The Behavioural, Neuroanatomical, and Molecular Effects of Chronic Social Stress on Mice That Lack Zinc Transporter 3 and Vesicular Zinc

University of Calgary PRISM: University of Calgary's Digital Repository

Graduate Studies The Vault: Electronic Theses and Dissertations

2018-12-04 The Behavioural, Neuroanatomical, and Molecular Effects of Chronic Social Stress on Mice That Lack Zinc Transporter 3 and Vesicular Zinc

McAllister, Brendan Barrymore

McAllister, B. B. (2018). The behavioural, neuroanatomical, and molecular effects of chronic social stress on mice that lack zinc transporter 3 and vesicular zinc (Unpublished doctoral thesis). University of Calgary, Calgary, AB. doi:10.11575/PRISM/34901 http://hdl.handle.net/1880/109238 doctoral thesis

University of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission. Downloaded from PRISM: https://prism.ucalgary.ca

UNIVERSITY OF CALGARY

The Behavioural, Neuroanatomical, and Molecular Effects of Chronic Social Stress on Mice

That Lack Zinc Transporter 3 and Vesicular Zinc

Brendan Barrymore McAllister

A THESIS

SUBMITTED TO THE FACULTY OF GRADUATE STUDIES

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE

DEGREE OF DOCTOR OF PHILOSOPHY

GRADUATE PROGRAM IN PSYCHOLOGY

CALGARY, ALBERTA

DECEMBER, 2018

ABSTRACT In certain neurons, zinc ions are stored in synaptic vesicles by a dedicated vesicular zinc transporter, called zinc transporter 3 (ZnT3). Vesicular zinc can then be released synaptically, in an activity-dependent fashion, to transmit signals by modulating a plethora of targets. To understand the function of vesicular zinc in the central nervous system, a useful tool is the ZnT3 knockout (KO) mouse, which lacks ZnT3 and, as a result, lacks vesicular zinc. Behavioural characterization of these mice has revealed subtle abnormalities in cognition and sensory processing. In addition, a pattern is becoming apparent, wherein ZnT3 KO mice behave normally under standard laboratory conditions but fail to exhibit neural plasticity and behavioural adaptation in response to certain treatments or experiences. The experiments described in this thesis were designed to assess how ZnT3 KO mice would respond to the experience of repeated social defeat (RSD) stress, a method of modeling depression-like behaviour in rodents, and to test the hypothesis that ZnT3 KO mice would fail to exhibit stress-induced neural plasticity, resulting in an altered behavioural response to stress. The primary finding was that, compared to wild type (WT) mice, ZnT3 KO mice exhibited reduced social avoidance of a novel conspecific following RSD, suggesting reduced susceptibility to the depression-like behaviour of social withdrawal. Both genotypes were equally susceptible to anxiety-like behaviour following RSD, however. To investigate the mechanisms behind the seemingly protective effect of eliminating vesicular zinc on stress-induced social avoidance, several neuroanatomical parameters were examined. No evidence was found that microglial activation, hippocampal neurogenesis, or hippocampal brain-derived neurotrophic factor (BDNF) levels could account for the difference in behavioural outcome. However, some evidence was found that altered structure of the corpus callosum or reduced BDNF levels in the nucleus accumbens may contribute to the protection against social avoidance in ZnT3 KO mice. Further work will be required to validate and extend these findings, in order to more fully understand the mechanisms behind the altered behavioural response to chronic stress – and the altered capacity for experience-dependent neuroplasticity – in mice that lack vesicular zinc.

iii

PREFACE Chapter 1 includes text from a review article authored by B. B. McAllister and R. H. Dyck and published in Neuroscience and Biobehavioral Reviews (McAllister & Dyck, 2017; https://doi.org/10.1016/j.neubiorev.2017.06.006). This material is used with permission from the publisher and copyright holder (Elsevier), under their “Personal use” policy. Chapter 2 is a slightly modified version of an article authored by B. B. McAllister, D. K. Wright, R. C. Wortman, S. R. Shultz, and R. H. Dyck, and published in Neurobiology of Stress (McAllister et al., 2018a; https://doi.org/10.1016/j.ynstr.2018.10.003). This material is used with permission from the publisher (Elsevier), under their “Personal use” policy, and the authors and copyright holders (see Appendix B). Chapter 4 is an extended version of a manuscript by B. B. McAllister, N. Bihelek, and R. H. Dyck that is available as a preprint on BioRxiv (McAllister et al., 2018b; https://doi.org/10.1101/421453). This material is used with permission from the authors and copyright holders (see Appendix B). iv

ACKNOWLEDGEMENTS I have worked with many people during my eight years in the Dyck lab. I could simply provide a list of them here, because – to a person – I have benefited from working with and learning from them all, and all are worthy of acknowledgement for making those eight years a very enjoyable period of my life that did not feel nearly so long as it now seems, looking back upon it. Instead, I will highlight the contribution and influence of only a few, but all should know that their impact was and is deeply appreciated. I would like to thank and acknowledge several members of the lab who were there when I first arrived, from whose experience I benefited: Patrick Wu, Taryn Bemister, Michael Smith, and, in particular, Veronika Kiryanova, who guided me on my first project, and contributed, probably more than anyone else, to the positive environment of the lab. During the first few years, I was also fortunate to have had the opportunity to work with Dr. Simon Spanswick, a great scientist and (I would say) an even better teacher, from whom I learned a great deal. Thanks must also go to the many who contributed to the experiments described in this thesis, mostly by assisting with the never-ending project of genotyping, but also in many smaller – and, in some cases, larger – ways: Sarah Thackray, Sarah Bryden, Nicoline Bihelek, Jacqueline Boon, Lisa Wilcox, Corrine Stahl, Abril Valverde Rascón, Sukhjinder Rehal, Colten Chipak, Andrea Herzog, Katy Sandoval, and Angela Pochakom. I must also acknowledge that the rationale for these experiments was influenced heavily by the work of a fellow graduate student in the lab, Mike Chrusch. I also acknowledge the support and contributions from two members of my supervisory committee, Dr. Tuan Trang and Dr. Mike Antle. In addition to serving on my committee, and assisting me in many ways over the years, it was Dr. Antle who first sparked my interest in neuroscience with his Brain and Behaviour course, and then cemented it with Physiological Psychology. That being the case, he is ultimately to thank (or perhaps to blame) for this thesis and its contents. Finally, and most of all, I thank my supervisor, Dr. Richard Dyck, for giving me the chance to work with him eight years ago, and for being an exemplary mentor and supporter all the time since. Your confidence in me has helped me to reach this point. It is not something I will forget.

DEDICATION To my mother, Lauri.

TABLE OF CONTENTS ABSTRACT...... ii PREFACE ...... iii ACKNOWLEDGEMENTS...... iv DEDICATION...... v TABLE OF CONTENTS...... vi LIST OF TABLES...... x LIST OF FIGURES...... xi LIST OF SYMBOLS, ABBREVIATIONS, & NOMENCLATURE...... xii CHAPTER ONE: GENERAL INTRODUCTION...... 1 1.1 ZINC BIOLOGY, VESICULAR ZINC, AND ZNT3...... 1 1.1.1 Zinc Homeostasis and Transport...... 2 1.1.2 Vesicular Zinc, Zincergic Neurons, and ZnT3...... 6 1.1.3 Anatomy of the Zincergic System...... 8 1.1.3.1 Telencephalon...... 9 1.1.3.2 Diencephalon, brainstem, and spinal cord...... 11 1.1.4 Neuronal Zinc Signaling...... 15 1.1.4.1 Excitatory signaling...... 16 1.1.4.1.1 NMDA receptors...... 16 1.1.4.1.2 AMPA and kainate receptors...... 18 1.1.4.2 Inhibitory signaling...... 20 1.1.4.2.1 Glycine receptors...... 20 1.1.4.2.2 GABA receptors...... 21 1.1.4.3 Metabotropic signaling...... 22 1.1.4.4 BDNF and TrkB signaling...... 24 1.1.4.5 Synaptic plasticity...... 25 1.1.5 ZnT3 in Health and Disease...... 27 1.1.5.1 Aging and Alzheimer’s Disease...... 27 1.1.5.2 Neurodegeneration and cell death...... 28 1.1.5.3 Other disorders...... 30 1.1.6 Characterization of ZnT3 KO Mice...... 31 1.1.6.1 Neurophysiology...... 31 vii

1.1.6.2 Behaviour...... 32 1.1.6.3 Experience-dependent plasticity...... 34 1.2 STRESS ...... 35 1.2.1 Social Defeat Stress...... 37 1.2.2 Neurobiology of Stress and Stress Resilience...... 38 1.2.3 Interactions Between Stress and the Zincergic System...... 40 1.3 SUMMARY OF EXPERIMENTS...... 42 1.4 STATEMENT OF CONTRIBUTION...... 43 CHAPTER TWO: VESICULAR ZINC AND SOCIAL STRESS...... 44 2.1 INTRODUCTION...... 44 2.2 METHOD...... 46 2.2.1 Animals...... 46 2.2.3 Experimental Design...... 46 2.2.4 Behavioural Assessment...... 48 2.2.4.1 Social interaction...... 48 2.2.4.3 Elevated plus-maze...... 48 2.2.4.3 Novelty-suppressed feeding...... 49 2.2.4.4 Spatial Y-maze...... 50 2.2.4.5 Conditioned fear...... 50 2.2.5 Anatomical analysis...... 51 2.2.5.1 Immunofluorescence labeling...... 51 2.2.5.2 Hippocampal cell counting...... 52 2.2.5.3 Microglial analysis...... 52 2.2.5.4 MRI acquisition and analysis ...... 53 2.2.6 Statistical Analysis...... 54 2.3 RESULTS ...... 54 2.3.1 Behavioural Assessment...... 54 3.2.1.1 Social interaction...... 54 2.3.1.2 Elevated plus-maze...... 56 2.3.1.3 Novelty-suppressed feeding...... 56 2.3.1.4 Spatial Y-maze...... 57 2.3.1.5 Conditioned fear...... 58 2.3.2 Anatomical Analyses...... 58 viii

2.3.2.1 Body and organ weights...... 58 2.3.2.2 Hippocampal cell proliferation...... 59 2.3.2.3 Microglial analysis...... 59 2.3.2.4 MRI volumetric analysis...... 60 2.4 DISCUSSION...... 61 2.5 CONCLUSION...... 66 2.6 ACKNOWLEDGEMENTS & STATEMENT OF CONTRIBUTION...... 66 FIGURES...... 68 TABLES...... 76 CHAPTER THREE: VESICULAR ZINC AND FLUOXETINE...... 84 3.1 INTRODUCTION...... 84 3.2 METHOD...... 86 3.2.1 Animals...... 86 3.2.2 Experimental Design...... 86 3.2.3 Repeated Social Defeat...... 87 3.2.4 Fluoxetine Treatment...... 87 3.2.5 Behavioural Assessment...... 87 3.2.5.1 Social interaction...... 87 3.2.5.2 Novelty-suppressed feeding...... 87 3.2.6 Quantification of Cell Survival...... 87 3.2.7 Statistical Analysis...... 88 3.3 RESULTS...... 88 3.3.1 Behavioural Assessment...... 88 3.3.1.1 Social interaction...... 88 3.3.1.1.1 Replication of previous results...... 88 3.3.1.1.2 Effect of fluoxetine treatment...... 90 3.3.1.2 Novelty-suppressed feeding...... 93 3.3.2 Quantification of Cell Survival...... 95 3.4 DISCUSSION...... 96 3.5 CONCLUSION...... 100 3.6 ACKNOWLEDGEMENTS...... 100 FIGURES...... 101 TABLES...... 109 ix

CHAPTER FOUR: VESICULAR ZINC AND BDNF...... 117 4.1 INTRODUCTION...... 117 4.2 METHOD...... 119 4.2.1 Animals...... 119 4.2.2 Experimental Design...... 119 4.2.2.1 Experiment one...... 119 4.2.2.2 Experiment two...... 119 4.2.3 Sample Preparation...... 119 4.2.4 Western Blotting...... 120 4.2.5 Enzyme-Linked Immunosorbent Assay (ELISA) ...... 120 4.2.6 Social Interaction Test...... 121 4.2.7 Statistical Analysis...... 121 4.3 RESULTS...... 121 4.3.1 Experiment One...... 121 4.3.1.1 Hippocampus...... 121 4.3.1.2 Neocortex...... 122 4.3.2 Experiment Two...... 123 4.3.2.1 Nucleus accumbens...... 123 4.2.2.2 Hippocampus...... 124 4.3.2.3 Social interaction behaviour...... 124 4.4 DISCUSSION...... 125 4.5 CONCLUSION...... 129 4.6 ACKNOWLEDGEMENTS & STATEMENT OF CONTRIBUTION...... 130 FIGURES...... 131 TABLES...... 138 CHAPTER FIVE: GENERAL DISCUSSION...... 144 REFERENCES...... 153 APPENDIX A: PARTITIONED-HOUSING...... 188 A.1 Partitioned-Housing...... 188 A.2 Results and Discussion...... 188 FIGURES...... 190 APPENDIX B: COPYRIGHT PERMISSIONS...... 193 x

