2 fragment is virtually the only peak in the spectrum, and has a nearly 80% ion throughput compared to the parent. In

Figure 4.3, we show the MS2 spectrum for SAM-995 and the MS3 spectrum for MMP-2200

(the lactoside), which shows the same effect. With an added level of fragmentation (which means an added level of selectivity), we end up with the same spectrum. For added selectivity in biological matrices, our method involves the use of MS3 quantitation.

Due to the low concentrations of drug used in animal studies compared to in vitro studies, the inherent dilution of microdialysis techniques, and the complexity of biological samples, it was necessary to develop a quantitation method that optimized sensitivity and selectivity

136

Figure 4.2 Fragmentation spectra and CID breakdowns. a) The most abundant peak

+2 2 for Ang 1-6-Ser-NH2 is the [M+2H] parent peak. b) The MS spectrum for Ang 1-6-Ser-

NH2 shows 5 identifiable fragments, with about 40% ion throughput for the largest peaks

(CID breakdown, inset). c) The most abundant peak for Ang 1-6-Ser(O-Glc)-NH2 is the

+2 2 [M+2H] parent peak. d) Ang 1-6-Ser(O-Glc)-NH2 fragments highly efficiently. The MS spectrum is dominated by a single peak, [M-Glc+2H]+2, with about 80% ion throughput.

137

Figure 4.3 Added selectivity with glycosylation. a) MS2 spectrum for SAM-995, a non-glycosylated opioid-based peptide. Seven fragments are identifiable, but only five are specific. b) MS3 spectrum for MMP-2200, the glycosylated peptide. The MS3 spectrum is identical to the MS2 spectrum for SAM-995, but has an added level of selectivity for improved quantitation in complex matrices.

138 for our target molecules. As shown in Figure 4.4, LC-MS3 quantitation can be used to increase the detection limits of peptides in solution by increasing the selectivity of the quantitation. Figure 4.4a shows the chromatogram for 100 pM Ang 1-6-Ser(O-Glc)-NH2 in 0.1% TFA with full scan MS detection. The elution time is 4.18 minutes, but no peak is visible due to interference in a fresh solution. In a complex solution the interference would be considerably worse. The MS spectrum is shown in panel 4.4b. Again, the known masses of 525.8 Da [M+2H]+2 and 1050.5 Da [M+H]+ are not distinguishable from the baseline. Panel 4.4c shows the chromatogram with MS2 detection (selecting and fragmenting 525.8 Da), and the peak for the target compound is visible but not the most abundant. In the corresponding MS2 spectrum (4.4d) the only peak attributable to the target compound is 444.999 (the glucose loss). Panel 4.4e shows the chromatogram with

MS3 detection, selecting for 525.8 followed by 444.8. The chromatogram is solely the target compound, as is the MS3 spectrum (4.4f). In this way we manage to gain selectivity for samples that will be analyzed in a complex matrix, and have limits of detection in the low pM regime for glycosylated compounds.

Therefore, glycosylation results in several benefits: 1) highly specific fragmentation patterns that are useful for detection in complex matrices, and 2) with higher ion throughputs comes lower limits of detection (as shown in Table 4.2). In addition to potentially improving stability and BBB penetration, sugar modified compounds have improved fragmentation behavior in the mass spectrometer, leading to lower limits of detection.

During method development, we came across an error in the mass of our glycosylated

PACAP derivative. The expected exact monoisotopic mass is 3376.7779 Da, but the measured mass was 3373.7079 Da, a 3.07 Da deficiency. However, other related compounds synthesized at the same time from the same reagents were measured with

139

Figure 4.4 LC-MS3 quantitation adds selectivity and sensitivity. a) Chromatogram for full MS scan of 100 pM Ang 1-6-Ser(O-Glc)-NH2. The peak for the target compound at 4.19 minutes is obscured. b) The corresponding MS spectrum. The known peaks at

525 and 1050 are not visible. c) LC-MS2 chromatogram, selecting for 525.80 Da. The target peak is visible but not the largest in the chromatogram. d) The corresponding

MS2 spectrum. The 444.999 peak is the loss of glucose, but the other peaks cannot be attributed to the target compound. e) The LC-MS3 chromatogram, selecting for 525.80 and subsequently 444.84 Da. The peak at 4.19 minutes is the main constituent. f) The peaks in the corresponding spectrum can be attributed solely to Ang 1-6-Ser(O-Glc)-

NH2.

140 Table 4.2 Sensitivity and range of peptide quantification. Limits of detection and linear dynamic range (in pM and amol, using a 5 µL injection) for MS3 quantitation.

Asterisk denotes highest concentration tested, rather than known limit. LOD and LDR for the PACAP compounds have not been determined at this time.

Compound LOD (pM) LOD (amol) LDR (pM) r2

Ang 1-6-Ser-NH2 35 175 35 - 1127* 0.995

Ang 1-6-Ser(O-Glc)-NH2 2 10 9.5 - 950* 0.996

SAM-995 11 55 50 - 5000 0.995 10,000 - 500,000* 0.999

MMP-2200 1 5 50 - 9910 0.991 10,000 – 500,000* 0.992

141 the correct mass, so we designed an experiment to determine to location of the mass deficiency on the molecule, with moderate success. Beginning with Full MS scans, we selected and fragmented the peaks produced by this compound through several levels of fragmentation, up to MS5. This allowed for a sequencing-type approach in which the exact mass of peptide fragments could be measured and the deficiency could be linked to segments of the peptide. For measured m/z values, see Appendix Figure C.13. After calculating the exact monoisotopic masses for each fragment, the fragmentation pathways were mapped (Figure 4.5). While several of the fragments showed the same mass deficiency, there were identifiable internal fragments with exact masses within the 2 ppm error of the instrument. Interestingly, CID fragmentation of peptides tends to yield b and y ions, few of which were seen with the PACAP derivatives. The identification of almost solely internal fragments may indicate that PACAP cyclizes, preventing traditional fragmentation.

Tracking the deficiency and comparing the internal fragments allowed us to identify three residues as possible sources of the mass deficiency, as shown in red in the sequence at the bottom of Figure 4.5. As a deficiency of 3.08 Da has no easily identifiable source, and peptides synthesized in the same time frame with the same approach show no such error, the first step in proceeding with this compound is to synthesize a new batch of peptide, to see if the problem recurs. It is possible that the deficiency is a combination of mutations, further complicating identification.

However, disregarding this mass error, we did see that the glycosylated PACAP derivative shows the same behavior as Ang 1-6-Ser(O-Glc)-NH2 and MMP-2200, in that the ion throughput during the first fragmentation step with a glycosylated peptide is higher. For spectra and CID breakdowns of compounds not shown in this chapter, see Appendix C.

In short, for three classes of peptides we were able to show that glycosylation improves

142

Figure 4.5 Investigation of glycosylated PACAP’s mass deficiency. By tracking the

17 fragmentation pathways of [Leu ] PACAP 1-27-Ser(O-Glc)-NH2, it was possible to narrow down the possible locations of the mass deficiency. The exact mass for the whole molecule is 3.07 Da deficient, as are several of the fragment masses, as labeled in red. However, there were several internal fragments whose exact mass was a match

(labeled in green), leading to the conclusion that the mass deficiency is likely affiliated with one of the three residues shown in red in the sequence on the bottom.

143 fragmentation behavior by increasing the ion throughput in the first stage of fragmentation, leading to improved limits of detection.

4.3.2 Glycosylated compounds have improved in vitro stability The in vitro stability of the six compounds was assessed in rat serum at biological temperature (37 °C), to model in vivo degradation due to the species present in serum as closely as possible.

Compounds of interest were spiked into rat serum at t = 0. Aliquots were removed from the solution at set time points, spiked with 1 µL glacial acetic acid to quench peptidase activity, and 1 µL internal standard (20 µM dAdLE). Samples were desalted using a

ZipTip® procedure optimized for the compounds of interest. Samples were diluted to 100

µL in 50:50 ACN:H2O:0.1% FA and injected directly onto the instrument using a 20-µL injection loop at a flow rate of 8 µL/min, resulting in a 2.5-minute bolus. Area under the curve (AUC) for the characterized MS2 fragments (Table 4.1) was used for quantification.

Statistical analysis was performed in GraphPad Prism 5 (La Jolla, CA). Trials with exponential curves with r2 < 0.7 were not included, and errors are reported as s.e.m.

Figure 4.6 shows the degradation results for the Angiotensin compounds. The two

Angiotensin derivatives are shown with the best fit exponential decay. The t50 of Ang 1-6-

Ser-NH2 was determined to be 10.8 ± 3 min (mean ± s.e.m., n = 5 trials), and the t50 for the glycosylated derivative was 39.5 ± 12 min (mean ± s.e.m., n = 5 trials). These values are statistically different (Student’s t-test, p < 0.05).

Figure 4.7a shows the degradation trials for SAM-995 and MMP-2200 over two hours, and the corresponding best-fit exponential decays. However, only one trial for SAM-995 degradation was successful (meeting the requirement that an exponential decay model fit

2 have an r of at least 0.7). The t50 based on the best-fit exponential curve was calculated to be 0.67 min. The t50 for MMP-2200 was calculated to be 8.33 ± 7.60 minutes (mean ± s.e.m., n = 3 trials). This is a large error, so more trials are necessary. Additionally,

144

Figure 4.6 In vitro degradation of Ang peptides in rSer at 37 °C. a) Workflow for degradation experiments. b) Angiotensin 1-6-Ser-NH2 and Ang 1-6-Ser-(OGlc)-NH2, n

= 5. Shown are best fit exponential decay for each peptide derivative. The t50 of Ang

1-6-Ser-NH2 was determined to be 10.8 ± 3 min (mean ± s.e.m.), and the t50 for the glycosylated derivative was 39.5 ± 12 min (mean ± s.e.m.).

145

Figure 4.7 In vitro degradation of SAM-995 and MMP-2200 in rSer at 37 °C. a) Over two hours. n = 1 and 3, respectively. Shown are best fit exponential decay for each peptide derivative. The best-fit t50 of SAM-995 was calculated to be 0.67 min, and the t50 for MMP-2200 was calculated to be 8.33 ± 7.60 (mean ± s.e.m.). Most notably, the

MMP-2200 degradation plateaus around 40%, compared to about 20% for SAM-995, which may translate to a higher persistence in blood in vivo.

146 previous studies indicated that MMP-2200 should have a much longer lifetime (in one study, MMP-2200 was detected in vivo more than 3 hours after dosing).28 However, these preliminary results are promising in that glycosylation appears to improve both the in vitro stability (larger t50) and the persistence, as determined by the higher plateau of the degradation data.

Figure 4.8a shows the degradation for the two PACAP compounds (n = 1). Again, the glycosylated derivative appears to have a slightly longer lifetime, though only one trial was performed in rat serum at 37 °C. Only one trial was performed, as LC method development was ongoing, and there was evidence that degradation was already occurring in the standard solutions sans protease. To investigate this effect further, we performed degradation studies under several conditions: 1) in 0.1% TFA at room temperature, 2) in aCSF at room temperature, 3) in aCSF at 37 °C, and finally, 4) the study shown in Figure 4.7a in rSer at 37 °C. The results of these are shown in Figure 4.8b and

17 17 c for [Leu ] PACAP 1-27-NH2 and [Leu ] PACAP 1-27-Ser(O-Glc)-NH2, respectively. We saw evidence that even in 0.1% TFA at room temperature, there was degradation over the course of one hour. In aCSF at biological temperature, only 30-40% of each compound remained in solution. The cause of this instability of the PACAP compounds, even when peptidases are not present, is uncertain, as are the degradation products.

Identification of the degradation products was attempted via several approaches, but was unsuccessful. Both PACAP compounds were analyzed immediately after thawing and after 60 minutes of incubation at 37 °C, using nano LC-MS2 data-dependent scanning and subjected to a SEQUEST search. However, even the non-glycosylated peptide was not identified by the sequencing software, despite the chromatographic peak being visible.

This is likely because MS2 fragmentation of PACAP does not yield the series of b and y ions that are common in CID fragmentation, and which are used to sequence peptides. It

147

Figure 4.8 In vitro degradation of PACAP derivatives. a) in rSer at 37 °C, n = 1.

Shown are best fit exponential decay for each peptide derivative. b) Degradation of

17 17 [Leu ] PACAP 1-27-NH2 under different conditions. c) Degradation of [Leu ] PACAP

1-27-Ser(O-Glc)-NH2 under different conditions.

148 is possible that PACAP is cyclizing, stabilizing against backbone fragmentation and preventing software identification. Due to the inability to identify the cause of PACAP instability or degradation products, PACAP derivatives did not move forward into the in vivo trials.

We have been able to show that glycosylation improves the in vitro stability of peptides.

The improved in vitro stability with glycosylation shows great promise for the development of new peptide-based drugs with longer in vivo lifetime. The ability to design effective peptide-based drugs with lifetimes sufficient for treatment would mean better drugs with higher specificity for target receptors and fewer side effects than existing drugs.

4.3.3 “Shotgun microdialysis” for direct comparison of in vivo lifetime and BBB penetration of peptide derivatives To assess the effect of glycosylation on in vivo stability and BBB penetration, we developed a method that involves the co-injection of compounds and the quantitation of their concentration in the blood and in the CSF over time. A common challenge in animal research is inter-animal variability. Different animals will often have slightly different responses to the same treatment, making it challenging to compare the behavioral or chemical responses of multiple animals to a single drug, let along multiple drugs. We have developed a method, coined “shotgun microdialysis”, in which we inject a single animal with all compounds of interest and monitor the CSF concentrations using microdialysis. This allows us to directly compare the compounds within a single animal, and account for the variability between animals.

In microdialysis, a probe is surgically implanted into the target region (in this case the striatum). The probe has a semi-permeable membrane with a molecular weight cut-off.

Perfusate is flowed through the probe, and species in the surrounding matrix can diffuse through the membrane as a function of the concentration gradient. Microdialysis allows

149 for sampling from a region without altering the volume of that region, which is of particular relevance when studying the CSF.

The work flow for in vivo experiments is shown in Figure 4.9. Animals were injected intravenously SAM-995 and MMP-2200 (n = 4) at 10 mg/kg via tail vein injection. Blood draws were taken at t = -10, 1, 5, 10, 20, 30, 40, 50, 60, 70, 80, and 90 minutes (where t

= 0 corresponds to injection). Microdialysis fractions were time-locked with blood draws at ten-minute increments such that blood draws correlate with the median time of the microdialysis fraction. Blood draws were centrifuged for 2 minutes in a tabletop mini- centrifuge to separate the serum and the red blood cells. Serum was pulled off and diluted

100x into a solution of 50:50 aCSF:“preservation solution” for matrix matching with the dialysate and standards, and immediately frozen on dry ice. Dialysate samples were collected for 10 minutes at a flow rate of 0.5 µL/min, with the microdialysis tee coupling a line containing preservation solution to the dialysate line immediately behind the probe, leading to a solution volume of 10 µL after 10 minutes. The dialysate samples were immediately frozen on dry ice.

All samples were desalted using µ-C18 Zip Tips® (EMD Millipore) according to the aforementioned protocol, yielding a final solution comprised of 10 µL 60:40

ACN:H2O:0.1% TFA (v/v). For reverse-phase liquid chromatography, samples should start in aqueous solution, thus samples were vacuum-centrifuged until 1 µL of solution remained, and reconstituted in 5 µL H2O with 0.1% TFA (v/v). Samples (5 µL) were injected onto the column and separated using a gradient elution. Target compounds were quantified by summing the height of known fragment peaks in the LC-MS3 spectra.

The serum concentration and CSF concentration estimates were calculated using a calibration curve, accounting for the dilution factors and the probe recoveries for each

150

Figure 4.9 Workflow for collection and analysis of in vivo samples. Animals are surgically implanted with a vascular access port (VAP) for access to the carotid artery, and a microdialysis probe in the striatum. Animals are injected with a cocktail of 10 mg/kg of each compound in saline via the tail vein. Time-locked blood and dialysate samples are collected, normalized to the time of injection as t = 0. aCSF, isotonic with the brain environment, is used as perfusate for microdialysis. A preservation solution comprised of 10% acetic acid, 140 pM dAdLE (internal standard), and 2% acetonitrile in water (v/v) is coupled to the dialysate immediately using a microdialysis tee.

Dialysate is stored on dry ice until analysis. Serum samples are collected by drawing

100 µL of whole blood through the VAP and centrifuging to separate out the serum.

Serum samples are spiked with acetic acid and internal standard before freezing. Both types of samples go through the same preparation steps immediately prior to analysis, including ZipTip© cleanup. Samples are then analyzed with nano-LC-MS3 and identified fragment peaks are used for quantification.

151 experiment. Standards were matrix-matched to samples and underwent the same sample preparation steps.

In vivo experiments with SAM-995 and MMP-2200 are ongoing. Figure 4.10 shows the in vivo data collected thus far (n = 4 animals). The preliminary results are quite promising.

As shown in Figure 4.10a&b, the normalized signal of MMP-2200 is notably higher than that of SAM-995 throughout the 60-minute collection in both serum (a) and CSF (b).

Additionally, in both cases, at 60-minutes post-injection, the signal for MMP-2200 was still considerably higher than the -10 min baseline sample, indicating that MMP-2200 does have improved stability in the blood and increased persistence in the brain, compared to

SAM-995. By calculating the area under the curve for these two plots, we can directly compare the differences in the two compounds (Figure 4.10c&d). In serum, the AUC for

MMP-2200 is significantly higher than SAM-995 (paired t-test, p < 0.05). In CSF, the AUC for MMP-2200 is visibly higher but with large error. The difference is not statistically significant (paired t-test, p > 0.05). However, these results are highly promising, indicating that MMP-2200 is more stable than SAM-995 in the blood and has improved BBB penetration.

We saw that glycosylation increases in vivo lifetime, and that it also appears to improve blood-brain barrier penetration. This is a highly promising result, as it indicates a method for the modification of peptide-based drugs that will allow for the design of more effective drugs due to increased in vivo stability and bioavailability.

4.4 Conclusions

We have shown here that glycosylation of peptides alters their behavior in multiple ways, all of which have beneficial effects. First, most peptides fragment along the peptide backbone, resulting in several identifiable fragments, typically b and y ions in CID fragmentation. However, this splits the gas-phase concentration of the target compound

152

Figure 4.10 In vivo quantitation of SAM-995 and MMP-2200. In a) and c) serum, and b and d) CSF. Four animals were co-injected with 10 mg/kg SAM-995 and MMP-2200.

Compounds were quantified in both serum and dialysate. The glycosylated derivative appears to reach higher concentrations and persist for longer in both the blood and the

CSF. Calculation of area under the curve (AUC) for c) serum and d) CSF show a visible increase, but only in serum is the AUC statistically higher for the lactoside (paired t-test, p < 0.05).

153 amongst several peaks. Particularly when using MS3 quantitation, this loss of intensity in the first fragmentation step can lead to higher limits of detection. Glycosylated peptides have the benefit of yielding a single fragment in MS2, removal of the sugar moiety with high efficiency. This directly affects limits of detection- we report here a 10-100x increase in LOD with glycosylation.

Additionally, we have shown that glycosylation has beneficial effects in vivo. In three classes of compounds, there was evidence that glycosylation improved in vitro lifetime. In the two compounds tested in vivo, addition of a sugar moiety improved lifetime in vivo, and increased BBB penetration and persistence in the CSF. In the future, we will address further means of improving the stability of compounds, and will optimize our methods and compounds for in vivo trials with larger peptides such as PACAP derivatives. Our shotgun microdialysis methodology will be expanded with the goal of injecting multiple drugs across several classes of compounds and spanning a larger molecular weight range.

Additionally, while our current method is sufficient to compare the BBB penetration of compounds within an animal, there are studies that recommend adjustments to our microdialysis method for more representative BBB behavior.

Traditionally, pharmaceutical companies have leaned away from peptide-based drugs as new treatment options, due to their poor survival in vivo and poor penetration properties.

We have shown that glycosylation of peptides can combat both these issues. Using this information, it will be possible to develop new pharmaceuticals with the safety and selectivity of peptides, but without the drawbacks.

4.5 Author Contributions

CLK developed mass spectrometric methods and performed all sample preparation and analysis. MJB performed all surgical procedures and assisted in sample collection. CL

154 assisted in sample collection and preparation. ARM assisted with degradation experiments. Peptides were synthesized by EMJ, CS, and LS.

4.6 Acknowledgements

Special thanks to Linda Breci and George Tsaprailis of the Arizona Proteomics

Consortium at the University of Arizona, for their advice and use of the Thermo LTQ Velos-

Orbitrap instrument, which was funded through NIH/NCRR 1 S10 RR028868-01.

155 Chapter 5

Identification of optimal structural modifications for the improvement of peptide-based drug delivery properties: an Angiotensin 1-7 study of in vivo

stability and BBB penetration

Abstract One peptide of particular interest in the development of CNS-active pharmaceuticals is

Angiotensin 1-7, an endogenous peptide which has been shown to have neuroprotective effects in cases of ischemic stroke. We have synthesized a series of Ang 1-7 derivatives for the purpose of investigating how different structural modifications improve the stability and penetrative properties of a compound. Here, we investigate the effects of these structural modifications using the in vitro and in vivo methods presented in Chapter 4. We successfully screened six Angiotensin 1-7 derivatives for in vitro and in vivo stability and

BBB penetration, and have identified a series of modifications that lead to improved drug delivery and stability for peptide-based drugs in the future.

156 5.1 Introduction

Angiotensin 1-7 (Ang 1-7) is a seven-residue peptide derived from Angiotensinogen as a part of the Renin-Angiotensin system (RAS).1 RAS plays a role in the moderation of several systems, including the cardiovascular and renal systems.1 There is also evidence that RAS may link diabetes and insulin resistance, and have effects on emotional and behavioral responses.2–5 Many angiotensin compounds have been isolated to date, and all are derived from Angiotensinogen. Many of the angiotensins cause vasoconstriction

1 via activation of the AT1 and AT2 GPCRs. Our understanding of the RAS system is still expanding, as new components of the system are continuously being discovered. The cascade begins with Angiotensinogen being clipped by Renin to produce Ang I, which is the first ten residues of Angiotensinogen (DRVYIHPFHL). Angiotensin converting enzyme

(ACE) and ACE2 further break down Ang I into Ang II (DRVYIHPF), Ang 1-9

(DRVYIHPFH), and Ang 1-7 (DRVYIHP).1,6 See Santos for a more extensive representation of the current understanding of the angiotensin compounds and their biological synthetic pathway.1

Ang 1-7 is produced by several routes, including from Ang II by ACE2, from Ang I by endopeptidases, and from Ang 1-9 by ACE. Ang 1-7 has been shown to decrease blood pressure, increase baroreflex, and increase cardiac output.1 This may be because Ang

4,7,8 1-7 is believed to act on the Mas receptor rather than the AT1 and AT2 GPCRs. Ang

1-7 and other Mas receptor activators have also been shown to provide cardioprotective effects, protective effects in hepatic fibrosis, and neuroprotective effects in cases of ischemic stroke.1,8–10

Due to these effects, Ang 1-7 shows potential as a peptide-based therapeutic.11

Endogenous peptide-based pharmaceuticals show great promise due to their high specificity and biocompatibility. Peptide-based drugs can typically be broken down into

157 non-hazardous degradation products, as they are comprised of naturally-occurring amino acids. Unfortunately, Ang 1-7 suffers from the same downfalls as many other peptide- based drugs, namely rapid degradation in vivo. The natural degradation process means that compounds based on endogenous species tend to be rapidly degraded in the body, often too quickly to be of use as a therapeutic treatment.12

Additionally, drugs meant to target the central nervous system must be able to penetrate the blood-brain barrier (BBB), which few peptides can do.13,14 The blood brain barrier is comprised of the cerebral capillary endothelial cells that form tight junctions.15 The BBB serves to protect the CNS from harmful compounds like neurotoxins, bacteria, and some macromolecules, while simultaneously allowing necessary compounds like sugars, salts, and small-molecule precursors to neurotransmitters to penetrate.15 Few small molecule drugs can cross the BBB, and even fewer peptide drugs.16

In order to address these problems, researchers have been working toward methods for improving the stability, bioavailability, and penetration properties of peptide-based drugs via approaches such as structural modification and conjugation.17 Two post-translational modifications that occur naturally are amidation of the carboxyl terminus, and glycosylation.17–19 The C-terminus of a peptide is often targeted by peptidases, and by replacing or masking it, peptidases can no longer target this site. Glycosylation can improve in vivo lifetime by altering the conformation of the peptide or hindering peptidases, and there is evidence that it may also improve BBB penetration.17,20,21 Other approaches to improving peptide lifetime in vivo and bioavailability include cyclization via thioether or disulfide bridges, halogenation, enzyme inhibition, and replacement of native residues with D-amino acids.17,22–25

Our collaborators have developed a series of Ang 1-7 derivatives in order to directly compare the effect of certain modifications on peptide stability and in vivo lifetime (Figure

158 5.1). Compound 1 is Native Angiotensin 1-7 (DRVYIHP), for direct comparison with modified compounds. Compound 2 is modified by amidation of the C-terminus to hinder peptidase activity, as previously discussed. There is evidence to support that glucose improves in vivo lifetime and BBB penetration,17,19,24 thus several sugar-modified compounds were synthesized using a serine attachment site. Compound 3 has the seventh residue (proline) replaced with a serine as an attachment site for a sugar, but does not have a sugar moiety and thus serves as a control for the effect of the serine replacement on peptide stability. The final three compounds (Compounds 4-6) are addition of glucose, cellobiose, and lactose onto the serine, respectively. To explore the effect of a disaccharide, lactose is being investigated (Gal-β(1-4)-Glc), as there is evidence to suggest that this modification may lead to increased stability without a loss of receptor affinity.26 Cellobiose (Glc-β(1-4)-Glc) is also being investigated. Cellobiose is a disaccharide produced by plants but not animals, and is hypothesized to last longer in vivo due to its non-native structure.27–29 Compounds 1 and 2 are only 1 Da different in mass, and Compounds 5 and 6 are isobaric. For ease of differentiation in the instrument,

Compounds 2 and 6 were synthesized using deuterated valine (Val-d8).

Due to the inherent cost of in vivo animal experiments, it is necessary to develop inexpensive screening methods for peptide stability prior to in vivo experiments. As presented in Chapter 4, we have developed a separation-free in vitro technique for quantifying the lifetime of our modified compounds in rat serum. By spiking the compounds into rat serum at biological temperature, we can mimic the degradation of compounds as a function of species in the serum, such as endogenous peptidases. This method is much more representative of biological degradation than techniques that rely only on one or two peptidases, such as trypsin. With this method, we can track the

159 .

Figure 5.1 Ang 1-7 modifications for improved in vivo lifetime and blood-brain barrier penetration. Compound 1: Native Ang 1-7 (DRVYIHP). Compound 2 is the same sequence but with a C-terminus amidation, which hinders peptidase activity by masking the binding site (DRVYIHP-NH2). Compound 3 replaces the seventh residue

(proline) with a serine, and is also amidated (DRVYIHS-NH2). Compounds 4-6 have glucose, cellobiose, and lactose sugar moieties attached to the serine (respectively), and are all amidated on the C-terminus. Sugar modification is believed to increase both in vivo lifetime and BBB permeability. Compounds 1 and 2 are only 1 Da different in mass, and Compounds 5 and 6 are isobaric. For ease of differentiation in the instrument, Compounds 2 and 6 were synthesized using Val-d8.

160 concentration of the compounds over time with a targeted detection scheme that allows us to determine the rate of degradation in vitro using direct-injection MS2 quantification.

As described in Chapter 4, we have developed an in vivo method that allows for simultaneous stability and penetration data to be collected from both the blood and the cerebrospinal fluid (CSF) of an animal. This method allows us to quantify both the stability and BBB penetration of our compounds in vivo. Our method involves time-locked blood draws and microdialysis for the determination of BBB permeability of each of the compounds and lifetime in both the blood and the CSF using nano-flow LC-MS3. With our targeted detection method, all six Ang 1-7 compounds can be co-injected (coined “shotgun microdialysis”) for direct comparison, which controls for inter-animal variability.

Using both the in vitro and in vivo methods developed and presented in Chapter 4, we can determine which structural modifications of peptides lead to the best in vivo characteristics for the improvement of peptide-based drug design.

5.2 Materials and Methods

5.2.1 Chemicals and reagents All chemicals were purchased through Sigma Aldrich (St.

Louis, MO, USA), unless otherwise stated. Internal standard (d-Alanine d-Leucine enkephalin, dAdLE) was purchased through American Peptide Company, Inc. (Sunnyvale,

CA).

5.2.2 Animals Male Sprague-Dawley rats (275-325 grams) were used for in vivo experiments. Animals were purchased from Harlan Laboratories (Indianapolis, IN) and were housed in a temperature and humidity controlled room with 12 h reversed light/dark cycles with food and water available ad libitum. All animals were treated as approved by the Institutional Animal Care and Use Committee, University of Arizona and in accordance

161 with the NIH Guidelines for the Care and Use of Laboratory Animals. Both the number of animals used and their suffering were minimized.

5.2.3 Peptide synthesis Peptides and glycosylated peptides were synthesized by the Polt

Lab (University of Arizona) from commercial material via solid-phase synthesis, as previously reported.19,27,30

5.2.4 Mass spectrometric identification and quantification of compounds In vitro degradation experiments were performed on an Applied Biosystems QStar Elite mass spectrometer, using direct injection and MS2 fragments for quantification. The area under the curve for the main MS2 fragments was used for quantification over the course of the

2.5-minute bolus. Values were normalized to the 0-minute fraction as 100% signal. LC-

MS3 methods were developed for in vivo samples using purified peptide standards and a

Proxeon nano-LC coupled to a Thermo LTQ-Orbitrap instrument using Advion TriVersa

NanoMate chip-based nanoelectrospray technology. The compounds were fragmented twice, yielding an MS3 spectrum, and the identifiable fragments were used for quantification. Table 5.1 shows the fragment masses and identifications used for quantification with both methods.

5.2.5 In vitro stability of Angiotensin derivatives Degradation studies were performed in rat serum (rSer) in a 37 °C water bath. Peptides were spiked in for a final concentration of ~100 µM and time points were determined by apparent degradation rate after n = 2 trials. At each time point, 10 µL rSer + peptide was removed and spiked with 1 µL glacial acetic acid to quench peptidase activity, 1 µL of 100 µM d-alanine d-leucine enkephalin

(dAdLE) as an internal standard, and 1 µL 25 mM octyl sulfonate (8S) as a pairing agent.

Samples were desalted using µ-C18 Zip Tips® (EMD Millipore) according to the following protocol, on a 10 µL scale: three conditioning steps with acetonitrile (ACN), followed by three conditioning steps with 0.1% trifluroacetic acid (TFA) in nanopure water (v/v). Ang

162 Table 5.1 Fragment masses and identification for quantification of Ang 1-7 derivatives. In vitro stability studies quantified MS2 fragments, whereas quantification in BBB penetration studies was performed using LC-MS3 quantification for an additional level of selectivity.

Compound Parent Main MS2 fragment MS3 fragments

+2 +2 +2 1 450.24 [M+2H] 392.71 [b6+2H] 370.20 [a6-NH3+2H] +2 378.71 [a6+2H] + 647.35 [b5+H]

+2 +2 +2 2 453.84 [M+2H] 396.26 [b6+2H] 382.74 [a6+2H] + 655.40 [b5+H] + 677.43 [b5+Na]

+2 +2 +2 3 444.74 [M+2H] 392.71 [b6+2H] 378.22 [a6+2H] + 647.37 [b5+H]

+2 +2 +2 4 525.77 [M+2H] 444.74 [-Glc+2H] 378.71 [a6+2H] +2 392.71 [b6+2H] +2 435.73 [-Glc-H2O+2H] + 647.35 [b5+H]

+2 +2 +2 5 606.79 [M+2H] 444.74 [-Cel+2H] 378.71 [a6+2H] +2 392.71 [b6+2H] +2 435.73 [-Cel-H2O+2H] + 647.35 [b5+H]

+2 +2 +2 6 610.81 [M+2H] 448.76 [-Lac+2H] 382.74 [a6+2H] +2 396.73 [b6+2H] +2 439.76 [-Lac-H2O+2H] + 655.40 [b5+H]

163 1-7 derivatives were found to bind best with 15 binding steps. A single wash with 0.1%

TFA in water desalted the column, and the procedure was completed with 10 elution steps in 60:40 ACN:H2O:0.1% TFA (v/v). Samples were then brought to 100 µL in 50:50

ACN:H2O with 0.1% formic acid for electrospray compatibility. Samples were analyzed using tandem MS on an AB Sciex QStar Elite mass spectrometer with a 20 µL injection loop. Samples were run in duplicate, averaged for each time point, and normalized to the

0-minute time point as 100% signal. Analyses were performed in GraphPad Prism 5 (La

Jolla, CA), and degradation trials with exponential fits with r2 < 0.7 were discarded.