LIST OF TABLES Table 2.1 Additional behavioural measures...... 76 Table 2.2 Comparison of control, susceptible, and resilient mice...... 77 Table 2.3 Microglial analysis...... 79 Table 2.4 ANOVA results...... 80 Table 3.1 Additional behavioural measures from the first social interaction test...... 109 Table 3.2 Additional behavioural measures from the novelty-suppressed feeding tests...... 110 Table 3.3 ANOVA results from the first social interaction test...... 112 Table 3.4 ANOVA results (effects of genotype, stress, and drug)...... 113 Table 4.1 Experiment two: protein levels...... 138 Table 4.2 Experiment two: results of the social interaction test...... 139 Table 4.3 ANOVA results (effects of age, sex, and genotype)...... 140 Table 4.4 ANOVA results for experiment two (effects of stress and genotype)...... 142

LIST OF FIGURES Figure 2.1 Timeline depicting the experimental design...... 68 Figure 2.2 Social interaction behaviour of WT and ZnT3 KO mice following RSD...... 69 Figure 2.3 Anxiety-like behaviour of WT and ZnT3 KO mice following RSD...... 70 Figure 2.4 Conditioned fear memory in WT and ZnT3 KO mice following RSD...... 71 Figure 2.5 Hippocampal neurogenesis in WT and ZnT3 KO mice following RSD...... 72 Figure 2.6 Microglial morphology in WT and ZnT3 KO mice following RSD...... 73 Figure 2.7 MRI volumetric analysis in WT and ZnT3 KO mice following RSD...... 75 Figure 3.1 Timeline depicting the experimental design...... 101 Figure 3.2 Replication of social interaction results from Chapter Two...... 102 Figure 3.3 Results from the second social interaction test, after 4 weeks of fluoxetine...... 103 Figure 3.4 Results from the second social interaction test, excluding susceptible mice...... 104 Figure 3.5 Comparison of social interaction behaviour in the first and second tests...... 105 Figure 3.6 Anxiety-like behaviour in the novelty-suppressed feeding (NSF) tests...... 106 Figure 3.7 Weight loss over the 16 h food restriction period prior to the NSF tests...... 107 Figure 3.8 Hippocampal cell survival following RSD and chronic fluoxetine...... 108 Figure 4.1 Example Western blots for BDNF and TrkB...... 131 Figure 4.2 Effects of age, sex, and ZnT3 status on BDNF levels in the hippocampus...... 132 Figure 4.3 Effects of age, sex, and ZnT3 status on TrkB levels in the hippocampus...... 133 Figure 4.4 Effects of age, sex, and ZnT3 status on BDNF levels in the neocortex...... 134 Figure 4.5 Effects of age, sex, and ZnT3 status on TrkB levels in the neocortex...... 135 Figure 4.6 Effects of stress and ZnT3 status on BDNF levels in the nucleus accumbens.....136 Figure 4.7 Social interaction behaviour of WT and ZnT3 KO mice following RSD...... 137 Figure A1 Social interaction with a novel conspecific by partition-housed controls...... 190 Figure A2 Social interaction with a novel CD-1 mouse by partition-housed controls...... 191 Figure A3 Behaviour in the NSF test by partition-housed controls...... 192 xii

LIST OF SYMBOLS, ABBREVIATIONS, NOMENCLATURE Aβ: amyloid-beta; ANOVA: analysis of variance; BBB: blood-brain barrier; BDNF: brain- derived neurotrophic factor; BLA: basolateral amygdala; BNST: bed nucleus of the stria terminalis; BrdU: 5-bromo-2′-deoxyuridine; CaEDTA: calcium-saturated EDTA; CI: confidence interval; CNS: central nervous system; CORT: cortisol/corticosterone; DAPI: 4′,6- diamidino-2-phenylindole; DCN: dorsal cochlear nucleus; dHPC: dorsal hippocampus; ELISA: enzyme-linked immunosorbent assay; EPM: elevated plus-maze; EPSC: excitatory postsynaptic current; EPSP: excitatory postsynaptic potential; ERK: extracellular signal- regulated kinase; GABA: γ-aminobutyric acid; GPR39: G-protein-coupled receptor 39; HFS: high-frequency stimulation; Iba1: ionized calcium-binding adapter molecule 1; IP: intraperitoneal; IPSC: inhibitory postsynaptic current; IPSP: inhibitory postsynaptic potential; KO: knockout; LSD: Least Significant Difference; LTD: long-term depression; LTP: long-term potentiation; MAPK: mitogen-activated protein kinase; MF-CA3: mossy fiber-CA3; MMP: matrix metalloproteinase; MT: metallothionein; MRI: magnetic resonance imaging; MWT: Morris water task; NAc: nucleus accumbens; NMDA: N-methyl-D- aspartate; NSF: novelty-suppressed feeding; PBS: phosphate buffered saline; pERK: phosphorylated ERK; PFA: paraformaldehyde; PH: partitioned-housing; PFC: prefrontal cortex; ROI: region of interest; RSD: repeated social defeat; SC-CA1: Schaffer collateral- CA1; SEM: standard error of the mean; SSRI: selective serotonin reuptake inhibitor; SGZ: subgranular zone; TrkB: tropomyosin receptor kinase B; TrkB.T: truncated TrkB; vHPC: ventral hippocampus; VTA: ventral tegmental area; WT: wild type; ZIP: Zrt, Irt-like protein; ZnT: zinc transporter.

CHAPTER 1: General Introduction 1

CHAPTER ONE: GENERAL INTRODUCTION 1.1 ZINC BIOLOGY, VESICULAR ZINC, AND ZNT3 The transition metal zinc is an essential micronutrient – in its cationic form (Zn2+), zinc is biologically ubiquitous, being vital to the physiology of lifeforms ranging from bacteria to plants and fungi to animals. That zinc is a critical factor in human health and development has been known as far back as to the late 1960s, when pioneering work by Ananda Prasad and colleagues (1961) on severely malnourished patients in Iran identified dietary zinc deficiency as a possible cause of developmental abnormalities including dwarfism and hypogonadism. Normally, the human body contains 2-3 g of zinc, about 0.1% of which is replenished daily through dietary sources (Maret & Sandstead, 2006). About 90% of this zinc is contained in skeletal muscle and bone (Kambe et al., 2015). The remainder is found in the liver, skin, and other organs – including the brain. Throughout the body, zinc plays a diverse set of roles than can be broadly divided into three categories (Frederickson, 1989; Kambe et al., 2015). The first is as a structural component of proteins. As much as 10% of the human proteome consists of zinc-binding proteins (Andreini et al., 2006); this includes proteins that contain the common zinc-finger structural motif, as well as others. Second, zinc acts as a cofactor for the function of many enzymes. A prime example is alcohol hydrogenase, which both includes zinc as a structural component and uses zinc as a cofactor in the process of metabolizing alcohol into acetaldehyde (Vallee & Auld, 1990). Third – and in similar fashion to another metal, calcium – zinc can serve as a cellular signaling factor, transmitting both intracellular and intercellular signals. Of particular intrigue to neuroscientists, this third category includes the ability of zinc to mediate or modulate communication between neurons – in other words, zinc can act essentially as a neurotransmitter. Such an ability requires that zinc be concentrated in presynaptic terminals and released into the synaptic cleft in a regulated fashion, where it binds to and modulates the function of receptors before ultimately being transported out of the cleft or in some other way inactivated. Partial evidence for this idea has been available for many years. Since the 1980s, it has been known that zinc is stored in synaptic vesicles (Pérez-Clausell & Danscher, 1985), that activity-dependent release of zinc can occur in the brain (Assaf & Chung, 1984; Howell et al., 1984; Aniksztejn et al, 1987), and that zinc is CHAPTER 1: General Introduction 2 capable of modulating neuronal function through its effects on receptors (Smart & Constanti, 1983; Peters et al., 1987; Westbrook & Mayer, 1987). But it was not until the discovery and characterization of a dedicated vesicular zinc transporter, called zinc transporter 3 (ZnT3), in the mid-to-late 1990s (Palmiter et al., 1996; Wenzel et al., 1997) – and the subsequent generation of a transgenic mouse that lacks ZnT3 (Cole et al., 1999) – that the mechanism behind presynaptic concentration and activity-dependent release of zinc was finally elucidated, filling one of the major criteria for zinc to be considered as a neurotransmitter. Experiments conducted in the years since have provided further evidence, and it is now beyond doubt that zinc can serve as a synaptically-released intercellular signal.

1.1.1 Zinc Homeostasis and Transport In biology, zinc exists solely in its zinc(II) valance state, placing it in the category of biologically-important divalent cations along with calcium, magnesium, and others. Among this category, the affinity with which zinc binds to protein ligands is very high, second only to copper (Maret, 2014a). Accordingly, most zinc ions in the body are tightly or irreversibly bound in protein or enzyme structures, and therefore have limited capacity to participate in signaling functions. A certain proportion of zinc exists outside this tightly-bound state, however. This pool of zinc ions has been referred to by many names: labile, mobile, highly exchangeable, loosely-bound. The term free zinc will be used in this thesis. As noted by Maret (2014b), this term is an operational definition; the precise coordination environment of “free” zinc ions is not known, and it is possible that these ions are bound to low molecular-weight ligands. But as an operational definition indicating zinc ions that are free to serve fast signaling functions, the term is useful. Within cells, the appropriate concentration of free zinc must be carefully maintained. The concentration of zinc in the cytosol must be high enough for zinc to sufficiently occupy its necessary binding sites, but low enough to prevent zinc from displacing lower affinity divalent cations from their own metalloproteins (Colvin et al., 2010). The destructive effects that zinc can have when its concentration strays too far from normal physiological levels are well-documented. This becomes particularly relevant during pathological events – such as ischemia or severe seizures (Koh et al., 1996; Lee et al., 2000) – that are associated with high extracellular concentrations of glutamate and zinc. Under these conditions, zinc can enter cells through calcium-permeable glutamate receptors or CHAPTER 1: General Introduction 3 voltage-dependent calcium channels and accumulate in excess (Sensi et al., 2011). Damaging effects can also occur when zinc is liberated from intracellular binding sites during oxidative stress, as zinc can act as an intracellular signal that initiates cell death – though it can also initiate mechanisms that are protective against oxidative stress (Aras & Aizenman, 2011). Due to the high affinity with which zinc binds protein ligands, the appropriate cytosolic concentration of free zinc is quite minimal. In contrast to the nanomolar level estimated to be found in the extracellular space (Frederickson et al., 2006a), cytosolic free zinc is maintained in the low picomolar range (Colvin et al., 2010). To achieve this, mechanisms exist for buffering and muffling zinc, analogous to the way in which the cytosolic concentration of calcium is carefully controlled (Colvin et al., 2010). Buffering refers to processes that maintain the appropriate cytosolic free zinc concentration under steady-state conditions, whereas muffling refers to processes that control the concentration in response to transient changes, such as an increase in the cytosolic inflow of zinc. One key set of mechanisms for zinc muffling are zinc transporting proteins, which extrude zinc from the cell or transfer zinc between the cytosol and various organelles, which can serve as high-concentration zinc storage compartments. When these compartments are considered, the total intracellular concentration of zinc is actually quite high, measuring in the hundreds of micromolar (Krezel & Maret, 2006; Colvin et al., 2008). The importance of maintaining zinc homeostasis is further evinced by the wide array of dedicated zinc transporting proteins that traffic zinc within cells (Kambe et al., 2015). These proteins are highly evolutionarily conserved and are found in all kingdoms of life. Two families – consisting of two dozen members in humans – comprise the full set. The first is the 14-member Zrt, Irt-like protein (ZIP) family, encoded by the solute carrier 39A (SLC39A) family of genes. The ZIPs are broadly involved in cytosolic import of zinc, though to varying degrees they transport other metals, including iron, manganese, and cadmium. Many of these proteins (including ZIP1-6, 8-10, 12 & 14) are expressed on the plasma membrane, and therefore import zinc ions into the cell. Another subset (including ZIP7-9, 11 & 13) are expressed on intracellular membranes, and mobilize zinc ions stored in compartments such as the Golgi apparatus, endoplasmic reticulum, and nucleus. The other family, which has 10 members, is the zinc transporter (ZnT) proteins, encoded by the SLC30A gene family. These proteins function in general opposition to the ZIPs, extruding zinc from the cytosol. Transporting zinc out of the cell is the function of the CHAPTER 1: General Introduction 4 ubiquitously-expressed ZnT1. This protein is vital for normal cellular function; loss of it results in embryonic non-viability (Andrews et al., 2004). In the brain, ZnT1 concentrates at postsynaptic membranes (Sindreu et al., 2014), where it binds to the GluN2A subunit of the N-methyl-D-aspartate (NMDA) receptor and influences dendritic spine size (Mellone et al., 2015). Extruding zinc from the cytosol into intracellular compartments is the role filled by most of the other ZnTs. Examples include ZnT2, ZnT3, and ZnT4 – which transport zinc into endosomes, lysosomes, and secretory vesicles – and ZnT5, ZnT6 and ZnT7, which transport zinc into the Golgi apparatus and trans-Golgi network. Though the structure of the ZnTs is not definitively known, some facts can be inferred from the homologous transporter YiiP, found in Escherichia coli, for which the crystal structure has been revealed (Lu and Fu, 2007; Lu et al., 2009). YiiP, along with the ZnT proteins, is a member of the cation diffusion facilitator (CDF) superfamily. Based on the structure of YiiP, ZnTs are predicted to be six transmembrane domain (TMD) proteins with intracellular N- and C-terminal domains (Kambe et al., 2015). TMDs I, II, IV and V form a channel where histidine and aspartate residues provide a zinc-binding site; other zinc-binding sites are present on the C-terminal domain and on the histidine-rich cytosolic loop between TMIV and TMV, which may be involved in sensing zinc levels. The ZnTs function as Zn2+/H+ antiporters, exchanging zinc for the protons found in acidified intracellular compartments (Ohana et al., 2009). They form homodimers, except for ZnT5 and ZnT6, which function as heterodimers; ZnT6 on its own cannot transport zinc (Fukunaka et al., 2009). In the case of human ZnT3, homodimers are formed by a covalent bond between tyrosine residues on the C-terminal domain; altering this dimerization alters the intracellular targeting of ZnT3 and its transport capacity (Salazar et al., 2009). Recently, it has been shown that – in addition to forming homodimers – ZnT1, ZnT2, ZnT3, and ZnT4 can form heterodimers, altering their localization and function (Golan et al., 2015). Another key mechanism of cytosolic zinc muffling and buffering are metallothionein (MT) proteins. MTs are cysteine-rich, giving them ample zinc-binding capacity. Of the four major MT isoforms in the mouse (there are at least a dozen in humans; Li & Maret, 2008), three of them are expressed in the central nervous system (CNS), with MT-I and MT-II being more widely distributed throughout the body and MT-III being CNS-specific (Masters et al., 1994). Each MT protein can bind up to seven ions of zinc; collectively, they bind 5- 15% of the zinc in the cytosol (Kambe et al., 2015). Interestingly, the MT zinc binding sites CHAPTER 1: General Introduction 5 differ in their affinity, with four exhibiting a picomolar affinity and three a nanomolar affinity or less (Krezel & Maret, 2007). MTs can thus buffer zinc over a range of concentrations and can donate zinc ions from their lower-affinity sites to other higher- affinity proteins. Because the cysteine thiol ligands that bind zinc are redox-sensitive, the zinc ions that are buffered by MTs are liable to be released during oxidative stress (Maret & Vallee, 1998). Another fascinating aspect of MT function is that some isoforms are regulated transcriptionally by the cytosolic zinc concentration. This occurs through metal regulatory element-binding transcription factor-1 (MTF-1), a cytosolic zinc-sensing protein which, upon binding to zinc, translocates to the nucleus and binds to the metal response element (MRE) present in the regulatory region of some genes, including the genes for MT-I and MT-II (Andrews, 2000). The ZnT1-encoding gene also has an MRE, making it responsive to MTF-1 (Langmade et al., 2000). This provides mechanisms by which the cell – in response to elevated, potentially-injurious levels of cytosolic zinc – can increase its capacity for zinc binding and zinc extrusion. ZnT3 is not regulated by MTF-1, nor is MT-III, the dominant MT isoform found in neurons. At the organismal level, zinc homeostasis is maintained by a balance between uptake of zinc from dietary sources and zinc excretion. Dietary zinc uptake occurs primarily in the duodenum and jejunum of the small intestine, where enterocytes express ZIP4 on their apical, luminal surfaces and ZnT1 and ZIP5 on their basolateral membranes (Kambe et al., 2015). ZIP4 is upregulated during zinc deficiency, controlling the amount of zinc taken up by these cells (Dufner-Beattie et al., 2003). Mutations in the Zip4 gene in humans can cause severe zinc deficiency, resulting in a disorder called acrodermatitis enteropathica that is characterized by alopecia, dermatitis, diarrhea, and growth retardation. Postnatal genetic deletion of intestinal ZIP4 produces a similar phenotype in mice (Geiser et al., 2012). Once in the intestinal enterocytes, zinc is exported into the portal blood by ZnT1 (McMahon & Cousins, 1998). From the blood, zinc is distributed throughout the organs and tissues of the body. If the level of zinc in the blood becomes excessive, ZIP5 may be involved in transporting zinc back into enterocytes, from whence it can be transported to the intestinal lumen (Geiser et al., 2013). In the blood, zinc is bound by proteins such as albumin and α2-macroglobulin. Zinc excretion from the body is mediated by gastrointestinal secretion, sloughing of mucosal and skin cells, and renal secretion (Kambe et al., 2015). In order to access the brain, zinc must cross either the blood-brain barrier (BBB) or the blood-cerebrospinal fluid barrier. Serum albumin is not required for zinc transport into CHAPTER 1: General Introduction 6 the brain, as rats that lack serum albumin show normal brain zinc uptake (Takeda et al., 1997). ZIP4 is expressed in the choroid plexus epithelium as well as in brain capillary structures; it might therefore be involved in zinc transport across both structures (Belloni- Olivi et al., 2009). Whatever the mechanisms, zinc can cross both barriers, with the BBB of the cerebral capillaries representing the major route of entry (Franklin et al., 1992).

Following injection of 65ZnCl2, 65Zn concentrates in the choroid plexus within an hour and then, over the course of several days, concentrates in the brain parenchyma (Takeda et al., 1994).

1.1.2 Vesicular Zinc, Zincergic Neurons, and ZnT3 A considerable pool of free zinc exists in the brain, stored within the axon terminals of neurons. This was known for many years prior to the discovery of ZnT3 or the characterization of neuronal zinc release, because – unlike the majority of zinc that is tightly protein-bound – this portion of the brain’s zinc is reactive to histochemical procedures. Free zinc was first histochemically visualized in the brain – and in other parts of the body – by researchers in Germany in the 1950s, including Friedrich Timm, using his eponymous sulphide-silver stain, and Helmut Maske (Maske, 1955; Timm, 1958; as cited in Pérez-Clausell & Danscher, 1985). Timm’s staining procedure produces a pattern that is particularly notable for its intensity in the mossy fibers and hilus of the hippocampal dentate gyrus; ultrastructural analysis later showed that the staining is localized to axon terminal boutons (Haug, 1967; Ibata & Otsuka, 1969). Since that time, the histochemical distribution of free zinc in the brain has been extensively characterized, using improved methods such as Gorm Danscher’s modification of Timm’s stain (Danscher, 1981) as well as the zinc-selenium stain (Danscher, 1982). Histochemically-reactive zinc is found in a subset of the brain’s glutamatergic neurons (Beaulieu et al., 1992; Sindreu et al., 2003). Ultrastructurally, it is localized to clear round synaptic vesicles of neurons that form asymmetric synapses (in the rat: Pérez- Clausell & Danscher, 1985; in the cat: Dyck et al., 1993; in the monkey: Ichinohe & Rockland, 2005a). Due to this localization, this pool of free zinc is often referred to as vesicular zinc (or sometimes synaptic zinc), while neurons that contain this zinc are referred to as zinc-enriched or zincergic. The proportion of vesicles in which vesicular zinc can be detected varies based on the staining procedure, as well as across brain regions and possibly across species. In the zinc-rich mossy fibers of the hippocampus, it has been CHAPTER 1: General Introduction 7 reported that up to 60-80% of the clear round vesicles contain zinc in the mouse, with a smaller proportion of vesicles containing zinc in the monkey (Wenzel et al., 1997). Other reports put the number as low as 15-20% in the mouse (Lavoie et al., 2011) and 10% in the rat (Pérez-Clausell & Danscher, 1985). In some forebrain regions, such as the neocortex and hippocampus, vesicular zinc accounts for about 20% of the total zinc content (Cole et al., 1999); in the whole brain, about 10% of the total zinc is estimated to reside in the vesicular pool. In the mid-to-late 1990s, Richard Palmiter and colleagues demonstrated that maintenance of the vesicular zinc pool is dependent on ZnT3. The mouse ZnT3 gene, located on chromosome 5 (in contrast to chromosome 2 in humans), was initially identified due to its homology with rat ZnT2. Once identified and cloned, ZnT3 mRNA was detected in the brain and testis (Palmiter et al., 1996). In the brain, the pattern of mRNA expression corresponds closely with the distribution of histochemically-visualized zincergic cells, with the exception of the paraventricular thalamic nucleus, where ZnT3 mRNA is clearly expressed but zincergic cells have not been detected. As with histochemical zinc staining, the most intense area of ZnT3 protein expression is the hippocampal mossy fibers. ZnT3 is localized to the boutons of the mossy fibers, where it is found on the membranes of clear round synaptic vesicles that also contain free zinc (Wenzel et al., 1997). The final piece of evidence that vesicular zinc is inextricably linked with the function of ZnT3 was the demonstration that – except for the choroid plexus, which apparently relies on a different mechanism to store zinc – free, histochemically-reactive zinc is absent from the brains of ZnT3 KO mice (Cole et al., 1999). This has also been verified using microprobe synchrotron X-ray fluorescence (Linkous et al., 2008) and zinc-binding fluorescent dyes (Lee et al., 2011). ZnT3-rich vesicles form a distinct subpopulation from vesicles that express synaptophysin (Salazar et al., 2004a); this fits with the observation that only a subset of vesicles contain zinc. Given this, it is interesting to consider whether zinc-rich vesicles represent a pool that is physiologically distinct from other synaptic vesicles. In the mossy fiber terminals, zinc-rich vesicles do not differ in volume from zinc-poor vesicles, but the presence or absence of vesicular zinc does seem to affect calcium-dependent release mechanisms in some vesicles, with the elimination of vesicular zinc possibly reducing the synchronous release of vesicles in response to presynaptic calcium transients (Lavoie et al., 2011). Zinc-rich vesicles may also differ from zinc-poor vesicles in their glutamate content, CHAPTER 1: General Introduction 8 as there is evidence that the co-targeting of ZnT3 and vesicular glutamate transporter 1 (VGLUT1) to vesicles increases the ability of VGLUT1 to take up glutamate, potentially increasing glutamate content in zinc-rich vesicles (Salazar et al., 2005). In addition, zinc- rich vesicles, relative to zinc-poor ones, appear to be preferentially released as neuronal activity increases (Lavoie et al., 2011). Several factors can influence the ZnT3 expression and zinc content of vesicles. Endosomal ZnT3 is targeted to synaptic vesicles primarily by the adaptor protein 3 (AP-3) complex (Salazar et al., 2004a); this is why mocha mutant mice, which lack AP-3, exhibit reduced ZnT3 expression and, consequently, reduced vesicular zinc (Kantheti et al., 1998; Stoltenberg et al., 2004). Expression of the AP-3 complex, and in turn ZnT3, is affected by factors including apolipoprotein E (Lee et al., 2010) and estrogen (Lee et al., 2004). The latter suppresses ZnT3 protein expression and provides a potential mechanism for sex differences in the zincergic system. Through in vitro experiments, primarily on PC12 cells, it has been shown that AP-3 is also involved in targeting chloride channel 3 (ClC-3) and VGLUT1 to vesicles. Both ClC-3 and VGLUT1 can functionally interact with ZnT3, with the presence of either protein enhancing the ability of ZnT3 to take up zinc, increasing vesicular zinc content (Salazar et al., 2004b, 2005). At the much broader, organismal level, there is evidence that ZnT3 and vesicular zinc content can be influenced by dietary factors. A prolonged zinc-deficient diet reduces the intensity of vesicular zinc staining, indicating that the vesicular zinc pool is sensitive to dietary zinc intake (Takeda et al., 2003; Grabrucker et al., 2014). Severe zinc deficiency may even affect the density of vesicles themselves – at least in the mossy fibers – through an unknown mechanism (Lu et al., 2000). In rats, maternal and developmental deficiency of omega-3 polyunsaturated fatty acid results in an elevation of ZnT3 mRNA levels and vesicular zinc staining that is detectable in adulthood (Jayasooriya et al., 2015).