Degradation studies were also run in artificial cerebrospinal fluid (aCSF) at 37 °C in order to validate that degradation was due to peptidases in serum and was not method-or preparation-dependent.

5.2.6 Carotid artery catheterization Carotid artery catheterizations were performed by staff at Envigo Laboratories. Animals were anesthetized with 1.5-2% isoflurane mixed in medical grade oxygen (1.5 L/minute) and two incisions were made: (1) a 0.5 cm midline incision between the scapulae; (2) a 2.0 cm incision to the right of the midline of the ventral neck. The right carotid artery was then isolated from surrounding tissues. Using 4-0 silk suture, two loose ties were placed at the caudal and cranial ends of the vessel. The cranial suture was then tied off and a bulldog clamp was placed above the caudal tie to temporarily occlude blood flow. A small incision was made in the vessel and a piece of

164 and its incision closed with 4-0 silk sutures. Animals were monitored during recovery for

5.2.7 Microdialysis surgery Upon arrival at our facility, animals were allowed to acclimate for 5-7 days. Animals were anesthetized with 1.5-2% isoflurane mixed in medical grade oxygen (1.5 L/minute) and positioned inside a stereotaxic frame (David Kopf Instruments,

Tujunga, CA). A midsaggital incision was made along the skull and a CMA 11 guide cannula (CMA Microdialysis, Kista, Sweden) was implanted in the striatum at the following coordinates AP: +1.0 mm and ML: +3.2 mm relative to bregma and DV -3.4 mm from the surface of the brain. The guide was then replaced with a CMA 11 probe containing a 4.5 mm cuprophane (Cupr) membrane with a molecular weight cut-off of 6 kDa. After the probe was positioned, the animals remained under anesthesia for the remainder of the experiment.

5.2.8 Microdialysis and blood draws For microdialysis, a dual syringe pump (PHD 2000,

Harvard Apparatus, Holliston, MA) was set to a flow rate of 0.5 L/minute. Samples were collected and preserved in a solution containing 10% v/v acetic acid, 2% v/v acetonitrile, and 140 pM d-alanine d-leucine enkephalin (dAdLE). dAdLE was used as an internal standard. Baseline blood (~100 L) and dialysate samples were collected. Subsequent time-locked samples (with blood being drawn at the median of the dialysate time range) were then collected every 10 minutes for 90 minutes, following a tail vein injection at t = 0 min. In addition, blood samples were also collected at both 1- and 5-minutes post-injection.

CSF samples were immediately frozen on dry ice. Blood samples were spun down to separate serum from the red blood cells. Serum was drawn off, and 10 L of serum was

165 spiked with glacial acetic acid to quench peptidase activity and internal standard to a final concentration of 5% HAc and 140 pM IS before being frozen on dry ice.

5.3 Results and Discussion

5.3.1 Mass spectrometric identification and quantification of compounds In vitro studies were performed using an Applied Biosystems QStar Elite mass spectrometer, using quantitation of MS2 fragments. For the in vivo studies, due to the small sample size

(10 µL) and expected low concentrations, we used a Proxeon nano-LC coupled to a

Thermo LTQ-Orbitrap instrument, and quantified MS3 fragments for increased specificity.

The high number of small peptides present in biological solutions such as blood, paired with the minimal clean-up applied with our method, means that the matrix is highly complex and high specificity is required to differentiate our small peptides from the hundreds that are present. Both retention time and MS3 fragment identification were necessary assure that we were quantifying the target molecules. The characteristic MS2 and MS3 fragments for each compound, and their identities, are listed in Table 5.1.

Peptide fragmentation pathways are fairly well-characterized for small peptides, and collison-induced dissociation (CID) fragmentation typically yields b and y ions. Figure 5.2 shows the annotated MS1-3 spectra for Angiotensin 1-7 (Compound 1). Most of the fragment ions identified are b ions, and fragmentation starts from the C-terminus. In full scan MS mode (Figure 5.2a), the main peak is the [M+2H]+2 state of Ang 1-7. In MS2, which means selecting the [M+2H]+2 parent peak and applying energy to cause fragmentation, we see five identifiable fragment peaks. This is common for the fragmentation of peptides. As stated earlier, it is necessary to use MS3 fragmentation for high selectivity in samples collected during in vivo experiments. Thus panel c of Figure

5.3 shows the MS3 spectrum, which is collected by selecting the [M+2H]+2 parent peak and fragmenting it, followed by selection of the most abundant fragment in the MS2

166

Figure 5.2 Annotated MS1-3 spectra for Angiotensin 1-7. a) Full MS spectrum. The most abundant peak is the parent [M+2H]+2. b) MS2 spectrum, collected by fragmenting

+2 3 +2 the [M+2H] ions. c) MS spectrum, collected by isolating and fragmenting the [b6+2H] ions in MS2.

167 +2 3 spectrum (392.71, [b6+2H] ) and fragmenting again. For Ang 1-7, the MS spectrum has two identifiable peaks, which are used for quantitation.

Fragmentation of glycosylated peptides has been studied far less, but we have seen some very interesting and promising behavior. Figure 5.3 shows the annotated MS1-3 spectra for Ang 1-6-Ser(O-Glc)-NH2 (Compound 4), the first of the glycosylated peptides we studied. In the full MS, the [M+2H]+2 is again the most abundant ion. However, the MS2 spectrum is where things get interesting. Fragmentation of the [M+2H]+2 parent peak of this glycosylated peptide gives a single fragment in the MS2, which is loss of the sugar moiety with high efficiency. Fragmentation of the most abundant peak in the MS2 spectrum for Ang 1-6-Ser(O-Glc)-NH2 provides the same spectrum as the initial fragmentation of Ang 1-7. The benefits of this are multiple: 1) the high efficiency of the first transition means that the overall ion throughput through the multiple stages of fragmentation is higher, which directly correlates to lower limits of detection, and 2) there are more identifiable peaks in the MS3 spectrum that can be used for quantitation, making it a more reliable method.

ZipTip® cleanup is a common practice in proteomics analysis, as it is highly useful for desalting biological samples. However, in developing methods, we determined that C18

ZTs have a poor recovery for our targets of interest, due to their hydrophilicity. In order to improve the recovery but maintain the benefit of desalting, we added the pairing agent octyl sulfonate (8S). Octyl sulfonate increased the recovery of the angiotensin compounds during the ZipTip® desalting process (Table 5.2). Octyl sulfonate ionically pairs with the highly protonated Ang compounds and its effects last through ZT and column separations, but is removed during ionization, so the masses of our compounds are not altered. An additional benefit, unanticipated but logical, is that 8S also considerably improves the chromatography, particularly of Ang 1-7 (Figure 5.4). In 0.1% TFA, addition of 8S

168

1-3 Figure 5.3 Annotated MS spectra for Ang 1-6-Ser(O-Glc)-NH2. a) Full MS spectrum. The most abundant peak is the parent [M+2H]+2. b) MS2 spectrum, collected by fragmenting the [M+2H]+2 ions. Note the high efficiency of fragmentation; only one major peak appears in the MS2 spectrum. c) MS3 spectrum, collected by isolating and fragmenting the [M-Glc+2H]+2 ions in MS2.

169 Table 5.2 Addition of octyl sulfonate pairing agent improves recovery during

ZipTip® purification. Due to the low recovery seen for our Ang peptides during ZT purification, we opted to use the pairing agent octyl sulfonate (8S). Addition of 8S improved recovery of all compounds in each of the three solutions of interest. Recovery improves most in H2O, but is not as improved in rat serum, likely due to the viscosity of the serum rather than binding effects.

Percent Recovery, Percent Recovery, Without 8S With 8S

Compound H2O aCSF rSer H2O aCSF rSer

1. Ang 1-7 15.4 24.1 11.1 100 96.8 11.8

2. Ang 1-7-NH2 11.8 12.4 3.8 72.5 75.1 9.1

4. Ang 1-6-Ser(O-Glc)-NH2 2.6 5.0 4.8 81.1 55.0 7.8

5. Ang 1-6-Ser(O-Cb)-NH2 5.0 3.6 1.8 56.6 71.1 6.6

6. Ang 1-6-Ser(O-Lac)-NH2 2.7 5.6 8.8 76.9 54.8 8.7

170

Figure 5.4 Improved chromatography with addition of octyl sulfonate pairing agent. Addition of octyl sulfonate during the ZT purification step had the added benefit of improving the chromatography as well. a) LC-MS3 chromatogram for Native Ang 1-7 without octyl sulfonate. b) LC-MS3 chromatogram for Native Ang 1-7 with octyl sulfonate. The pairing agent increases the hydrophobicity of Ang 1-7, yielding a later retention time and more efficient separation due to more time spent in the stationary phase.

171 increases the retention time of Ang 1-7 and considerably improves the efficiency. Using the equation,

푡 푁 = 5.54( 푟 )2 푤0.5

where N is the number of theoretical plates, tr is the retention time, and w0.5 is the width at half the maximum peak height, N increases from around 1000 plates to just under 200,000 theoretical plates.

Figure 5.5 shows the selected ion chromatograms and MS3 spectra for each of the

Angiotensin derivatives at 1 nM, and also the internal standard dAdLE (bottom, 70 pM).

Colors match Figure 5.1 and columns are stacked in order (1-6). Both retention time and the spectra are used for identification, and fragment peak heights are summed for quantitation. Note that several of the chromatographic peaks are tailing, which is due to the high concentration. Additionally, retention time for Ang 1-6-Ser(O-Glc)-NH2 is shifted compared to Figure 5.4. This is due to matrix effects. The sample shown in Figure 5.5 is matrix-matched with our biological samples, which causes some band-broadening of the peaks and altered retention time compared to pure standards in 0.1% TFA. However, building the calibration curves with standards that are matrix-matched and have gone through the same ZT cleanup procedure circumvents identification and concentration calculation errors due to matrix effects and variability in ZT recovery.

Table 5.3 shows the limits of detection (LODs) and linear dynamic ranges (LDRs) for each of the compounds. LODs span 1-100 pM in a 5 µL injection. The best LOD was for

Compound 4 at 1.9 pM, and the worst was for Compound 1, at 111 pM. In several cases, the full extent of the LDR was not examined, as the calibration curves remained linear beyond the expected concentration range of samples, and there was concern about overloading the column. As noted earlier, the LODs for the glycosylated peptides are 1-2

172

Figure 5.5 Selected ion chromatograms and MS3 spectra. Rows are stacked in order, with Compound 1 on top. The color scheme matches Figure 5.1. The left column shows the selected ion chromatograms for each compound. Standards were prepared at 1 nM in aCSF:preservation solution for matrix matching. The right column shows characteristic MS3 spectra for each compound. The bottom row is the internal standard, dAdLE, at 70 pM.

173 Table 5.3 Sensitivity and range of peptide quantification. Limits of detection and linear dynamic range (in pM and amol, using a 5 µL injection) for MS3 quantitation of

Angiotensin 1-7 derivatives. Asterisk denotes highest concentration tested, rather than known limit.

Compound LOD (pM) LOD (amol) LDR (pM) r2

1. Ang 1-7 111 555 111 - 1113* 0.992

2. Ang 1-7-NH2, Val-d8 34 170 34-1104* 0.979

3. Ang 1-6-Ser-NH2 35 175 35-1127* 0.995

4. Ang 1-6-Ser(O-Glc)-NH2 1.9 9.5 9.5 - 950 0.996

5. Ang 1-6-Ser(O-Cb)-NH2 4.1 20.5 13-825* 0.998

6. Ang 1-6-Ser(O-Lac)-NH , 2 4.1 20.5 13-820* 0.991 Val-d8

174 orders of magnitude lower than those of the non-glycosylated peptides, due to the high efficiency of the MS2 fragmentation step. For glycosylated peptides, a larger total percent of the ions are advanced to the second fragmentation step, leading to higher signal in MS3 for the same concentration, and thus a lower limit of detection.

5.3.2 Modified compounds have increased in vitro lifetime The in vitro stability of the six Angiotensin 1-7 derivatives was assessed using rat serum at 37 °C, in order to model degradation due to the species present in a biological system as closely as possible.

Figure 5.6a presents the work flow for the degradation sample preparation. Compounds of interest (at a stock concentration of 1 mg/mL in 1% HAc) were spiked into rat serum at t = 0 minutes for a concentration of ~100 µM. Aliquots (10 µL) were removed from the solution at set time points, spiked with 1 µL glacial acetic acid to quench peptidase activity,

1 µL of 25 mM octyl sulfonate, and 1 µL internal standard (20 µM dAdLE). Samples were desalted used a ZipTip® procedure optimized for the compounds of interest. Samples were diluted to 100 µL in 50:50 ACN:H2O:0.1% FA and injected directly onto the instrument using a 20-µL injection loop at a flow rate of 8 µL/min, resulting in a 2.5-minute bolus. Area under the curve (AUC) for the characterized MS2 fragments was used for quantification. Statistical analysis was performed in GraphPad Prism 5 (La Jolla, CA).

Trials with exponential curves with r2 < 0.7 were not included, and errors are reported as s.e.m. As shown in Figure 5.6b and Table 5.4, after 26.9 ± 3 minutes, 50% of the Native

Ang 1-7 had degraded (n = 6). The replacement of the carboxyl terminus with an amide increased the t50 to 33.3 ± 9 minutes. Many peptidases attack the carboxyl terminus and masking this via amidation hinders this mechanism.17–19 By replacing the carboxyl terminus with an amide, these peptidases are less successful at recognizing the target site on our compounds. Interestingly, the replacement of the 7th residue (proline) with a serine decreased the half-life to 10.8 ± 3 minutes. Proline motifs add conformational

175

Figure 5.6 In vitro stability of Ang 1-7 derivatives. a) Work flow for analysis of in vitro stability. Compounds are spiked into rat serum at 37 °C and aliquots are taken at set time points. Aliquots are quenched with acid and spiked with internal standard

(dAdLE), then desalted with C18 ZipTips®. Samples are run in duplicate with direct injection using a 20-µL injection loop on an Applied Biosystems QStar Elite.

Quantification is based on the area under the curve for the identified MS2 fragments over the course of the injection. b) Results of degradation experiments. Reported results are average ± s.e.m. Statistical significance determined by one-way ANOVA with

Bonferroni’s Multiple Comparison Test. Native Ang 1-7 had a half-life of 26.9 ± 3 min in vitro. The t50 values of Compounds 1-4 were not found to be statistically different (p

> 0.05). Modification with disaccharides lactose and cellobiose increased the t50 significantly (p < 0.01), to 3.6 ± 0.4 h and 2.4 ± 0.3 h, respectively. The disaccharide modifications were not statistically different from each other (p > 0.05).

176 Table 5.4 Lifetime of Angiotensin 1-7 derivatives in vitro.

Asterisks denote significance from the native compound as determined by one-way ANOVA with Bonferroni post-hoc, p < 0.05.

Compound t50 (mean ± s.e.m.) n 1 Native Ang 1-7 26.9 ± 3 min 6

2 Ang 1-7-NH2, Val-d8 33.3 ± 9 min 4

3 Ang 1-6-Ser-NH2 10.8 ± 3 min 5

4 Ang 1-6-Ser(O-Glc)- NH2 39.5 ± 12 min 5

5 Ang 1-6-Ser(O-Cb)- NH2 2.4 ± 0.3 h * 4

6 Ang 1-6-Ser(O-Lac)- NH2, Val-d8 3.6 ± 0.4 h * 3

177 restrictions that serine does not, likely leading to steric hindrance of peptidases.

Additionally, there is evidence that proline bonds are resistant to many peptidases, so it is reasonable that replacement of proline with serine would decrease in vitro lifetime.31 Due to this effect, it may be interesting in the future to synthesize a derivative with the serine protected by a proline on the C-terminal side, though whether this affects binding would be necessary to investigate.

Addition of a glucose moiety onto the serine recovered and even slightly improved the lifetime to 39.5 ± 12 min. Addition of a sugar moiety increases the t50 in all cases, and when the sugar modification is a disaccharide, the lifetime is increased considerably (t50 for Compound 5 and Compound 6 are 2.4 ± 0.3 h, and 3.3 ± 0.4 h, respectively). This is partially believed to be due to the steric hindrance of the sugar moiety during peptidase activity. While improved activity of disaccharide derivatives of endogenous peptides has been reported, few have studied the changes in stability due to these modifications.21 The improved in vitro stability with disaccharide glycosylation shows great promise for the development of new peptide-based drugs with longer in vivo lifetime. If the BBB penetration and activity of these compounds can be shown to be improved over native peptides, the future of pharmaceutical drug design could be very different. The ability to design effective peptide-based drugs with lifetimes sufficient for treatment would mean better drugs with higher specificity for target receptors and fewer side effects than existing drugs.

Degradation products of native Ang 1-7 and Ang 1-6-Ser(O-Glc)-NH2 were investigated.

Aliquots of degradation solutions (compounds at ~100 µM in rSer) were removed and quenched at t = 0 and t = 20 minutes for native Ang 1-7, and t = 0 and t = 3 hours for Ang

1-6-Ser(O-Glc)-NH2. Samples were prepared as stated previously, and were used for identification of the main degradation products to ensure biological compatibility and to

178 check for unexpected degradation products. The samples were run on the Thermo LTQ-

Orbitrap instrument using a data-dependent MS2 scanning method. Samples were screened for product identification using a SEQUEST proteomics search of a rat serum database modified to include Ang 1-7 and Ang 1-6-Ser, including modification via amidation and glycosylation. The identified degradation products of Compound 1

(DRVYIHP) were: DRVYIH, VYIHP, YIHP, DRVYI, RVYIHP, and RVYI. The sole identified degradation product of Compound 4 (DRVYIHS[O-Glc]-NH2) was DRVYI. Likely there are fewer degradation products because the glucose modification causes steric hindrance of the peptidases, and sites near the glucose are inaccessible. We saw no evidence that the modification via amidation or glycosylation yielded harmful degradation products, or products that are not produced by the natural degradation of endogenous Ang 1-7.

5.3.3. Non-selective protein binding may decrease free fraction of peptides The top panel of Figure 5.7 shows our Angiotensin 1-7 derivatives in the first five minutes of the degradation experiment (rSer, 37 °C). For most of the compounds (all but the lactoside derivative), within one minute approximately 25% of the compound appears to have degraded. This rate of degradation is not consistent with the rate over the course of the experiment, leading us to question if there was another possible cause. An additional experiment was performed (Figure 5.7b, black) where Ang 1-7 was incubated at 37 °C in artificial cerebrospinal fluid (aCSF), a solution that is isotonic with CSF but does not contain any peptidases. In theory, samples will not exhibit degradation in this solution, and in this case our experimental results matched our expectations. However, it was theorized that non-specific protein binding may be occurring, which reduces the free fraction of our peptides in solution. As we are only quantifying the rate of disappearance of the whole peptides, rather than degradation products, this would have the effect of increasing the rate of apparent degradation. To test this theory, we spiked aCSF with

179

Figure 5.7 Apparent degradation in the first minutes may be due to protein binding. A) The top graph expands the first five minutes of the degradation experiment shown in Figure 5.6. For all species other than the lactoside derivative, there is apparent degradation on the order of 25% in the first minute. B) In order to investigate whether this is due to degradation or binding to proteins in solution, we repeated the degradation experiment with Ang 1-7 in aCSF, and aCSF with BSA in biological concentration. We saw that the presence of BSA mimics the in vitro degradation of Ang

1-7 despite the lack of peptidases, indicating that the initial signal drop is likely due to peptides undergoing non-specific binding with proteins in the solution, rather than degradation.

180 bovine serum albumin (50 g/L). Albumin is the most abundant plasma protein in mammals, and has high genetic homology between species.32,33 The human reference range for albumin in blood is 35-50 g/L, and rat serum albumin concentrations are on the same scale.34,35 As can be seen in Panel B of Figure 5.7, the apparent degradation in aCSF spiked with BSA (blue) matches degradation seen in rat serum (red) in the first five minutes of our degradation experiment. This indicates that the initial drop in signal is likely due to protein binding rather than degradation due to peptidases.

It is interesting to note that we see this effect in only five of the six compounds. We do not see it in the lactoside derivative. This may indicate that the lactoside is not as prone to non-specific protein binding, which would lead to a larger free concentration in blood and potentially a larger effect for a smaller dosage. This coupled with the longer t50 for the lactoside derivative speak favorably towards its potential as a peptide modification for more effective pharmaceuticals.

5.3.4 “Shotgun microdialysis” allows for the direct comparison of in vivo lifetime and BBB penetration of Ang 1-7 derivatives A common challenge in animal research is inter-animal variability. It is difficult to control for all the variables in complex biological systems, and even genetically identical animals do not react identically. Different animals often have slightly different responses to the same treatment. Keeping that in mind, it is often challenging to compare the behavioral or chemical responses of multiple animals to any given drug. We have developed a method, coined “shotgun microdialysis”, in which we inject a single animal with all six compounds of interest and monitor the CSF concentrations using microdialysis. This allows us to directly compare the compounds within a single animal, and account for the variability between animals.

In microdialysis, a probe is surgically implanted into the target region (in this case the striatum). The probe has a semi-permeable membrane with a molecular weight cut-off.

181 Perfusate is flowed through the probe, and species in the surrounding matrix can diffuse through the membrane as a function of the concentration gradient. Microdialysis allows for sampling from a region without altering the volume of that region, which is of particular relevance when studying the CSF.

Probe recovery studies were performed with a series of microdialysis probes to optimize for the best recovery of our compounds (Table 5.5). Probes were submerged in a stirring solution of four Ang compounds in aCSF and recovery was calculated by comparing the solution concentration to the concentration of the dialysate, accounting for dilution. A flow rate of 0.5 μL/min was used. Six probes were tested, with differences in membrane length, outer diameter (OD), and membrane material. The 2 mm CMA 11 probe (CMA

Microdialysis, Harvard Apparatus, Holliston, MA) was originally selected, as it had the best recovery for the four compounds tested with the target membrane length and smaller OD.

Initial studies were performed in mice, requiring the shorter 2 mm probes. The studies presented herein occurred later in the project and were performed in rats, at which time we began to use 4.5 mm CMA 11 probes.

As microdialysis is inherently a diffusion-limited technique, there is a delicate balance between probe recovery and temporal resolution. The slower the perfusate is flowed through the probe, the higher the recovery will be, but the sampling time is increased. We quantified the probe recovery for CMA 11 probes at five flow rates for four of the target compounds (Figure 5.8). It was decided that a flow rate of 0.5 μL/minute optimizes for both percent recovery and temporal resolution. The pairing of a second line that introduces preservation solution yields a temporal resolution of 10 minutes for the collection of 10 μL of solution. Increasing the flow rate would increase the temporal resolution but would also decrease the percent recovery. As we expect concentrations

182 Table 5.5 Microdialysis probe selection Several microdialysis probes were tested for in vitro recovery of target compounds. Probes different in membrane length, outer diameter (OD), membrane material, and MW cut-off. The percent recovery for four of the six compounds was tested. Asterisk denotes the probes selected for initial in vivo experiments in mice. Further experiments in rats used the same probe with a 4.5 mm membrane.

Percent Recovery Probe Length Cut-off (OD, mm) (mm) (kDa) Membrane 1. Ang 1-7 2. –NH2 4. O-Glc 6. O-Lac

CMA 11 (0.24) 1 6 Cupr 7.7 3.6 4.3 3.3

*CMA 11 (0.24) 2 6 Cupr 13.8 13.5 7.3 6.6

CMA 7 (0.24) 2 6 Cupr 7.5 5.5 6.1 5.2

CMA 12 Elite 2 20 PAES 16.6 12.5 10.6 4.4 (0.50)

AgnThos MAB 4 6 Cupr 14.6 9.3 7.9 6.0 4.15.5.Cu (0.20)

AgnThos MAB 4 6 PES 11.9 11.0 7.1 6.7 4.15.4.PES (0.20)

183

Figure 5.8 Effect of flow rate on probe recovery. Probe recovery was quantified for four compounds at fie flow rates. Probes used were CMA 11 probes, 2 mm (see Table

5.5). It was determined that a flow rate of 0.5 μL/min would be used for in vivo studies.

This flow rate allows for the collection of 10 μL of solution in 10 minutes, taking into account the additional flow of the preservation solution. Using a flow rate of 0.5 μL/min optimizes for both percent recovery and temporal resolution.

184 toward the bottom of our LDR, particularly for the non-modified compounds, we decided to prioritize recovery slightly above temporal resolution.

To directly compare the in vivo lifetime and BBB penetration of all six Angiotensin derivatives, we developed a method that involves dual blood draws and microdialysis (as discussed in Chapter 4). The compounds were injected intravenously at 10 mg/kg via a single tail vein injection. Blood draws were taken at t = -10, 1, 5, 10, 20, 30, 40, 50, 60,

70, 80, and 90 minutes (where t = 0 corresponds to injection). Microdialysis fractions were time-locked with blood draws at ten-minute increments such that blood draws correlate with the median time of the microdialysis fraction. Blood draws were centrifuged for 2 minutes in a tabletop mini-centrifuge to separate the serum and the red blood cells. Serum was pulled off and diluted 100x into a solution of 50:50 aCSF:“preservation solution” for matrix matching with the dialysate and standards, and immediately frozen on dry ice.

Dialysate samples were collected for 10 minutes at a flow rate of 0.5 µL/min, with the microdialysis tee coupling a line containing preservation solution to the dialysate line immediately behind the probe, leading to a solution volume of 10 µL after 10 minutes. The dialysate samples were immediately frozen on dry ice. Post-experiment, samples were stored in a -80 °C freezer until analysis.

All samples were desalted using µ-C18 Zip Tips® (EMD Millipore) according to the aforementioned protocol, yielding a final solution comprised of 10 µL 60:40

185 The serum concentration and CSF concentration estimates were calculated using a calibration curve, accounting for the dilution factors and the probe recoveries for each experiment. Standards were matrix-matched to samples and underwent the same sample preparation steps. Probe recoveries for probes used in the in vivo experiments were calculated by submerging the probe into a vial of standards post-experiment (Table 5.6).

Probe recoveries tended to average around 2%. Note that the probe recoveries are highly variable between probes, and are lower than the recovery values seen in our method development, despite being longer probes (4.5 mm). This is due to biofouling, which is a large concern in microdialysis, and is why we used the post-experiment recovery values to back-calculate the CSF concentrations for each individual probe.

Figure 5.9 shows the results of the in vivo experiments, with the color scheme matching

Figure 5.1 (n = 8 animals). The y-axis is normalized signal, where each compound is normalized to the signal of Native Ang 1-7 at t = 10 min. This compound was selected because we wanted to compare the effect of modifications to the original peptide. Time t

= 10 minutes after injection was selected because in several cases the 1- or 5- minute fraction for Ang 1-7 was lost due to 1) failed blood draw, 2) LC system over-pressure, or

3) signal below LOQ. In serum (Figure 5.8a), we saw that Compounds 1 and 2 are overlaid and virtually indistinguishable, indicating that amidation did not notably alter the lifetime in vivo. Compounds 3-6 all reach higher apparent concentrations at t = 1 min and have mostly been cleared by t = 30 min. While on the surface this seems to contradict the in vitro stability data that indicated that glycosylated peptides would last much longer in vivo, one must take into account that the in vitro technique mimics only the activity of species present in serum. The stability method does not account for kidney or liver clearance, which are believed to be the leading causes of plasma clearance of angiotensins.36 It is interesting to note that Compound 3, the serine-substituted derivative, seems to survive

186 Table 5.6 Post-experiment probe recoveries for probes used in vivo. After in vivo experiments, recovery of probes was tested in a standard solution. The recovery for each compound was used to back-calculate the concentration in the CSF. Low recoveries are due to biofouling.

Percent recovery of each Angiotensin 1-7 derivative Probe 1 2 3 4 5 6

1 6.20 12.31 6.45 6.86 8.23 5.19

2 4.55 8.56 3.66 2.67 1.54 1.49

3 0.73 1.26 0.53 0.52 0.82 0.94

4 2.10 3.23 1.78 0.86 0.79 2.51

5 2.51 1.27 1.19 1.70 1.13 0.45

6 0.33 0.64 0.39 0.41 0.60 0.72

7 1.72 2.37 1.70 1.76 2.27 2.13

8 10.66 5.25 2.08 2.93 3.40 1.83

Average ± 3.60 ± 1.2 4.36 ± 1.5 2.22 ± 0.7 2.21 ± 0.7 2.35 ± 0.9 1.91 ± 0.5 s.e.m.

187

Figure 5.9 In vivo results following injection of six Ang 1-7 derivatives. Six Ang compounds were injected (10 mg/kg, i.v.) and quantified via dual blood draws and microdialysis. a) Signal for the six compounds in serum, normalized to Native Ang 1-7 at t = 10 min. The amidated derivative showed behavior similar to the native.

Compounds 3-6 reached higher apparent concentrations in the serum, but concentrations for all compounds returned to baseline by 30 minutes, likely due to kidney clearance. a, right) Zoom in on t = 0 to 20-minute time frame. b) Concentration estimates for the six compounds in CSF, accounting for dilution factors and probe recovery, and normalized to Native Ang 1-7 at t = 10 min. The sugar modified compounds reached higher concentrations in the CSF, and persisted for longer. b, right) Zoom in on t = 0 to 20-minute time frame. n = 8 animals.

188 better than the in vitro results indicated. However, there is some concern that impurities in the sugar modified compounds or degradation/oxidation thereof may cause falsely high readings of this synthetic precursor. For example, looking back at Figure 5.3, the MS spectrum for Compound 4 shows a small peak at m/z 444.74, coinciding with loss of glucose, which also happens to be the parent mass of Compound 3. It is unknown at this time if this is an impurity or caused by oxidation. In future experiments, the serine derivative will be modified, likely by deuterating a residue other than the Val, to distinguish between the injected compound and any impurities or “products” of the other derivatives.

The graph on the right of Figure 5.8a is a zoom in on the t = 0 – 20 min region for easier comparison between the derivatives.

Figure 5.9b is the normalized signal for the derivatives as quantified in the CSF, again normalized to Native Ang 1-7 at t = 10 min. The glycosylated compounds show a clear improvement in both penetration and persistence in the brain compared to Native Ang 1-

7, reaching higher normalized signals and lasting longer. The penetration of the glycosylated compounds does appear to be slightly slower, as they all reach their maximum concentration in the 10-minute fraction, whereas the non-glycosylated compounds peak at 1 minute. This is possibly due to an altered mechanism of entry into the brain, determined by the sugar moiety rather than the peptide backbone. The right panel in Figure 5.8b is a zoom of the region from t = 0 to 20 minutes. Here the increased penetration of the glycosylated compounds is clear (error bars are mean + s.e.m.).