1.1.3 Anatomy of the Zincergic System Before describing the anatomy of the zincergic system in more detail, a few general statements can be made. First, the vast majority of zincergic neurons are found in the telencephalon, from where they project mostly to other telencephalic regions. Second, neurons in cortical structures do innervate some subcortical structures with zincergic projections, but this pattern is not reciprocated, with only one established exception. Third, a considerable portion of the zincergic terminals in a given region generally stem from CHAPTER 1: General Introduction 9 intrinsic intraregional projections, with additional terminals provided by commissural projections from the region’s contralateral counterpart as well as projections from other – predominantly, though not exclusively, ipsilateral – structures. Finally, even in regions where zincergic neurons are abundant, these cells represent only a subset of the region’s glutamatergic neurons. Accordingly, regions that send zincergic projections to a given target also send zinc-negative projections to the same target. For example, not all CA3 pyramidal neurons are zincergic (Slomianka et al., 1992), and in CA1 stratum radiatum – where the axons of CA3 pyramidal neurons synapse on the dendrites of CA1 neurons – only about 45% of the axospinous terminals contain vesicular zinc (Sindreu et al., 2003). One noteworthy exception to this rule is the granule cells of the dentate gyrus, because all of the mossy fiber boutons appear to contain vesicular zinc (Sindreu et al, 2003).

1.1.3.1 Telencephalon When visualized with Timm’s stain, the hippocampal mossy fibers are the most visually striking example of zincergic projections, but zincergic terminals are also plentiful in other areas of the hippocampus, as well as in many other forebrain regions including the neocortex, entorhinal cortex, claustrum, striatum, amygdala, septum, tenia tecta, piriform cortex, olfactory tubercle, anterior olfactory nucleus, and olfactory bulb (Friedman & Price, 1984; Pérez-Clausell & Danscher, 1985; Slomianka, 1992; Frederickson et al., 1992; Mengual et al., 1995; Pérez-Clausell, 1996; Jo et al., 2000b; Ichinohe & Rockland, 2005a, 2005b). Within these regions, zincergic terminals are not distributed homogenously. Instead, they are limited to, or vary in density across, different compartments. Zinc staining in the neocortex is generally characterized by a banded appearance, with heavy staining in layers 1, 2/3, and 5, moderate zinc staining in layer 6, and light staining in layer 4 (cat: Dyck et al., 1993; rat: Pérez-Clausell, 1996; monkey: Dyck et al., 2003; mouse: Brown & Dyck; 2004; ferret: Khalil & Levitt, 2013). This pattern is similar across species, though with some interspecies variation in the staining of sublamina. Through a modification of the zinc staining procedure, it is possible to visualize the cell bodies from which vesicular zinc-containing terminals originate (Howell & Frederickson, 1990; Slomianka et al., 1990; Brown & Dyck, 2003). This involves administering a selenium compound to an animal, either systemically or directly via intracerebral microinjection, and then allowing the animal to survive for a period of time prior to histological preparation. The selenium ions form crystals with free zinc ions – CHAPTER 1: General Introduction 10 providing the basis for the zinc-selenium stain – and over the survival period the zinc- selenium crystals are trafficked to the cell body by retrograde transport. This procedure reveals that zincergic neurons are present in almost all regions where zincergic terminals are found. However, a few structures are notable for containing zincergic terminals stemming from other regions, while possessing no zincergic cell bodies of their own, including the striatum (Slomianka et al., 1990; Sørensen et al., 1995; Brown & Dyck, 2004), olfactory bulbs (Jo et al., 2000b, 2002), and most of the septal complex (Slomianka et al., 1990; Sørensen et al., 1993; Mandava et al., 1993). In the neocortex, the location of zincergic cell bodies is broadly reflective of the location of zinc-containing terminals. Zincergic cell density is highest in layers 2/3, with moderate density in layer 6 and lower density in layer 5. No zincergic cells are found in layer 4 (Slomianka et al., 1990; Garrett et al., 1991; Casanovas-Aguilar et al., 1998; Brown & Dyck, 2004). Zinc-containing cells in the neocortex have been identified predominantly as pyramidal neurons, though some in layer 6 may be inverted pyramids or other cell types (Miró-Bernié et al., 2003; Brown & Dyck, 2005). Of the neocortical areas, the zincergic network has been most extensively studied in visual cortex. The zincergic terminals in visual cortex arise largely from cells within visual cortex itself, both from the ipsi- and contralateral hemisphere (Garrett et al., 1992; Casanovas-Aguilar, 1995, 1998). Visual cortex also receives zincergic projections from the frontal and orbitofrontal cortices, and a smaller number of projections from almost all remaining cortical areas – including the auditory, sensorimotor, parietal, retrosplenial, and perirhinal cortices – and from the claustrum (Garrett et al., 1992; Casanovas-Aguilar et al., 1998). The pattern of zincergic innervation described for visual cortex can be generalized to other neocortical regions, with minor differences. For example, the orbitofrontal cortex also receives projections from the amygdala, tenia tecta, and piriform cortex, and the frontal cortex receives projections from the hippocampus (Casanovas-Aguilar et al., 1998) and from the basolateral nucleus of the amygdala (Cunningham et al., 2007). The amygdala has also been the subject of detailed retrograde tracing experiments aimed at mapping out the connectivity of the zincergic network. The distribution of zincergic cells, like the pattern of terminal staining, varies across different nuclei. Zincergic cell bodies are plentiful in the nuclei of the basolateral complex and in the posterior cortical nuclei but are rare in the anterior amygdala and absent from the lateral olfactory tract nucleus, accessory olfactory tract nucleus, central nucleus, and much of the medial nucleus CHAPTER 1: General Introduction 11

(Christensen & Geneser, 1995; Brown & Dyck, 2004). In the corticomedial amygdaloid complex, zincergic terminals originate from other amygdala neurons – primarily these are intrinsic intra- and internuclear projections, but some are commissural projections from the contralateral amygdala – as well as from neurons in various cortical regions and in hippocampus, specifically from the ipsilateral CA1 region. The specific locations of the cells of origin vary depending on the corticomedial nuclei in question (Christensen & Frederickson, 1998). The distribution of zincergic terminals and cell bodies has been mapped in particular detail in the case of the hippocampal formation, which includes the hippocampus proper in addition to the dentate gyrus, entorhinal cortex, and subiculum (Slomianka, 1992). The entorhinal cortex contains both zincergic cells and terminals. In lateral entorhinal cortex, terminal staining is most prominent in layers 2/3, whereas in medial entorhinal cortex, staining is strongest in deep layer 3 and in layer 5. The entorhinal cortex projects to the dentate gyrus via the perforant path; notably, the lateral perforant path contains zincergic projections, while the medial perforant path does not. In the dentate gyrus, zincergic neurons are found in the granule cell layer, giving rise to the extremely zinc-rich mossy fiber pathway. Zincergic cells are also found deep in the pyramidal cell layer of CA3 and in the superficial pyramidal cell layer of CA1. Zincergic terminal staining, stemming from CA3 and CA1 pyramidal cells, is prominent in stratum oriens and stratum radiatum, the latter of which contains the Schaffer collateral pathway projecting from CA3 to CA1. Finally, the subiculum is notable in that it is among the most zinc-rich of any brain regions, but lacks zincergic cell bodies of its own, though such cells are located in layers 2/3 and 5 of the surrounding pre-, pro-, and parasubiculum. The subiculum itself receives zincergic projections from all three of these regions, as well as from the cingulate, perirhinal, and entorhinal cortices, and from the ipsilateral anterodorsal nucleus of the thalamus; a rare example of a subcortical structure that sends zincergic projections to a cortical one (Long & Frederickson, 1994; Long et al., 1995). The most notable source of zincergic projections to the subiculum is pyramidal cells in CA1 and the prosubiculum.

1.1.3.2 Diencephalon, brainstem, and spinal cord In sharp contrast with the telencephalon, much of the diencephalon is only sparsely innervated by zincergic projections and, with only a few exceptions, entirely lacking in zincergic cell bodies. Moderate-to-dense terminal staining is detectable in a few regions, CHAPTER 1: General Introduction 12 however, including the bed nucleus of the stria terminalis (BNST) and some associated nuclei, the zona incerta, the lateral habenula, and parts of the thalamus and hypothalamus (Pérez-Clausell et al., 1989; Frederickson et al., 1992; Mengual et al., 2001). In the thalamus, several nuclei stain in select regions or in their entirety, including the laterodorsal, anterodorsal, and anteroventral nuclei (Mengual et al., 2001). In the hypothalamus, zincergic terminals are found in the outer shell of the ventromedial nucleus, in the terete nucleus, and in the premammillary nucleus (Frederickson et al., 1992). These zincergic terminals are the product, at least in part, of amygdaloid efferents that pass through the stria terminalis (Pérez-Clausell et al., 1989). Some of the zincergic projections to the ventromedial hypothalamus likely pass through the fimbria as well, and originate in the hippocampal formation near the subiculum. In the case of the BNST, the zincergic terminals also stem primarily from cells in the amygdala and hippocampus (Howell et al., 1991). Like the diencephalon, the brainstem and cerebellum contain relatively little vesicular zinc, with the notable exception of the dorsal cochlear nucleus (DCN) in the auditory brainstem (Frederickson et al., 1988), in which the granule neurons send zincergic projections to form the parallel fiber system in the molecular zone. In the cerebellum, a few cell types contain vesicular zinc, specifically the Golgi, stellate, and basket cells (Wang et al., 2002). As these cells are GABAergic, they mark a departure from the rule, observed in the forebrain, that zincergic neurons are all glutamatergic. In the spinal cord, zincergic terminals are abundant in the dorsal and ventral gray matter, with the greatest density in the dorsal horn (Jo et al., 2000a; Schrøder et al., 2000). The distribution of zincergic terminals varies further across the different spinal cord lamina, though all lamina exhibit at least some zinc staining. In contrast to the forebrain, the majority (~60-90%, depending on the region) of the zincergic neurons in the spinal cord are GABAergic; the rest are likely glutamatergic and, in fewer cases, glycinergic (Danscher et al., 2001; Wang et al., 2001). Neurons that express ZnT3 and contain histochemically- reactive zinc are also found in the peripheral nervous system, including in the dorsal root ganglia, superior cervical ganglia, and lumbar sympathetic ganglia, though it is not confirmed in all cases that zinc is stored in synaptic vesicles in the axon terminals of these neurons (reviewed by Wang & Dahlström, 2008). Finally, it appears that photoreceptor cells, and possibly other retinal cell types, are zincergic. This conclusion is based on the localization of ZnT3 (Redenti & Chappell, 2004) CHAPTER 1: General Introduction 13 and histochemically-reactive zinc (Ugarte & Osborne, 1998; Wang et al., 2006) in the retina, as well as on fluorescent imaging experiments. ZnT3 expression and zinc staining are observed in the outer plexiform layer, where photoreceptor cells synapse on the dendrites of bipolar and horizontal cells. Using a zinc-sensitive fluorescent dye, zinc release has been visualized in this layer following depolarization of retinal cells, providing evidence that photoreceptor cells are zincergic (Redenti & Chappell, 2005). ZnT3 expression and zinc staining are also observed in the inner plexiform layer, where – among other synaptic interactions – bipolar cells synapse on the dendrites of ganglion cells. Further, ZnT3 is localized to the inner nuclear layer, which contains the cell bodies of horizontal, bipolar, and amacrine cells, suggesting that some or all of these cell types could be zincergic. However, no evidence of zinc release was observed in the inner plexiform layer following retinal cell depolarization (Redenti & Chappell, 2005). Zinc staining is also observed in some ganglion cells (Ugarte & Osborne, 1998; Wang et al., 2006) and ZnT3 is expressed in the ganglion cell layer (Redenti & Chappell, 2004).