This data can also be presented as shown in Figure 5.10, where the area under the curve

(AUC) for each compound was calculated from -10 to 60 minutes (mean + s.e.m.). Panel a shows the AUC values for all six compounds in the serum. While there does appear to be an increase in AUC for most of the compounds compared to Native Ang 1-7, indicating increased in vivo lifetime, the lactoside derivative was the only compound with an AUC

189

Figure 5.10 Calculated area under the curve (AUC) for each of the Angiotensin derivatives in vivo. AUC was calculated from Figure 5.8 for each compound from -10

- 60 minutes in a) serum, and b) CSF. All modified compounds show slightly increased

AUC in serum compared to Compound 1, indicating improved stability, but only

Compound 6 is statistically significant (repeated measures ANOVA, Dunnett’s multiple comparison test, p < 0.05). Modified compounds also show increased AUC in the CSF, indicating improved penetration or persistence, but again the lactoside is the only statistically significant increase (repeated measures ANOVA, Dunnett’s multiple comparison test, p < 0.05). n = 8 animals.

190 significantly higher than Ang 1-7 (repeated measures ANOVA, Dunnett’s multiple comparison test, p < 0.05). Interestingly, the glucoside seemed to have an improved lifetime in vivo compared to the cellobioside, which was unexpected. Perhaps the foreign structure of the non-native sugar was recognized and thus cleared from the blood more quickly.

Figure 5.10b shows the AUCs for the six derivatives in the CSF. Again, there appears to be an increase for most of the derivatives, indicating increased penetration and/or persistence in the brain, though the lactoside derivative was the only compound with a significant increase compared to native (repeated measures ANOVA, Dunnett’s multiple comparison test, p < 0.05). The cellobiose derivative showed improved penetration and persistence compared to Native Ang 1-7, but not to the extent of the lactoside, and was not statistically significant (p > 0.05). As cellobiose is not produced endogenously in mammals, we theorize that the cellobioside did not penetrate the BBB as well as the lactoside because of its slightly alien structure, or because it was being cleared from the blood more quickly, as indicated by the serum data.

In short, we were able to co-inject all six Ang 1-7 derivatives and quantify them in both rat serum and dialysate. We were able to directly compare the in vivo lifetime, BBB penetration, and persistence in the brain between all six compounds in the same animals, circumventing the challenge of inter-animal variability. The sugar modified compounds showed improved penetration across the BBB and persistence in the CSF, particularly the compound modified with lactose. We have shown that it is possible to compare multiple drugs in a single animal, and that modification of peptides can overcome their shortcomings for the development of peptide-based drugs.

191 5.4 Conclusions

Six Angiotensin 1-7 derivatives were synthesized to investigate the effect of certain structural modifications on the in vivo stability and BBB penetration of peptide-based drug candidates. Glycosylated derivatives, particularly one compound with a lactose modification, showed promising stability and penetration properties for in vivo therapeutic targets. Our in vitro method allowed for the screening of lifetime of peptides in rat serum to aid in the development of more stable derivatives. The addition of a sugar moiety was shown to increase biological half-life and improve permeation of the blood brain barrier.

The lactoside derivative (Ang 1-6-Ser[O-Lac]-NH2) had the most promising properties of the six compounds, increasing the in vitro lifetime from 26.9 ± 3 minutes for Native Ang 1-

7 to 3.3 ± 0.4 hours. The in vitro screening method, while overestimating the lifetime of compounds in vivo, is successful in comparing the lifetimes. This allows us to rapidly and simply screen new derivatives for improvements in stability without undergoing expensive animal experiments. The in vivo method involved dual blood draws and microdialysis, and successfully quantified in vivo lifetime of compounds in the blood, and penetration across the BBB and persistence in the CSF. Sugar modified compounds showed improved BBB penetration and persistence in the CSF, with the lactoside showing a statistically significant improvement over Native Ang 1-7 (p < 0.05).

It would be interesting in the future to investigate a few more derivatives, including a compound with the serine protected by a proline, or derivatives with other disaccharide moieties. Additionally, modification of Compound 3 would allow for clearer distinctions to be made between compounds. Currently, the lactoside derivative (Compound 6) shows the greatest promise in terms of stability and penetration. Further studies will investigate the physiological effects of these derivatives, including binding assays and behavioral studies.

192 The increased lifetime and BBB penetration of our modified compounds show great promise for the world of pharmaceutical development. By modifying our peptides, we have been able to show that there is a way around the “fatal flaws” of peptide-based therapeutics, namely their rapid degradation in vivo and poor membrane permeability.

Using this information, it will be possible to develop new pharmaceuticals with the safety and selectivity of peptides, but without the drawbacks.

5.5 Author Contributions

CLK developed mass spectrometric methods and performed sample preparation and analysis. MJB performed all surgical procedures and assisted in sample collection. CL assisted in sample collection and preparation. ARM assisted with degradation experiments. Ang 1-7 derivatives were synthesized by EMJ, CS, and LS.

5.6 Acknowledgements

Special thanks to Linda Breci and George Tsaprailis of the Arizona Proteomics

Consortium at the University of Arizona, for their advice and use of the Thermo LTQ Velos-

Orbitrap instrument, which was funded through NIH/NCRR 1 S10 RR028868-01.

193 Chapter 6

Conclusions and future directions

Throughout this work, we have shown applications of tandem mass spectrometry for the quantification of neurologically-relevant molecules, from small molecule neurotransmitters to moderately sized peptide-based drug candidates. Using tandem MS allows for sensitive and selective detection in complex matrices, which is necessary to gain the neurochemical information that will allow us to solve biological problems. Chapter 2 introduced a high-throughput derivitization-based method that can be used to quantify biogenic amines in complex matrices such as whole-head homogenate. By coupling a liquid-liquid extraction with direct-injection MS2 analysis, we were able to quantify biogenic amines in the low nM regime without a chromatographic separation. This improves the sample throughput compared to existing methods that require LC. Additionally, there are fewer limitations on the species that can be analyzed, as MS is more widely applicable that EC, which is commonly used for this purpose. Chapter 3 expanded upon the applications of this technique, showing how it can be used to explore changes in biogenic amine content as a function of certain stimuli. We applied our method to investigate the effects the parasite Nosema ceranae on the biogenic amine content of honey bee brains.

As Nosema is known to cause behavioral changes in infected honey bees, quantification of biogenic amines that regulate behavior can provide information that was not previously available about how the parasite alters brain chemistry in honey bees.

Chapters 4 and 5 introduced methods for the quantification of peptide-based drugs in whole blood and CSF, as well as methods to study in vitro degradation and in vivo blood- brain barrier penetration. While peptide-based drugs show great promise in terms of selectivity to receptors, their instability and low membrane permeability in vivo have traditionally limited drug development. Our methods allow for the screening of potential

194 peptide-based therapeutics for candidates with sufficient properties to warrant further investigation. This can improve the throughput of development of peptide-based pharmaceuticals.

In all these cases, tandem mass spectrometry was used to quantify neurochemicals in complex matrices, allowing for sensitive and selective quantitation. The information gleaned from these methods may allow us to understand the mechanisms of the brain more clearly, in order to develop new medical approaches or improve existing treatments.

This research is ongoing. The remainder of Chapter 6 will elucidate some of the future directions of our research.

6.1 Expanding the portfolio of BzCl-DI-MS

6.1.1 Increasing the number of amine compounds being investigated As it stands, our detection method for biogenic amines quantifies only dopamine, histamine, serotonin, octopamine, and L-DOPA. To truly fill in the gaps where other methods fall short, our method should be expanded to include non-electroactive compounds, such as GABA, glutamate and acetylcholine. Both are neurotransmitters that control neural circuit activity.

GABA is a key inhibitory neurotransmitter in vertebrates. GABA and GABA receptors are implicated in neuropathic pain.1,2 While methods for the quantitation of GABA have previously been published, most involve assays or derivatization followed by column separations.3 A 1995 study quantified GABA in vivo using microdialysis-CE/MS, but had

LODs in the high µM regime.4 A recent study quantified both GABA and glutamate in microdialysis samples using LC-MS/MS with an LOD of 1 nM and an impressive 3-minute separation.5 Our method would nearly rival this in speed and detection limit, without requiring as much instrumentation.

It would also be of benefit to branch into the biosynthetic precursors and products of the compounds for which we have already developed methods. For example, we currently

195 quantify dopamine and its synthetic precursor, L-DOPA. It would be beneficial to expand quantitation of this pathway, say to also quantify tyrosine (L-DOPA’s precursor), or to quantify downstream products of dopamine, such as norepinephrine, 3,4- dihydroxyphenylacetic acid (DOPAC), or 3-methoxytryptamine (3-MT). This could give us a better idea of if or where dysregulation is occurring in neurological disorders. Figure 6.1 shows the pathways for the five molecules quantified with this method. Most of the biosynthetic precursors or products have potential for quantitation by this method (i.e. have primary or secondary amines or phenols, which can be labeled with BzCl). The five compounds currently quantified are shown in boxes.

Additionally, the method presented herein targeted only neurologically relevant molecules.

As BzCl derivatization can be used on any species containing a phenol, or primary or secondary amine, this method can be applied across many different fields of study. The only inherent challenge of translating this method to new applications is accounting for changes in matrix, which affects quantitation substantially. As shown in Chapter 2, complex matrices (such as Drosophila whole-head homogenate, which contains the cuticle) can add challenges to our quantification.

6.1.2 Application to additional factors in honey bee hive health Herein we presented data on the effect of Nosema ceranae infection on five biogenic amines in the whole brains of honey bees, Apis mellifera. As noted in the methods section of Chapter 3, bees employed for this study were collected between March and July only, despite the researchers involved collecting and analyzing samples from bees year-round. This is because we saw preliminary differences in the whole brain biogenic amine content based on the time of year. Figure 6.2 shows a preliminary analysis of the five biogenic amines quantified in bees during different seasons. The samples shown here were all control bees at 7 days old that were fed a consistent diet. Our samples span three of the four

196 Figure 6.1 Biosynthetic pathways for several biogenic amines. Biogenic amines currently quantified with our method are shown in boxes. Any compound with a primary or secondary amine, or phenol can be labeled with benzoyl chloride derivatization.

197

Figure 6.2 Biogenic amine content in whole honey bee brains shows differences based on season. All data shown in for control bees, Day 7, fed a consistent diet. The number value in the graphs is the n for each sample set. Asterisks denote statistical significance, as determined by one-way ANOVA with Bonferroni post hoc, *p < 0.05,

***p < 0.005. In several cases, the biogenic amine content of whole honey bee brains are seen to change with season.

198 seasons. As this is preliminary data, we do not have a small number of samples in each category; for Spring, n = 9 bees; for Summer, n = 4 bees, and for Winter, n = 8 bees.

Several of the biogenic amines showed statistical differences between seasons (one-way

ANOVA, Bonferonni post-hoc, p < 0.05), though statistical analysis of such a small number of samples will not necessarily hold up to a larger number of samples.

Histamine, octopamine, and dopamine show a consistent decrease over the course of the year (spring > summer > winter), though these results are only statistically significant for histamine (spring to winter p < 0.005, summer to winter p < 0.05) and dopamine (spring to winter p < 0.05). Serotonin and L-DOPA show similar trends to each other, with the highest biogenic amine content seen in the spring, lowest in the summer, and an increase again in winter. Serotonin is significantly decreased in summer bees compared to spring bees (p < 0.05)., though again, only four bee brains were quantified here.

Decreased survival during winter months is often seen in managed honey bee colonies.6

Additionally, our collaborator at the ARS-USDA noted that there was a higher mortality during the winter months. It would certainly be worth further investigation if this

“longitudinal study”, so to speak, showed consistent changes in biogenic amine content with season in healthy control bees.

In many managed colonies, winter survival rates are affected by mites, fungi, bacteria, or viruses.6,7 In particular, the ectoparasitic mite Varroa destructor seems at its most destructive in the winter.8,9 Varroa is known to decrease the lifespan of bees and has been blamed for recent considerable winter losses.7–9 Another direction we plan to take this project is analyzing the effect of Varroa on biogenic amines in the honey bee whole brain. There is evidence to suggest that Varroa infestation alters the homing ability of bees, which leads to the possibility that biogenic amines involved in behavior may be altered.10 We did some preliminary analysis of the effect of Varroa on biogenic amine

199 content (Figure 6.3), controlling for age, diet, and season. We saw a statistical difference in serotonin content only, though dopamine decreased in infected bees as well (t-test, p <

0.05). However, the number of samples was again low (n = 9 for controls and n = 16 for mite-infested bees). Additionally, this data does not account for the number of mites per bee.

The samples shown in Figure 6.3 were all collected during spring. Seasonal survival rates have been seen to vary with Varroa infestation (with a notable decrease in the winter, as previously discussed), so it would be interesting to do a study of the biogenic amine content of honey bees infested with Varroa mites over the course of a year, or specifically during winter. Further experiments will investigate the effect of Varroa parasitism on biogenic amine content in honey bee brains.

6.2 Glycosylation and other modifications for improved pharmaceutical properties of peptide-based drugs In Chapter 4 we introduced in vitro and in vivo methods for the mass spectrometric quantitation of some pharmaceutical properties of peptide-based drugs, including stability and BBB penetration. In Chapter 5 we applied these methods to a series of Ang 1-7 derivatives to elucidate the effects of specific modifications and optimize for the most effective drug design. Moving forward will involve the expansion of these methodologies to include larger and more complex peptides and analysis of new modifications.

Another class of compounds that may soon by under investigation is derivatives of vasoactive intestinal peptide (VIP), which has some sequence homology with PACAP and has been shown to have some neuroprotective potential.11,12 Certainly there will be challenges in optimizing microdialysis and quantitation methods for these larger molecules that will come to light.

200

Figure 6.3 Varroa destructor may affect biogenic amine content in honey bee whole brains. Bees sampled were controlled for age, diet, and season. Statistical significance was found for serotonin only (t-test, p < 0.05), however the sample size is low (values in white). Data does not account for number of mites per bee. Bees in this study were collected during spring.

201 6.2.1 Investigation of apparent PACAP degradation Initial studies on two PACAP derivatives indicated that the compounds were too unstable in solution to move into in vivo trials. However, several attempts were made to identify the degradation products by both infusion and LC-MS, and no products could be isolated.

Figure 6.4a shows the full MS scan for an LC separation of 3 nM PACAP 1-27-Ser(O-Glc)-

NH2 both fresh (black trace) and after an hour-long incubation at 37 °C (red trace). No new peaks appear in the full MS, which would be expected if degradation were producing new compounds. Figure 6.4b shows the extracted ion chromatogram (EIC) for 675.45 Da, the known m/z for the +5 charge state of glycosylated PACAP, for the same separation shown in 6.4a. The peak at 16.23 minutes, which corresponds to PACAP, is distinctly decreased, which led to our initial hypothesis that degradation was occurring. Notably, the peak at 16.23 minutes is not a large component of the full MS chromatogram, despite being a pure sample, so it is possible that degradation products would not be visible in the full MS. For this reason, we proceeded to do an infusion study (Figure 6.4c), where we infused a 3 μM solution 1) freshly prepared in 50:50 ACN:H2O:0.1% FA (black trace), 2) the same solution incubated for 1 hour at 37 °C (red trace), and 3) the same solution incubated for an additional 24 hours at room temperature (blue trace). In all three cases we see peaks attributable to our PACAP derivative, and negligible changes in the spectrum, with the exception of small contaminants leaked from the Eppendorf tube in the long-term sample.

Compared to the in vitro degradation data presented in Chapter 4, the presence of our

PACAP derivative in the infusion solution is anomalous; we should have seen the compound disappear entirely over this time course. This information, coupled with the lack of identifiable degradation products, leads to our current theory. We believe that rather than degrading, the PACAP compounds are adsorbing to the tubes in which

202

17 Figure 6.4 No degradation products of PACAP 1-27-S-(O-Glc)-NH2 [ Leu] have been identified. a) Chromatograms for the full MS scans of PACAP 1-27-S-(O-Glc)-

17 NH2 [ Leu] freshly thawed (black) and after an hour-long incubation at 37 °C (red). No discernible degradation peaks appear after incubation. b) Extracted ion chromatograms

203 17 for 675.45 Da (PACAP 1-27-S-(O-Glc)-NH2 [ Leu], +5 charge state) in a freshly thawed solution (black) and after an hour-long incubation at 37 °C (red). PACAP (16.23 minutes) clearly loses signal over this time. c) A 3 μM solution was infused and the spectrum was summed over one minute. Shown are the spectra for PACAP 1-27-S-

17 (O-Glc)-NH2 [ Leu] freshly prepared (black), incubated for one hour at 37 °C (red), and incubated for an additional 24 hours at 25 °C (blue). No degradation products are visible. Our current theory is that the apparent degradation of PACAP is actually due to adsorption to the sample tubes.

204 samples are prepared. This would explain differential behavior based on concentration, where low concentration (100 pM) solutions appear to degrade too rapidly to build calibration curves, but higher concentration solutions (such as the 3 μM infusion solution) do not show distinct losses. Additionally, the presence of ACN in the infusion solution would have prevented some of the nonspecific adsorption. As additional support for this theory, our collaborators have not seen notable concentration losses when running LC of samples in the μM range. We are currently looking into compounds that can be spiked into our samples to prevent nonspecific adsorption without altering the solvent.

6.2.2 Positive and negative controls for blood-brain barrier penetration will strengthen our methodology As discussed in Chapter 4, microdialysis has been used for BBB penetration studies for several decades, and optimized methods for minimizing damage to the BBB have been published.13 An intensive review was published in 1997 that outlines the key considerations in developing a microdialysis method that minimizes damage and preserves natural BBB function during this inherently invasive approach.13

In general, the following practices are recommended: use isotonic perfusate and try to avoid a temperature gradient between the perfusate and the body temperature. Use low flow rates to avoid high pressure inside the probe, which may hinder diffusion. Use smaller diameter probes, preferably concentric, to minimize damage during insertion. Minimize surgical trauma by allowing recovery time between implantation and measurement.

Benveniste et al. saw altered cerebral blood flow and glucose metabolism within two hours of implantation that was recovered by 24 hours, which is the recommended length of recovery.14 Under these conditions, one study investigated the BBB transport of the hydrophilic drug atenolol and the lipophilic drug acetaminophen. Both drugs were found to have similar penetration to that reported in previous studies with different experimental techniques.15 Also, anesthesia can interfere with some physiological processes,16 thus

205 should be avoided if possible. One way to circumvent anesthetizing animals during the experiment is to implant a guide cannula in advance.

Certain aspects of our approach do not fall within these guidelines, namely probe implantation the day of the experiment, and performing an experiment while the animal is under anesthesia. The carotid catheterization prevented us from following these guidelines, as it is best to minimize the number of times the animal is placed under anesthesia and the longer the catheter is implanted, the higher the likelihood of blockage or accidental removal. In the future, we would likely change one of two things, either following the previously accepted guidelines and no longer using a carotid catheter, or implementing the use of BBB+ and BBB- negative controls.

The BBB penetration properties of several compounds have been studied in depth, allowing them to be used as controls for studies using new techniques for probing the

BBB.17 Some compounds with known permeability that have been used as controls are acetaminophen,15 [14C]-alpha-amino-isobutyric acid,18 atenolol,15,19,20 and ibuprofen.21,22

However, if the compounds are to be used as controls during drug experiments, we would not want them to cause any physiological effects that might alter the experimental results, such as increased heart rate.

For this reason, we originally attempted to add loratadine as a BBB- control. Loratadine, more commonly known as Claritin, is a non-sedating antihistamine that crosses the blood- brain barrier only minimally.23–27 In one experiment, we injected 10 mg/kg loratadine into the tail vein of an animal undergoing a shotgun microdialysis 90 minutes after the drug injection. This allowed us to collect a baseline over the course of the shotgun experiment, as well as collect information on the BBB penetration of loratadine. However, despite using MS3 quantitation, we detected an interfering molecule at much higher concentration than we would expect to see the injected drug, even in baseline fractions. While MS3

206 quantitation is typically highly selective, the fragmentation pathways of small molecules and the subsequent masses are less selective than those of larger molecules.

The method we developed to quantify loratadine used the ion transitions 383.15  337.17 and quantified the MS3 fragments 259.17, 267.09, 281.17, 294.17, 302.17, and 322.17

Da. Figure 6.5 shows the peak area for these fragments over the course of the experiment. In this experiment, the animal was dosed with a shotgun mixture at t = 0, and with 10 mg/kg loratadine at 95 minutes. The horizontal line at 2.2 a.u. is the signal measured for the highest concentration standard we used, at 10 nM, and nearly all the fractions, both serum and dialysate, are higher than this value.

For comparison’s sake, the concentration of endogenous Ang 1-7 measured in our baseline fractions was around 300 pM. Even more strikingly, these values do not account for the 100x dilution of serum samples prior to analysis. Taking this into account, we measured approximately 10-35 μM “loratadine” during baseline collections. To compare, the maximum concentration measured for our Angiotensin lactoside derivative was about

2 μM in serum, immediately after injection. Taking these values into account, our ability to distinguish a small amount of loratadine crossing the BBB is highly questionable.

We are looking into using different compound as a BBB- control, possibly the beta-blocker atenolol.

6.3 Potential routes for the elucidation of enkephalin dynamics in vivo In Appendix

B we present a method for the quantitation of methionine- and leucine-enkephalin in rat brain dialysate, and show that we can stimulate release in the anterior cingulate cortex

(ACC) with reverse microdialysis of a high concentration potassium solution.

207

Figure 6.5 An interfering species prevents the use of loratadine as a BBB- control in animal experiments. An LC-MS3 method was developed for the quantitation of loratadine. Both serum and dialysate samples were collected during a shotgun microdialysis experiment, where the shotgun drug mixture (10 mg/kg) was injected i.v. at t = 0 minutes, and loratadine (10 mg/kg) was injected i.v. at 95 minutes. In all samples, the measured peak area was greater than that of our highest concentration standard, which is higher than the expected concentration of loratadine at any time after injection (accounting for dilution). There is an interferent that prevents the use of this quantitative method for this compound as a BBB- control.

208 6.3.1 Quantitation of enkephalins in chronic pain models The next step is to quantify changes in endogenous opioid peptides (EOPs) in instances of chronic pain, in order to better understand how signaling changes in healthy versus pained animals.

Chronic pain is perhaps the most significant malady of western society, with over 100 million Americans suffering from some form of chronic pain.28,29 An understanding of the biochemical mechanisms which regulate EOP levels in the central nervous system

(CNS),30 the interconnected nature of the opioid and other neurotransmitter systems, 31,32 and the specific regions of the CNS where EOPs dynamically respond to physical and pharmacological stimuli are needed in the continuing effort to develop better and safer analgesics for the treatment of chronic pain conditions.33,34 Spinal nerve ligation (SNL) is a surgically-induced model of chronic pain.35 By studying differences in baseline EOPs in animals with and without SNL, we can try to understand signaling dynamics in order to determine better treatments for people suffering from chronic pain.

Preliminary results of SNL experiments are shown in Figure 6.6, mean ± s.e.m. for the on- column concentrations (pM) of met-enkephalin, leu-enkephalin, and endomorphin II.

Results include three sham animals (animals that underwent the surgery but not the spinal ligation) and five SNL animals. Both leucine enkephalin and endomorphin II were significantly increased in SNL animals (t-test, p < 0.05). These preliminary results suggest a role for leucine enkephalin and endomorphin II in chronic pain signaling. Further studies should expand the number of animals tested and investigate the effect of different analgesics on enkephalin levels.

6.3.2 Pharmaceutical probing of enkephalin dynamics Gabapentin is a small-molecule drug often used for its antinociceptive properties, particularly in cases of neuropathic pain.36–41 Gabapentin was originally synthesized to mimic gamma-aminobutyric acid

(GABA) and was used as an anticonvulsant.42,43 Gabapentin mediates voltage-gated

209

Figure 6.6 SNL surgery alters baseline concentration of EOPs. Spinal nerve ligation

(SNL) is a model for chronic pain. EOPs were quantified in sham animals (n = 3) and

SNL animals (n = 5). Baseline concentration of leucine enkephalin and endomorphin II were found to be significantly increased in SNL animals (t-test, p < 0.05).

210 calcium channel activity by binding to the α2δ subunit, rather than interacting with GABA receptors, despite the structural similarity.44–46 There is evidence that gabapentin reduces the expression of pro-inflammatory cytokines by activating interleukin-10-heme- oxygenase-1 (IL-10-HO-1).47,48 In recent years, gabapentin’s effects in SNL chronic pain model have been under investigation. Chapman et al. reported decreased spontaneous spinal neuron activity with gabapentin treatment in SNL rats using electrophysiology.49

Hooker et al. used blood-oxygen-level dependent (BOLD) fMRI to monitor several brain regions in SNL rats treated with gabapentin, and reported decreased BOLD signals in cingulate cortex after injection, implying decreased activity in this region.50 However, the effect of gabapentin on enkephalin signaling is unknown at this time.

Based on our observation of release of the enkephalins with potassium stimulation, we decided to treat animals with gabapentin to investigate if enkephalins in the ACC are affected. In this preliminary study, 8 animals (3 sham, 5 SNL) were injected intravenously with 50 mg/kg gabapentin at t = 0. We did not see measurable changes in met-enkephalin, leu-enkephalin, or endomorphin II post-injection in either sham or SNL animals, indicating that gabapentin does not affect the release of enkephalins in the ACC (Figure 6.7), despite evidence that it acts on that region of the brain. This may indicate that gabapentin’s mechanism of action does not overlap with enkephalin signaling. Further pharmaceutical probing is necessary to help elucidate the mechanisms of enkephalin release in the ACC.

6.4 Concluding remarks

In this work, we have shown the varied applications of tandem MS for the quantification of neurologically relevant molecules in complex biological matrices. The ability to measure these compounds under different biological circumstances or in response to certain stimuli allows us to probe the inner workings of the brain and solve biological problems. Tandem

MS can be used to improve the throughput, sensitivity, or selectivity of many methods,

211

Figure 6.7 Injection of gabapentin does not affect enkephalin release. Measured percent baseline for a) methionine enkephalin, b) leucine enkephalin, and c) endomorphin II after gabapentin injection at t = 0. Baseline is defined as the average of the three fractions collected prior to injection. Gabapentin does not appear to alter enkephalin levels in the ACC. Error bars are s.e.m. Sham n = 3, SNL n = 5 animals.

212 allowing for the measurement of compounds that have been previously inaccessible. As shown in Chapters 2 and 3, tandem MS can allow for higher-throughput measurements of complex samples than current techniques, without the chemical limitations. Chapters 4 and 5 address the complex challenge of the blood-brain barrier and how to improve the in vivo lifetime and BBB penetration of peptide-based pharmaceuticals.

Mass spectrometry is a powerful tool for the analysis and quantitation of neurologically- relevant molecules. Few techniques have the sensitivity and selectivity of mass spectrometry, let alone the potential to analyze such a wide variety of compounds. As technology and sample preparation techniques improve, the sensitivity and applicability of mass spectrometry can only advance. Mass spectrometry will certainly continue to be on the leading edge of technological advances and improvements in our understanding of the intricate workings of the brain, allowing us to develop better models and better therapeutic approaches.

213 APPENDIX A

Procedure: Benzoyl Chloride Labeling Reaction

Preparation

Step 1. Calculate volume of each reagent needed. For each sample, you will need:

 100 μL 2% benzoyl chloride per sample (v/v in dry acetonitrile)  200 μL buffer per sample (200 mM carbonate buffer, pH 11.0)  400 μL dichloromethane per sample  400 μL basified water per sample (deionized water adjusted to pH 8.0 with glacial ammonium hydroxide)

Step 2. Prepare labeling solution: 2% BzCl in dry ACN (store with Drierite until color of indicator changes, then replace)

Ex. 6 μL BzCl in 294 μL ACN Note: this solution degrades quickly. Do not keep longer than 4-6 hours.

Hydrolyze excess with H2O prior to disposal.

Step 3. Prepare standard solutions at 1.0 mM in 0.1 N perchloric acid. Refrigerate when not in use. Standards may be used up to one month after preparation.

Step 4. Prepare five 100 μL dilutions of standards at 1.0 μM in HClO4.

Step 5: Begin with biological sample homogenized in 100 μL 0.1 N HClO4 in a test tube

Reaction Procedure

Step 6: To each 100 μL sample (or standard), add 200 μL carbonate buffer

Step 7: Immediately label by addition of 100 μL 2% BzCl, as biogenic amines oxidize under basic conditions. Solution will turn cloudy.

For internal standard, use deuterated benzoyl chloride. Prepare solution at 1 μM, spike final solution (after extraction, drying, and reconstitution) into samples for a final concentration of 50 nM.

Step 8: Vortex solution for 5 seconds.

Extraction Procedure

Step 9: Add 200 μL dichloromethane to reaction mixture with gastight syringe. Vortex 5 seconds. Pull off organic (bottom) layer and place in new test tube.

214 Step 10: Repeat step 9, adding organic layer to tube from Step 9.

Step 11: Wash DCM extracted layer with 400 μL basified water. Discard wash.

Step 12: Wash DCM extracted layer with 400 μL basified water. Remove organic layer into a fresh test tube.

Step 13: Spin down sample (~20 min) to dryness on a vacuum concentrator.

Step 14: Reconstitute sample in 100 μL ACN:H2O:formic acid 1:1:0.1% (v/v). For biological samples, it is typically necessary to dilute samples 5x to bring signals into the linear range of the instrument.

Step 15: Combine reacted standards. Prepare calibration curve over range 1-1000 nM. Validate linearity before running samples.

215 APPENDIX B Quantification of endogenous opioid peptide dynamics in the anterior

cingulate cortex by online-preservation microdialysis

Abstract Endogenous opioids and their receptors are important in multiple biological circuits including those associated with pain, reward, addiction, drug abuse, stress, and affective disorders. The dynamic monitoring of endogenous opioid peptides (EOPs) in the mammalian central nervous system has been difficult due to their low concentrations, high susceptibility to degradation, and the significant expense and specialization of available monitoring techniques. To address the role of EOPs in chronic pain, and expand dynamic in vivo measurements of EOPs to cortical regions, we developed on-line preservation microdialysis with nano liquid-chromatography and tandem mass-spectrometry (nano LC-

MSn). This approach was employed to prevent peptide degradation during collection from awake and freely moving rats. A commercially available microchip-based electrospray ionization source facilitated ultra-trace measurements of enkephalins (limit of detection

1.5 amol) and overcame difficulties associated with previously employed LC-MS instrumentation. This allowed in vivo measurement of leucine and methionine enkephalin, and endomorphin II in the rat anterior cingulate cortex at estimated concentrations of 46 pM, 160 pM, and 1.4 nM, respectively. Additionally, significant stimulated release occurred for enkephalin (> 150% baseline). This approach can be used to quantitatively monitor release of endogenous opioids in physiological settings for pharmacological and behavioral challenge of EOP dynamics.