1.1.3.3 Development of the zincergic system The developmental time course of the zincergic system has been studied in detail, using in situ hybridization for ZnT3 and zinc-selenium staining to visualize zincergic terminals and cell bodies. In rats, faint zinc staining is detectable at birth in the piriform, cingulate, and motor cortices, as well as in the septum and hippocampus (Valente et al., 2002). There is a notable increase in zinc staining from P3 to P5. By this point, zinc staining is apparent in the telencephalic structures where it is later seen in the adult brain, though the distribution within structures remains immature. Another dramatic increase occurs from P9 to P12, at which point zinc staining is very close to its mature appearance. Staining intensity peaks between the second and third postnatal week, and then slowly declines to the level observed in adulthood. In mice, ZnT3 expression is detectable during late embryonic development, with diffuse expression in anterior cerebral areas, and clearly- defined expression in the ventricular zone, subventricular zone, cortical plate, piriform cortex, and olfactory bulbs (Valente & Auladell, 2002). ZnT3 expression intensifies in the neocortex, hippocampus, amygdala, and parts of the entorhinal and perirhinal cortices throughout early postnatal development. Diffuse ZnT3 expression is transiently observed in the septum, striatum, thalamus, and cerebellum, as well as the corpus callosum and anterior commissure, but disappears from these regions by, or shortly after, P9. ZnT3 CHAPTER 1: General Introduction 14 expression in proliferative regions, including the subventricular zone, ependymal layer, and cerebellar external granule cell layer, transiently increases during early postnatal development, then also begins to decrease around P9. From P12 onward, the brain has essentially adopted its mature pattern of ZnT3 expression. In contrast to the adult brain – in which the pattern of vesicular zinc-containing cells is closely correlated with the pattern of cells expressing ZnT3 – there is significant divergence between the two patterns in the early postnatal brain (Valente & Auladell, 2002). This suggests that ZnT3 might serve a developmental function that differs from its primary role in adulthood as a vesicular zinc transporter, and it indicates that other transporters are involved in generating a transient, non-ZnT3-dependent pool of histochemically-reactive zinc in some neurons. Interestingly, prior to P9 – but not after – some zinc-containing cells express the glial cell marker GFAP, which explains why ZnT3 expression is observed in the white matter and neuroepithelia at this stage. The developmental distribution of zincergic terminals or cells has been studied in the olfactory bulb, piriform cortex (Friedman & Price, 1984), amygdala (Mizukawa et al., 1989), striatum (Vincent & Semba, 1989), and barrel cortex (Czupryn & Skangiel-Kramska, 1997; Garrett & Slomianka, 1992), as well as in the hippocampus of the rat (Zimmer & Haug, 1978), mouse (Slomianka & Geneser, 1997), and rabbit (Sanchez-Andres et al., 1997). Understanding the maturation of the zincergic system in the hippocampus is particularly critical, considering that many electrophysiological experiments designed to examine the signaling functions of vesicular zinc use hippocampal slices prepared from young animals. Since mice are commonly used for this purpose, the description here focuses on the hippocampus of the mouse. In the dentate gyrus, no zinc staining is observed at P0. By P15, the pattern of terminal staining corresponds with that seen in adulthood, but only about two-thirds of the neurons in the granule cell layer stain for zinc. It is not until P28 that all such cells are stained. Zinc staining in the mossy fibers can first be visualized at P3 and appears mature by P15. Staining in stratum radiatum, containing the Schaffer collateral fibers, follows a similar time course as the mossy fibers, though it is evident a little earlier. In terms of cell body labeling, the CA3 pyramidal cells achieve a mature appearance by P7, considerably earlier than the CA1 pyramidal cells, which exhibit a mature appearance by P21. Most of the parahippocampal regions, including the subiculum, presubiculum, parasubiculum, and entorhinal areas, exhibit a mature appearance by P21, and all appear fully mature by P28. CHAPTER 1: General Introduction 15

1.1.4 Neuronal Zinc Signaling Since the discovery that free zinc is contained in synaptic vesicles, it has been speculated that zinc is released into the synaptic cleft by vesicular exocytosis. And it has long been established – primarily through experiments on brain slices and cultured cells – that exogenously-applied zinc can bind to, and modulate the function of, a plethora of targets, including glutamate, γ-aminobutyric acid (GABA), and glycine receptors (reviewed by Smart et al., 2004). Thus, there are many potential mechanisms through which synaptically-released zinc could exert a neurotransmitter-like effect. Synaptically-released zinc can also translocate into neurons (Li et al., 2001b; Suh, 2009; Aiba et al., 2013), where it can modulate intracellular signaling through a host of mechanisms. The extent to which extracellular or intracellular zinc signaling occurs under normal, physiological conditions is not fully understood, but recent studies have begun to provide a clearer understanding, specifically of how much zinc is found in the extracellular space and which receptors zinc ions act on. The tonic extracellular concentration of zinc is estimated to be in the low nanomolar (< 25 nM) range (Frederickson et al., 2006a), with even lower concentrations around synapses, as discussed in the next section. These low concentrations are sufficient for zinc to exert tonic modulatory effects on some receptors with very high affinity for zinc. When zincergic neurons are activated, the extracellular zinc concentration is increased, potentially allowing zinc to exert phasic signaling effects on lower-affinity targets. The exact concentrations achieved during different levels of neuronal activity is still a matter of debate, however. Concentrations of 100-300 µM are sometimes cited, based in part on early research looking at zinc release during very intense excitation (Assaf & Chung, 1984), but there is little evidence that this level of zinc release occurs under normal physiological conditions. Most estimates put the number around 1-10 µM following stimulation of the mossy fibers (reviewed by Frederickson et al., 2006b), though the concentration has also been estimated to be in the low nanomolar range, not much above the tonic extracellular level (Kay, 2003). There is evidence that zinc release following tetanic stimulation of the mossy fibers is substantial enough that zinc can spill-over from synapses in stratum lucidum and diffuse over 100 µm into the adjacent stratum oriens over a timespan of seconds, where its effect can be detected on NMDA receptors (Ueno et al., 2002). In contrast to this, others have CHAPTER 1: General Introduction 16 argued that vesicular zinc does not freely diffuse across the synaptic cleft, but is instead externalized, adhering to extracellular membranes to form a “zinc veneer”, with zinc ions in the veneer potentially contributing to tonic zinc signaling (Kay, 2006; Kay & Tóth, 2006, Nydegger et al., 2010, 2012). Whether or not some or all of the released zinc contributes to an extracellular veneer, the preponderance of the evidence supports the interpretation that zinc ions can move across the synaptic cleft, allowing zinc signaling to occur at both pre- and postsynaptic receptors. The elimination of vesicular zinc, and therefore activity-dependent zinc release, makes the ZnT3 KO mouse a useful tool for studying the signaling roles of vesicular zinc. Over the last several years, the development of new tools – including increasingly effective zinc chelators and fluorescent probes – and their subsequent application to ZnT3 KO mice has produced the best evidence to date that ZnT3-dependent vesicular zinc release can affect neuronal function through a variety of receptors. Simultaneously, new advancements have challenged the applicability of some previous findings. In electrophysiological experiments, the effects of endogenous zinc signaling are often inferred by applying a zinc chelating agent that binds free zinc ions, thereby blocking zinc signaling. Tricine and calcium-saturated EDTA (CaEDTA) are two commonly used chelators – both are membrane-impermeant, which restricts their chelating effect specifically to the extracellular space. However, both these chelators are now known to be insufficient to prevent fast, high-affinity zinc binding (Pan et al., 2011; Anderson et al., 2015). Tricine does not bind zinc strongly enough to prevent zinc from accessing sites with low-nanomolar zinc affinity, even when the chelator is applied at relatively high concentrations. CaEDTA is a much stronger chelator, but it is slower and therefore unable to block fast zinc signals of the sort that likely occur following vesicular zinc release (Kay, 2003; Paoletti et al., 2009; Pan et al., 2011). The use of these chelators may thus lead to an underestimation of the impact of zinc in some experimental settings. A more optimal choice for extracellular zinc chelation is ZX1, a relatively new, membrane-impermeant zinc chelator that is fast and binds zinc with higher-affinity than CaEDTA (Pan et al., 2011; Anderson et al., 2015).

1.1.4.1 Excitatory signaling 1.1.4.1.1 NMDA receptors Zinc is well-placed and well-equipped to serve as a modulator of excitatory synaptic transmission, as it is localized in vesicles in glutamatergic terminals and can influence the CHAPTER 1: General Introduction 17 function of all three classes of ionotropic glutamate receptors: NMDA, AMPA, and kainate receptors. The first of these is the best-known target of zinc among the neurotransmitter receptors. Depending on its concentration, zinc has the potential to interact with NMDA receptors through multiple sites (reviewed by Paoletti et al., 2009). A high-affinity, voltage- independent binding site on the N-terminal domain of the GluN2A subunit is responsible for allosteric inhibition of NMDA receptors by low nanomolar levels of zinc, through a reduction in the channel opening probability. Similarly, zinc can bind to the N-terminal domain of the GluN2B subunit but with affinity in the low micromolar range. At higher concentrations still (>20 µM), zinc can act as a voltage-dependent pore blocker of NMDA receptors, regardless of their subunit composition. Numerous experiments on brain slices have demonstrated that exogenous application of micromolar concentrations of zinc can inhibit excitatory neurotransmission mediated by NMDA receptors (Vogt et al., 2000; Izumi et al., 2006; Vergnano et al., 2014). Conversely, applying an extracellular chelator to block endogenous zinc signaling potentiates NMDA receptor-mediated activity (Lu et al., 2000; Vogt et al., 2000; Molnár & Nadler, 2001). The evidence for a tonic effect of extracellular zinc on NMDA receptors is mixed, though the most recent evidence favors such an effect. Vergnano et al. (2014) examined tonic zinc inhibition of NMDA-dependent excitatory postsynaptic currents (EPSCs). They found no effects of zinc chelation (with tricine or CaEDTA) at either mossy fiber-CA3 (MF- CA3) synapses or Schaffer collateral-CA1 (SC-CA1) synapses, suggesting that the extracellular zinc concentration at rest is insufficient for zinc to tonically occupy the high- affinity GluN2A binding site. However, a more recent experiment, using the superior zinc chelator ZX1 and a new extracellular ratiometric zinc sensor, estimated tonic zinc levels to be around 1-10 nM in the DCN (Anderson et al., 2015), sufficient to inhibit extra-synaptic NMDA receptors in wild type (WT) mice (and also in ZnT3 KO mice, suggesting that spontaneous vesicular zinc release is not a critical determinant of the tonic zinc level, at least in the DCN). At NMDA receptors located closer to the synapse, however, there was no inhibitory effect of tonic zinc; this suggests that tonic zinc levels are lower around the synapse, leaving the binding site on these receptors unoccupied. Thus, these results do not necessarily contradict the results of Vergnano et al. (2014), who found that GluN2A- containing NMDA receptors were not inhibited by tonic zinc. The results also fit with findings by Izumi et al. (2006) that tonic zinc has no effect on excitatory postsynaptic potentials (EPSPs) mediated by synaptic NMDA receptors, but does affect other NMDA- CHAPTER 1: General Introduction 18 receptor dependent functions, probably by inhibiting extra-synaptic GluN2A-containing NMDA receptors. Why extracellular tonic zinc levels might be lower near synapses is not known. It could be the case that the expression of zinc-importing ZIPs decreases as distance from the synapse increases. Whatever the cause, by leaving the GluN2A binding site unoccupied, lower zinc levels around the synapse would provide more opportunity for activity-dependent zinc signaling. Indeed, it has been demonstrated repeatedly that binding of released zinc to the high-affinity GluN2A site is relevant for NMDA receptor function under physiological conditions. Mutant mice have been created in which the affinity for zinc at this site is dramatically reduced, eliminating the inhibition of NMDA receptors by nanomolar concentrations of zinc (Nozaki et al., 2011). Using these mice and ZnT3 KO mice, Vergnano et al. (2014) clearly demonstrated that vesicular zinc release inhibits NMDA receptors through binding to the GluN2A subunit. This was observed both at MF-CA3 and SC-CA1 synapses. Using ZX1, Pan et al. (2011) showed that a single mossy fiber action potential releases enough zinc to inhibit NMDA EPSCs. ZnT3-dependent zinc release has also been shown to inhibit GluN2A-containing NMDA receptors in the DCN (Anderson et al., 2015). In this system, stimulation of the zincergic parallel fibers leads to NMDA-mediated EPSCs in postsynaptic interneurons via activation of extra-synaptic NMDA receptors by glutamate spillover from the synapse. Chelation of zinc with ZX1 enhanced the EPSCs. Whether released vesicular zinc can inhibit NMDA receptors by binding to the lower-affinity sites is a matter of more debate. Assuming the extracellular concentration of zinc achieved during mossy fiber stimulation is 1-10 µM (Frederickson et al., 2006b), then zinc should bind to the GluN2B site. Indeed, there is evidence that endogenous zinc release has inhibitory effects on NMDA-mediated neurotransmission that are not accounted for by its effect on the voltage-independent GluN2A site (Vogt et al., 2000; Molnár & Nadler, 2001). On the other hand, Vergnano et al. (2014) did not find that endogenous zinc release was substantial enough to affect the GluN2B site. Other findings (see below) that vesicular zinc release can modulate micromolar-affinity targets provide indirect evidence that released zinc should be able to bind to the low-micromolar affinity GluN2B subunit, and perhaps even the lower affinity, non-subunit-dependent site.