216 B.1 Introduction

The opioid system comprises four related G protein-coupled receptors, including the μ

(MOP), δ (DOP), κ (KOP) and nociceptin receptor (NOP) and their endogenous ligands endorphins, enkephalins, dynorphins, and the nociceptin family of peptides.1 These neuropeptides are produced by post-translational proteolytic cleavage from precursor proteins.2 Additionally, endomorphin I (EM I) and endomorphin II (EM II), tetrapeptides with high affinity and selectivity for the MOP, have been discovered though their precursors have not yet been identified.3 EOPs are linked to many fundamental neurochemical processes including modulation of pain, reward,4 learning and memory,5 and feeding and metabolism.6 Thus, the techniques capable of quantifying dynamic changes are critical to probe the role of endogenous opioid peptides in specific brain circuits underlying both physiological7,8 and pathological states including chronic pain, substance abuse, and depression.9,10 Chronic pain is perhaps the most significant malady of western society, with over 100 million Americans suffering from some form of chronic pain.11,12 An understanding of the biochemical mechanisms which regulate EOP levels in the central nervous system (CNS),13 the interconnected nature of the opioid and other neurotransmitter systems, 14,15 and the specific regions of the CNS where EOPs dynamically respond to physical and pharmacological stimuli are needed in the continuing effort to develop better and safer analgesics for the treatment of chronic pain conditions.8,16

A variety of approaches have been used in attempts to measure EOPs.17–21 However, due to their low concentrations (< 1 nM) in most tissues,22 only a handful of techniques are appropriately suited for quantification of dynamic changes in extracellular EOP levels.8

Microdialysis is a relatively noninvasive way to collect EOPs from the CNS, in that it is non-lethal (compared to techniques such as radio immunoassay), and allows sampling at time intervals relevant to physiological studies (5-30 minutes).23,24 The development of

217 nano LC-MSn represents a relatively new approach,25,26 whereas radio immunoassay

(RIA) has seen a more widespread use for dynamic monitoring of EOPs and other neuropeptides over the past 4 decades.17,23,27,28 Despite the utility of RIA as a bioanalytical tool, mass spectrometry monitoring of neuropeptides in microdialysate has seen much development in the past decade due to the promise of unsurpassed chemical resolution and neuropeptide structural information.29–33 Only a small subset of relevant works report in vivo microdialysis in mammalian models with mass spectrometry quantification of

EOPs.33–37 Studies of dynamic enkephalin release employing nano LC-MSn have been successfully demonstrated in rat striatal tissue in part because of relatively high concentrations of EOPs in the region38 and the expertise of the experimentalists who have pioneered this technique.39 However, it has become increasingly clear that the perception and experience of pain and pain relief is related to EOP signaling in cortico-limbic and reward pathways linking the anterior cingulate cortex (ACC) and subregions of the striatum.4,16,40,41 The need to extend dynamic EOP monitoring to the cortex leads to additional analytical challenges arising from the smaller cross-section of tissue accessed by the microdialysis probe, more moderate distributions of opioid receptors and rapid clearance of low concentrations of EOPs by endogenous peptidases.42,43

In this work, we report dynamic measurements of the EOPs methionine-enkephalin (ME), leucine-enkephalin (LE) and endomorphin II (EM II) obtained via microdialysis sampling in the rat ACC with nano LC-MSn quantification. Our utilization of an entirely commercial nano LC-MS system with chip-based electrospray ionization (ESI chip) has reduced EOP detection limits (LE: 1.5 amol; ME 2.3 amol; EM II: 75 amol) to below those previously reported using hand-fabricated ESI probes.33,34 The microdialysis system was modified to include an on-line preservation system that minimized peptide degradation after initial recoveries of ME and LE were observed below quantifiable limits (< 1 pM). We identify

218 post-dialysis enzymatic activity as a leading factor by studying the in vitro degradation of

EOPs in rat cerebral spinal fluid (CSF). Enzymatic degradation is minimized with high concentration acetic acid (HAc), and the resulting enkephalin signal increased due to online preservation of EOPs. Improved quantification was achieved through the addition of [DAla2,DLeu5]-enkephalin (dAdLE) as an internal standard analog through the online- preservation apparatus. The collective advances allowed us to study the dynamics of ME and LE levels in the ACC of awake and freely moving rats in response to exocytotic stimulation.

B.2 Materials and Methods

B.2.1 Chemicals All chemicals used to perform this work were purchased from Sigma

Aldrich (St. Louis, MO) unless otherwise indicated. All peptides were acquired from

American Peptide Company (Sunnyvale, CA). Peptide standards were of 98% purity.

B.2.2 Statistical analysis & validation criteria Statistical tests were selected in accordance with the design of experiment for each of the studies reported in this work. All statistical calculations were facilitated using GraphPad Prism 6.1 software (GraphPad

Software, Inc. La Jolla, CA) unless otherwise noted. Error is reported as s.e.m. throughout this work, unless otherwise noted. Instrument detection limits were established by determining the standard deviation of signals for each analyte from blank injections during multiple experiments (n = 8) over 3 separate days. The minimum statistically detectable signal was determined using the one-tailed t-value for 7 degrees of freedom at alpha =

0.01, and calibrated using the minimum standard signal corresponding to 1 pM concentration. For all other statistical tests, significance was defined as alpha = 0.05.

Comparison of baseline levels of ME and LE with and without online preservation were performed by unpaired t-test (variance of populations not assumed to be equal) with the

Holme-Sidak correction for multiple comparisons, with n = 3. Correlation of ME and MSE

219 data was performed in Microsoft Excel using the Pearson product-moment correlation coefficient. For degradation studies of ME and LE in serum and CSF two-way repeated- measures ANOVA with Tukey multiple-comparisons posttest were performed with n = 3 replicate analyses in each set. In vivo endogenous opioid data were compared via two- way repeated-measures ANOVA with Holme-Sidak post-test to assess the effect of stimulation on the extracellular concentration of EOPs at each time point. Baseline enkephalin levels were measured as an average over three 20-minute fractions. The total

ME and LE response factors were calculated relative to the signal from dAdLE added to the samples via online calibration. Results are reported as percent change from baseline based on normalization of these response factors to the first baseline fraction collected.

Data sets varied in number of animals due to the lack of MSE and EM II analysis in the instrumental method for some early experiments in which only LE and ME were studied, thus for K+ stimulation data sets, n = 7, 6, and 3 for LE, ME, and EN II, respectively. All vehicle data is n = 4. ME levels are reported based on the relative change total abundance of methionine enkephalin species (sum of ME and MSE levels). Validation of in vivo experiment included the following criteria: instrument calibration sensitivity in terms of MS3 extracted ion chromatogram integrated peak area (AU) fell between 750 and 3000 AU for

ME and LE and 75 and 300 AU for EM II, collected sample volume was within the 10 - 20

µL, rat behavior was not anomalous (i.e. erratic spinning and/or apparent neurologic damage), individual microdialysis probe recoveries for ME and LE were measured to be between 1.0 – 5.0 % and 2.0 – 10.0 % respectively in a post-experiment calibration (vide supra). Experiments failing to meet these criteria were not included in statistical analysis or reported in this work.

B.2.3 Animals Male Sprague-Dawley rats (300 - 325 g, Harlan Laboratories Inc.,

Indianapolis, IN, USA) were housed in a climate-controlled room on a standard 12-hour

220 light/dark cycle. Food and water were available ad libitum. All experiments were performed in accordance to policies and procedures set forth by the International

Association for the Study of Pain and the National Institutes of Health guidelines for the handling and use of laboratory animals. Approval was obtained from the Institutional

Animal Care and Use Committee of the University of Arizona prior to all experimentation.

Every effort was made to minimize animal pain and distress as well as to minimize the number of animals used.

B.2.4 Intracranial (ACC) cannula implantation for microdialysis Implantation of probes was modified from the previous procedure.40 Animals were anesthetized using ketamine/xylazine injection (80/12 mg/kg, IP). Ketamine was obtained from Western

Medical Supply, Arcadia, CA. They were mounted in a stereotaxic apparatus (Model 902,

David Kopf Instruments, Tujunga CA). Holes were drilled in the skull above the ACC and a 26 AWG guide cannula (AZ-08; Eicom, San Diego, CA) was implanted into the left ACC

(anteroposterior (AP), bregma +2.6 mm; mediolateral (ML), midline +0.8 mm; dorsoventral

(DV), skull -2.0 mm). The cannula was fixed into place with dental acrylic cement (BASi,

West Lafayette, IN). Stainless steel plugs were inserted in each guide cannula to keep them free of debris. Following surgery, all animals were housed individually and were allowed to recover for 7 - 9 days prior to experimentation.

B.2.5 CSF collection from cisterna magna for in vitro experiments CSF collection was performed as previously reported.44 Briefly, two inch segments of PE-60 tubing were prepared (Scientific Commodities Inc., Lake Havasu, AZ, USA). Needle tips were removed from 23 AWG syringes needles (BD Precision Glide, Franklin Lakes, NJ, USA).

Catheters were fashioned using needles, tubing, and a gel loading pipette tip (Fisher

Scientific, Pittsburgh, PA, USA). These were assembled using super glue and the adhesive was allowed to dry overnight. On the day of the collection rats were anesthetized

221 with isoflurane (2% in air, 2 L/min) and placed in a stereotaxic frame. A 1.5 cm longitudinal incision from the back ridge of the skull to C1 was made and the muscles were retracted to expose the atlanto-occipital membrane. A prepared catheter and micropipette was used to puncture the membrane and collect the CSF (70-150 µL), free of blood, from the cisterna magna.

B.2.6 In vivo microdialysis procedure Microdialysis experiments were done in awake and freely moving animals. Microdialysis probes (AZ-8; Eicom) were inserted into the guide cannula so the 2.0 mm semi-permeable membrane protrudes from the guide into the ACC. The microdialysis probe was perfused with aCSF (147.0 mM NaCl, 2.8 mM KCl,

1.2 mM MgCl2 and 1.2 mM CaCl2) at a rate of 0.5 μL/min using a gastight syringe, syringe pump drive and hive syringe pump controller (MDN-0250, MD-1001, MDN-1020; BASi).

Following a ~120-minute washout period, 20-minute fractions were collected: three baselines fractions, one high KCl infusion fraction, and 3 post-stimulation fractions were collected into pre-chilled (4 °C) 0.5 mL siliconized eppendorf tubes. The aCSF used for high KCl was composed of 79.8 mM NaCl, 70 mM KCl, 1.2 mM MgCl2, and 1.2 mM CaCl2.

Infusion was started 10 minutes prior to the designated 20-minute high-KCl stimulation period to account for the time needed for the sample to move through the microdialysis set-up tubing. High KCl was perfused for a total of 30 minutes through the microdialysis probe into the ACC. Samples were immediately frozen on dry ice as soon as the total volume was collected. Following the completion of in vivo measurements, the microdialysis probe was flushed with aCSF for 1 hour and then subjected to a post- calibration routine. A solution containing 700 pM LE and ME as well as 7 nM EM II in water was used to evaluate the in vitro recovery of peptides and to confirm the function of the probe. Occasionally the probe membrane becomes blocked during an experiment, so this step was performed to verify the functionality of the probe and to check that the in vitro

222 recovery for that particular probe is in agreement with the expected values. Following each in vivo microdialysis experiment, rats were euthanized with CO2 overdose and their brains were collected in 10% formalin fixative. Coronal brain sections (30 μm thick) were cut using a Microm HM 525 cryostat and the correct ACC location of the implanted cannulas was verified.

B.2.7 Online-preservation system An online preservation system was developed to increase the recovery of enkephalins (Figure B.3a & b). It consists of a micro-tee connector (CMAP000043; CMA Microdialysis/Harvard Apparatus, Holliston, MA), FEP tubing (840 9501; CMA Microdialysis), and tubing adaptors (340 9500; CMA

Microdialysis). To allow free movement of animals while minimizing the risk of fluid lines becoming tangled or stretched a multichannel liquid swivel was incorporated (CMA

Microdialysis). On-line preservation fluid consists of aqueous 10% (v/v) acetic acid, 20

µM bestatin, 1 µM thiorphan, 50 µM D,L-methionine, and 2% (v/v) acetonitrile as well as the dAdLE at 50 pM. Bestatin is a protease inhibitor, and thiorphan is an enkephalinase inhibitor. These compounds were added as an additional weapon against peptidase activity in the collected fractions. D,L-methionine was added to sacrificially combat the effect of oxidants on ME. Online-preservation occurs after the dialysate is recovered from the implanted microdialysis probe so there in no concern of the solution diffusing across the microdialysis membrane and having a deleterious effect on the animal or influence the integrity of the experiment (Figure B.3c). Given the low flow rates for the microdialysis experiment (0.5 µL/min), the preservation solution was introduced at the same flow rate.

The introduction of the preservation fluid effectively dilutes the dialysate by a factor of two, but it also increases the rate with which dialysate is collected by a factor of two. Temporal resolution was set at 20 minutes, wherein 20 µL were collected (10 µL dialysate and 10

µL preservation fluid). Of the 20 µL, 10 µL were prepared and injected into the instrument,

223 leaving 10 µL in case of sample preparation or instrument failure. The apparatus has the disadvantage of being bulkier than a microdialysis probe as it has an additional inlet line.

However, problems associated with this were easily resolved by introducing an automated swivel mechanism that promoted unencumbered movement of the rats during experimentation (Figure B.3a).

B.2.8 Sample preparation for nano LC-MS3 analysis Microdialysate was frozen and left on dry ice immediately after sample was collected. When all the samples were collected, they were stored at – 80 °C until analysis (< 24 hours from collection). Fractions were thawed at room temperature in batches of 4 and were desalted using a C18 ZipTip®

(ZTC18S096; EMD Millipore, Billerica, MA). The ZipTip® was briefly equilibrated with three 10 µL volumes of acetonitrile (ACN) followed by three 10 µL volumes of aqueous

0.1% triflouroacetic acid (TFA). A 10 µL volume of sample was drawn up and dispensed

10 times to load peptides onto the C18 resin. The sample is then washed with two volumes of 0.1% TFA and then eluted in 10 µL of 60% ACN, 40% water with 0.1% TFA (v/v). The sample is dried to less than 1 µL using a SpeedVac concentrator (Savant, Inc. USA), and reconstituted in 10 µL 0.1 % TFA to ensure proper loading on the capillary chromatography system.

B.2.9 Nano LC-MS3 analysis of opioids Samples are separated using a Proxeon Easy- nLC II (ThermoFisher Scientific, Waltham, MA) with a 2.0 cm pre-column (Easy-column,

ID 100 µm, 5 µm particle, C18-A1) and a 10 cm analytical column (Easy-column ID 75 µm,

3 µm particle, C18-A2), with 10 µL sample injections. The gradient program written for the separation includes a series of wash steps to minimize carryover. The mobile phases used are A: 0.1% v/v formic acid in H2O and B: 0.1% v/v formic acid in acetonitrile. The pre-column is equilibrated with 10 µL solvent A, and the analytical column is equilibrated with 6 µL solvent A. The gradient begins with 5% B and ramps to 25% B over 3 minutes.

224 It then ramps from 25% B to 95% B over 4 minutes, where it holds for 4 minutes. The wash steps begin with a sudden drop back to 5% B held for 2 minutes, a jump to 95% B for 2 minutes, another drop to 5% for 2 minutes, and a final jump to 60% B for 1 minute.

The wash portion of the gradient program prevents carry-over from the column, and the autosampler is washed with 100 µL of solvent A after injection. Electrospray ionization of the samples was achieved using a Triversa Nanomate chip-based ESI system (Advion,

Inc. Ithaca, NY). An Orbitrap Velos Pro hybrid ion-trap-orbitrap mass spectrometer

(ThermoFisher Scientific, Waltham, MA) was used for quantification. Mass analysis is conducted in the LIT with radial ejection of ions for sensitive detection. Tandem MS is carried out with two isolation-and-fragmentation steps (MS3) for all enkephalin species.

LE and ME are analyzed through initial ion-trap isolation of m/z 556 and 574, respectively and then fragmented at a collision energy sufficient to produce a maximum intensity for

+ + the a4 ion (m/z 397). The a4 ion is then isolated and fragmented to produce a characteristic MS3 spectrum which is dominated by the fragment ions with m/z 380, 323,

279 (Table B.1). Despite the fragments of ME and LE being the same, the method of data collection does allow for distinction between the two species due to the difference in the original parent molecule selected for fragmentation. Fragmentation pathways are analogous for MSE and for dAdLE and slightly different for the tetrapeptide EM II.

B.2.10 Flow-injection MS study of opioid degradation To study the factors affecting the loss of opioid sample material in biological samples collected from rodent CSF, a separation-free injection system was devised for an electrospray ionization quadrupole ion-trap mass spectrometer (LCQ, Thermo Finnigan LLC, San Jose, CA). A bolus of sample material containing opioid peptides within CSF were introduced via a six port valve with fluid flow delivered via a syringe pump (KD Scientific). The sample was introduced in 10 microliter injection volumes, at a flow rate of 2.0 mL min-1, after the sample treatment

225 period had concluded. Enkephalins (LE and ME, 1.0 µM) were incubated in serum, aCSF, or CSF with or without treatment at ambient temperature to mimic the conditions of microdialysis collections for times varying from one minute to 60 minutes. After samples had been incubated for the prescribed amount of time they were prepared for mass spectrometry analysis by withdrawing 10 microliters of solution and spiking with 1 microliter of a 10 µM solution of dAdLE in 50% HAc and subjecting them to C18 ZipTip® treatment as described previously. These solutions, once eluted from the ZT, were diluted to 100 µL in 50:50 acetonitrile/water with 0.1% formic acid. Tandem mass spectrometry analysis (MS3) was conducted during sample introduction to yield specific, quantitative signals proportional to the remaining enkephalin concentrations in the samples at each time point. CSF was collected from 5 rats and stored on ice for same-day analysis by DI-

MS. Artificial cerebral spinal fluid (aCSF) consisting of the un-buffered isotonic solution used for microdialysis perfusion, was used as an enzyme-free control.

B.3 Results and Discussion

B.3.1 Chip-based ESI for improved nano LC-MS3 quantification of endogenous opioids A successful strategy for quantification of EOPs developed by Kennedy et al.34 was adapted for the study presented here. This method can be reduced to three important steps 1) microdialysis recovery of peptides, 2) nano-LC separation of small volume (< 10

µL) samples, 3) electrospray-ionization and tandem mass-spectrometry which allows identification and quantification of neuropeptides (Figure B.1a). Although this method is promising, particularly for measurements of enkephalins and dynorphin fragments, it has seen little adaptation by other laboratories for dynamic monitoring of neuropeptides in vivo.

Additionally, it has been applied to only a few brain regions such as the striatum. This is due to the inherent difficulty in making measurements of only a few attomoles of peptides and also because custom separation capillaries and electrospray emitters are often used

226

Figure B.1 Analysis of endogenous opioid peptides. a) EOP measurements are achieved using in vivo microdialysis with nano LC-MSn identification & quantification. b)

This method employs advanced robotic sample delivery and chip-based ESI technology to increase the sensitivity of mass spectrometry and increase reproducibility, thus reducing error. Micrographs of ESI Chip nozzles adapted with permission from Advion, Inc. c) The

MS3 spectra of leucine enkephalin (LE) and the internal standard dAdLE are shown with ion transitions labeled. d) This technique is highly quantitative with a linear dynamic range

(LDR) for LE extending from 1 - 70 pM.

227 and require expertise that is not easily transferred to other laboratories.8,32,34 Interestingly, the combination of these systems utilizing a 75 µm diameter C18 column demonstrated detection limits matching or exceeding those displayed by custom 25 µm diameter columns.34 This suggests that increased ionization efficiency afforded by the ESI chip overcomes the reported advantage in decreased column diameter and reduces the difficulty associated with hand-fabricated instrumentation. It would be interesting to see a combination of 25 µm diameter columns combined with ESI-chip technology for the potential of zeptomole detection limits in nano LC-MSn.

Several key changes were introduced to expand the applicability of the methodology by exploiting commercially available microdialysis, nano-LC, and chip-based ESI technology.

To deal with the small implantation site within the ACC and to minimize disruption of tissue, smaller probes with a cuprophane membrane were used (Figure B.3c). Perhaps most significantly, the linear ion trap (LIT) mass spectrometer was interfaced to a commercial nano-LC system through an ESI chip coupler (Figure B.1b). This adaptation of nano LC-

MS3 quantitation to an entirely commercial based system offers the possibility to greatly expand the use of this methodology for the quantitative study of neurotransmitters in vivo.

Several researchers introduce agents into the microdialysis perfusate to improve recovery or to study a wider variety of analytes at increasingly low concentrations.45,46 A good example is the addition of ascorbic acid as an antioxidant which is already present in the

CNS at concentrations between 100 – 500 µM.47 The addition of anti-inflammatory agents to the dialysate can also work to mitigate the impact of microdialysis probe implantation, albeit at the cost of chemically perturbing the CNS.48 This can have the distinct disadvantages of reverse dialysis of added chemicals into the CNS which can cause physiological disruption or injury to the animal. In this case, a perfusate containing 5%

HAc would certainly be dangerous to the animal. A similar argument can be made about

228 the introduction of peptidase inhibitors such as bestatin or thiorphan, which would affect enkephalin metabolism, or the addition of dAdLE which has potent opioid receptor binding properties.18,49

Due to the inherent variability that arises from sample handling and mass spectrometric analysis, we decided to use an internal standard and calculate a normalized response factor that would account for much of this variance. The synthetic enkephalin dAdLE was chosen due to its similarity to both ME and LE in its structure and gas-phase chemistry and is allowed through our use of online preservation to prevent differentiated peptidase activity between dAdLE and EOPs (Figure B.1c). The MS3 fragmentation spectra of ME and LE have been previously presented for quantification.34 The fragmentation sequence for dAdLE has analogous gas-phase behavior where fragment ions relevant to quantification have a +14 Da shift in m/z relative to the endogenous enkephalin fragment due to the D-alanine residue replacing glycine (Table B.1). While the addition of dAdLE as an internal standard proved extremely valuable for relative quantification of EOPs in microdialysis fractions (vide infra), external calibration of enkephalins showed a robust

2 linear range as represented in Figure B.1d (linear dynamic range for LE 1.0 - 70 pM, R =

0.9863, n = 3 replicates at each level over a 24 hours). This compares favorably with results from other nano LC-MSn systems.33,34,37 This online-preservation apparatus represents an important modification to existing microdialysis strategies by including a commercially available mixing tee directly after the microdialysis probe so online preservation fluid can be introduced rapidly after EOPs cross the membrane (0.94 µL dead volume, and 1.87 minutes transit time). This results in an 8 - 20 fold increase in signal for

ME and LE, despite the 2x dilution factor, and allows EOPs from the ACC to be quantified using a commercial nano-LC-MS3 system. Furthermore, dAdLE was present in all samples, indicating the successful introduction of the online preservation fluid, as well as

229 Table B.1 Descriptive summary of specific ion transitions involved in MS3 quantification of enkephalins. All values have units m/z. The symmetry between the ion transition results from the high-degree of similarity in the gas-phase collision-induced

2 fragmentation of each of the species. For ME, LE, and MSE the MS fragmentation

+ results in the a4 ion with the same structure at m/z 397 owing to the loss of the c-terminal amino acid residue upon fragmentation and subsequent spontaneous loss of CO. For

DADLE, the internal standard enkephalin analog, the m/z of the listed ions in the MS2 and MS3 transition is shifted in mass by 14 Da, corresponding to the replacement of the

+ Gly residue with D-Ala. † EM II initially fragments to form the b3 ion (408 m/z).

Description MSE ME LE EM II DADLE

MS (M+H)+ 590 574 556 572 570

2 + † MS a4 397 397 397 408 411

234, 278, 234, 278, 234, 278, 233, 245 248, 292, MS3 fragments 295, 323, 295, 323, 295, 323, 261, 311, 309, 337, 380 380 380 339, 381 394

230 allowing normalization of peptide signals for improved relative quantification despite high variance arising in animal experiments.

B.3.2 Initial in vivo measurements in the ACC and degradation of opioid peptides

As pain is a phenomenon of the conscious mind, we designed an experimental setup to perform microdialysis in awake and freely moving rats. Initial studies were performed on

ME and LE only, until the methodology was optimized, at which time the analysis was expanded to include methionine enkephalin sulfoxide (the oxidation product of ME), endomorphin II, and dynorphin A 1-8. The repertoire of EOPs and metabolites monitored was expanded to include methionine sulphoxide enkephalin, as it was present at quantifiable levels, despite efforts to minimize oxidation. We believe that the sulphoxide arises primarily during sample evaporation and reconstitution following desalting based upon the variability in the relative amounts of ME to MSE across individual samples. That a portion of the MSE measured could arise endogenously cannot be ruled out, and would require a more efficient prevention of methionine residue oxidation. Interestingly the potent µ-opioid agonist EM II was observed and quantified at ~ 1 nM levels in the ACC while EM I was not observed. Previously endomorphins I and II had been quantified post mortem in both bovine and human brains, but the difficulty in identifying their origin in the genome has some to question if these peptides are a function of the measurement process.3,50 This is the first in vivo microdialysis report employing mass spectrometry monitoring to measure an endomorphin in the rat brain. This claim is substantiated by repeated measurements, wherein the MS3 spectra matches the EM II standard, as it occurs within 2 seconds of the retention time of the standard, and the fragmentation spectrum matches the standard. An earlier report utilizing push-pull perfusion and HPLC with electrochemical detection reported EM II levels in rat spinal cord as 1.3 ± 0.3 nM, supporting both our values and our identification of the peptides in vivo.51 Our finding

231 shows an equivalent level of EM II in the cortical tissue of a living rat, which challenges the hypothesis that EM I is the dominant endomorphin in the brain, whereas EM II is dominant in the spinal cord.52 Relatively short-duration stimulation (30 minute) with 70 mM K+ through the dialysis membrane caused both ME and LE enkephalin levels to increase significantly (p < 0.0200) to 172 ± 42 % and 165 ± 23 % of baseline, respectively.

Despite the extremely low instrument detection limits (LE: 1.5 amol, ME: 2.3 amol) and the known distribution of µ-opioid receptors in the frontal cortex,13 the concentrations of

ME and LE in microdialysis fractions were consistently below the linear dynamic range or undetectable. It has previously been hypothesized that a combination of enzymatic degradation of enkephalins, adsorption, and oxidation (particularly of ME) can lead to low peptide recoveries.34 This seemed a plausible problem given the 50 kDa molecular weight cutoff of the membrane can allow passage of proteins and enzymes, the known oxygen permeability of microdialysis tubing, and addition to the general problem of peptide adsorption during sample handling. While this bore filter is larger than usual, it was chosen due to improved recovery of the target compounds. Given the low flow rates required for maximizing microdialysis relative recoveries (0.5 microliters/min.) there was ~15 minutes between analyte passage across the dialysis membrane and fraction collection. This is enough time for significant sample loss to occur due to enzymatic activity, as suggested by the known half-lives of enkephalins in enzymatically active environments such as blood and CSF.49 While previous studies have shown successful recovery and preservation of enkephalins at measureable levels under other conditions, the goal of these studies was to provide a working methodology for monitoring ME and LE in the ACC of awake and freely moving rats. To preserve the integrity of the experiment we decided to study enzymatic degradation of peptides as it is a significant problem in a variety of neuropeptide measurement (both in vivo and post mortem) and develop a dynamic solution. Flow- injection analysis tandem mass spectrometry (DI-MSn) enabled quantitative monitoring of

232 peptides during enzymatic degradation in rodent serum or CSF. Using a quadrupole ion trap mass spectrometer with MS3 monitoring this separation-free approach allowed analysis of samples with a 2.1-minute period (95% duty cycle). Selective, yet sensitive quantitation over a linear range of 90 nM-1.8 µM for ME and LE with detection limits of 23 and 15 nM respectively was achieved. This allowed the exploration of enzymatic degradation in rat blood serum and systematic determination of the most appropriate online preservation medium for in vivo microdialysis in the ACC (Figure B.2). The technique was first characterized in serum using an immediate extraction and analysis of

LE and ME (Figure B.2a & b). However, this was limited by the time required for sample preparation and analysis (~10 minutes) and necessitated staggered parallel experiments.

To improve the throughput and temporal resolution of the experiment, HAc quenching was performed at the desired time by removal of a portion of serum and spiking to a final concentration of 5% HAc and analyzing the samples 1 hour later (Figure B.2a & b, Ser).

This resulted in rapid enzymatic degradation of enkephalins; the data were fit by a single exponential decay model with serum half-lives for LE and ME determined to be 8.3 and

4.7 minutes respectively (r2 = 0.974 & 0.957). When aliquots of serum were pulled and

HAc was not used, the enkephalin levels had completely degraded in the hour prior to analysis (Figure B.2a & b, Ser + 60 min.). To investigate preemptive preservation, 5%

HAc was added to serum with enkephalins and allowed to incubate for 1 hour.

Degradation was not observed within at least a 45-minute time frame (Figure B.2a & b,

Ser + HAc + 60 min.). Statistical analysis (two-way repeated-measures ANOVA) was used to evaluate the enkephalin degradation data, confirming that the major effect on variance was treatment of the samples (LE: 67% total variance, F3,8 = 975.7, MS = 31061, p < 0.0001; ME: 64% total, variance F3,8 = 641.1, MS = 29595, p < 0.0001). The effect of

HAc preservation on enkephalin concentrations is significant and readily apparent between 5 and 45 minutes (Tukey post-test, see methods). These experiments

233

Figure B.2 Degradation studies of enkephalins in biological fluids by DI-MSn. The stability of Leu and Met-Enkephalin spiked into rodent serum were studied under different treatment conditions (a & b, respectively). Enkephalins were added to rat serum and degradation was allowed to proceed and the remaining enkephalin levels were measured at specific time points (Ser). When peptides were incubated in serum for 60 minutes prior to the study, degradation was already complete before analysis could take place

(Ser + 60 min.). Addition of 5% acetic acid (HAc) to the serum preserved enkephalins by inhibiting degradation during a 60-minute period before beginning measurements (Ser

+ HAc + 60 min.). Carrying out these degradation experiments using cerebral spinal fluid

(CSF) from rats showed that Leu and Met-Enkephalin degradation occurs in CSF over

234 20 minutes (c & d, + CSF). These data were fit with a single exponential decay to highlight the degradation. Addition of 5% HAc prevents the degradation (+CSF + HAc) showing that enkephalin stability is the same as in an isotonic aCSF solution (aCSF) and these data are fit around the line at 100% concentration. All error bars are s.e.m.

Statistical significance is indicated by asterisks when p < 0.05.

235 demonstrate the robust ability of DI-MSn to analyze peptide degradation in rapid (~2 minute), separation-free experiments. Furthermore, the temporal resolution of the degradation study was reduced to under 5 minutes using HAc quenching and allowing the samples to be stored for at least an hour at room temperature without further degradation.

The effect of HAc preservation on enkephalins in sera is dramatic, showing no degradation over 45 minutes of incubation, whereas unquenched, unpreserved samples were almost completely degraded by the time they could be analyzed.

The rates at which enkephalins degrade in CSF directly impacts the monitoring of enkephalins in vivo (Figure B.2c & d). For in vivo microdialysis experiments, degradation to levels below the limits of quantitation (~1 pM) can occur during and after microdialysis collection for many of the same reasons that degradation occurs in serum. Here, we utilize the method developed for serum degradation and compare the effect of CSF with and without HAc preservation on the stability of enkephalins over a one-hour time period at room temperature. These conditions mimic those encountered during the transit of enkephalins through microdialysis tubing before entering chilled fraction collectors. The results indicate clear degradation of both ME and LE in CSF as compared with aCSF (two- way repeated measures ANOVA with Tukey post-test, p < 0.05 at all times after 5 minutes). Degradation data in CSF were fit with a single exponential model (solid red line) to highlight decrease in enkephalin levels. Under these conditions degradation appears to plateau after 30 minutes, preventing extraction of robust fits with the single exponential model; however, the effective time over which the neuropeptides are metabolized is evident. Addition of HAc to CSF preserves ME and LE, showing no statistical difference from the aCSF negative control (two-way repeated measures ANOVA, Tukey post-test,

+CSF vs. aCSF, p > 0.05).

236 B.3.3 Online-preservation of endogenous peptides To achieve preservation through the inhibition of enzymatic activity, to reduce amino acid oxidation, and to minimize adsorption at surfaces, a modification to the traditional microdialysis apparatus was made

(Figure B.3a). A second line infusing a preservation fluid composed of aqueous 10% (v/v) acetic acid, 20 µM bestatin, 1 µM thiorphan, 50 µM D,L-methionine, and 2% (v/v) acetonitrile as well as the dAdLE at 50 pM was coupled to the dialysate flow immediately after the microdialysis probe via a micro-tee union (Figure B.3b). The cause of degradation is uncertain, but believed to be enzymatic, though various components in the preservation solution are targeted to combat the various potential causes of degradation

(enzymes or proteases, oxidation, and adsorption). As the internal standard is introduced with the preservation solution, degradation of the internal standard does not occur.

The effect of on-line preservation on measured enkephalin levels in dialysate was apparent in the magnitude of the enkephalin signals (Figure B.3d). Measured concentrations of LE in the dialysate significantly increased from 0.95 ± 0.24 pM to 7.2 ±

2.2 pM (n = 3 animals, t4=4.772, p = 0.0088). ME increased from 0.29 ± 0.12 pM to 5.9 ±

1.6 pM (n = 3 animals, t4=23.996, p < 0.0001) which is a larger relative increase than for

LE suggesting a more rapid degradation due to a combination of enzyme activity and methionine oxidation. Note that the concentration coming from the microdialysis probe should be 50% lower when a second fluid line is introduced due to the dilution factor.

However, the significant increase we see when using the micro-tee union means that the measured concentrations in standard microdialysis were artificially low due to sample degradation. This result shows that sample degradation is still a limiting factor in endogenous peptide analysis in vivo. While sample degradation could certainly be prevented with the addition of chemical inhibitors introduced into the region, the goal of

237

Figure B.3 Online-preservation apparatus scheme allows for improved microdialysis recoveries from the ACC. The apparatus a) utilizes two syringe pumps

(P1, P2): isotonic aCSF delivered through P1 and preservation fluid through P2.

Because of the additional fluid lines, a multi-fluid-line swivel (S) was used. A micro-tee adaptor (T) is placed after the microdialysis probe (µ) to combine the preservation fluid with the dialysate to achieve preservation before the sample travels through the fraction collector (F). b) A photograph of the online preservation micro-T and dialysis probe are shown. c) The implantation site of the microdialysis probe (red highlight) is shown with stereotaxic coordinate on the sagittal (left) and coronal (center background) slice diagrams. d) The result of online preservation was a large increase in measured concentration, presumably due to the inhibition of enzyme activity as is evidenced in figure 2 (n = 3 animals, error bars s.e.m., asterisks denote p < 0.05).

238 these experiments is to interrogate the system with as little external disruption to the natural processes and environment as possible.

B.3.4 Online-preservation microdialysis allowed quantifiable measurement of EOPs in the ACC Using the online-preservation technique, the levels of ME and LE were quantifiable in ACC using nano LC-MSn for the first time (Figure B.4). The structures of these peptides are shown with key structural differences highlighted (Figure B.4a, highlighted chemical structures). The peptides eluted in order of increasing hydrophobicity, as expected in reverse-phase chromatography (Figure B.4b). Crucially, these peptides could be both identified and quantified based on the spectral fingerprint.

MS3 spectra from in vivo microdialysis samples of each peptide are shown with the corresponding ion transitions leading to the collected spectra (Figure B.4c). An attempt was made to quantify dynorphin A 1-8, which has been previously quantified by nano LC-

MSn in striatal tissue, though it was not detected in numerous samples and these negative results were not included in this report.

B.3.5 Identification and quantification of methionine enkephalin sulphoxide As the thiol of ME is susceptible to oxidation, further experiments were tailored to account for the possibility that ME levels were falsely low due to oxidation. Oxidation of the methionine residue in ME results in methionine sulfoxide enkephalin (MSE). This species undergoes gas-phase fragmentation analogous to that of ME and LE (Figure B.4c, top spectra). The amount of MSE was observed to have a significant positive correlation with ME over a linear range of ~1 pM to ~100 pM (n = 3, r = 0.999986, p = 0.0034). MSE was observed to account for measureable amounts of the total methionine-enkephalin-related signal observed for in vivo samples. It was found that summing the absolute signals resulted in more representative relative quantification.

239

Figure B.4 Representative nano LC-MS3 data from in vivo microdialysis sampling of the rat ACC. Four endogenous species were monitored and detected from the animals: 1 methionine sulphoxide enkephalin (MSE), 2 methionine enkephalin (ME), 3 leucine enkephalin (LE), and 4 endomorphin II (EM II). The structures of these peptides are shown (a) with key structural differences highlighted (colored regions). The internal standard, 5 dAdLE is continuously present in the online preservation fluid. b) A baseline

Extracted Ion Chromatogram is shown for each of the five peptides involved in quantification. The grey shaded areas under the peaks represent the integrated region for quantification. c) Mass spectra from in vivo samples corresponding to the shaded area confirm the identity of the analyte being quantified. The corresponding ion transitions are shown to the right of the MS3 spectrum for each peptide and the m/z of the specific fragment ions used to quantify each neuropeptide are labeled in the mass spectrum.

240 B.3.6 Endomorphin II is present in dialysate from the ACC To expand the number of peptides being quantified in the ACC beyond ME, LE, and MSE the tetrapeptides EM I and EM II were investigated. Previously endomorphins I and II, endogenous peptides with morphine-like activity, had been quantified post mortem in both bovine and human brains, but the difficulty in identifying their origin in the genome has some to question if these peptides are a function of the measurement process.3,50 This is the first in vivo microdialysis report employing mass spectrometry monitoring to measure an endomorphin in the rat brain. An MS3 method was developed using the following ion transitions: EM I 611 → 447 → 233, 260, 284. 429; EM II 572 → 408→ 233, 261, 311,

339, 381 (Figure B.4c, bottom spectra). Instrument detection limits for EM I and EM II were 10 and 7.5 pM, respectively. EM II was quantified in three out of seven animals that underwent 70 mM K+ stimulation, and all control (Table B.2). The level of EM II did not change significantly upon K+ stimulation (Figure B.5c). Measurable EM I was not observed in any of the animals investigated.

B.3.7 Stimulated release of enkephalins in the ACC The release of enkephalins was observed in response to perfusion of 70 mM K+ solution through the microdialysis probe

(Figure B.5). The stimulation took place over 30 minutes with the high K+ solution arriving at the fraction collector between 0 and 30 minutes on the experimental timescale (Figure

B.5 a, b, & c, grey bar). For normalized LE concentrations the interaction between time and stimulating treatment was observed to have a significant effect (F5,45 = 3.093, MS =

1258 p = 0.0175). This can be clearly attributed to the LE levels increasing to 165 ± 23 % of baseline upon high K+ stimulation as compared with vehicle (n = 7 animals, Holme-

Sidak post-test, t54 = 3.068, p = 0.0200). Similarly, normalized ME concentrations increased with a significant effect from the stimulation (F1,8 = 8.277, MS = 65724, p =

0.0206). Subject matching over time was also found to be significant (F8,40 = 2.709, MS =

241 Table B.2 Relevant amount & concentration values for each of the opioid peptides monitored in this study. Instrument detection limits (IDL) were calculated in terms of both total amount (in amol) and loaded sample concentration

(pM). Both LE and ME levels were measured on-column within the linear range of

1-100 pM. In vitro probe recoveries were used to estimate in vivo concentration in the anterior cingulate cortex. Note that the ME concentrations reported were a result of the sum of the total ME and MSE signals. The in vivo enkephalin levels are estimated in the low-to-mid pM range with ME being the more abundant of the two.

The EM II concentration was measured near the detection limit on-column.

Combined with the relatively low microdialysis recovery the estimate concentration in vivo is estimated at 1 nM. All errors are s.e.m.

LE ME EM II

Mass IDL 1.5 2.3 75 (amol)

Conc. IDL 0.150 0.230 7.5 (pM)

Measured Dialysate 4.4 ± 1.3 3.8 ± 1.2 19 ± 8.0 Conc. (pM)

In vitro probe 9.6 ± 1.1 2.4 ± 0.4 1.4 ± 0.3 recoveries (%)

ACC Estimate 46 ± 14 160 ± 50 1400 ± 600 Conc. (pM)

242 Figure B.5 Stimulated release of LE and ME. Stimulation of the ACC by reverse- microdialysis of 70 mM K+ solution for 30 min (grey bar in a-c). a) LE shows significant release in the first fraction collected after stimulation (n = 7 animals) compared to vehicle (n = 4 animals, two-way ANOVA with Holme-Sidak post-test, p = 0.0200). b)

ME shows similar results (n = 6 animals stim, n = 4 animals vehicle, two-way ANOVA,

Holme-Sidak post-test, p = 0.0109). c) EM II was observed in several animals (n = 3 animals stimulation, n = 4 vehicle) but did not respond to the secretagogue. Error bars are s.e.m. Asterisk denotes statistical significance, p < 0.05.

243 7941, p 0.0175). In the first fraction following stimulation, ME significantly increased in to

172 ± 42 % of baseline (n = 6 animals, s.e.m.) as compared with vehicle (Holme-Sidak post-test, t48 = 3.301, p = 0.0109). The following fraction showed near statistical significance for ME concentration as it increased to 170 ± 39 % of baseline (t48 = 2.729, p

= 0.0520, error as s.e.m.). Six of the seven animals considered in this study were included in statistical analysis of total ME levels due to instrumentation problems resulting in zero- value measurements for baseline and post-stimulation fractions. These results indicate that, in response to exocytotic stimulus, both ME and LE are released from cells in the

ACC of rats and that the relative concentrations of these opioid neuropeptides can be monitored by microdialysis with online preservation and calibration. Levels of these peptides begin to decline within 20 minutes of cessation of the high K+ stimulus (which is the temporal resolution of this experiment), indicating rapid enkephalin metabolism in the extracellular space.

B.4 Conclusions

Continuing advancement in dynamic in vivo monitoring of EOPs and other neuropeptides is fraught with challenges arising from the inherently low concentrations of these species in the extracellular space (< 1 nM) and the rapid time-frame (< 10 minutes) in which multiple pathways for peptide degradation can occur (e.g. enzymatic activity, oxidation, adsorption). We have adapted a methodology for nano LC-MS3 quantification of enkephalins and other EOPs. We used a commercially available system in contrast to previous studies where custom nano LC instrumentation (hand-fabricated capillary LC columns and ESI emitters) were used.34,45,26 Furthermore the nature of the instrumentation allows instrument detection limits for leucine and methionine enkephalin to be reduced to ultra-low levels of 1.5 amol for LE, 2.3 amol for ME, 75 amol for EM II.

244 To accurately understand neuropeptide action in vivo, the physiological system should be perturbed as little as possible, both chemically and physically. Given these concerns, and the confirmation of our hypothesis that enzymatic activity from the CSF causes degradation of enkephalins, a strategy to achieve online-preservation and calibration of enkephalins was devised.

The results indicate that ME and LE likely act as neurotransmitters in the ACC and they demonstrate the feasibility of studying dynamic changes in enkephalin EOPs in the ACC as a result of a pharmacological or behavioral challenge. The on-column concentrations could be measured directly for each experiment using an external calibration and the measured microdialysis probe recoveries, estimates of in vivo concentration can be made

(Table B.2). These estimated concentrations in the ACC (LE: 46 pM; ME: 160 pM) are similar to those reported in parts of the striatum which is not surprising given the relative density of µ-opioid receptors.24,51,53 All three compounds (ME, LE, and EM II) have affinities for the µ-opioid receptors, with Ki values of 25.2, 27.7, and 0.69 nM, respectively.54 The relative abundance of ME to LE is ~ 3.5 in vivo, which approaches the

5:1 ratio of ME to LE sequences encoded by the proenkephalin gene. EM II levels did not significantly change despite the use of a known secretagogue, which may serve as additional evidence that endomorphins originate outside of the CNS or are not synthesized and secreted as other endogenous opioid neuropeptides.55 Taken together the methodological innovation presented in this work have allowed for dynamic and direct

EOP measurements to be expanded to the frontal cortex of rats where studies suggest an important role for neuropeptides in the perception and relief of chronic pain. The use of smaller probes (2 mm) and the use of our online-preservation technique allow for sensitive measurements to be made in smaller brain regions than was previously possible. This method may also be applicable to other brain regions, and the mapping of dynamic action

245 of enkephalins and other EOPs in vivo is a necessary step in translating study in animal models to comprehensive clinical treatments.

B.5 Author Contributions

NDL prepared all figures and wrote the manuscript, designed the new methods, and performed statistical analysis of the data. NDL & CLK performed the chromatography and mass spectrometry and maintained the instruments. CLK was in charge of revisions of the manuscript. DSK and EL performed animal procedures and microdialysis collection.

DSK & EN contributed to the statistical analysis and design of experiments. EN provided expertise in neurochemical matters and wrote a portion of the introduction and methods.

B.6 Acknowledgements

We wish to thank Janice Oyarzo for assisting with her expertise in animal surgical procedues. The service of Linda Breci and George Tsaprailis, who direct the Arizona

Proteomics Consortium, was crucial for training and access to nano LC-MS instrumentation. Scott Derigne provided engineering expertise and support in maintaining and repairing the mass spectrometers. Daniel Eikel of Advion Inc. is acknowledged for providing permissions and micrographs and of the ESI-Chip. Funding for the instrumentation was provided through NIH/NCRR 1 S10 RR028868-01. Support for researchers and research activities was available through NIH DA35425 (MLH) and NIH-

NIDA DA034975 (FP).

246 APPENDIX C: Compound structures, mass spectra, and chromatograms

Figure C.1 SAM-995 spectra. a) Structure of SAM-995, sequence YTGFLS-NH2. b)

Full mass spectrum. The main peak is the protonated peak, and the sodiated peak is also identified. c) The protonated parent peak, 686.35, was selected and fragmented to collect the MS2 spectrum. MS3 spectra were collected for the selection and subsequent fragmentation of several peaks in the MS2. Selection and fragmentation of

+ 686.35 followed by the 582.29 [b5+H] ion yielded the spectrum seen in panel d.

3 + Selection and fragmentation of MS of 686.35 followed by 651.31 [M-NH2-H2O+H] yielded the spectrum seen in panel e. It was determined that the spectrum shown in panel d is more optimal for reproducible quantification. The peaks shown in red were used for quantification.

247

Figure C.2 MMP-2200 spectra. a) Full mass spectrum. The main peak is the protonated peak, and the sodiated peak is also identified. The structure of MMP is inset. b) The protonated parent peak, 1010.46, was selected and fragmented to collect the MS2 spectrum. c) The MS3 spectrum was collected by selecting and subsequently fragmenting the 1010.46 peak followed by the 686.35 peak [M-Lac+H]+. The MS3 peaks used for quantification are highlighted in red.

248

Figure C.3 LC gradient for SAM-995, MMP2200, and six Angiotensin 1-7 derivatives. Gradient runs from 0-37 minutes, though data is only collected from minutes 0-25. The remaining gradient is used to wash any remaining material off the column to minimize carryover between runs. Solvent A is H2O with 0.1% trifluoroacetic acid (v/v), and Solvent B is acetonitrile with 0.1% trifluoroacetic acid (v/v).

249

Figure C.4 Chromatograms and MS3 spectra for SAM-995 and MMP2200. a) SAM-

995 elutes at 16.25 minutes. b) The MS3 spectrum for SAM-995, with the peaks used to quantify highlighted in red. c) MMP2200 elutes at 15.46 minutes. d) The MS3 spectrum for MMP2200, with the peaks used to quantify highlighted in red.

250

Figure C.5 Angiotensin 1-7 spectra. a) Full mass spectrum. The main peak is the doubly protonated peak. The structure of native Angiotensin 1-7 is inset. b) The doubly protonated parent peak, 450.24, was selected and fragmented to collect the MS2 spectrum. c) The MS3 spectrum was collected by selecting and subsequently

+2 3 fragmenting the 450.24 peak followed by the 392.71 peak [b6+2H] . The MS peaks used for quantification are highlighted in red.

251

Figure C.6 Ang 1-7-NH2 spectra. a) Full mass spectrum. The main peak is the doubly protonated peak. The structure of native Angiotensin 1-7 is inset. b) The doubly protonated parent peak, 453.84, was selected and fragmented to collect the MS2 spectrum. c) The MS3 spectrum was collected by selecting and subsequently

+2 3 fragmenting the 453.84 peak followed by the 396.26 peak [b6+2H] . The MS peaks used for quantification are highlighted in red.

252

Figure C.7 Ang 1-6-Ser-NH2 spectra. a) Full mass spectrum. The main peak is the doubly protonated peak. The structure is inset. b) The doubly protonated parent peak,

444.74, was selected and fragmented to collect the MS2 spectrum. c) The MS3 spectrum was collected by selecting and subsequently fragmenting the 444.74 peak

+2 3 followed by the 392.71 peak [b6+2H] . The MS peaks used for quantification are highlighted in red.

253

Figure C.8 Ang 1-6-Ser(O-Glc)-NH2 spectra. a) Full mass spectrum. The main peak is the doubly protonated peak. The structure is inset. b) The doubly protonated parent peak, 525.77, was selected and fragmented to collect the MS2 spectrum. c)

The MS3 spectrum was collected by selecting and subsequently fragmenting the

525.77 peak followed by the 444.74 peak [M-Glc+2H]+2. The MS3 peaks used for quantification are highlighted in red.

254

Figure C.9 Ang 1-6-Ser(O-Cb)-NH2 spectra. a) Full mass spectrum. The main peak is the doubly protonated peak. The structure is inset. b) The doubly protonated parent peak, 606.79, was selected and fragmented to collect the MS2 spectrum. c) The MS3 spectrum was collected by selecting and subsequently fragmenting the 606.79 peak followed by the 444.74 peak [M-Cb+2H]+2. The MS3 peaks used for quantification are highlighted in red.

255

Figure C.10 Ang 1-6-Ser(O-Lac)-NH2 spectra. a) Full mass spectrum of an impure mixture. The main identifiable peaks are the doubly protonated peak and the [M-

Glc+2H]+2 peaks. The structure is inset. b) The doubly protonated parent peak, 610.82, was selected and fragmented to collect the MS2 spectrum. c) The MS3 spectrum was collected by selecting and subsequently fragmenting the 610.82 peak followed by the

448.76 peak [M-Lac+2H]+2. The MS3 peaks used for quantification are highlighted in red.

256

17 Figure C.11 [Leu ] PACAP 1-27-NH2 spectra. a) Full mass spectrum. The main identifiable peaks are [M+4H]+4 and [M+5H]+5 peaks. The structure is shown above. Both peaks were fragmented to determine which charge state would be used, based on the ion throughput. b) The [M+5H]+5 peak, 626.75, was selected and fragmented to collect the MS2 spectrum. The ion throughput at different applied fragmentation energies is plotted in the CID breakdown, inset. c) The [M+4H]+4 peak, 783.18, was selected and fragmented to collect the MS2 spectrum. The ion throughput at different applied fragmentation energies is plotted in the CID breakdown, inset. d) It was determined that the [M+5H]+5 peak produced a better MS2 spectrum, so the MS3 spectrum was collected by selecting and subsequently fragmenting the 626.75 peak followed by the 750.65 peak

+4 3 [b26+4H] . The MS peaks used for quantification are highlighted in red.

257

17 Figure C.12 [Leu ] PACAP 1-27-Ser(O-Glc)-NH2 spectra. a) Full mass spectrum. The main identifiable peaks are [M+4H]+4 and [M+5H]+5 peaks. The structure is shown above. b) The [M+5H]+5 peak, 675.95, was selected and fragmented to collect the MS2 spectrum.

The ion throughput at different applied fragmentation energies is plotted in the CID breakdown, inset. c) The MS3 spectrum was collected by selecting and subsequently fragmenting the 675.95 peak followed by the 643.54 peak [M-Glc+5H]+5. The MS3 peaks used for quantification are highlighted in red.

258

Figure C.13 LC gradient for PACAP derivatives. Gradient runs from 0-42 minutes, though data is only collected from minutes 0-30. The remaining gradient is used to wash any remaining material off the column to minimize carryover between runs.

Solvent A is H2O with 0.1% trifluoroacetic acid (v/v), and Solvent B is acetonitrile with

0.1% trifluoroacetic acid (v/v).

259

Figure C.14 LC-MS3 chromatograms and spectra for PACAP derivatives. a) [17Leu]

3 17 PACAP 1-27-NH2 elutes at 17.38 minutes. b) The MS spectrum for [ Leu] PACAP 1-

17 27-NH2, with the peaks used to quantify highlighted in red. c) [ Leu] PACAP 1-27-

3 17 Ser(OGlc)-NH2 elutes at 16.24 minutes. d) The MS spectrum for [ Leu] PACAP 1-27-

Ser(OGlc)-NH2, with the peaks used to quantify highlighted in red.

260 REFERENCES Chapter 1: Introduction: Quantifying neurochemicals with mass spectrometry 1. Giangrande, P. L. F. The history of blood transfusion. Br. J. Haematol. 110, 758– 767 (2000). 2. Ma, H. et al. Correction of a pathogenic gene mutation in human embryos. Nature (2017). doi:10.1038/nature23305 3. Fundamental Neuroscience. (Elsevier, 2008). 4. Mayo Foundation for Medical Education and Research (MFMER). Parkinson’s Disease. National Institute on Aging (2017). Available at: http://www.mayoclinic.org/diseases-conditions/parkinsons- disease/basics/definition/con-20028488. 5. Stefanis, L. Alpha-synuclein aggregation and synaptic pathology in Parkinson’s disease and Dementia with Lewy Bodies. Neurobiol. Aging 39, S4 (2016). 6. Girault, J.-A. & Greengard, P. The Neurobiology of Dopamine Signaling. Arch. Neurol. 61, 641 (2004). 7. Dementia with Lewy Bodies: Clinical and Biological Aspects. Clinical research (Springer Japan, 2017). doi:10.1007/978-4-431-55948-1 8. Changeux, J.-P. Nicotine addiction and nicotinic receptors: lessons from genetically modified mice. Nat. Rev. Neurosci. 11, 389–401 (2010). 9. López-Figueroa, A. L. et al. Serotonin 5-HT1A, 5-HT1B, and 5-HT2A receptor mRNA expression in subjects with major depression, bipolar disorder, and schizophrenia. Biol. Psychiatry 55, 225–33 (2004). 10. Brisch, R. The role of dopamine in schizophrenia from a neurobiological and evolutionary perspective: Old fashioned, but still in vogue. Front. Psychiatry 5, 1– 11 (2014). 11. Albert, P. R. & Benkelfat, C. The neurobiology of depression- revisiting the serotonin hypothesis. II. Genetic , epigenetic and clinical studies. Phil Trans R Soc B 368, 3–6 (2013). 12. Petty, F. GABA and mood disorders: a brief review and hypothesis. J. Affect. Disord. 34, 275–281 (1995). 13. Brambilla, P., Perez, J., Barale, F., Schettini, G. & Soares, J. C. GABAergic dysfunction in mood disorders. Mol. Psychiatry 8, 721–737 (2003). 14. Schwyzer, R. ACTH: a short introductory review. Ann. N. Y. Acad. Sci. 297, 3–26 (1977). 15. Sayers, G., Sayers, M. A., Lewis, H. L. & Long, C. N. H. Effect of Adrenotropic Hormone on Ascorbic Acid and Cholesterol Content of the Adrenal. Exp. Biol. Med. 55, 238–239 (1944). 16. Burns, T. W., Merkin, M. & Sayers, M. A. Concentration of Adrenocorticotrophic Hormone in Rat, Porcine, and Human Pituitary Tissue. Endocrinology 44, 439–444 (1948).

261 17. Sayers, M. A., Sayers, G. & Woodbury, L. A. The assay of adrenocorticotrophic hormone by the adrenal ascorbic acid-depletion method. Endocrinology 42, 379– 393 (1948). 18. Bessey, O. A. A method for the determination of small quantities of ascorbic acid and dehydroascorbio acid in turbid and colored solutions in the presence of other reducing substances. J. Biol. Chem. 126, 771–784 (1938). 19. Pollard, T. D. A Guide to Simple and Informative Binding Assays. Mol. Biol. Cell 21, 4061–4067 (2010). 20. Goldsmith, S. J. Radioimmunoassay: Review of basic principles. Semin. Nucl. Med. 5, 125–152 (1975). 21. Lequin, R. M. Enzyme immunoassay (EIA)/enzyme-linked immunosorbent assay (ELISA). Clin. Chem. 51, 2415–2418 (2005). 22. Berson, S. A. & Yalow, R. S. General Principles of Radioimmunoassay. Clin. Chim. Acta 22, 51–69 (1968). 23. Moon, B. U., De Vries, M. G., Cordeiro, C. A., Westerink, B. H. C. & Verpoorte, E. Microdialysis-coupled enzymatic microreactor for in vivo glucose monitoring in rats. Anal. Chem. 85, 10949–10955 (2013). 24. BIO-RAD. Introduction to ELISA- Basics Guide. (2017). Available at: https://www.bio-rad-antibodies.com/an-introduction-to-elisa.html. 25. Spackman, D. H., Stein, W. H. & Moore, S. Automatic recording apparatus for use in the chromatography of amino acids. Anal. Chem. 30, 1190–1206 (1958). 26. Giddings, J. C. Unified Separation Science. (John Wiley & Sons, Inc., 1991). 27. Needham, S. R. & Valaskovic, G. A. in Protein Analysis using Mass Spectrometry: Accelerating Protein Biotherapeutics from Lab to Patient (eds. Lee, M. S. & Ji, Q. C.) 45–54 (John Wiley & Sons, Inc., 2017). 28. Corkery, L. J., Pang, H., Schneider, B. B., Covey, T. R. & Siu, K. W. M. Automated nanospray using chip-based emitters for the quantitative analysis of pharmaceutical compounds. J. Am. Soc. Mass Spectrom. 16, 363–369 (2005). 29. Yuill, E. M., Sa, N., Ray, S. J., Hieftje, G. M. & Baker, L. A. Electrospray ionization from nanopipette emitters with tip diameters of less than 100 nm. Anal. Chem. 85, 8498–8502 (2013). 30. Sagar, K. A. & Smyth, M. R. Simultaneous determination of levodopa, carbidopa and their metabolites in human plasma and urine samples using LC-EC. J. Pharm. Biomed. Anal. 22, 613–24 (2000). 31. Wightman, R. M., May, L. J. & Michael, A. C. Detection of dopamine dynamics in the brain. Anal. Chem. 60, 769A–779A (1988). 32. Mefford, I. N., Gilberg, M. & Barchas, J. D. Simultaneous determination of catecholamines and unconjugated 3,4-dihydroxyphenylacetic acid in brain tissue by ion-pairing reverse-phase high-performance liquid chromatography with electrochemical detection. Anal. Biochem. 104, 469–472 (1980). 33. Kissinger, P. T., Refshauge, C., Dreiling, R. & Adams, R. N. An Electrochemical Detector for Liquid Chromatography with Picogram Sensitivity. Anal. Lett. 6, 465– 477 (1973).

262 34. Flores, A. J. et al. Differential effects of the NMDA receptor antagonist MK-801 on dopamine receptor D1- and D2-induced abnormal involuntary movements in a preclinical model. Neurosci. Lett. 564, 48–52 (2014). 35. Hwang, D.-F., Chang, S.-H., Shiua, C.-Y. & Chai, T. High-performance liquid chromatographic determination of biogenic amines in fish implicated in food poisoning. J. Chromatogr. B Biomed. Sci. Appl. 693, 23–30 (1997). 36. Olives Barba, A. I., Cámara Hurtado, M., Sánchez Mata, M. C., Fernández Ruiz, V. & López Sáenz De Tejada, M. Application of a UV-vis detection-HPLC method for a rapid determination of lycopene and β-carotene in vegetables. Food Chem. 95, 328–336 (2006). 37. Sagirli, O., Çetin, S. M. & Önal, A. Determination of gabapentin in human plasma and urine by high-performance liquid chromatography with UV-vis detection. J. Pharm. Biomed. Anal. 42, 618–624 (2006). 38. Galievsky, V. A., Stasheuski, A. S. & Krylov, S. N. Improvement of LOD in Fluorescence Detection with Spectrally Non-Uniform Background by Optimization of Emission Filtering. Anal. Chem. acs.analchem.7b03400 (2017). doi:10.1021/acs.analchem.7b03400 39. Bronsema, K. J., Bischoff, R. & Van De Merbel, N. C. High-sensitivity LC-MS/MS quantification of peptides and proteins in complex biological samples: The impact of enzymatic digestion and internal standard selection on method performance. Anal. Chem. 85, 9528–9535 (2013). 40. Mabrouk, O. S., Li, Q., Song, P. & Kennedy, R. T. Microdialysis and mass spectrometric monitoring of dopamine and enkephalins in the globus pallidus reveal reciprocal interactions that regulate movement. J. Neurochem. 118, 24–33 (2011). 41. Grouzmann, E. & Lamine, F. Determination of catecholamines in plasma and urine. Best Pract. Res. Clin. Endocrinol. Metab. 27, 713–723 (2013). 42. Asanuma, M., Miyazaki, I. & Ogawa, N. Dopamine- or L-DOPA-induced neurotoxicity: the role of dopamine quinone formation and tyrosinase in a model of Parkinson’s disease. Neurotox. Res. 5, 165–76 (2003). 43. Ohtaki, H., Nakamachi, T., Dohi, K. & Shioda, S. Role of PACAP in ischemic neural death. J. Mol. Neurosci. 36, 16–25 (2008). 44. Dejda, A., Sokołowska, P. & Nowak, J. Z. Neuroprotective potential of three neuropeptides PACAP, VIP and PHI. Pharmacol. Rep. 57, 307–20 (2005). 45. Weaver, E. M. & Hummon, A. B. Imaging mass spectrometry: From tissue sections to cell cultures. Adv. Drug Deliv. Rev. 65, 1039–1055 (2013). 46. Liu, X., Weaver, E. M. & Hummon, A. B. Evaluation of therapeutics in three- dimensional cell culture systems by MALDI imaging mass spectrometry. Anal. Chem. 85, 6295–6302 (2013). 47. Franke, H., Galla, H. J. & Beuckmann, C. T. Primary cultures of brain microvessel endothelial cells: A valid and flexible model to study drug transport through the blood-brain barrier in vitro. Brain Res. Protoc. 5, 248–256 (2000). 48. Appelt-Menzel, A. et al. Establishment of a Human Blood-Brain Barrier Co-culture Model Mimicking the Neurovascular Unit Using Induced Pluri- and Multipotent Stem Cells. Stem Cell Reports 8, (2017).

263 49. Powell, M. F. et al. Peptide Stability in Drug Development. II. Effect of Single Amino Acid Substitution and Glycosylation on Peptide Reactivity in Human Serum. Pharmaceutical Research 10, 1268–1273 (1993). 50. Wright, B. L. C., Lai, J. T. F. & Sinclair, A. J. Cerebrospinal fluid and lumbar puncture: A practical review. J. Neurol. 259, 1530–1545 (2012). 51. Sakka, L., Coll, G. & Chazal, J. Anatomy and physiology of cerebrospinal fluid. Eur. Ann. Otorhinolaryngol. Head Neck Dis. 128, 309–316 (2011). 52. Laude, N. D. Addressing the Neurochemical Problem: Sensitive and Selective Measurements of Neurotransmitters, Neuropeptides, and Synaptic Vesicles. Dissertation (2015). 53. Westerhout, J., Ploeger, B., Smeets, J., Danhof, M. & Lange, E. C. M. Physiologically Based Pharmacokinetic Modeling to Investigate Regional Brain Distribution Kinetics in Rats. AAPS J. 14, 543–553 (2012). 54. De Lange, E. C. M. Utility of CSF in translational neuroscience. J. Pharmacokinet. Pharmacodyn. 40, 315–326 (2013). 55. Assad, D. X., Borges, G. A., Avelino, S. R. & Guerra, E. N. S. Additive cytotoxic effects of radiation and mTOR inhibitors in a cervical cancer cell line. Pathol. - Res. Pract. 0–1 (2017). doi:10.1016/j.prp.2017.10.019 56. Niccoli, S., Boreham, D. R., Phenix, C. P. & Lees, S. J. Non-radioactive 2-deoxy-2- fluoro-D-glucose inhibits glucose uptake in xenograft tumours and sensitizes HeLa cells to doxorubicin in vitro. PLoS One 12, e0187584 (2017). 57. Pan, U. N. et al. Protein-Based Multifunctional Nanocarriers for Imaging, Photothermal Therapy, and Anticancer Drug Delivery. ACS Appl. Mater. Interfaces 9, 19495–19501 (2017). 58. Greene, L. A. & Tischler, A. S. Establishment of a noradrenergic clonal line of rat adrenal pheochromocytoma cells which respond to nerve growth factor. Proc. Natl. Acad. Sci. 73, 2424–2428 (1976). 59. Burbach, J. P. H. in Neuropeptides Methods and Protocols (ed. Merighi, A.) 789, 1–36 (Humana Press, 2011). 60. Uchida, D., Arimura, A., Somogyvári-Vigh, A., Shioda, S. & Banks, W. a. Prevention of ischemia-induced death of hippocampal neurons by pituitary adenylate cyclase activating polypeptide. Brain Res. 736, 280–6 (1996). 61. Bucher, E. S. & Wightman, R. M. Electrochemical Analysis of Neurotransmitters. Annu. Rev. Anal. Chem. (Palo Alto. Calif). 8, 239–61 (2015). 62. Finnegan, J. M. et al. Vesicular quantal size measured by amperometry at chromaffin, mast, pheochromocytoma, and pancreatic beta-cells. J. Neurochem. 66, 1914–1923 (1996). 63. Pardridge, W. M. Blood-brain barrier biology and methodology. J. Neurovirol. 5, 556–569 (1999). 64. Cecchelli, R. et al. Modelling of the blood-brain barrier in drug discovery and development. Nat. Rev. Drug Discov. 6, 650–661 (2007). 65. Keithley, R. B. et al. Higher sensitivity dopamine measurements with faster-scan cyclic voltammetry. Anal. Chem. 83, 3563–3571 (2011).