1.1.4.1.2 AMPA and kainate receptors The reported effects of zinc on AMPA receptors run the gamut of possible findings, CHAPTER 1: General Introduction 19 with some studies reporting potentiation of AMPA currents by micromolar zinc (Rassendren et al., 1990; Lin et al., 2001; Blakemore & Trombley, 2004) and others reporting inhibition at micromolar concentrations (Zhang et al., 2002). The effect may also be subunit-dependent and biphasic; one study showed that homomeric GluA3 receptors, but not other AMPA receptors, are potentiated by low micromolar (< 10 µM) zinc, whereas all AMPA receptors seem to be inhibited by high micromolar (> 100 µM) zinc (Dreixler & Leonard, 1994). Vesicular zinc is known to be present at synapses expressing AMPA receptors, though interestingly AMPA receptors are less abundant at zinc-positive terminals than at zinc-negative ones, at least at SC-CA1 synapses (Sindreu et al., 2003). Surprisingly, then, in many experiments in which AMPA-mediated currents have been recorded from slices, no effect of zinc was detected (e.g., Pan et al., 2011; Perez-Rosello et al., 2013; Vergnano et al., 2014). The strongest evidence to date that favours an effect of endogenous zinc on AMPA receptors was provided by Kalappa et al. (2015). Chelation of extracellular zinc with ZX1 increased AMPA-mediated EPSCs in the DCN and at the hippocampal SC-CA1 synapse, indicating an inhibitory effect of zinc on AMPA receptors. Further, this inhibition was dependent on vesicular zinc release. It should be noted, however, that prior experiments using ZX1 did not detect an effect of zinc chelation on AMPA-mediated currents at MF-CA3 synapses (Pan et al., 2011). Like NMDA and AMPA receptors, the effect of zinc on kainate receptors is subunit- dependent (Mott et al., 2008). GluK4- and GluK5-containing receptors are inhibited by similar concentrations of zinc as the GluN2B-containing NMDA receptor (i.e., less than 10 µM), whereas GluK1 or GluK2 homomers are less zinc-sensitive, requiring concentrations above 50 µM to be inhibited. The exception seems to be GluK3-containing receptors, which are potentiated by concentrations of zinc ranging from low micromolar to about 100 µM; at concentrations above that, GluK3 homomers are inhibited (Veran et al., 2012). In hippocampal slices, exogenous application of zinc inhibits kainate-mediated EPSCs in CA3 cells with an IC50 of about 15 µM (Mott et al., 2008). In addition, repeated theta-pattern stimulation of the mossy fibers potentiates kainate-mediated field EPSPs; this effect was also inhibited by zinc, whereas elimination of endogenous extracellular zinc by chelation potentiated frequency facilitation. CHAPTER 1: General Introduction 20

1.1.4.2 Inhibitory signaling 1.1.4.2.1 Glycine receptors Zinc can modulate both major classes of inhibitory neurotransmitter receptors found in the CNS: glycine and GABA receptors. Regarding the former, the effect of zinc is dependent on concentration (reviewed by Lynch, 2004). At nanomolar to low-micromolar levels, zinc allosterically potentiates glycine receptor function by binding to the extracellular N-terminal domain of the α1 subunit, decreasing the rate of dissociation of glycine from the receptor, and increasing the channel opening probability and mean burst duration. At higher concentrations (>10 µM), zinc has a transient potentiating effect, especially if glycine is present at the receptor before zinc, but then inhibits receptor function by binding to a pH-sensitive site that is distinct from the nanomolar-sensitive one. This binding stabilizes the closed conformation of the channel. When expressed in cultured cells, the W170S missense mutation of the human glycine receptor α1-subunit gene results in a loss of the potentiating effect of zinc and a simultaneous increase in the attenuating effect of high zinc concentrations. Other channel properties are not affected in heteromeric α1β glycine receptors, which are the most common type in the CNS (Zhou et al., 2013). Individuals with this mutation exhibit hyperekplexia, characterized by an enhanced startle response and hypertonia. In addition, transgenic mice with a substitution of the D80 residue in the α1-subunit gene, which also eliminates the potentiating effect of zinc on α1-containing glycine receptors, exhibit a hyperekplexia-like neuromuscular phenotype and increased startle response (Hirzel et al., 2006). These studies provide evidence that zinc signaling at glycine receptors is physiologically relevant and important. Extracellular zinc is capable of tonically influencing inhibitory neurotransmission through glycine receptors. Experiments on slices containing hypoglossal motoneurons show that tonic extracellular zinc levels potentiate spontaneous inhibitory postsynaptic currents (IPSCs) (Hirzel et al., 2006). More recently, it has been observed that tonic zinc has an inhibitory effect in slices from the DCN. Here, chelating zinc with ZX1 disinhibited the principal fusiform neurons by eliminating the potentiating effect of zinc on glycine receptors (Perez-Rosello et al., 2015). Consistent with tonic inhibition, the effect of zinc was not dependent on ZnT3. CHAPTER 1: General Introduction 21

1.1.4.2.2 GABA receptors Though zincergic neurons in the CNS are most commonly observed to be glutamatergic, neurons that are both GABAergic and zinc-containing have been identified in the cerebellum and spinal cord (Danscher et al., 2001; Wang et al., 2001, 2002), and there is even evidence that some hippocampal mossy fiber terminals contain both vesicular zinc and GABA (in adult rats: Ruiz et al., 2004; in neonatal rats: Safiulina et al, 2006).

There are two classes of GABA receptors, GABAA and GABAB, of which the former has been better characterized with respect to the effects of zinc. Interestingly, it appears that zinc inhibits these receptors, which at the circuit level would be predicted to have a disinhibitory or net excitatory effect. This contrasts with most other neurotransmitter systems, in which the modulatory effect of zinc is net inhibitory. However, the inhibitory effect of zinc on GABAergic transmission recorded from slices is somewhat equivocal, as discussed below. In cell cultures, the effect of zinc on GABAA receptors depends on their subunit composition.

Heteromeric GABAA receptors containing α-subunits and β-subunits are non-competitively inhibited by concentrations of zinc ranging from submicromolar to 300 µM, though receptors containing the γ-subunit are rendered insensitive to this effect (Draguhn et al., 1990; Smart et al., 1991). Zinc can act at three extracellular sites on the GABA receptor; one located in the channel pore, and the other two located at the interface between the α and β subunits (Hosie et al., 2013). The presence of the γ-subunit disrupts two of these binding sites, accounting for the decreased inhibitory effect of zinc.

In the hippocampus, endogenously-released zinc appears to inhibit GABAA receptors on CA3 neurons (Ruiz et al., 2004), as chelating zinc with CaEDTA potentiated IPSCs recorded from these cells in slice preparations. The source of the zinc (and GABA) is likely the mossy fibers, as the effect of zinc chelation was observed following stimulation of the dentate granule cells, but not the stratum radiatum interneurons. However, the only available evidence from ZnT3 KO mice does not support an inhibitory effect of endogenously-released zinc on GABAA-mediated IPSPs in CA3 pyramidal cells (Lopantsev et al., 2003). When single GABAA-mediated IPSPs (or GABAB-mediated IPSPs) were evoked in hippocampal slices from ZnT3 KO mice, the IPSP amplitude and duration were normal. Only with intense repetitive stimulation was a difference between WT and ZnT3 KO mice revealed, and in this case the GABA-mediated IPSPs recorded from ZnT3 KO slices were actually smaller in amplitude. If zinc was exerting an inhibitory effect on GABAA receptors, CHAPTER 1: General Introduction 22 then one would expect GABAergic transmission to be potentiated, rather than diminished, in the ZnT3 KO mice.

1.1.4.3 Metabotropic signaling In contrast to the abundance of ionotropic targets that have been characterized for zinc, less is known about how zinc affects metabotropic signaling. One landmark discovery in this area was the recognition of a Gq-protein-coupled receptor in colonocytic cells that, in response to micromolar concentrations of extracellular zinc, causes an increase in the cytosolic calcium concentration via mobilization of intracellular calcium stores through the

Gq-PLC-IP3 pathway (Hershfinkel et al., 2001). The identity of this receptor, for which zinc is the primary endogenous ligand, has since been confirmed as the orphan G-protein- coupled receptor 39 (GPR39) (Holst et al., 2007; Yasuda et al., 2007; Besser et al., 2009). Furthermore, there is evidence that this receptor is functional in the brain; an exciting development in the field of zinc neurobiology. In hippocampal slices, repeated stimulation of the mossy fibers increases intracellular calcium in CA3 neurons through the Gq-PLC pathway. This effect is reduced by zinc chelation with CaEDTA, and it is also reduced in both ZnT3 KO and GPR39 KO mice (Besser et al., 2009; Chorin et al., 2011), though some residual effect remains, likely due to activation of mGluRs by glutamate. This demonstrates that vesicular zinc release and the subsequent binding of zinc to GPR39 contributes to the rise in intracellular calcium in CA3 pyramidal neurons in the hippocampus. Contrarily, in examining the same phenomenon in MF-CA3 synapses, Evstratova and Tóth (2011) did not detect an effect of zinc chelation with CaEDTA on the rise in intracellular calcium, nor did they observe a reduction in the calcium transient in ZnT3 KO mice, which would suggest that zinc signaling has no role in this phenomenon. The reason for the inconsistency between the results of the two groups is not clear, though methodological differences – including differences in the method of applying the calcium-sensitive die, and the location that the calcium transients were recorded from (cell bodies in the CA3 pyramidal layer by Besser et al. versus the proximal apical dendrite by Evstratova & Tóth) – could be the source. Gq activation via the zinc-GPR39 pathway increases phosphorylation of Ca2+/calmodulin-dependent protein kinase II (CaMKII) and extracellular signal-regulated kinase (ERK) 1/2 (Besser et al., 2009). Downstream of ERK1/2, the potassium-chloride cotransporter 2 (KCC2) is also affected by zinc-induced GPR39 activation (Chorin et al., CHAPTER 1: General Introduction 23

2011). Zinc signaling increases KCC2 expression in CA3 neurons, likely through a reduction in endocytosis and/or an increase in rapid recycling of the transporter to the plasma membrane. KCC2 participates in setting the chloride gradient, and increasing its activity lowers the reversal potential for GABAA currents, increasing GABA-mediated inhibitory drive. Interestingly, intracellular influx of zinc has the opposite effect on KCC2 activity (Hershfinkel et al., 2009). GPR39 KO mice are more susceptible to kainate-induced seizures, possibly due to the inability to increase KCC2 expression and inhibitory drive in response to excitatory stimulation (Gilad et al., 2015). Activation of GPR39 by release of vesicular zinc can also affect the presynaptic neuron, through a mechanism that is dependent on endocannabinoid signaling (Perez- Rosello et al., 2013). In the DCN, zinc activates GPR39 on postsynaptic fusiform cells, leading to synthesis of 2-arachidonolylglycerol (2-AG) through a PLC-dependent pathway. This acts as retrograde signal, activating presynaptic CB1 receptors and leading to short- term depression of the synapse through a decrease in release probability. Importantly, tonic zinc levels are insufficient to induce 2-AG synthesis; activity-dependent vesicular zinc release is required. Vesicular zinc’s ability to induce endocannabinoid signaling provides a mechanism by which the effect of zinc can switch from inhibitory to excitatory based on the frequency of neuronal stimulation. In the DCN, application of ZX1 during 5 or 20 Hz stimulation of the zincergic parallel fibers potentiates AMPA EPSC amplitude in the postsynaptic cartwheel cells, confirming zinc’s inhibitory effect on AMPA receptors. However, during high-frequency (50 Hz) stimulation, ZX1 has the opposite effect on AMPA- mediated currents, indicating a potentiating effect of zinc. This potentiating effect is achieved by recruitment of endocannabinoid signaling, which reduces presynaptic release probability and enhances synaptic facilitation (Kalappa & Tzounopoulos, 2017). Presynaptic effects have also been observed in the case of ERK1/2 signaling in the hippocampal mossy fiber path (Sindreu et al., 2011). In the presynaptic mossy fibers, the activated phosphorylated form of ERK1/2 (pERK) is reduced in ZnT3 KO mice, both at baseline and, more pronouncedly, after pharmacological or behavioural treatments that normally increase pERK. In addition, mitogen-activated protein kinase (MAPK) phosphatase activity is increased in ZnT3 KO mice. This is possibly due to disinhibition caused by the lack of zinc, as zinc inhibits MAPK phosphatase. Normally, vesicular exocytosis leads to an increase in presynaptic mossy fiber pERK. Sindreu et al. propose that this is due to direct inhibition of MAPK phosphatase by zinc, which may be transiently CHAPTER 1: General Introduction 24 elevated in the cytosol following presynaptic reuptake. Inhibition of MAPK phosphatase would then result in an increase in pERK, due to reduced dephosphorylation. Alternatively, zinc might exert its presynaptic effect by triggering a retrograde signal – as in the GPR39- dependent endocannabinoid signaling pathway described above – or directly through activation of receptors on the presynaptic membrane. Given some of the findings presented in the next section, tropomyosin receptor kinase B (TrkB) might be a candidate.

1.1.4.4 BDNF and TrkB signaling There is evidence that zinc can influence neuronal function through TrkB, a tyrosine kinase receptor of which brain-derived neurotrophic factor (BDNF) is the canonical endogenous ligand. Zinc has been posited to affect TrkB by translocating into the cell and activating an intracellular signaling cascade. In cell cultures and hippocampal slices, exogenous application of zinc at micromolar concentrations leads to transactivation (i.e., activation independent of BDNF) of TrkB, as indicated by TrkB phosphorylation (Huang et al., 2008). Transactivation requires entry of zinc into the cell, where zinc is thought to inhibit C-terminal Src kinase. This reduces autoinhibition – and therefore increases activation – of Src family kinases, which are involved in phosphorylation of TrkB. In hippocampal slices, mossy fiber potentiation by exogenous zinc was blocked in mice lacking TrkB (Huang et al., 2008), providing some evidence of the importance of TrkB transactivation by zinc. However, a caveat to these findings is that it is unclear whether they apply in the intact brain. This was investigated by Helgager et al. (2014), who found – contrary to what one might expect if zinc transactivates TrkB in vivo – that TrkB phosphorylation was not reduced in the mossy fibers of ZnT3 KO mice (Helgager et al., 2014). Indeed, it was increased, as was total hippocampal BDNF content. Another way zinc may influence TrkB is through an extracellular, BDNF-dependent pathway. So far, experiments on this phenomenon have only been conducted using cell cultures – and, in some cases, using high concentrations of zinc over extended periods of time – so it is uncertain how relevant the findings are to the function of the intact brain. In this putative pathway, micromolar concentrations of zinc promote activation of matrix metalloproteinases (MMPs), which cleave extracellular proBDNF and thereby increase the amount of mature BDNF at the synapse. MMPs potentially promote proBDNF release as well (Hwang et al., 2005). BDNF then activates TrkB receptors, promoting phosphorylation of downstream targets such as C-terminal Src kinase, ERK1/2, and Akt. Increases in CHAPTER 1: General Introduction 25 extracellular tissue plasminogen activator and plasminogen also seem to be downstream of the zinc-MMP-BDNF-TrkB pathway (Hwang et al., 2011), as is activation of protein kinase A (PKA) (Poddar et al., 2015). PKA then initiates another cascade, by phosphorylating striatal-enriched tyrosine phosphatase (STEP) and ERK2, the former of which has multiple PKA sites. Phosphorylation of STEP makes ERK2 more persistently active, by decreasing the rate of ERK2 inactivation (dephosphorylation) by STEP.

1.1.4.5 Synaptic plasticity The contribution of vesicular zinc to forms of synaptic plasticity, such as long-term potentiation (LTP) and long-term depression (LTD), has primarily been studied in the hippocampus. Within the hippocampus, different mechanisms are involved in LTP induction at different synapses (Malenka & Bear, 2004). At SC-CA1 synapses, LTP induction occurs following high-frequency stimulation (HFS) or theta burst activity, and it is dependent on NMDA receptors. Though the effect of zinc on excitatory neurotransmission seems to be almost uniformly inhibitory, its effect on NMDA-dependent LTP is somewhat more complex. Exogenous application of zinc at very high concentrations (100-300 µM) inhibits LTP when administered prior to induction, most likely through inhibition of NMDA receptors (Xie & Smart, 1994; Izumi et al., 2006; Lorca et al., 2011). In contrast, a lower concentration of zinc (5 µM, but not 30 µM) potentiates HFS-induced LTP and lowers the stimulation frequency required for induction (Takeda et al., 2009). This effect is blocked by an intracellular zinc chelator, suggesting that cellular zinc influx may be required for this potentiation. Similarly, concentrations of zinc ranging from 5-50 µM potentiate LTP induced by theta burst stimulation (Lorca et al., 2011), though in this case the effect was found to be dependent on postsynaptic P2X receptor activation. Specifically, zinc appears to enhance P2X4-evoked currents by binding to the receptor and increasing its affinity for its endogenous ligand, ATP (Coddou et al., 2007; Lorca et al., 2011). In short, evidence from in vitro experiments indicates that high micromolar zinc concentrations inhibit LTP induction, while low micromolar concentrations potentiate it. However, in an experiment examining the same phenomenon in vivo, local perfusion of 0.1-1 µM zinc was found to inhibit LTP (Takeda et al., 2015), which calls into question the relevance of the in vitro findings. Experiments looking at the effects of endogenous zinc on SC-CA1 LTP add further complexity to the scenario. When endogenous zinc signaling is blocked by chelation with CaEDTA, LTP induction by HFS is prevented in hippocampal slices (Izumi et al., 2006; CHAPTER 1: General Introduction 26

Takeda et al., 2009). Izumi et al. suggest that this is due to disinhibition of extrasynaptic GluN2A-containing NMDA receptors, leading to untimely NMDA receptor activation, which blocks LTP induction. Higher concentrations of CaEDTA also inhibit CA1 LTP in vivo (Takeda et al., 2015). However, binding of endogenous zinc to the GluN2A high-affinity subunit normally attenuates LTP at SC-CA1 synapses, as inferred by the fact that mice that lack the high-affinity zinc binding site show enhanced LTP (Vergnano et al., 2014). Thus, zinc appears necessary for SC-CA1 LTP induction, but may also attenuate the strength of this LTP even at very low concentrations. ZnT3 KO mice have not yet been used to examine the contribution of vesicular zinc release to synaptic plasticity at these synapses. MF-CA3 synapses do not show classic NMDA-dependent LTP (Malenka & Bear, 2004), but do exhibit a form of postsynaptic, NMDA-dependent LTP that involves potentiation of NMDA-mediated EPSCs following short bursts of stimulation (Rebola et al., 2008). Similar to its effects at SC-CA1 synapses, endogenous zinc binding at the NMDA receptor inhibits this form of LTP (Vergnano et al., 2014). Also like the SC-CA1 synapse, very high concentrations of exogenous zinc can block LTP in CA3 cells (Xie & Smart, 1994). MF-CA3 synapses are also distinct in their ability to exhibit presynaptic NMDA- independent LTP, in which release probability is increased following HFS. At these synapses, chelation of extracellular zinc prevents LTP induction (Li et al., 2001a). Interestingly, chronic dietary zinc deficiency also impairs MF-CA3 LTP, as does depleting intracellular zinc with a membrane-permeable chelator (Lu et al., 2000). Thus, as with SC- CA1 synapses, LTP at MF-CA3 synapses seems to be dependent, in some form, on released zinc. Surprisingly, then, MF-CA3 LTP is reportedly normal in ZnT3 KO mice, except that post-tetanic potentiation is briefly elevated compared to WT controls (Lavoie et al., 2011). The lack of an apparent effect on LTP might be explained by another recent study that, using ZnT3 KO mice, has revealed fascinating, bidirectional contributions of vesicular zinc release to synaptic plasticity at MF-CA3 synapses (Pan et al., 2011). This study showed that zinc is necessary for the induction of presynaptically-mediated LTP. In ZnT3 KO mice, or following zinc chelation with ZX1, this form of plasticity is lost. However, zinc normally inhibits the induction of a form of postsynaptic, NMDA-independent, calcium-dependent LTP at these synapses. Thus, while one form of LTP is lost in ZnT3 KO mice, another form is simultaneously unmasked. Notably, unmasking of this form of LTP is more apparent in ZnT3 KO mice than following acute zinc chelation in WT mice, indicating that long-term CHAPTER 1: General Introduction 27 compensatory mechanisms are probably involved. The opposing nature of these two mechanisms might account for why Lavoie et al. (2011) did not observe differences in LTP in ZnT3 KO mice. Very little has been published on the possible contributions of zinc signaling to LTD. Interestingly, LTD may be more sensitive to disruption by zinc than LTP, at least at the SC-CA1 synapse. Exogenous application of zinc (1-10 µM) was found to inhibit LTD induction by prolonged, low-frequency stimulation, whereas an inhibitory effect of zinc on LTP induction was only apparent at greater concentrations (Izumi et al., 2006). Given the effective concentrations, as well as other pharmacological evidence, zinc’s effect on LTD was likely due to inhibition of GluN2B-containing NMDA receptors. Also, whereas chelation of extracellular zinc with CaEDTA prevented LTP induction, LTD induction remained intact, suggesting that this form of plasticity is not dependent on zinc. LTD has not been examined in ZnT3 KO mice.

1.1.5 ZnT3 in Health and Disease 1.1.5.1 Aging and Alzheimer’s Disease The evidence for a link between zinc, ZnT3, and Alzheimer’s disease – both from mouse models and from examination of human brain tissue – is extensive. Zinc is involved in amyloid-beta (Aβ) precipitation and plaque formation, and zinc, ZnT3, and other ZnT proteins are all found in amyloid plaques (Bush et al., 1994; Lee et al., 1999; Stoltenberg et al., 2007; Zhang et al., 2008, 2010; Zheng et al., 2010; Lee et al., 2012). In APP/PS1 mice, a model of Alzheimer’s disease, amyloid plaques are found more abundantly in cortical layers that are vesicular zinc-rich (i.e., layers 2/3 and 5) compared to layers with less zincergic innervation (Stoltenberg et al., 2007). In human brain tissue, Aβ oligomers co-localize with synaptic markers. Exogenous Aβ oligomers applied to mouse brain slices exhibit activity- dependent localization to synapses; this phenomenon is diminished in slices from ZnT3 KO mice (Deshpande et al., 2009). In Tg2576 transgenic mice, another model of Alzheimer’s disease, genetic deletion of ZnT3 increases the ratio of soluble-to-insoluble Aβ, decreases the plaque load (Lee et al., 2002), and decreases the formation of cerebral amyloid angiopathy (Friedlich et al., 2004). The decrease in plaque load is particularly evident in female mice, which tend to have greater plaque loads to begin with relative to males (Lee et al., 2002). Interestingly, as female mice age they show an increase in mossy fiber vesicular zinc, total brain zinc, and ZnT3 protein levels (Lee et al., 2002, 2004, 2012). In female CHAPTER 1: General Introduction 28

Tg2576 mice, however, mossy fiber vesicular zinc and ZnT3 expression decrease with age, and ZnT3 expression becomes observable in reactive astrocytes (Lee et al., 2012). Others have found that, in both mice and humans, ZnT3 protein expression and levels of zinc in the hippocampus decline with age, and that the decline in ZnT3 expression is even more pronounced in the cortices of Alzheimer’s disease patients relative to healthy controls (Adlard et al., 2010). Similarly, expression of ZnT3 mRNA declines with age (Oleson et al., 2016), and decreased ZnT3 mRNA has been observed post-mortem in the brains of people with Alzheimer’s disease (Beyer et al., 2009). Decreased ZnT3 in prefrontal and parietal cortex has also been found to correlate with cognitive impairment in people with various forms of dementia (Whitfield et al, 2014). Further, in a study comparing Alzheimer’s disease patients to individuals who are cognitively intact despite exhibiting Alzheimer’s disease neuropathology (i.e., Aβ plaques and neurofibrillary tangles), it was observed that the Alzheimer’s patients had decreased levels of ZnT3 in the hippocampus and increased levels of vesicular and total zinc, whereas the cognitively intact individuals exhibited normal hippocampal ZnT3 expression and less of an increase in total zinc (Bjorklund et al., 2012). In addition, Aβ plaques did not associate with postsynaptic densities in the cognitively intact group, unlike in the Alzheimer’s disease patients. In mice, age-related cognitive decline can be ameliorated with 6 weeks of treatment with clioquinol, a zinc chaperone that is proposed to normalize zinc homeostasis (Adlard et al., 2015). This treatment also restores SC-CA1 LTP induction, which is impaired in aged mice. Furthermore, age-related cognitive decline is exacerbated in ZnT3 KO mice (Adlard et al., 2010). Taken together, these findings indicate that dysregulated homeostasis of vesicular zinc is associated with age- and Alzheimer’s disease-associated cognitive decline, whereas maintaining zinc homeostasis or correcting it with drug treatment protects cognitive ability.