264 66. Trouillon, R. & Ewing, A. G. Single cell amperometry reveals glycocalyx hinders the release of neurotransmitters during exocytosis. Anal. Chem. 85, 4822–4828 (2013). 67. Ross, A. E. & Venton, B. J. Nafion-CNT coated carbon-fiber microelectrodes for enhanced detection of adenosine. Analyst 137, 3045–51 (2012). 68. Swamy, B. E. K. & Venton, B. J. Carbon nanotube-modified microelectrodes for simultaneous detection of dopamine and serotonin in vivo. Analyst 132, 876–884 (2007). 69. Samaranayake, S. et al. In vivo histamine voltammetry in the mouse premammillary nucleus. Analyst 140, 3759–65 (2015). 70. Cooper, S. E. & Venton, B. J. Fast-scan cyclic voltammetry for the detection of tyramine and octopamine. Anal. Bioanal. Chem. 394, 329–36 (2009). 71. Schmidt, A. C., Dunaway, L. E., Roberts, J. G., McCarty, G. S. & Sombers, L. a. Multiple scan rate voltammetry for selective quantification of real-time enkephalin dynamics. Anal. Chem. 86, 7806–7812 (2014). 72. Gaddum, J. H. Substances released in nervous activity. Pharmacol. Anal. Cent. Nerv. action 1–6 (1962). 73. Gardner, E. L., Chen, J. & Paredes, W. Overview of chemical sampling techniques. J. Neurosci. Methods 48, 173–197 (1993). 74. Philippu, A. & Kraus, M. M. in In Vivo Neuropharmacology and Neurophysiology 121, 207–236 (2017). 75. Lisi, T. L., Westlund, K. N. & Sluka, K. A. Comparison of microdialysis and push- pull perfusion for retrieval of serotonin and norepinephrine in the spinal cord dorsal horn. J. Neurosci. Methods 126, 187–194 (2003). 76. Kottegoda, S., Shaik, I. & Shippy, S. A. Demonstration of low flow push-pull perfusion. J. Neurosci. Methods 121, 93–101 (2002). 77. Myers, R. D., Adell, A. & Lankford, M. F. Simultaneous comparison of cerebral dialysis and push-pull perfusion in the brain of rats: A critical review. Neurosci. Biobehav. Rev. 22, 371–387 (1998). 78. Ungerstedt, U. Microdialysis--principles and applications for studies in animals and man. J. Intern. Med. 230, 365–73 (1991). 79. Benveniste, H. Brain microdialysis. J. Neurochem. 52, 1667–79 (1989). 80. Rice, M. E. & Nicholson, C. in Neuromethods, Vol 27: Voltammetric Methods in Brain Systems (ed. A. Boulton, G. Baker, R. N. A.) 27–79 (Humana Press Inc., 1995). 81. Li, Q., Jon-Kar Zubieta & Kennedy, R. T. Practical aspects of in vivo detection of neuropeptides by microdialysis coupled off-line to capillary LC with multistage MS. Anal. Chem. 81, 2242–2250 (2009). 82. De Lange, E. C. M., De Boer, B. A. G. & Breimer, D. D. Microdialysis for pharmacokinetic analysis of drug transport to the brain. Adv. Drug Deliv. Rev. 36, 211–227 (1999). 83. Kennedy, R. T. Emerging trends in in vivo neurochemical monitoring by microdialysis. Curr. Opin. Chem. Biol. 17, 860–7 (2013).

265 84. Watson, C. J., Venton, B. J. & Kennedy, R. T. In vivo measurements of neurotransmitters by microdialysis sampling. Anal. Chem. 78, 1391–1399 (2006). 85. Nandi, P. & Lunte, S. M. Recent trends in microdialysis sampling integrated with conventional and microanalytical systems for monitoring biological events: A review. Anal. Chim. Acta 651, 1–14 (2009). 86. Griffiths, J. A Brief History of Mass Spectrometry. 80, 5678–5683 (2008). 87. Yates III, J. R. A century of mass spectrometry: from atoms to proteomes. Nat. Methods 8, 633–637 (2011). 88. Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F. & Whitehouse, C. M. Electrospray ionization- principles and practice. Mass Spectrom. Rev. 9, 37–70 (1990). 89. Ferreira, C. R. et al. Ambient ionization mass spectrometry for point-of-care diagnostics and other clinical measurements. Clin. Chem. 62, 99–110 (2016). 90. Staniforth, M. & Stavros, V. G. Recent advances in experimental techniques to probe fast excited-state dynamics in biological molecules in the gas phase: dynamics in nucleotides, amino acids and beyond. Proc. Math. Phys. Eng. Sci. 469, 20130458 (2013). 91. Banerjee, S. & Mazumdar, S. Electrospray Ionization Mass Spectrometry: A Technique to Access the Information beyond the Molecular Weight of the Analyte. Int. J. Anal. Chem. 2012, 1–40 (2012). 92. Gross, J. H. Mass Spectrometry. (Springer-Verlag, 2011). 93. Yates, J. R. 3rd et al. Future prospects for the analysis of complex biological systems using micro-column liquid chromatography-electrospray tandem mass spectrometry. Analyst 121, 65R–76R (1996). 94. Sturm, R. M., Dowell, J. A. & Li, L. Rat Brain Neuropeptidomics: Tissue Collection, Protease Inhibition, Neuropeptides Extraction, and Mass Spectrometric Analysis. Methods Mol Biol 615, 217–226 (2010). 95. Gonçalves, C. & Alpendurada, M. F. Solid-phase micro-extraction-gas chromatography-(tandem) mass spectrometry as a tool for pesticide residue analysis in water samples at high sensitivity and selectivity with confirmation capabilities. J. Chromatogr. A 1026, 239–250 (2004). 96. Blesa, J., Phani, S., Jackson-Lewis, V. & Przedborski, S. Classic and New Animal Models of Parkinson’s Disease. J. Biomed. Biotechnol. 2012, (2012). 97. Mogil, J. S., Davis, K. D. & Derbyshire, S. W. The necessity of animal models in pain research. Pain 151, 12–17 (2010). 98. Duty, S. & Jenner, P. Animal models of Parkinson’s disease: A source of novel treatments and clues to the cause of the disease. Br. J. Pharmacol. 164, 1357– 1391 (2011). 99. Williams, S. M., Haines, J. L. & Moore, J. H. The use of animal models in the study of complex disease: All else is never equal or why do so many human studies fail to replicate animal findings? BioEssays 26, 170–179 (2004). 100. Markou, A., Chiamulera, C., Geyer, M. A., Tricklebank, M. & Steckler, T. Removing obstacles in neuroscience drug discovery: the future path for animal models. Neuropsychopharmacology 34, 74–89 (2009).

266 101. Shanks, N., Greek, R. & Greek, J. Are animal models predictive for humans? Philos. Ethics, Humanit. Med. 4, 2 (2009). 102. Baumans, V. Use of animals in experimental research: an ethical dilemma? Gene Ther. 11 Suppl 1, S64-6 (2004). 103. Radbruch, A. & Isaacs, J. Animal models in infection and inflammation - Chance and necessity. Eur. J. Immunol. 39, 1991–1993 (2009). 104. Kemp, M. W. & Massey, R. C. The use of insect models to study human pathogens. Drug Discov. Today 4, 105–110 (2007). 105. Kaiser, K. What’s new?: From gene to phenotype in Drosophila and other organisms. BioEssays 12, 297–301 (1990). 106. Pandey, U. B. & Nichols, C. D. Human Disease Models in Drosophila melanogaster and the Role of the Fly in Therapeutic Drug Discovery. Drug Deliv. 63, 411–436 (2011). 107. Bilen, J. & Bonini, N. M. Drosophila as a Model for Human Neurodegenerative Disease. Annu. Rev. Genet. 39, 153–171 (2005). 108. Cotter, G., Doyle, S. & Kavanagh, K. Development of an insect model for the in vivo pathogenicity testing of yeasts. FEMS Immunol. Med. Microbiol. 27, 163–169 (2000). 109. Carey, A. F. & Carlson, J. R. Insect olfaction from model systems to disease control. Proc. Natl. Acad. Sci. 108, 12987–12995 (2011). 110. Greenspan, R. J. & Dierick, H. A. ‘Am not I a fly like thee?’ From genes in fruit flies to behavior in humans. Hum. Mol. Genet. 13, 267–273 (2004). 111. Bier, E. Drosophila, the golden bug, emerges as a tool for human genetics. Nat Rev Genet 6, 9–23 (2005). 112. Feany, M. B. & Bender, W. W. A Drosophila model of Parkinson’s disease. Nature 404, 394–398 (2000). 113. St Johnston, D. The Art and Design of Genetic Screens: Drosophila melanogaster. Nat. Rev. Genet. 3, 176–188 (2002). 114. Jennings, B. H. Drosophila-a versatile model in biology & medicine. Mater. Today 14, 190–195 (2011). 115. Southwick, E. E. & Southwick, L. Estimating the Economic Value of Honey Bees (Hymenoptera: Apidae) as Agricultural Pollinators in the United States. J. Econ. Entomol. 85, 621–633 (1992). 116. Aizen, M. A. & Harder, L. D. The Global Stock of Domesticated Honey Bees Is Growing Slower Than Agricultural Demand for Pollination. Curr. Biol. 19, 915–918 (2009). 117. Chen, Y., Evans, J. D., Smith, I. B. & Pettis, J. S. Nosema ceranae is a long-present and wide-spread microsporidian infection of the European honey bee (Apis mellifera) in the United States. J. Invertebr. Pathol. 97, 186–188 (2008). 118. van Dooremalen, C. et al. Winter survival of individual honey bees and honey bee colonies depends on level of Varroa destructor infestation. PLoS One 7, e36285 (2012).

267 119. Genersch, E. et al. The German bee monitoring project: a long term study to understand periodically high winter losses of honey bee colonies. Apidologie 41, 332–352 (2010). 120. Bullock, T. H. Neuron Doctrine and Electrophysiology. Science (80-. ). 129, 997– 1002 (1959). 121. Brooks, D. J. Functional imaging studies on dopamine and motor control. J Neural Transm 108, 1283–1298 (2001). 122. Van Swinderen, B. & Andretic, R. Dopamine in Drosophila: setting arousal thresholds in a miniature brain. Proc. R. Soc. B Biol. Sci. 278, 906–913 (2011). 123. Herrera-Solis, A., Herrera-Morales, W., Nunez-Jaramillo, L. & Arias-Carrion, O. Dopaminergic Modulation of Sleep-Wake States. CNS Neurol. Disord. - Drug Targets 16, (2017). 124. Nutt, D. J., Lingford-Hughes, A., Erritzoe, D. & Stokes, P. R. A. The dopamine theory of addiction: 40 years of highs and lows. Nat. Rev. Neurosci. 16, 305–312 (2015). 125. Chrousos, G. P. Stress and disorders of the stress system. Nat. Rev. Endocrinol. 5, 374–381 (2009). 126. Tank, A. W. & Wong, D. L. Peripheral and central effects of circulating catecholamines. Compr. Physiol. 5, 1–15 (2015). 127. Wirth, A., Holst, K. & Ponimaskin, E. How serotonin receptors regulate morphogenic signalling in neurons. Prog. Neurobiol. 151, 35–56 (2017). 128. Puig, M. V. & Gulledge, A. T. Serotonin and Prefrontal Cortex Function: Neurons, Networks, and Circuits. Mol Neurobiol 44, 449–464 (2011). 129. Makos, M. a, Kim, Y.-C., Han, K.-A., Heien, M. L. & Ewing, A. G. In vivo electrochemical measurements of exogenously applied dopamine in Drosophila melanogaster. Anal. Chem. 81, 1848–54 (2009). 130. Ream, P. J., Suljak, S. W., Ewing, A. G. & Han, K.-A. Micellar electrokinetic capillary chromatography-electrochemical detection for analysis of biogenic amines in Drosophila melanogaster. Anal. Chem. 75, 3972–8 (2003). 131. Perry, M., Li, Q. & Kennedy, R. T. Review of recent advances in analytical techniques for the determination of neurotransmitters. Anal. Chim. Acta 653, 1–22 (2009). 132. Implications of the Blood-Brain Barrier and Its Manipulation. (Plenum Publishing Corporation, 1989). 133. Zlokovic, B. V. The Blood-Brain Barrier in Health and Chronic Neurodegenerative Disorders. Neuron 57, 178–201 (2008). 134. Abbott, N. J., Patabendige, A. A. K., Dolman, D. E. M., Yusof, S. R. & Begley, D. J. Structure and function of the blood-brain barrier. Neurobiol. Dis. 37, 13–25 (2010). 135. Rubin, L. L. & Staddon, J. M. The Cell Biology of the Blood-Brain Barrier. Annu. Rev. Neurosci. 22, 11–28 (1999). 136. Terasaki, T. & Hosoya, K. I. The blood-brain barrier efflux transporters as a detoxifying system for the brain. Adv. Drug Deliv. Rev. 36, 195–209 (1999).

268 137. Pardridge, W. M. Blood-brain barrier delivery. Drug Discov. Today 12, 54–61 (2007). 138. Pardridge, W. M. The blood-brain barrier: bottleneck in brain drug development. NeuroRx 2, 3–14 (2005). 139. Abbott, N. J., Rönnbäck, L. & Hansson, E. Astrocyte-endothelial interactions at the blood-brain barrier. Nat. Rev. Neurosci. 7, 41–53 (2006). 140. Banks, W. A. & Kastin, A. J. Peptides and the blood-brain barrier: Lipophilicity as a predictor of permeability. Brain Res. Bull. 15, 287–292 (1985). 141. Banks, W. A. Characteristics of compounds that cross the blood-brain barrier. BMC Neurol. 9, S3 (2009). 142. Urquhart, B. L. & Kim, R. B. Blood-brain barrier transporters and response to CNS- active drugs. Eur. J. Clin. Pharmacol. 65, 1063–1070 (2009). 143. Hervé, F., Ghinea, N. & Scherrmann, J.-M. CNS delivery via adsorptive transcytosis. AAPS J. 10, 455–72 (2008). 144. Gabathuler, R. Approaches to transport therapeutic drugs across the blood-brain barrier to treat brain diseases. Neurobiol. Dis. 37, 48–57 (2010). 145. Schinkel, A. H. P-Glycoprotein, a gatekeeper in the blood-brain barrier. Adv. Drug Deliv. Rev. 36, 179–194 (1999). 146. Wu, D. & Pardridge, W. M. Central nervous system pharmacologic effect in conscious rats after intravenous injection of a biotinylated vasoactive intestinal peptide analog coupled to a blood-brain barrier drug delivery system. J. Pharmacol. Exp. Ther. 279, 77–83 (1996). 147. Cornford, E. M. & Hyman, S. Blood-brain barrier permeability to small and large molecules. Adv. Drug Deliv. Rev. 36, 145–163 (1999). 148. Jones, E. M. & Polt, R. CNS active O-linked glycopeptides. Front. Chem. 3, 40 (2015). 149. Kaspar, A. A. & Reichert, J. M. Future directions for peptide therapeutics development. Drug Discov. Today 18, 807–817 (2013). 150. Fosgerau, K. & Hoffmann, T. Peptide therapeutics: Current status and future directions. Drug Discov. Today 20, 122–128 (2015). 151. Aldrich, J. V. & McLaughlin, J. P. Opioid Peptides: Potential for Drug Development. Drug Discov. Today. Technol. 9, e23–e31 (2012). 152. Aldrich, J. V. & McLaughlin, J. P. Peptide kappa opioid receptor ligands: potential for drug development. AAPS J. 11, 312–22 (2009). 153. Seelig, A., Gottschlich, R. & Devant, R. M. A method to determine the ability of drugs to diffuse through the blood-brain barrier. Proc Natl Acad Sci U S A 91, 68– 72 (1994). 154. Pardridge, W. M. Drug and gene delivery to the brain: The vascular route. Neuron 36, 555–558 (2002). 155. Gentry, C. L. et al. The effect of halogenation on blood-brain barrier permeability of a novel peptide drug. Peptides 20, 1229–1238 (1999). 156. Schinkel, A. H., Wagenaar, E., Mol, C. A. A. M. & Deemter, L. Van. P-Glycoprotein

269 in the Blood-Brain Barrier of Mice Influences the Brain Penetration and Pharmacological Activity of Many Drugs. 2517–2524 157. Egleton, R. D. & Davis, T. P. Development of neuropeptide drugs that cross the blood-brain barrier. NeuroRX 2, 44–53 (2005). 158. Egleton, R. D. & Davis, T. P. Bioavailability and transport of peptides and peptide drugs into the brain. Peptides 18, 1431–1439 (1997). 159. Sato, A. K., Viswanathan, M., Kent, R. B. & Wood, C. R. Therapeutic peptides: technological advances driving peptides into development. Curr. Opin. Biotechnol. 17, 638–642 (2006). 160. Rink, R. et al. To protect peptide pharmaceuticals against peptidases. J. Pharmacol. Toxicol. Methods 61, 210–218 (2010). 161. Witt, K. A., Gillespie, T. J., Huber, J. D., Egleton, R. D. & Davis, T. P. Peptide drug modifications to enhance bioavailability and blood-brain barrier permeability. Peptides 22, 2329–2343 (2001). 162. de Vries, L. et al. Oral and pulmonary delivery of thioether-bridged angiotensin-(1- 7). Peptides 31, 893–898 (2010). 163. Varamini, P. & Toth, I. Lipid- and sugar-modified endomorphins: novel targets for the treatment of neuropathic pain. Front. Pharmacol. 4, 155 (2013). 164. Lowery, J. J. et al. In vivo characterization of MMP-2200, a mixed δ/μ opioid agonist, in mice. J. Pharmacol. Exp. Ther. 336, 767–78 (2011). 165. Yue, X. et al. Effects of the novel glycopeptide opioid agonist MMP-2200 in preclinical models of Parkinson’s disease. Brain Res. 1413, 72–83 (2011). 166. Miyata, A. et al. Isolation of a neuropeptide corresponding to the N-terminal 27 residues of the pituitary adenylate cyclase activating polypeptide with 38 residues (PACAP38). Biochem. Biophys. Res. Commun. 170, 643–8 (1990). 167. Vaudry, D. et al. Pituitary Adenylate Cyclase-Activating Polypeptide and Its Receptors : 20 Years after the Discovery. Pept. Res. 61, 283–357 (2009). 168. Sherwood, N. M., Krueckl, S. L. & McRory, J. E. The origin and function of the pituitary adenylate cyclase-activating polypeptide (PACAP)/glucagon superfamily. Endocr. Rev. 21, 619–70 (2000). 169. Arimura, a et al. PACAP functions as a neurotrophic factor. Ann. N. Y. Acad. Sci. 739, 228–43 (1994). 170. Tamas, A. et al. Effect of PACAP in central and peripheral nerve injuries. Int. J. Mol. Sci. 13, 8430–8448 (2012). 171. Reglodi, D., Kiss, P., Lubics, A. & Tamas, A. Review on the Protective Effects of PACAP in Models of Neurodegenerative Diseases In Vitro and In Vivo. Curr. Pharm. Des. 17, 962–972 (2011). 172. Somogyvari-Vigh, A. & Reglodi, D. Pituitary Adenylate Cyclase Activating Polypeptide: A Potential Neuroprotective Peptide. Curr. Pharm. Des. 10, 2861– 2889 (2004). 173. Santos, R. A. S. Angiotensin-(1-7). Hypertension 63, 1138–1147 (2014). 174. Pereira, R. M. et al. The renin-angiotensin system in a rat model of hepatic fibrosis:

270 Evidence for a protective role of Angiotensin-(1-7). J. Hepatol. 46, 674–681 (2007). 175. Kluskens, L. D. et al. Angiotensin- (1–7) with Thioether Bridge : An Angiotensin- Converting Enzyme-Resistant, Potent Angiotensin-(1-7) Analog. Pharmacology 328, 4–7 (2009). 176. Lee, S. et al. Effect of a Selective Mas Receptor Agonist in Cerebral Ischemia In Vitro and In Vivo. PLoS One 10, e0142087 (2015). 177. Sasongko, L. et al. Imaging P-glycoprotein transport activity at the human blood- brain barrier with positron emission tomography. Clin. Pharmacol. Ther. 77, 503– 514 (2005). 178. Yano, Y., Budinger, T. F., Friedland, R. P., Derenzo, S. E. & Huesman, R. H. Brain Tumor Evaluation Using Rb-82 and Positron Emission Tomography. J Nucl Med 23, 532–538 (1982). 179. Yamamizu, K. et al. In Vitro Modeling of Blood-Brain Barrier with Human iPSC- Derived Endothelial Cells, Pericytes, Neurons, and Astrocytes via Notch Signaling. Stem Cell Reports 8, 634–647 (2017). 180. Dehouck, M. ‐P, Méresse, S., Delorme, P., Fruchart, J. ‐C & Cecchelli, R. An Easier, Reproducible, and Mass‐Production Method to Study the Blood–Brain Barrier In Vitro. J. Neurochem. 54, 1798–1801 (1990). 181. Xia, H., Cheng, Z., Cheng, Y. & Xu, Y. Investigating the passage of tetramethylpyrazine-loaded liposomes across blood-brain barrier models in vitro and ex vivo. Mater. Sci. Eng. C 69, 1010–1017 (2016). 182. de Lange, E. C., Danhof, M., de Boer, a G. & Breimer, D. D. Methodological considerations of intracerebral microdialysis in pharmacokinetic studies on drug transport across the blood-brain barrier. Brain Res. Brain Res. Rev. 25, 27–49 (1997). 183. de Lange, E. C. M., Danhof, M., de Boer, A. G. & Breimer, D. D. Critical factors of intracerebral microdialysis as a technique to determined the pharmacokinetics of drugs in rat brain. Brain Res. 666, 1–8 (1994). 184. Nakazono, T., Murakami, T., Sakai, S., Higashi, Y. & Yata, N. Application of microdialysis for study of caffeine distribution into brain and cerebrospinal fluid in rats. Chem. Pharm. Bull. (Tokyo). 40, 2510–5 (1992). 185. Behrens, H. L. & Li, L. Peptidomics. 615, 57–73 (2010). 186. Hui, L. et al. Discovery and functional study of a novel crustacean tachykinin neuropeptide. ACS Chem. Neurosci. 2, 711–722 (2011). 187. Bernay, B. et al. Discovering new bioactive neuropeptides in the striatum secretome using in vivo microdialysis and versatile proteomics. Mol. Cell. Proteomics 8, 946– 958 (2009). 188. Hou, X., Xie, F. & Sweedler, J. V. Relative Quantitation of Neuropeptides Over a Thousand-fold Concentration Range. J Am Soc Mass Spectrom. 23, 2083–2093 (2012). 189. Tawfik, V. L. & Flood, P. Electrical Synapses. Anesthesiology 124, 13–15 (2016).

271 Chapter 2: Separation-free quantitation of biogenic amines in brain tissue with mass spectrometry 1. Silla Santos, M. H. Biogenic amines: Their importance in foods. Int. J. Food Microbiol. 29, 213–231 (1996). 2. Önal, A. A review: Current analytical methods for the determination of biogenic amines in foods. Food Chem. 103, 1475–1486 (2007). 3. Landete, J. M., Ferrer, S., Polo, L. & Pardo, I. Biogenic amines in wines from three Spanish regions. J. Agric. Food Chem. 53, 1119–1124 (2005). 4. McClung, C. & Hirsh, J. The trace amine tyramine is essential for sensitization to cocaine in Drosophila. Curr. Biol. 9, 853–860 (1999). 5. Kahsai, L. & Winther, Å. M. E. Chemical neuroanatomy of the Drosophila central complex: Distribution of multiple neuropeptides in relation to neurotransmitters. J. Comp. Neurol. 519, 290–315 (2011). 6. Nässel, D. R. & Winther, Å. M. E. Drosophila neuropeptides in regulation of physiology and behavior. Prog. Neurobiol. 92, 42–104 (2010). 7. Dierick, H. A. & Greenspan, R. J. Serotonin and neuropeptide F have opposite modulatory effects on fly aggression. Nat. Genet. 39, 678–682 (2007). 8. Shalaby, A. R. Significance of biogenic amines to food safety and human health. Food Res. Int. 29, 675–690 (1997). 9. Jia, S., Ryu, Y., Kwon, S. W. & Lee, J. An in situ benzoylation-dispersive liquid- liquid microextraction method based on solidification of floating organic droplets for determination of biogenic amines by liquid chromatography-ultraviolet analysis. J. Chromatogr. A 1282, 1–10 (2013). 10. Page, L. B., Raker, J. W. & Berberich, F. R. Pheochromocytoma with Predominant Ephinephrine Secretion. Am. J. Med. 47, 648–652 (1969). 11. Goldenberg, M., Serlin, I., Edwards, T. & Rapport, M. M. Chemical Screening Methods for the Diagnosis of Pheochromocytoma. Am. J. Med. 16, 310–327 (1954). 12. Smythe, G. A., Edwards, G., Graham, P. & Lazarus, L. Biochemical Diagnosis of Pheochromocytoma by Simultaneous Measurement of Urinary Excretion of Epinephrine and Norepinephrine. Clin. Chem. 38, 486–492 (1992). 13. Manger, W. M. et al. Chemical Quantitation of Epinephrine and Norepinephrine in Thirteen Patients with Pheochromocytoma. Circulation 10, 641–652 (1954). 14. Grouzmann, E. & Lamine, F. Determination of catecholamines in plasma and urine. Best Pract. Res. Clin. Endocrinol. Metab. 27, 713–723 (2013). 15. Brooks, D. J. Functional imaging studies on dopamine and motor control. J Neural Transm 108, 1283–1298 (2001). 16. Van Swinderen, B. & Andretic, R. Dopamine in Drosophila: setting arousal thresholds in a miniature brain. Proc. R. Soc. B Biol. Sci. 278, 906–913 (2011). 17. Herrera-Solis, A., Herrera-Morales, W., Nunez-Jaramillo, L. & Arias-Carrion, O. Dopaminergic Modulation of Sleep-Wake States. CNS Neurol. Disord. - Drug Targets 16, (2017). 18. Nutt, D. J., Lingford-Hughes, A., Erritzoe, D. & Stokes, P. R. A. The dopamine

272 theory of addiction: 40 years of highs and lows. Nat. Rev. Neurosci. 16, 305–312 (2015). 19. Flores, A. J. et al. Differential effects of the NMDA receptor antagonist MK-801 on dopamine receptor D1- and D2-induced abnormal involuntary movements in a preclinical model. Neurosci. Lett. 564, 48–52 (2014). 20. Wirth, A., Holst, K. & Ponimaskin, E. How serotonin receptors regulate morphogenic signalling in neurons. Prog. Neurobiol. 151, 35–56 (2017). 21. Albert, P. R. & Benkelfat, C. The neurobiology of depression- revisiting the serotonin hypothesis. II. Genetic , epigenetic and clinical studies. Phil Trans R Soc B 368, 3–6 (2013). 22. Puig, M. V. & Gulledge, A. T. Serotonin and Prefrontal Cortex Function: Neurons, Networks, and Circuits. Mol Neurobiol 44, 449–464 (2011). 23. Chrousos, G. P. Stress and disorders of the stress system. Nat. Rev. Endocrinol. 5, 374–381 (2009). 24. Tank, A. W. & Wong, D. L. Peripheral and central effects of circulating catecholamines. Compr. Physiol. 5, 1–15 (2015). 25. Li, Q., Jon-Kar Zubieta & Kennedy, R. T. Practical aspects of in vivo detection of neuropeptides by microdialysis coupled off-line to capillary LC with multistage MS. Anal. Chem. 81, 2242–2250 (2009). 26. Hou, X., Xie, F. & Sweedler, J. V. Relative Quantitation of Neuropeptides Over a Thousand-fold Concentration Range. J Am Soc Mass Spectrom. 23, 2083–2093 (2012). 27. Sagar, K. A. & Smyth, M. R. Simultaneous determination of levodopa, carbidopa and their metabolites in human plasma and urine samples using LC-EC. J. Pharm. Biomed. Anal. 22, 613–24 (2000). 28. Wightman, R. M., May, L. J. & Michael, A. C. Detection of dopamine dynamics in the brain. Anal. Chem. 60, 769A–779A (1988). 29. Mefford, I. N., Gilberg, M. & Barchas, J. D. Simultaneous determination of catecholamines and unconjugated 3,4-dihydroxyphenylacetic acid in brain tissue by ion-pairing reverse-phase high-performance liquid chromatography with electrochemical detection. Anal. Biochem. 104, 469–472 (1980). 30. Kissinger, P. T., Refshauge, C., Dreiling, R. & Adams, R. N. An Electrochemical Detector for Liquid Chromatography with Picogram Sensitivity. Anal. Lett. 6, 465– 477 (1973). 31. Makos, M. a, Kim, Y.-C., Han, K.-A., Heien, M. L. & Ewing, A. G. In vivo electrochemical measurements of exogenously applied dopamine in Drosophila melanogaster. Anal. Chem. 81, 1848–54 (2009). 32. Ream, P. J., Suljak, S. W., Ewing, A. G. & Han, K.-A. Micellar electrokinetic capillary chromatography-electrochemical detection for analysis of biogenic amines in Drosophila melanogaster. Anal. Chem. 75, 3972–8 (2003). 33. Song, P., Mabrouk, O. S., Hershey, N. D. & Kennedy, R. T. In vivo neurochemical monitoring using benzoyl chloride derivatization and liquid chromatography-mass spectrometry. Anal. Chem. 84, 412–419 (2012).

273 34. Wong, J. T. et al. Benzoyl chloride derivatization with liquid chromatography – mass spectrometry for targeted metabolomics of neurochemicals in biological samples. J. Chromatogr. A 1446, 78–90 (2016). 35. Blesa, J., Phani, S., Jackson-Lewis, V. & Przedborski, S. Classic and New Animal Models of Parkinson’s Disease. J. Biomed. Biotechnol. 2012, (2012). 36. Torres, E. M. & Dunnett, S. B. Amphetamine induced rotation in the assessment of lesions and grafts in the unilateral rat model of Parkinson ’ s disease. Eur. Neuropsychopharmacol. 17, 206–214 (2007). 37. Schwarting, R. K. W. & Huston, J. P. Unilateral 6-hydroxydopamine lesions of meso-striatal dopamine neurons and their physiological sequelae. Prog. Neurobiol. 49, 215–266 (1996). 38. Chen, A. et al. Dispensable, Redundant, Complementary, and Cooperative Roles of Dopamine, Octopamine, and Serotonin in Drosophila melanogaster. Genetics 193, 159–176 (2013). 39. Greenspan, R. J. & Dierick, H. A. ‘Am not I a fly like thee?’ From genes in fruit flies to behavior in humans. Hum. Mol. Genet. 13, 267–273 (2004). 40. Yue, X. et al. Comparative Study of the Neurotrophic Effects Elicited by VEGF-B and GDNF in Preclinical in vivo Models of Parkinson’s Disease. Neuroscience 258, 385–400 (2014). 41. Bartlett, M. J. et al. Long-term effect of sub-anesthetic ketamine in reducing L- DOPA-induced dyskinesias in a preclinical model. Neurosci. Lett. 612, 121–125 (2015). 42. Torres, E. M. & Dunnett, S. B. in Animal Models of Movement Disorders: Volume 1 (eds. Lane, E. L. & Dunnett, S. B.) 62, 363–379 (2011). 43. Schwarting, R. K. W. & Huston, J. P. The unilateral 6-hydroxydopamine lesion model in behavioral brain research. Analysis of function defecits, recovery and treatments. Prog. Neurobiol. 50, 275–331 (1996). 44. Elgin, S. C. R. & Miller, D. W. in The genetics and biology of Drosophila, Vol 2a (eds. Ashburner, M. & Wright, T. R. F.) 112–121 (London: Academic, 1978). 45. Kume, K. A Drosophila dopamine transporter mutant, fumin (fmn), is defective in arousal regulation. Sleep Biol. Rhythms 4, 263–273 (2006). 46. Kume, K., Kume, S., Park, S. K., Hirsh, J. & Jackson, F. R. Dopamine Is a Regulator of Arousal in the Fruit Fly. J. Neurosci. 25, 7377–7384 (2005). 47. Kloppenburg, P., Ferns, D. & Mercer, A. R. Serotonin Enhances Central Olfactory Neuron Responses to Female Sex Pheromone in the Male Sphinx Moth Manduca sexta. J. Neurosci. 19, 8172–8181 (1999). 48. Gage, S. L., Daly, K. C. & Nighorn, A. Nitric oxide affects short-term olfactory memory in the antennal lobe of Manduca sexta. J. Exp. Biol. 216, 3294–3300 (2013). 49. Gage, S. L. & Nighorn, A. The role of nitric oxide in memory is modulated by diurnal time. Front. Syst. Neurosci. 8, 1–8 (2014). 50. Schotten, C. Ueber die Oxydation des Piperidins. Berichte der Dtsch. Chem. Gesellschaft 17, 2544–2547 (1884).

274 51. Baumann, E. Ueber eine einfache Methode der Darstellung von Benzoësäureäthern. Berichte der Dtsch. Chem. Gesellschaft 19, 3218–3222 (1886). 52. Kürti, L. & Czakó, B. Strategic Applications of Named Reactions in Organic Synthesis. (2005). 53. David, J. C. & Coulon, J. F. Octopamine in invertebrates and vertebrates. A review. Prog Neurobiol 24, 141–185 (1985). 54. Verlinden, H. et al. The role of octopamine in locusts and other arthropods. J. Insect Physiol. 56, 854–67 (2010). 55. Cooper, S. E. & Venton, B. J. Fast-scan cyclic voltammetry for the detection of tyramine and octopamine. Anal. Bioanal. Chem. 394, 329–36 (2009). 56. Hardie, S. L. & Hirsh, J. An improved method for the separation and detection of biogenic amines in adult Drosophila brain extracts by high performance liquid chromatography. J. Neurosci. Methods 153, 243–249 (2006). 57. Hudson, R. F. & Wardill, J. E. 350. The mechanism of hydrolysis of acid chlorides. Part I. The effect of hydroxyl ions, temperature, and substituents on the rate of hydrolysis of benzoyl chloride. J. Chem. Soc. 1729 (1950). doi:10.1039/jr9500001729 58. Hegstrand, L. R. & Eichelman, B. Determination of rat brain tissue catecholamines using liquid chromatography with electochemical detection. J. Chromatogr. B Biomed. Appl. 222, 107–111 (1981). 59. Harris, J. W. & Woodring, J. Effects of stress, age, season, and source colony on levels of octapamine, dopamine and serotonin in the honeybee (Apis mellifera L.) brain. 38, 29–35 (1992). 60. Taylor, D. J., Robinson, G. E., Logan, B. J., Laverty, R. & Mercer, A. R. Changes in brain amine levels associated with the morphological and behavioural development of the worker honeybee. J. Comp. Physiol. A 170, 715–721 (1992). 61. Harris, J. W. & Woodring, J. Elevated brain dopamine levels associated with ovary development in queenless worker honey bees (Apis mellifera L.). Comp. Biochem. Physiol. Part C Pharmacol. Toxicol. Endocrinol. 111, 271–279 (1995). 62. Wagener-Hulme, C., Kuehn, J. C., Schulz, D. J. & Robinson, G. E. Biogenic amines and division of labor in honey bee colonies. J. Comp. Physiol. A 184, 471–479 (1999). 63. Peso, M. et al. Physiology of reproductive worker honey bees (Apis mellifera): insights for the development of the worker caste. J. Comp. Physiol. A Neuroethol. Sensory, Neural, Behav. Physiol. 202, 147–158 (2016). 64. Søvik, E., Cornish, J. L. & Barron, A. B. Cocaine Tolerance in Honey Bees. PLoS One 8, e64920 (2013). 65. Mackenzie, S. M. et al. Mutations in the white gene of Drosophila melanogaster affecting ABC transporters that determine eye colouration. Biochim. Biophys. Acta - 1419, 173–185 (1999). 66. Morgan, T. H. Sex Limited Inheritance in Drosophila. Science (80-. ). 32, 120–122 (1910).

275 67. Borycz, J., Borycz, J. A., Kubów, A., Lloyd, V. & Meinertzhagen, I. A. Drosophila ABC transporter mutants white, brown and scarlet have altered contents and distribution of biogenic amines in the brain. J. Exp. Biol. 211, 3454–66 (2008). 68. Kuklinski, N. J., Berglund, E. C., Engelbrektsson, J. & Ewing, A. G. Biogenic Amines in Microdissected Brain Regions of Drosophila melanogaster Measured with Micellar Electrokinetic Capillary Chromatography—Electrochemical Detection. Anal. Chem. 82, 7729–7735 (2010). 69. Fang, H., Vickrey, T. L. & Venton, B. J. Analysis of biogenic amines in a single Drosophila larva brain by capillary electrophoresis with fast-scan cyclic voltammetry detection. Anal. Chem. 83, 2258–2264 (2011). 70. Yarali, A. et al. Genetic distortion of the balance between punishment and relief learning in Drosophila. J. Neurogenet. 23, 235–47 (2009). 71. Pörzgen, P., Park, S. K., Hirsh, J., Sonders, M. S. & Amara, S. G. The antidepressant-sensitive dopamine transporter in Drosophila melanogaster: a primordial carrier for catecholamines. Mol. Pharmacol. 59, 83–95 (2001). 72. Watson, D. G., Zhou, P., Midgley, J. M., Milligan, C. D. & Kaiser, K. The determination of biogenic amines in four strains of the fruit fly Drosophila melanogaster. 11, 1145–1149 (1993). 73. Fang, H., Pajski, M. L., Ross, A. E. & Venton, B. J. Quantitation of dopamine, serotonin and adenosine content in a tissue punch from a brain slice using capillary electrophoresis with fast-scan cyclic voltammetry detection. Anal. Methods 5, 2704–2711 (2013). 74. Denno, M. E., Privman, E. & Venton, B. J. Analysis of Neurotransmitter Tissue Content of Drosophila melanogaster in Different Life Stages. ACS Chem. Neurosci. 6, 117–123 (2015). 75. Borycz, J., Vohra, M., Tokarczyk, G. & Meinertzhagen, I. A. The determination of histamine in the Drosophila head. J. Neurosci. Methods 101, 141–148 (2000). 76. Borycz, J. A., Borycz, J., Kubów, A., Kostyleva, R. & Meinertzhagen, I. A. Histamine compartments of the Drosophila brain with an estimate of the quantum content at the photoreceptor synapse. J. Neurophysiol. 93, 1611–9 (2005). 77. Neckameyer, W. S., Woodrome, S., Holt, B. & Mayer, A. Dopamine and senescence in Drosophila melanogaster. Neurobiol. Aging 21, 145–152 (2000). 78. Stuart, A. E. From fruit flies to barnacles, histamine is the neurotransmitter of arthropod photoreceptors. Neuron 22, 431–3 (1999). 79. Nässel, D. R. Histamine in the brain of insects: a review. Microsc. Res. Tech. 44, 121–36 (1999). 80. Denno, M. E., Privman, E., Borman, R. P., Wolin, D. C. & Venton, B. J. Quantification of Histamine and Carcinine in Drosophila melanogaster Tissues. ACS Chem. Neurosci. 7, 407–414 (2016). 81. Borycz, J., Borycz, J. a, Loubani, M. & Meinertzhagen, I. a. Tan and Ebony Genes Regulate a Novel Pathway for Transmitter Metabolism At Fly Photoreceptor Terminals. J. Neurosci. 22, 10549–10557 (2002). 82. Hardie, R. C. Is histamine a neurotransmitter in insect photoreceptors? J. Comp. Physiol. A 161, 201–213 (1987).

276 Chapter 3: Nosema ceranae parasitism in honey bees (Apis mellifera) increases biogenic amines associated with foraging behavior and alters olfactory learning and memory 1. Goulson, D., Nicholls, E., Botias, C. & Rotheray, E. L. Bee declines driven by combined stress from parasites, pesticides, and lack of flowers. Science (80-. ). 347, (2015). 2. Chen, Y. P. et al. Morphological, Molecular, and Phylogenetic Characterization of Nosema ceranae , a Microsporidian Parasite Isolated from the European Honey Bee, Apis mellifera. J. Eukaryot. Microbiol. 56, 142–147 (2009). 3. Higes, M., Martín-Hernández, R., García-Palencia, P., Marín, P. & Meana, A. Horizontal transmission of Nosema ceranae (Microsporidia) from worker honeybees to queens (Apis mellifera). Environ. Microbiol. Rep. 1, 495–498 (2009). 4. Alaux, C. et al. Pathological effects of the microsporidium Nosema ceranae on honey bee queen physiology (Apis mellifera). J. Invertebr. Pathol. 106, 380–385 (2011). 5. Fries, I., Feng, F., da Silva, A., Slemenda, S. B. & Pieniazek, N. J. Nosema ceranae n. sp. (Microspora, Nosematidae), morphological and molecular characterization of a microsporidian parasite of the Asian honey bee Apis cerana (Hymenoptera, Apidae). Eur. J. Protistol. 32, 356–365 (1996). 6. Higes, M., Martín, R. & Meana, A. Nosema ceranae, a new microsporidian parasite in honeybees in Europe. J. Invertebr. Pathol. 92, 93–95 (2006). 7. Chen, Y., Evans, J. D., Smith, I. B. & Pettis, J. S. Nosema ceranae is a long-present and wide-spread microsporidian infection of the European honey bee (Apis mellifera) in the United States. J. Invertebr. Pathol. 97, 186–188 (2008). 8. Holt, H. L., Aronstein, K. A. & Grozinger, C. M. Chronic parasitization by Nosema microsporidia causes global expression changes in core nutritional, metabolic and behavioral pathways in honey bee workers (Apis mellifera). BMC Genomics 14, (2013). 9. McDonnell, C. M. et al. Ecto- and endoparasite induce similar chemical and brain neurogenomic responses in the honey bee (Apis mellifera). BMC Ecol. 13, (2013). 10. Mayack, C. & Natsopoulou, M. E. Nosema ceranae alters a highly conserved hormonal stress pathway in honeybees. Insect Mol. Biol. 24, 662–670 (2015). 11. Higes, M., García-Palencia, P., Martín-Hernández, R. & Meana, A. Experimental infection of Apis mellifera honeybees with Nosema ceranae (Microsporidia). J. Invertebr. Pathol. 94, 211–217 (2007). 12. Mayack, C. & Naug, D. Energetic stress in the honeybee Apis mellifera from Nosema ceranae infection. J. Invertebr. Pathol. 100, 185–188 (2009). 13. Wang, D.-I. & Moeller, F. E. Histological Comparisons of the Development of Hypopharyngeal Glands in Healthy and Nosema-Infected Worker Honey Bees. J. Invertebr. Pathol. 14, 135–142 (1969). 14. Alaux, C. et al. Interactions between Nosema microspores and a neonicotinoid weaken honeybees (Apis mellifera). Environ. Microbiol. 12, 774–782 (2010). 15. Jack, C. J., Uppala, S. S., Lucas, H. M. & Sagili, R. R. Effects of pollen dilution on

277 infection of Nosema ceranae in honey bees. J. Insect Physiol. 87, 12–19 (2016). 16. Vidau, C. et al. Differential proteomic analysis of midguts from Nosema ceranae- infected honeybees reveals manipulation of key host functions. J. Invertebr. Pathol. 121, 89–96 (2014). 17. Naug, D. & Gibbs, A. Behavioral changes mediated by hunger in honeybees infected with Nosema ceranae. Apidologie 40, 595–599 (2009). 18. Wang, D.-I. & Moeller, F. E. The Division of Labor and Queen Attendance Behavior of Nosema-Infected Worker Honey Bees. J. Econ. Entomol. 63, 1539–1541 (1970). 19. Dussaubat, C. et al. Flight behavior and pheromone changes associated to Nosema ceranae infection of honey bee workers (Apis mellifera) in field conditions. J. Invertebr. Pathol. 113, 42–51 (2013). 20. Goblirsch, M., Huang, Z. Y. & Spivak, M. Physiological and Behavioral Changes in Honey Bees (Apis mellifera) Induced by Nosema ceranae Infection. PLoS One 8, 1–8 (2013). 21. Lecocq, A., Jensen, A. B., Kryger, P. & Nieh, J. C. Parasite infection accelerates age polyethism in young honey bees. Sci. Rep. 6, 1–11 (2016). 22. Natsopoulou, M. E., McMahon, D. P. & Paxton, R. J. Parasites modulate within- colony activity and accelerate the temporal polyethism schedule of a social insect, the honey bee. Behav Ecol Sociobiol 70, 1019–1031 (2016). 23. Higes, M. et al. How natural infection by Nosema ceranae causes honeybee colony collapse. Environ. Microbiol. 10, 2659–2669 (2008). 24. Perry, C. J., Søvik, E., Myerscough, M. R. & Barron, A. B. Rapid behavioral maturation accelerates failure of stressed honey bee colonies. PNAS 112, 3427– 3432 (2015). 25. Barron, A. B. Death of the bee hive: understanding the failure of an insect society. Curr. Opin. Insect Sci. 10, 45–50 (2015). 26. Kralj, J. & Fuchs, S. Nosema sp. influences flight behavior of infected honey bee (Apis mellifera) foragers. Apidologie 41, 21–28 (2010). 27. Dussaubat, C. et al. Nosema spp. Infection Alters Pheromone Production in Honey Bees (Apis mellifera). J Chem Ecol 36, 522–525 (2010). 28. Wolf, S. et al. So Near and Yet So Far: Harmonic Radar Reveals Reduced Homing Ability of Nosema Infected Honeybees. PLoS One 9, 1–15 (2014). 29. Woyciechowski, M. & Kozlowski, J. Division of labor by division of risk according to worker life expectancy in the honey bee (Apis mellifera L.). Apidologie 29, 191–205 (1998). 30. Kuszewska, K. & Woyciechowski, M. Risky robbing is a job for short-lived and infected worker honeybees. Apidologie 45, 537–544 (2014). 31. Alaux, C., Crauser, D., Pioz, M., Saulnier, C. & Le Conte, Y. Parasitic and immune modulation of flight activity in honey bees tracked with optical counters. J. Exp. Biol. 217, 3416–3424 (2014). 32. Dosselli, R., Grassl, J., Carson, A., Simmons, L. W. & Baer, B. Flight behaviour of honey bee (Apis mellifera) workers is altered by initial infections of the fungal parasite Nosema apis. Sci. Rep. 6, 1–11 (2016).

278 33. Esch, H. in Honeybee Neurobiology and Behavior: A Tribute to Randolf Menzel 53– 64 (2012). 34. Fries, I. et al. Standard methods for Nosema research. J. Apic. Res. 52, 1–28 (2013). 35. Menzel, R., Manz, G., Menzel, R. & Greggers, U. Massed and Spaced Learning in Honeybees : The Role of CS, US, the Intertrial Interval, and the Test Interval. Learn. Mem. 8, 198–208 (2001). 36. Manneburg, M., Lahm, H.-W. & Fountoulakis, M. Oxidation of Cysteine and Methionine Residues during Acid Hydrolysis of Proteins in the Presence of Sodium Azide. Anal. Biochem. 224, 122–127 (1995). 37. Fountoulakis, M. & Lahm, H.-W. Hydrolysis and amino acid composition analysis of proteins. J. Chromatogr. A 826, 109–134 (1998). 38. Elkin, R. G. & Wasynczuk, A. M. Amino Acid Analysis of Feedstuff Hydrolysates by Precolunn Derivatization with Phenylisothiocyanate and Reversed-Phase High- Performance Liquid Chromotography. Cereal Chem. 64, 226–229 (1987). 39. Kwanyuen, P. & Burton, J. W. A Modified Amino Acid Analysis Using PITC Derivatization for Soybeans with Accurate Determination of Cysteine and Half- Cystine. J. Am. Oil Chem. Soc. 87, 127–132 (2010). 40. Kloppenburg, P., Ferns, D. & Mercer, A. R. Serotonin Enhances Central Olfactory Neuron Responses to Female Sex Pheromone in the Male Sphinx Moth Manduca sexta. J. Neurosci. 19, 8172–8181 (1999). 41. Gage, S. L., Daly, K. C. & Nighorn, A. Nitric oxide affects short-term olfactory memory in the antennal lobe of Manduca sexta. J. Exp. Biol. 216, 3294–3300 (2013). 42. Gage, S. L. & Nighorn, A. The role of nitric oxide in memory is modulated by diurnal time. Front. Syst. Neurosci. 8, 1–8 (2014). 43. Song, P., Mabrouk, O. S., Hershey, N. D. & Kennedy, R. T. In vivo neurochemical monitoring using benzoyl chloride derivatization and liquid chromatography-mass spectrometry. Anal. Chem. 84, 412–419 (2012). 44. Gauthier, M. & Grünewald, B. in Honeybee Neurobiology and Behavior: A Tribute to Randolf Menzel 155–170 (2012). 45. Charbonneau, L. R., Hillier, N. K., Rogers, R. E. L., Williams, G. R. & Shutler, D. Effects of Nosema apis, N. ceranae, and coinfections on honey bee (Apis mellifera) learning and memory. Sci. Rep. 6, 1–7 (2016). 46. Deyl, Z., Hyanek, J. & Horakova, M. Profiling of amino acids in body fluids and tissues by means of liquid chromatography. J. Chromatogr. 379, 177–250 (1986). 47. Spackman, D. H., Stein, W. H. & Moore, S. Automatic recording apparatus for use in the chromatography of amino acids. Anal. Chem. 30, 1190–1206 (1958). 48. Hill, R. L. Hydrolysis of proteins. Adv. Protein Chem. 20, 37–107 (1965). 49. Tuppy, V. H. ( ~ ber die beim tryptischen Abbau yon Pferde-Cytochrom c entstehenden Peptide. 108, (1962). 50. Matsubara, H. & Sasaki, R. M. High recovery of tryptophan from acid hydrolysates of proteins. Biochem. Biophys. Res. Commun. 35, 175–181 (1969).

279 51. Penke, B., Ferenczi, R. & Kovács, K. A new acid hydrolysis method for determining tryptophan in peptides and proteins. Anal. Biochem. 60, 45–50 (1974). 52. Inglis, A. S. & Liu, T.-Y. The Stability of Cysteine and Cystine during Acid Hydrolysis of Proteins and Peptides. J. Biol. Chem. 245, 112–116 (1970). 53. Spies, J. R. Determination of tryptophan in proteins. Anal. Chem. 39, 1412–6 (1967). 54. Manneberg, M., Lahm, H. W. & Fountoulakis, M. Quantification of cysteine residues following oxidation to cysteic acid in the presence of sodium azide. Anal. Biochem. 231, 349–53 (1995). 55. Namera, A., Yashiki, M., Nishida, M. & Kojima, T. Direct extract derivatization for determination of amino acids in human urine by gas chromatography and mass spectrometry. J. Chromatogr. B. Analyt. Technol. Biomed. Life Sci. 776, 49–55 (2002). 56. Bidlingmeyer, B. A., Cohen, S. A. & Tarvin, T. L. Rapid analysis of amino acids using pre-column derivatization. J. Chromatogr. 336, 93–104 (1984). 57. Wu, G. Amino acids: metabolism, functions, and nutrition. Amino Acids 37, 1–17 (2009). 58. Bicker, G. Biogenic Amines in the Brain of the Honeybee: Cellular Distribution, Development, and Behavioral Functions. Microsc. Res. Tech. 44, 166–178 (1999). 59. Scheiner, R., Baumann, A. & Blenau, W. Aminergic Control and Modulation of Honeybee Behavior. Curr. Neuropharmacol. 4, 259–276 (2006). 60. Brandes, C., Sugawa, M. & Menzel, R. HPLC measurements of catecholamines in single honeybee brain reveals caste specific differences bw worker and queens in A. mellifera. 9, 53–57 (1990). 61. Harris, J. W. & Woodring, J. Effects of stress, age, season, and source colony on levels of octapamine, dopamine and serotonin in the honeybee (Apis mellifera L.) brain. 38, 29–35 (1992). 62. Harris, J. W. & Woodring, J. Elevated brain dopamine levels associated with ovary development in queenless worker honey bees (Apis mellifera L.). Comp. Biochem. Physiol. Part C Pharmacol. Toxicol. Endocrinol. 111, 271–279 (1995). 63. Wagener-Hulme, C., Kuehn, J. C., Schulz, D. J. & Robinson, G. E. Biogenic amines and division of labor in honey bee colonies. J. Comp. Physiol. A 184, 471–479 (1999). 64. Sasaki, K. & Nagao, T. Distribution and levels of dopamine and its metabolites in brains of reproductive workers in honeybees. J. Insect Physiol. 47, 1205–1216 (2001). 65. Søvik, E., Cornish, J. L. & Barron, A. B. Cocaine Tolerance in Honey Bees. PLoS One 8, e64920 (2013). 66. Peso, M. et al. Physiology of reproductive worker honey bees (Apis mellifera): insights for the development of the worker caste. J. Comp. Physiol. A Neuroethol. Sensory, Neural, Behav. Physiol. 202, 147–158 (2016). 67. Schulz, D. J., Sullivan, J. P. & Robinson, G. E. Juvenile Hormone and Octopamine in the Regulation of Division of Labor in Honey Bee Colonies. Horm. Behav. 42,

280 222–231 (2002). 68. Schulz, D. J., Barron, A. B. & Robinson, G. E. A Role for Octopamine in Honey Bee Division of Labor. Brain Behav Evol 60, 350–359 (2002). 69. Barron, A. B., Schulz, D. J. & Robinson, G. E. Octopamine modulates responsiveness to foraging-related stimuli in honey bees (Apis mellifera). J. Comp. Physiol. A 188, 603–610 (2002). 70. Schulz, D. J. & Robinson, G. E. Octopamine influences division of labor in honey bee colonies. J. Comp. Physiol. A 187, 53–61 (2001). 71. Behrends, A. & Scheiner, R. Octopamine improves learning in newly emerged bees but not in old foragers. J. Exp. Biol. 215, 1076–1083 (2012). 72. Schulz, D. J. & Robinson, G. E. Biogenic amines and division of labor in honey bee colonies: behaviorally related changes in the antennal lobes and age-related changes in the mushroom bodies. J. Comp. Physiol. A 184, 481–488 (1999). 73. Fleming, J. C., Schmehl, D. R. & Ellis, J. D. Characterizing the impact of commercial pollen substitute diets on the level of Nosema spp. in Honey Bees (Apis mellifera L.). PLoS One 10, 1–14 (2015). 74. Ushitani, T., Perry, C. J., Cheng, K. & Barron, A. B. Accelerated behavioural development changes fine-scale search behaviour and spatial memory in honey bees (Apis mellifera L.). J. Exp. Biol. 219, 412–418 (2016). 75. Campbell, J., Kessler, B., Mayack, C. & Naug, D. Behavioural fever in infected honeybees: parasitic manipulation or coincidental benefit? Parasitology 137, 1487– 91 (2010). 76. Adamo, S. A. Parasites: evolutionʼs neurobiologists. J. Exp. Biol. 216, 3–10 (2013). 77. Perrot-Minnot, M.-J. & Cézilly, F. Investigating candidate neuromodulatory systems underlying parasitic manipulation: concepts, limitations and prospects. J. Exp. Biol. 216, 134–141 (2013). 78. Helluy, S. Parasite-induced alterations of sensorimotor pathways in gammarids: collateral damage of neuroinflammation? J. Exp. Biol. 216, 67–77 (2013). 79. Adamo, S. A. Parasitic Suppression of Feeding in the Tobacco Hornworm, Manduca sexta: Parallels With Feeding Depression After an Immune Challenge. Arch. Insect Biochem. Physiol. 60, 185–197 (2005). 80. Libersat, F., Delago, A. & Gal, R. Manipulation of Host Behavior by Parasitic Insects and Insect Parasites. Annu. Rev. Entomol. 54, 189–207 (2009). 81. Adamo, S. A. Modulating the Modulators: Parasites, Neuromodulators and Host Behavioral Change. Brain. Behav. Evol. 60, 370–377 (2002). 82. Dantzer, R., O’Connor, J. C., Freund, G. G., Johnson, R. W. & Kelley, K. W. From inflammation to sickness and depression: when the immune system subjugates the brain. Nat. Rev. Neurosci. 9, 46–57 (2008). 83. De Simoni, M. G. & Imeri, L. Cytokine-Neurotransmitter Interactions in the Brain. Biol Signals Recept 7, 33–44 (1998). 84. Adamo, S. A. Why should an immune response activate the stress response? Insights from the insects (the cricket Gryllus texensis). Brain. Behav. Immun. 24, 194–200 (2010).

281 85. de Jong-Brink, M., Bergamin-Sassen, M. & Soto, M. S. Multiple strategies of schistosomes to meet their requirements in the intermediate snail host. Parasitology 123, S129–S141 (2001). 86. Antúnez, K. et al. Immune suppression in the honey bee (Apis mellifera) following infection by Nosema ceranae (Microsporidia). Environ. Microbiol. 11, 2284–2290 (2009). 87. Li, W. et al. Silencing the Honey Bee (Apis mellifera) Naked Cuticle Gene (nkd) Improves Host Immune Function and Reduces Nosema ceranae Infections. Appl. Environ. Microbiol. 82, 6779–6787 (2016).

Chapter 4: Glycosylation of peptide-based drug candidates improves in vivo stability and penetration of the blood-brain barrier 1. Jones, E. M. & Polt, R. CNS active O-linked glycopeptides. Front. Chem. 3, 40 (2015). 2. Kaspar, A. A. & Reichert, J. M. Future directions for peptide therapeutics development. Drug Discov. Today 18, 807–817 (2013). 3. Fosgerau, K. & Hoffmann, T. Peptide therapeutics: Current status and future directions. Drug Discov. Today 20, 122–128 (2015). 4. Implications of the Blood-Brain Barrier and Its Manipulation. (Plenum Publishing Corporation, 1989). 5. Abbott, N. J., Patabendige, A. A. K., Dolman, D. E. M., Yusof, S. R. & Begley, D. J. Structure and function of the blood-brain barrier. Neurobiol. Dis. 37, 13–25 (2010). 6. Pardridge, W. M. Blood-brain barrier delivery. Drug Discov. Today 12, 54–61 (2007). 7. Seelig, A., Gottschlich, R. & Devant, R. M. A method to determine the ability of drugs to diffuse through the blood-brain barrier. Proc Natl Acad Sci U S A 91, 68– 72 (1994). 8. Pardridge, W. M. Drug and gene delivery to the brain: The vascular route. Neuron 36, 555–558 (2002). 9. Gentry, C. L. et al. The effect of halogenation on blood-brain barrier permeability of a novel peptide drug. Peptides 20, 1229–1238 (1999). 10. Schinkel, A. H., Wagenaar, E., Mol, C. A. A. M. & Deemter, L. Van. P-Glycoprotein in the Blood-Brain Barrier of Mice Influences the Brain Penetration and Pharmacological Activity of Many Drugs. 2517–2524 11. Egleton, R. D. & Davis, T. P. Development of neuropeptide drugs that cross the blood-brain barrier. NeuroRX 2, 44–53 (2005). 12. Egleton, R. D. & Davis, T. P. Bioavailability and transport of peptides and peptide drugs into the brain. Peptides 18, 1431–1439 (1997). 13. Banks, W. A. Characteristics of compounds that cross the blood-brain barrier. BMC Neurol. 9, S3 (2009). 14. Sato, A. K., Viswanathan, M., Kent, R. B. & Wood, C. R. Therapeutic peptides: technological advances driving peptides into development. Curr. Opin. Biotechnol.

282 17, 638–642 (2006). 15. Sasongko, L. et al. Imaging P-glycoprotein transport activity at the human blood- brain barrier with positron emission tomography. Clin. Pharmacol. Ther. 77, 503– 514 (2005). 16. Yano, Y., Budinger, T. F., Friedland, R. P., Derenzo, S. E. & Huesman, R. H. Brain Tumor Evaluation Using Rb-82 and Positron Emission Tomography. J Nucl Med 23, 532–538 (1982). 17. Yamamizu, K. et al. In Vitro Modeling of Blood-Brain Barrier with Human iPSC- Derived Endothelial Cells, Pericytes, Neurons, and Astrocytes via Notch Signaling. Stem Cell Reports 8, 634–647 (2017). 18. Franke, H., Galla, H. J. & Beuckmann, C. T. Primary cultures of brain microvessel endothelial cells: A valid and flexible model to study drug transport through the blood-brain barrier in vitro. Brain Res. Protoc. 5, 248–256 (2000). 19. Dehouck, M. ‐ P, Méresse, S., Delorme, P., Fruchart, J. ‐ C & Cecchelli, R. An Easier, Reproducible, and Mass‐ Production Method to Study the Blood–Brain Barrier In Vitro. J. Neurochem. 54, 1798–1801 (1990). 20. Cecchelli, R. et al. Modelling of the blood-brain barrier in drug discovery and development. Nat. Rev. Drug Discov. 6, 650–661 (2007). 21. de Lange, E. C., Danhof, M., de Boer, a G. & Breimer, D. D. Methodological considerations of intracerebral microdialysis in pharmacokinetic studies on drug transport across the blood-brain barrier. Brain Res. Brain Res. Rev. 25, 27–49 (1997). 22. Gardner, E. L., Chen, J. & Paredes, W. Overview of chemical sampling techniques. J. Neurosci. Methods 48, 173–197 (1993). 23. de Lange, E. C. M., Danhof, M., de Boer, A. G. & Breimer, D. D. Critical factors of intracerebral microdialysis as a technique to determined the pharmacokinetics of drugs in rat brain. Brain Res. 666, 1–8 (1994). 24. De Lange, E. C. M., De Boer, B. A. G. & Breimer, D. D. Microdialysis for pharmacokinetic analysis of drug transport to the brain. Adv. Drug Deliv. Rev. 36, 211–227 (1999). 25. Nakazono, T., Murakami, T., Sakai, S., Higashi, Y. & Yata, N. Application of microdialysis for study of caffeine distribution into brain and cerebrospinal fluid in rats. Chem. Pharm. Bull. (Tokyo). 40, 2510–5 (1992). 26. Witt, K. A., Gillespie, T. J., Huber, J. D., Egleton, R. D. & Davis, T. P. Peptide drug modifications to enhance bioavailability and blood-brain barrier permeability. Peptides 22, 2329–2343 (2001). 27. Bilsky, E. J. et al. Enkephalin glycopeptide analogues produce analgesia with reduced dependence liability. J. Med. Chem. 43, 2586–2590 (2000). 28. Lowery, J. J. et al. In vivo characterization of MMP-2200, a mixed δ/μ opioid agonist, in mice. J. Pharmacol. Exp. Ther. 336, 767–78 (2011). 29. Mabrouk, O. S., Falk, T., Sherman, S. J. & Kennedy, R. T. CNS Penetration of the Opioid Glycopeptide MMP-2200: A Microdialysis Study. 531, 99–103 (2013). 30. Lee, S. et al. Effect of a Selective Mas Receptor Agonist in Cerebral Ischemia In

283 Vitro and In Vivo. PLoS One 10, e0142087 (2015). 31. Pereira, R. M. et al. The renin-angiotensin system in a rat model of hepatic fibrosis: Evidence for a protective role of Angiotensin-(1-7). J. Hepatol. 46, 674–681 (2007). 32. Kluskens, L. D. et al. Angiotensin- (1–7) with Thioether Bridge : An Angiotensin- Converting Enzyme-Resistant, Potent Angiotensin-(1-7) Analog. Pharmacology 328, 4–7 (2009). 33. Yue, X. et al. Effects of the novel glycopeptide opioid agonist MMP-2200 in preclinical models of Parkinson’s disease. Brain Res. 1413, 72–83 (2011). 34. Mabrouk, O. S., Falk, T., Sherman, S. J., Kennedy, R. T. & Polt, R. CNS Penetration of the opioid glycopeptides MMP-2200: A microdialysis study. Neurosci. Lett. 531, 99–103 (2012). 35. Miyata, A. et al. Isolation of a neuropeptide corresponding to the N-terminal 27 residues of the pituitary adenylate cyclase activating polypeptide with 38 residues (PACAP38). Biochem. Biophys. Res. Commun. 170, 643–8 (1990). 36. Vaudry, D. et al. Pituitary Adenylate Cyclase-Activating Polypeptide and Its Receptors : 20 Years after the Discovery. Pept. Res. 61, 283–357 (2009). 37. Sherwood, N. M., Krueckl, S. L. & McRory, J. E. The origin and function of the pituitary adenylate cyclase-activating polypeptide (PACAP)/glucagon superfamily. Endocr. Rev. 21, 619–70 (2000). 38. Arimura, a et al. PACAP functions as a neurotrophic factor. Ann. N. Y. Acad. Sci. 739, 228–43 (1994). 39. Tamas, A. et al. Effect of PACAP in central and peripheral nerve injuries. Int. J. Mol. Sci. 13, 8430–8448 (2012). 40. Reglodi, D., Kiss, P., Lubics, A. & Tamas, A. Review on the Protective Effects of PACAP in Models of Neurodegenerative Diseases In Vitro and In Vivo. Curr. Pharm. Des. 17, 962–972 (2011). 41. Somogyvari-Vigh, A. & Reglodi, D. Pituitary Adenylate Cyclase Activating Polypeptide: A Potential Neuroprotective Peptide. Curr. Pharm. Des. 10, 2861– 2889 (2004). 42. Ohtaki, H., Nakamachi, T., Dohi, K. & Shioda, S. Role of PACAP in ischemic neural death. J. Mol. Neurosci. 36, 16–25 (2008). 43. Dejda, A., Sokołowska, P. & Nowak, J. Z. Neuroprotective potential of three neuropeptides PACAP, VIP and PHI. Pharmacol. Rep. 57, 307–20 (2005). 44. Mitchell, S. A., Pratt, M. R., Hruby, V. J. & Polt, R. Solid-phase synthesis of O-linked glycopeptide analogues of enkephalin. J. Org. Chem. 66, 2327–2342 (2001). 45. Polt, R., Szabo, L., Treiberg, J., Li, Y. & Hruby, V. J. General-Methods for Alpha- or Beta-O-Ser/Thr or Beta-O-Ser/Thr Glycosides and Glycopeptides. Solid-Phase Synthesis of O-Glycosyl Cyclic Enkephalin Analogs. J Am Chem Soc 114, 10249– 10258 (1992). 46. Paizs, B. & Suhai, S. Fragmentation pathways of protonated peptides. Mass Spectrom. Rev. 24, 508–48 (2005).

284 Chapter 5: Identification of optimal structural modifications for the improvement of peptide-based drug delivery properties: an Angiotensin 1-7 study of in vivo stability and BBB penetration 1. Santos, R. A. S. Angiotensin-(1-7). Hypertension 63, 1138–1147 (2014). 2. Ramalingam, L. et al. The renin angiotensin system, oxidative stress and mitochondrial function in obesity and insulin resistance. Biochim. Biophys. Acta 1863, 1106–1114 (2017). 3. Almeida-Santos, A. F. et al. Anxiolytic- and antidepressant-like effects of angiotensin-(1-7) in hypertensive transgenic (mRen2)27 rats. Clin. Sci. 130, 1247– 55 (2016). 4. Kangussu, L. M. et al. Reduced anxiety-like behavior in transgenic rats which chronically overexpress angiotensin-(1-7): Role of the Mas receptor. Behav. Brain Res. 331, 193–198 (2017). 5. Fontes, M. A. P., Martins Lima, A. & Santos, R. A. S. Brain angiotensin-(1-7)/Mas axis: A new target to reduce the cardiovascular risk to emotional stress. Neuropeptides 56, 9–17 (2016). 6. Donoghue, M. et al. A Novel Angiotensin-Converting Enzyme-Related Carboxypeptidase (ACE2) Converts Angiotensin I to Angiotensin 1-9. Circ. Res. 87, e1–e9 (2000). 7. Santos, R. A. S. et al. Angiotensin-(1-7) is an endogenous ligand for the G protein- coupled receptor Mas. Proc. Natl. Acad. Sci. U. S. A. 100, 8258–63 (2003). 8. Lee, S. et al. Effect of a Selective Mas Receptor Agonist in Cerebral Ischemia In Vitro and In Vivo. PLoS One 10, e0142087 (2015). 9. Pereira, R. M. et al. The renin-angiotensin system in a rat model of hepatic fibrosis: Evidence for a protective role of Angiotensin-(1-7). J. Hepatol. 46, 674–681 (2007). 10. Kluskens, L. D. et al. Angiotensin- (1–7) with Thioether Bridge : An Angiotensin- Converting Enzyme-Resistant, Potent Angiotensin-(1-7) Analog. Pharmacology 328, 4–7 (2009). 11. Recio, C., Maione, F., Iqbal, A. J., Mascolo, N. & De Feo, V. The potential therapeutic application of peptides and peptidomimetics in cardiovascular disease. Front. Pharmacol. 7, 1–11 (2017). 12. Fosgerau, K. & Hoffmann, T. Peptide therapeutics: Current status and future directions. Drug Discov. Today 20, 122–128 (2015). 13. Egleton, R. D. & Davis, T. P. Bioavailability and transport of peptides and peptide drugs into the brain. Peptides 18, 1431–1439 (1997). 14. Gabathuler, R. Approaches to transport therapeutic drugs across the blood-brain barrier to treat brain diseases. Neurobiol. Dis. 37, 48–57 (2010). 15. Abbott, N. J., Patabendige, A. A. K., Dolman, D. E. M., Yusof, S. R. & Begley, D. J. Structure and function of the blood-brain barrier. Neurobiol. Dis. 37, 13–25 (2010). 16. Pardridge, W. M. Blood-brain barrier delivery. Drug Discov. Today 12, 54–61 (2007). 17. Witt, K. A., Gillespie, T. J., Huber, J. D., Egleton, R. D. & Davis, T. P. Peptide drug

285 modifications to enhance bioavailability and blood-brain barrier permeability. Peptides 22, 2329–2343 (2001). 18. Kastin, A. (ed. . Handbook of Biologically Active Peptides. (2013). 19. Jones, E. M. & Polt, R. CNS active O-linked glycopeptides. Front. Chem. 3, 40 (2015). 20. Powell, M. F. et al. Peptide Stability in Drug Development. II. Effect of Single Amino Acid Substitution and Glycosylation on Peptide Reactivity in Human Serum. Pharmaceutical Research 10, 1268–1273 (1993). 21. Varamini, P. & Toth, I. Lipid- and sugar-modified endomorphins: novel targets for the treatment of neuropathic pain. Front. Pharmacol. 4, 155 (2013). 22. Rink, R. et al. To protect peptide pharmaceuticals against peptidases. J. Pharmacol. Toxicol. Methods 61, 210–218 (2010). 23. Sato, A. K., Viswanathan, M., Kent, R. B. & Wood, C. R. Therapeutic peptides: technological advances driving peptides into development. Curr. Opin. Biotechnol. 17, 638–642 (2006). 24. Egleton, R. D. & Davis, T. P. Development of neuropeptide drugs that cross the blood-brain barrier. NeuroRX 2, 44–53 (2005). 25. de Vries, L. et al. Oral and pulmonary delivery of thioether-bridged angiotensin-(1- 7). Peptides 31, 893–898 (2010). 26. Yue, X. et al. Effects of the novel glycopeptide opioid agonist MMP-2200 in preclinical models of Parkinson’s disease. Brain Res. 1413, 72–83 (2011). 27. Mitchell, S. A., Pratt, M. R., Hruby, V. J. & Polt, R. Solid-phase synthesis of O-linked glycopeptide analogues of enkephalin. J. Org. Chem. 66, 2327–2342 (2001). 28. Mizuma, T., Ohta, K., Koyanagi, A. & Awazu, S. Improvement of intestinal absorption of leucine enkephalin by sugar coupling and peptidase inhibitors. J. Pharm. Sci. 85, 854–857 (1996). 29. Fan, J. Q. & Lee, Y. C. Detailed studies on substrate structure requirements of glycoamidases A and F. J. Biol. Chem. 272, 27058–27064 (1997). 30. Polt, R., Szabo, L., Treiberg, J., Li, Y. & Hruby, V. J. General-Methods for Alpha- or Beta-O-Ser/Thr or Beta-O-Ser/Thr Glycosides and Glycopeptides. Solid-Phase Synthesis of O-Glycosyl Cyclic Enkephalin Analogs. J Am Chem Soc 114, 10249– 10258 (1992). 31. Vanhoof, G., Goossens, F., De Meester, I., Hendriks, D. & Scharpe, S. Proline motifs in peptides and their biological processing. FASEB J. 9, 736–744 (1995). 32. Whicher, J. & Spence, C. When is serum albumin worth measuring? Ann. Clin. Biochem. 24 ( Pt 6), 572–580 (1987). 33. Bradshaw, R. A. & Peters Jr., T. The Amino Acid Sequence of Peptide (1-24) of Rat and Human Serum Albumins. J. Biol. Chem. 244, 5582–5589 (1969). 34. Albumin, Serum. Mayo Clinic (2017). Available at: https://www.mayomedicallaboratories.com/test- catalog/Clinical+and+Interpretive/8436. 35. Zaias, J., Mineau, M., Cray, C., Yoon, D. & Altman, N. H. Reference values for

286 serum proteins of common laboratory rodent strains. J. Am. Assoc. Lab. Anim. Sci. 48, 387–390 (2009). 36. Allred, A. J., Diz, D. I., Ferrario, C. M. & Chappell, M. C. Pathways for angiotensin- (1-7) metabolism in pulmonary and renal tissues. Am. J. Physiol Ren. Physiol 279, F841–F850 (2000).

Chapter 6: Conclusions and future directions 1. Hwang, J. H. & Yaksh, T. L. The effect of spinal GABA receptor agonists on tactile allodynia in a surgically-induced neuropathic pain model in the rat. Pain 70, 15–22 (1997). 2. Enna, S. J. & McCarson, K. E. The Role of GABA in the Mediation and Perception of Pain. Adv. Pharmacol. 54, 1–27 (2006). 3. Faull, K. F., DoAmaral, J. R., Berger, P. A. & Barchas, J. D. Mass spectrometric identification and selected ion monitoring quantitation of gamma-amino-butyric acid (GABA) in human lumbar cerebrospinal fluid. J. Neurochem. 31, 1119–1122 (1978). 4. Takada, Y., Yoshida, M., Sakairi, M. & Koizumi, H. Detection of γ-aminobutyric acid in a rat brain usingin vivo microdialysis-capillary electrophoresis/mass spectrometry. Rapid Commun. Mass Spectrom. 9, 895–896 (1995). 5. Buck, K., Voehringer, P. & Ferger, B. Rapid analysis of GABA and glutamate in microdialysis samples using high performance liquid chromatography and tandem mass spectrometry. J. Neurosci. Methods 182, 78–84 (2009). 6. Genersch, E. et al. The German bee monitoring project: a long term study to understand periodically high winter losses of honey bee colonies. Apidologie 41, 332–352 (2010). 7. Genersch, E. Honey bee pathology: current threats to honey bees and beekeeping. Appl. Microbiol. Biotechnol. 87, 87–97 (2010). 8. Dainat, B., Evans, J. D., Chen, Y. P., Gauthier, L. & Neumann, P. Dead or alive: deformed wing virus and Varroa destructor reduce the life span of winter honeybees. Appl. Environ. Microbiol. 78, 981–7 (2012). 9. van Dooremalen, C. et al. Winter survival of individual honey bees and honey bee colonies depends on level of Varroa destructor infestation. PLoS One 7, e36285 (2012). 10. Kralj, J. & Fuchs, S. Parasitic Varroa destructor mites influence flight duration and homing ability of infested Apis mellifera foragers. Apidologie 37, 577–587 (2006). 11. Vaudry, D. et al. Pituitary Adenylate Cyclase-Activating Polypeptide and Its Receptors : 20 Years after the Discovery. Pept. Res. 61, 283–357 (2009). 12. Dejda, A., Sokołowska, P. & Nowak, J. Z. Neuroprotective potential of three neuropeptides PACAP, VIP and PHI. Pharmacol. Rep. 57, 307–20 (2005). 13. de Lange, E. C., Danhof, M., de Boer, a G. & Breimer, D. D. Methodological considerations of intracerebral microdialysis in pharmacokinetic studies on drug transport across the blood-brain barrier. Brain Res. Brain Res. Rev. 25, 27–49 (1997).

287 14. Benveniste, H., Drejer, J., Schousboe, A. & Diemer, N. H. Regional cerebral glucose phosphorylation and blood flow after insertation of a microdialysis fiber through the dorsal hippocampus in the rat. J. Neurochem. 49, 729–734 (1987). 15. de Lange, E. C. M., Danhof, M., de Boer, A. G. & Breimer, D. D. Critical factors of intracerebral microdialysis as a technique to determined the pharmacokinetics of drugs in rat brain. Brain Res. 666, 1–8 (1994). 16. Claassen, V. Techniques in the Behavioral and Neural Sciences: Neglected Factors in Pharmacology and Neuroscience Research. (Elsevier, 1994). 17. Seelig, A., Gottschlich, R. & Devant, R. M. A method to determine the ability of drugs to diffuse through the blood-brain barrier. Proc Natl Acad Sci U S A 91, 68– 72 (1994). 18. Benveniste, H., Drejer, J., Schousboe, A. & Diemer, N. H. Elevation of the extracellular concentration of glutamate and aspartate in rat hippocampus during transient cerebral ischemia monitored by intracerebal microdialysis. J. Neurochem. 43, 1369–1374 (1984). 19. Agon, P., Goethals, P., Haver, D. & Kaufman, J.-M. Permeability of the blood-brain barrier for atenolol studied by positron emission tomography. J. Pharm. Pharmacol. 43, 597–600 (1991). 20. Gengo, F. M., Pagan, S. C., Hopkins, L. N., Wagner, D. & Schuster, D. P. Nonlinear Distribution of Atenolol Between Plasma and Cerebrospinal Fluid. Pharm. Res. 6, 248–251 (1989). 21. Parepally, J. M. R., Mandula, H. & Smith, Q. R. Brain uptake of nonsteroidal anti- inflammatory drugs: Ibuprofen, flurbiprofen, and indomethacin. Pharm. Res. 23, 873–881 (2006). 22. Gulyaeva, N. et al. Relative hydrophobicity and lipophilicity of drugs measured by aqueous two-phase partitioning, octanol-buffer partitioning and HPLC. A simple model for predicting blood-brain distribution. Eur. J. Med. Chem. 38, 391–396 (2003). 23. Dockhorn, R. J. et al. Safety and efficacy of loratadine (Sch-29851): a new non- sedating antihistamine in seasonal allergic rhinitis. Ann. Allergy 58, 407–11 (1987). 24. Clissold, S. P., Sorkin, E. M. & Goa, K. L. Loratadine. Drugs 37, 42–57 (1989). 25. Neuwelt, E. A. Mechanisms of Disease: The Blood-Brain Barrier. Neurosurgery 54, 131–142 (2004). 26. González, M. A. & Estes, K. S. Pharmacokinetic overview of oral second-generation H1 antihistamines. Int. J. Clin. Pharmacol. Ther. 36, 292–300 (1998). 27. Kay, G. G. The effects of antihistamines on cognition and performance. J. Allergy Clin. Immunol. 105, S622–S627 (2000). 28. Relieving Pain in America: A Blueprint for Transforming Prevention, Care, Education, and Research. The Clinical Journal oF Pain 28, (National Academies Press, 2011). 29. Substance Abuse and Mental Health Services Administration, C. for B. H. S. and Q. The DAWN Report: Highlights of the 2011 Drug Abuse Warning Network (DAWN) Findings on Drug-Related Emergency Department Visits. (2013).

288 30. McGinty, J. F., van der Kooy, D. & Bloom, F. E. The distribution and morphology of opioid peptide immunoreactive neurons in the cerebral cortex of rats. J. Neurosci. 4, 1104–1117 (1984). 31. Valverde, O. et al. Delta9-tetrahydrocannabinol releases and facilitates the effects of endogenous enkephalins: reduction in morphine withdrawal syndrome without change in rewarding effect. Eur. J. Neurosci. 13, 1816–24 (2001). 32. Navratilova, E. & Porreca, F. Reward and motivation in pain and pain relief. Nat. Neurosci. 17, 1304–1312 (2014). 33. Murphy, N. P. Dynamic measurement of extracellular opioid activity: Status quo, challenges, and significance in rewarded behaviors. ACS Chem. Neurosci. 6, 94– 107 (2015). 34. Becerra, L., Navratilova, E., Porreca, F. & Borsook, D. Analogous responses in the nucleus accumbens and cingulate cortex to pain onset (aversion) and offset (relief) in rats and humans. J. Neurophysiol. 110, 1221–6 (2013). 35. Ho Kim, S. & Mo Chung, J. An experimental model for peripheral neuropathy produced by segmental spinal nerve ligation in the rat. Pain 50, 355–363 (1992). 36. Attal, N. et al. EFNS guidelines on the pharmacological treatment of neuropathic pain: 2010 revision. Eur. J. Neurol. 17, 1113-e88 (2010). 37. Moore, R. A., Wiffen, P. J., Derry, S. & McQuay, H. J. in Cochrane Database of Systematic Reviews (ed. Moore, M.) (John Wiley & Sons, Ltd, 2009). doi:10.1002/14651858.CD007938 38. Gilron, I. et al. Morphine, gabapentin, or their combination for neuropathic pain. N. Engl. J. Med. 352, 1324–34 (2005). 39. Levendoglu, F., Ogün, C. O., Ozerbil, O., Ogün, T. C. & Ugurlu, H. Gabapentin is a first line drug for the treatment of neuropathic pain in spinal cord injury. Spine (Phila. Pa. 1976). 29, 743–51 (2004). 40. Taylor, C. P. et al. A summary of mechanistic hypotheses of gabapentin pharmacology. Epilepsy Res. 29, 233–49 (1998). 41. Mehta, S., McIntyre, A., Janzen, S., Loh, E. & Teasell, R. Systematic Review of Pharmacologic Treatments of Pain After Spinal Cord Injury: An Update. Arch. Phys. Med. Rehabil. 97, 1381–1391.e1 (2016). 42. Sirven, J. I. New uses for older drugs: the tales of aspirin, thalidomide, and gabapentin. Mayo Clin. Proc. 85, 508–11 (2010). 43. Goldlust, A., Su, T. Z., Welty, D. F., Taylor, C. P. & Oxender, D. L. Effects of anticonvulsant drug gabapentin on the enzymes in metabolic pathways of glutamate and GABA. Epilepsy Res. 22, 1–11 (1995). 44. Maneuf, Y. P., Luo, Z. D. & Lee, K. Α2Δ and the Mechanism of Action of Gabapentin in the Treatment of Pain. Semin. Cell Dev. Biol. 17, 565–570 (2006). 45. Gee, N. S. et al. The novel anticonvulsant drug, gabapentin (neurontin), binds to the α2δ subunit of a calcium channel. J. Biol. Chem. 271, 5768–5776 (1996). 46. Field, M. J., Hughes, J. & Singh, L. Further evidence for the role of the alpha(2)delta subunit of voltage dependent calcium channels in models of neuropathic pain. Br J Pharmacol 131, 282–286 (2000).

289 47. Bao, Y. H. et al. Gabapentin enhances the morphine anti-nociceptive effect in neuropathic pain via the interleukin-10-heme oxygenase-1 signalling pathway in rats. J. Mol. Neurosci. 54, 137–146 (2014). 48. Stoicea, N. et al. Opioid-induced hyperalgesia in chronic pain patients and the mitigating effects of gabapentin. Front. Pharmacol. 6, 1–6 (2015). 49. Chapman, V., Suzuki, R., Chamarette, H. L., Rygh, L. J. & Dickenson, A. H. Effects of systemic carbamazepine and gabapentin on spinal neuronal responses in spinal nerve ligated rats. Pain 75, 261–72 (1998). 50. Hooker, B. A. et al. Gabapentin-induced pharmacodynamic effects in the spinal nerve ligation model of neuropathic pain. Eur. J. Pain 18, 223–37 (2014).

Appendix 1. Bodnar, R. J. Endogenous opiates and behavior: 2012. Peptides 50, 55–95 (2013). 2. Yoo, J. H., Kitchen, I. & Bailey, A. The endogenous opioid system in cocaine addiction: What lessons have opioid peptide and receptor knockout mice taught us? Br. J. Pharmacol. 166, 1993–2014 (2012). 3. Terskiy, A. et al. Search of the human proteome for endomorphin-1 and endomorphin-2 precursor proteins. Life Sci. 81, 1593–1601 (2007). 4. Navratilova, E., Xie, J. Y., King, T. & Porreca, F. Evaluation of reward from pain relief. Ann. N. Y. Acad. Sci. 1282, 1–11 (2013). 5. Wollemann, M. & Benyhe, S. Non-opioid actions of opioid peptides. Life Sci. 75, 257–270 (2004). 6. Olszewski, P. K., Alsiö, J., Schiöth, H. B. & Levine, A. S. Opioids as facilitators of feeding: Can any food be rewarding? Physiol. Behav. 104, 105–110 (2011). 7. Pasternak, G. W. Opioids and their receptors: Are we there yet? Neuropharmacology 76, 198–203 (2014). 8. Murphy, N. P. Dynamic measurement of extracellular opioid activity: Status quo, challenges, and significance in rewarded behaviors. ACS Chem. Neurosci. 6, 94– 107 (2015). 9. Borbély, É., Scheich, B. & Helyes, Z. Neuropeptides in learning and memory. Neuropeptides 47, 439–450 (2013). 10. Dacher, M. & Nugent, F. S. Opiates and plasticity. Neuropharmacology 61, 1088– 1096 (2011). 11. Pizzo, P. A., Clark, N. M. & Carter Pokras, O. Relieving Pain in America: A Blueprint for Transforming Prevention, Care, Education, and Research. Institute of Medicine (2011). doi:10.3109/15360288.2012.678473 12. Substance Abuse and Mental Health Services Administration, C. for B. H. S. and Q. The DAWN Report: Highlights of the 2011 Drug Abuse Warning Network (DAWN) Findings on Drug-Related Emergency Department Visits. (2013). 13. McGinty, J. F., van der Kooy, D. & Bloom, F. E. The distribution and morphology of opioid peptide immunoreactive neurons in the cerebral cortex of rats. J. Neurosci. 4, 1104–1117 (1984).

290 14. Valverde, O. et al. Delta9-tetrahydrocannabinol releases and facilitates the effects of endogenous enkephalins: reduction in morphine withdrawal syndrome without change in rewarding effect. Eur. J. Neurosci. 13, 1816–24 (2001). 15. Navratilova, E. & Porreca, F. Reward and motivation in pain and pain relief. Nat. Neurosci. 17, 1304–1312 (2014). 16. Becerra, L., Navratilova, E., Porreca, F. & Borsook, D. Analogous responses in the nucleus accumbens and cingulate cortex to pain onset (aversion) and offset (relief) in rats and humans. J. Neurophysiol. 110, 1221–6 (2013). 17. Marinelli, P. W., Lam, M., Bai, L., Quirion, R. & Gianoulakis, C. A microdialysis profile of dynorphin A1-8 release in the rat nucleus accumbens following alcohol administration. Alcohol. Clin. Exp. Res. 30, 982–990 (2006). 18. Feng, Y. et al. Current research on opioid receptor function. Curr. Drug Targets 13, 230–46 (2012). 19. Henriksen, G. & Willoch, F. Imaging of opioid receptors in the central nervous system. Brain 131, 1171–1196 (2008). 20. Nadal, X., La Porta, C., Bura, S. A. & Maldonado, R. Involvement of the opioid and cannabinoid systems in pain control: New insights from knockout studies. Eur. J. Pharmacol. 716, 142–157 (2013). 21. Harlan, R. E., Shivers, B. D., Romano, G. J., Howells, R. D. & Pfaff, D. W. Localization of preproenkephalin mRNA in the rat brain and spinal cord by in situ hybridization. J. Comp. Neurol. 258, 159–184 (1987). 22. Burbach, J. P. H. in Neuropeptides Methods and Protocols (ed. Merighi, A.) 789, 1–36 (Humana Press, 2011). 23. Maidment, N. T., Brumbaugh, D. R., Rudolph, V. D., Erdelyi, E. & Evans, C. J. Microdialysis of extracellular endogenous opioid peptides from rat brain in vivo. Neuroscience 33, 549–557 (1989). 24. Shen, H., Lada, M. W. & Kennedy, R. T. Monitoring of met-enkephalin in vivo with 5-min temporal resolution using microdialysis sampling and capillary liquid chromatography with electrochemical detection. J. Chromatogr. B Biomed. Appl. 704, 43–52 (1997). 25. Emmett, M. R., Andrén, P. E. & Caprioli, R. M. Specific molecular mass detection of endogenously released neuropeptides using in vivo microdialysis/mass spectrometry. J. Neurosci. Methods 62, 141–147 (1995). 26. Jakubowski, J. A., Hatcher, N. G. & Sweedler, J. V. Online microdialysis-dynamic nanoelectrospray ionization-mass spectrometry for monitoring neuropeptide secretion. J. Mass Spectrom. 40, 924–31 (2005). 27. Olive, M. F. & Maidment, N. T. Opioid regulation of pallidal enkephalin release: bimodal effects of locally administered mu and delta opioid agonists in freely moving rats. J. Pharmacol. Exp. Ther. 285, 1310–1316 (1998). 28. Nandi, P. & Lunte, S. M. Recent trends in microdialysis sampling integrated with conventional and microanalytical systems for monitoring biological events: A review. Anal. Chim. Acta 651, 1–14 (2009). 29. Behrens, H. L. & Li, L. Peptidomics. 615, 57–73 (2010).

291 30. Hui, L. et al. Discovery and functional study of a novel crustacean tachykinin neuropeptide. ACS Chem. Neurosci. 2, 711–722 (2011). 31. Bernay, B. et al. Discovering new bioactive neuropeptides in the striatum secretome using in vivo microdialysis and versatile proteomics. Mol. Cell. Proteomics 8, 946– 958 (2009). 32. Hou, X., Xie, F. & Sweedler, J. V. Relative Quantitation of Neuropeptides Over a Thousand-fold Concentration Range. J Am Soc Mass Spectrom. 23, 2083–2093 (2012). 33. Mabrouk, O. S., Li, Q., Song, P. & Kennedy, R. T. Microdialysis and mass spectrometric monitoring of dopamine and enkephalins in the globus pallidus reveal reciprocal interactions that regulate movement. J. Neurochem. 118, 24–33 (2011). 34. Li, Q., Jon-Kar Zubieta & Kennedy, R. T. Practical aspects of in vivo detection of neuropeptides by microdialysis coupled off-line to capillary LC with multistage MS. Anal. Chem. 81, 2242–2250 (2009). 35. Zhou, Y., Mabrouk, O. S. & Kennedy, R. T. Rapid preconcentration for liquid chromatography-mass spectrometry assay of trace level neuropeptides. J. Am. Soc. Mass Spectrom. 24, 1700–1709 (2013). 36. Lee, J. E. et al. Quantitative Peptidomics for Discovery of Circadian-Related Peptides from the Rat Suprachiasmatic Nucleus. (2013). 37. Difeliceantonio, A. G., Mabrouk, O. S., Kennedy, R. T. & Berridge, K. C. Enkephalin surges in dorsal neostriatum as a signal to eat. Curr. Biol. 22, 1918–1924 (2012). 38. Merighi, A., Salio, C., Ferrini, F. & Lossi, L. Neuromodulatory function of neuropeptides in the normal CNS. J. Chem. Neuroanat. 42, 276–287 (2011). 39. Kennedy, R. T. Emerging trends in in vivo neurochemical monitoring by microdialysis. Curr. Opin. Chem. Biol. 17, 860–7 (2013). 40. Navratilova, E. et al. Pain relief produces negative reinforcement through activation of mesolimbic reward-valuation circuitry. Proc. Natl. Acad. Sci. U. S. A. 109, 20709– 13 (2012). 41. Vogt, B. A., Sikes, R. W. & Vogt, L. J. in Neurobiology of Cingulate Cortex and Limbic Thalamus: A Comprehensive Handbook (eds. Vogt, B. A. & Gabriel, M.) 313–344 (Birkhauser, 1993). 42. Vogt, B. a, Wiley, R. G. & Jensen, E. L. Localization of Mu and delta opioid receptors to anterior cingulate afferents and projection neurons and input/output model of Mu regulation. Experimental neurology 135, 83–92 (1995). 43. Medja, F. et al. Thiorphan, a neutral endopeptidase inhibitor used for diarrhoea, is neuroprotective in newborn mice. Brain 129, 3209–3223 (2006). 44. Meske, D. S. et al. Opioid and noradrenergic contributions of tapentadol in experimental neuropathic pain. Neurosci. Lett. 562, 91–96 (2014). 45. Schmerberg, C. M. & Li, L. Mass spectrometric detection of neuropeptides using affinity-enhanced microdialysis with antibody-coated magnetic nanoparticles. Anal. Chem. 85, 915–22 (2013). 46. Buczynski, M. W. & Parsons, L. H. Quantification of brain endocannabinoid levels: methods, interpretations and pitfalls. Br. J. Pharmacol. 160, 423–42 (2010).

292 47. Sanderink, G. J., Artur, Y. & Siest, G. Human aminopeptidases: a review of the literature. J. Clin. Chem. Clin. Biochem. 26, 795–807 (1988). 48. Jaquins-Gerstl, A. et al. Effect of dexamethasone on gliosis, ischemia, and dopamine extraction during microdialysis sampling in brain tissue. Anal. Chem. 83, 7662–7667 (2011). 49. Bourgoin, S. et al. Effects of kelatorphan and other peptidase inhibitors on the in vitro and in vivo release of methionine-enkephalin-like material from the rat spinal cord. J. Pharmacol. Exp. Ther. 238, 360–6 (1986). 50. Hackler, L., Zadina, J. E., Ge, L. J. & Kastin, A. J. Isolation of relatively large amounts of endomorphin-1 and endomorphin-2 from human brain cortex. Peptides 18, 1635–1639 (1997). 51. Lisi, T. L. & Sluka, K. a. A new electrochemical HPLC method for analysis of enkephalins and endomorphins. J. Neurosci. Methods 150, 74–79 (2006). 52. Wang, Q. P. et al. Endomorphin-2 immunoreactivity in the cervical dorsal horn of the rat spinal cord at the electron microscopic level. Neuroscience 113, 593–605 (2002). 53. Perry, M., Li, Q. & Kennedy, R. T. Review of recent advances in analytical techniques for the determination of neurotransmitters. Anal. Chim. Acta 653, 1–22 (2009). 54. Janecka, A., Fichna, J. & Janecki, T. Opioid receptors and their ligands. Curr. Top. Med. Chem. 4, 1–17 (2004). 55. Perlikowska, R. & Janecka, A. Bioavailability of Endomorphins and the Blood-brain Barrier- A Review. Med. Chem. (Los. Angeles). 10, 2–17 (2014).

293