1.1.5.2 Neurodegeneration and cell death Though much research has been conducted on zinc dyshomeostasis and how it can contribute to cell death and neurodegeneration (reviewed by Sensi et al., 2011), there seems to be little evidence that disrupted ZnT3 function or zinc signaling are implicated in neurodegenerative disorders, aside from the connections with Alzheimer’s disease. An exception is a recent study showing that protein levels of ZnT3 (as well as ZnT6) are reduced in the spinal cords of people with sporadic amyotrophic lateral sclerosis (Kaneko et CHAPTER 1: General Introduction 29 al., 2015). Further work will be required to determine whether this is a causal factor in the etiology of the disease or, as seems quite possible, a result of motor neuron loss or some other aspect of the disease process, though Kaneko et al. (2015) did find that spinal cord ZnT3 protein levels were not altered in a mouse model of motor neuron loss, suggesting that motor neuron loss by itself may not be enough to explain the reduction in ZnT3. In general, reduced or eliminated ZnT3 expression seems to precipitate – or at least correlate with – certain deleterious health conditions, as described throughout this section. In a few instances, however, elimination of ZnT3 or chelation of vesicular zinc seems to improve outcomes. In a mouse model of multiple sclerosis, ZnT3 KO mice show better motor function, less demyelination, reduced BBB disruption, and reduced immune cell infiltration into the spinal cord (Choi et al., 2016). A lack of zinc or ZnT3 is also protective against cell death following injury of the retina or optic nerve. In a model of light-induced retinal damage, ZnT3 KO mice and mice maintained on a zinc-deficient diet exhibit less loss of photoreceptor cells (Bai et al., 2013). In the case of optic nerve injury, retinal zinc homeostasis is perturbed, resulting in zinc accumulation first in the amacrine cells and, ultimately, in the retinal ganglion cells (see section 2.2). ZnT3 KO mice show reduced zinc accumulation in the amacrine and retinal ganglion cells and increased ganglion cell survival following nerve crush injury (Li et al., 2017). Even more remarkably, post-injury axon regeneration is greatly enhanced in ZnT3 KO mice relative to WT mice. Similar pro- survival and pro-regenerative effects can be produced by injecting zinc chelators into the eye. Eliminating vesicular zinc may reduce neuronal death following certain types of insult to the brain, probably by eliminating the translocation of released zinc and reducing zinc accumulation in the postsynaptic cells. Following hypoglycemia and subsequent glucose reperfusion, ZnT3 KO mice show reduced cell death of hippocampal CA1 neurons (Suh et al., 2008). However, there is also evidence that ZnT3-dependent zinc release is not necessary for zinc accumulation and cell death. Following kainate-induced seizures, zinc accumulation and cell death were greater in the CA1 neurons of ZnT3 KO mice compared to WT mice – though the same was not true for cells in CA3 (Lee et al., 2000). Similarly, following traumatic brain injury, ZnT3 KO mice exhibit aberrant zinc staining in the somata of neurons and more cell death than WT mice (Doering et al., 2010). Vesicular zinc release has also been implicated in the phenomenon of spreading depression, a wave of neuronal and glial depolarization that – while not injurious under CHAPTER 1: General Introduction 30 normal conditions – is increased in frequency in the wake of brain injury and likely contributes to neuronal damage. Oxygen-glucose deprivation leads to an increase in the extracellular zinc concentration, coincident with the occurrence of spreading depression, but this increase is not observed in ZnT3 KO mice (Carter et al., 2011). Further, intracellular accumulation of zinc – which is normally observed following spreading depression, probably due to uptake of released vesicular zinc – is not observed in ZnT3 KO mice following spreading depression. It is possible that intracellular accumulation of zinc may contribute to neuronal injury, though it is also possible that zinc accumulation may have beneficial effects by contributing to ischemic preconditioning (Lee et al., 2008). In addition, released zinc may, by acting on extracellular sites (e.g., NMDA receptors), dampen neuronal excitability and protect against excitotoxicity. Indeed, vesicular zinc release following spreading depression appears to reduce the frequency of spreading depression events and the rate of spreading depression propagation (Aiba et al., 2012). Zinc only inhibits spreading depression under conditions of normal oxygen content, however, as is the case when spreading depression is induced by potassium chloride injection. Under hypoxic conditions, as when spreading depression is induced by oxygen-glucose deprivation, zinc no longer has this inhibitory effect (Aiba & Shuttleworth, 2013).

1.1.5.3 Other disorders There is evidence that polymorphisms of SLC30A3, the gene that encodes ZnT3, are associated with increased risk of developing schizophrenia. SLC30A3 was found to be consistently downregulated in two schizophrenia patient cohorts (Maycox et al., 2009). A subsequent analysis, using a candidate gene approach, of four different single-nucleotide polymorphisms in the SLC30A3 gene showed a significant association with schizophrenia at the allelic, genotype, and haplotype level (Perez-Becerril et al., 2013). Interestingly, breaking down the analysis by gender revealed a stronger association in females but no significant association in males, indicating that the effect of SLC30A3 genetic status on the risk of developing schizophrenia is specific to females. Recently, this finding was replicated in a larger cohort (Perez-Becerril et al., 2016). The reason for the sex-specificity of this effect is unclear. Estrogen has been shown to decrease ZnT3 expression in female mice (Lee et al., 2004). If this is also the case in humans, then it could explain why females might be more susceptible to a further decrease in ZnT3 function potentially caused by a SLC30A3 polymorphism. CHAPTER 1: General Introduction 31

There also seems to be a genetic association between SLC30A3 and the occurrence of febrile seizures (Hildebrand et al., 2015). It was found that a rare SLC30A3 variant was present in three patients with febrile seizures (1% of the sample), but no one in the control group, and only ~0.1% of the sample in two exome databases. Consistent with the research showing a broadly inhibitory effect of zinc on neurotransmission, and with the observation that ZnT3 KO mice are more susceptible to seizures (Cole et al., 2000), it was observed that ZnT3 KO mice have increased susceptibility to heat-mediated clonic-tonic seizures. Finally, the identified human ZnT3 variant, when transfected into rat PC12 cells, failed to traffic to the membranes of synaptic-like microvesicles and led to a near-total loss of zinc in the synaptic-like microvesicle fraction. This raises the intriguing possibility that at least some SLC30A3 variants in humans might result not merely in moderate changes in transporter efficacy, but rather in near-complete loss of transporter function, not unlike the complete loss of function that occurs in ZnT3 KO mice.

1.1.6 Characterization of ZnT3 KO Mice 1.1.6.1 Neurophysiology As described in the section 1.1.4, electrophysiological function is predominantly normal in ZnT3 KO mice, though with some abnormalities emerging under certain conditions (Lopantsev et al., 2003; Besser et al., 2009; Chorin et al., 2011; Evstratova & Tóth, 2011; Lavoie et al., 2011; Pan et al., 2011; Perez-Rosello et al., 2013; Vergnano et al., 2014; Anderson et al., 2015; Kalappa et al., 2015; Kalappa & Tzounopoulos, 2017). ZnT3 KO mice also appear to exhibit some abnormalities in spontaneous neuronal activity. The frequency of spontaneous EPSCs is diminished in ZnT3 KO mice, though miniature EPSC frequency is normal (Lopantsev et al., 2003). In addition, Lavoie et al. (2011) found that the strength of miniature and spontaneous EPSCs recorded from CA3 pyramidal cells was decreased in ZnT3 KO mice, though the effect was specific to large-amplitude spontaneous and miniature EPSCs. Consistent with the broadly inhibitory effect of vesicular zinc signaling, ZnT3 KO and heterozygous mice are more susceptible to seizures induced by kainic acid (Cole et al., 2000). Interestingly, ZnT3 KO mice are somewhat less susceptible to seizures induced by low doses of bicuculline, a GABAA antagonist, which suggests that GABAergic tone may be increased in these mice, possibly compensating for the loss of the inhibitory effects of zinc on glutamatergic neurotransmission. In light of this suggestion, it is notable that ZnT3 KO CHAPTER 1: General Introduction 32 mice exhibit decreased levels of GABA transaminase and succinic semialdehyde dehydrogenase – two enzymes involved in GABA metabolism – in the barrel cortex (Nakashima et al., 2011). Unlike some seizure-prone strains of mice, ZnT3 KO mice do not exhibit spontaneous seizures or seizures in response to handling or loud noise (Cole et al., 2000), though they are more susceptible to heat-mediated seizures (Hildebrand et al., 2015).

1.1.6.2 Behaviour Given the distribution of ZnT3 and vesicular zinc in the brain, with an abundance in the neocortex and limbic structures, it would be reasonable to assume that zinc signaling is involved in modulating cognitive and emotional behaviours. The generation of the ZnT3 KO mouse provided a valuable tool with which to test this hypothesis. Surprisingly, the initial behavioural characterization of the ZnT3 KO mice revealed extremely little in the way of abnormality (Cole et al., 2001). This study was not without limitations; most notably, mice were tested and analyzed in mixed sex groups, precluding the detection of any sex-specific effects. However, the lack of abnormality over a wide variety of behavioural domains – including motor coordination, thermal nociception, auditory startle response and prepulse inhibition, olfaction, spontaneous exploration, anxiety, passive avoidance, cued and contextual fear memory, spatial navigation, long term memory, reversal learning, and working memory – provided convincing evidence that the loss of vesicular zinc does not result in any notable impairment or abnormality detectable by standard laboratory tests of mouse behaviour. Recent results have pushed back against this initial conclusion, however. The core finding that ZnT3 KO mice do not show gross behavioural abnormalities remains mostly unchallenged, but the evidence now supports subtle abnormalities in certain behavioural domains. One such domain, largely dependent on the amygdala and hippocampus, is fear memory. Examining male mice, Martel et al. (2010) confirmed that ZnT3 KO mice perform normally in a standard fear conditioning test involving multiple tone/shock pairings. However, in a “weak” conditioning paradigm consisting of a single tone/shock pairing, ZnT3 KO mice froze less than controls when re-exposed to the conditioning environment (i.e., contextual fear memory) or to the tone in a novel environment (i.e., cued fear memory). ZnT3 KO mice also exhibited less freezing than controls during a cued fear memory test that used a more complex auditory signal, which might suggest disrupted auditory processing. Further, ZnT3 KO mice were impaired at cued and contextual fear memory CHAPTER 1: General Introduction 33 when tone/shock pairings were administered with the tone and shock separated by a 15 s interval (i.e., trace fear conditioning). In contrast to these results, Sindreu et al. (2011) found that ZnT3 KO mice performed normally in a contextual fear conditioning test in which only a single shock was administered. However, the mice were impaired at discriminating between different contexts. Whether or not ZnT3 KO mice show impaired fear extinction is presently equivocal. The cued fear response has been found to extinguish faster in ZnT3 KO mice relative to WT mice (Martel et al., 2010), but when extinction of the contextual fear response was examined, there was no difference in the extinction rate, though ZnT3 KO mice did freeze less than WT mice during the first session of extinction training further supporting impaired contextual fear memory (Martel et al., 2011). In another set of experiments, Martel et al. (2011) demonstrated further abnormalities in cognition, this time on abilities that are dependent on the hippocampus and perirhinal cortex. Male ZnT3 KO mice were impaired at long-term memory, as inferred from the novel object preference test with a 24 h interval between initial object exposure and testing. When tested with a much shorter latency, however, ZnT3 KO mice perform normally (Sindreu et al., 2011; Wu & Dyck, 2018), suggesting a similar pattern of findings as described for fear memory: ZnT3 KO mice can perform a task when the conditions are fairly “easy”, but are impaired in a more difficult version of the same test. The same has also been observed in a whisker-dependent texture discrimination task, which male ZnT3 KO mice can perform when the difference between textures is large, but not when required to discriminate between fine differences (Wu & Dyck, 2018). Finally, ZnT3 KO mice show mild deficits at performing the Morris water task (MWT), a test of spatial navigation and memory. In the standard version of the MWT with a fixed-location platform, ZnT3 KO mice perform normally. But if the platform is transferred to a novel location, as is often done during the reversal learning phase of the test, they show a slight impairment at finding it. Cole et al. (2001) observed that ZnT3 KO mice took longer to find the platform than controls only on the second trial (of nine) during reversal learning. Martel et al. (2011) found that ZnT3 KO mice took longer to find the platform on the second day of the reversal phase but achieved normal performance by the third day. The impairment on the second day was at least in part because the mice spent more time investigating the previous platform location, suggesting either perseveration or an impaired ability to remember the new platform location. Impaired cognitive flexibility or working memory is also consistent with the finding by Sindreu et al. (2011) that ZnT3 KO CHAPTER 1: General Introduction 34 mice were impaired in a rewarded-alternation spatial T-maze test, failing to reliably alternate at a rate greater than chance. Finally, Adlard et al. (2010) have also reported that ZnT3 KO mice perform normally when tested in the MWT. This is only the case at 3 months of age, however; by 6 months of age, these mice perform significantly worse than age-matched WT controls, suggesting that a lack of vesicular zinc exacerbates the cognitive decline that normally occurs with age. Mixed-sex groups were used in this experiment, so it is unclear whether the effect is more pronounced in either sex. A very recent study by Yoo et al. (2016) has noted the most pronounced behavioural abnormalities in ZnT3 KO mice to date. Interestingly, the effects were strongly sex- dependent; many behavioural abnormalities were seen in males, but none was seen in females. The behavioural effects, observed at 4-5 weeks of age, included decreased social interaction and preference for social novelty, increased repetitive behaviour in the marble burying test, and decreased locomotion and increased anxiety-like behaviour in the open field test. In addition, ZnT3 KO mice had larger cortices; more NeuN-positive neurons in frontal cortex; greater expression of SMI32, a marker of neural processes; and greater expression of BDNF, TrkB, MMP-9, and Shank3. The difference in brain size was first observed at 15 days of age and became more notable by 5 weeks. Finally, greater levels of intracellular free zinc, and greater MT1 and MT2 expression were observed. Taken together, these results are suggestive of an autism-like phenotype in ZnT3 KO mice. While certainly intriguing, these results differ considerably from the pattern of findings established by other studies to date, in which ZnT3 KO mice exhibit – at most – subtle abnormalities. Since Yoo et al. (2016) examined ZnT3 KO mice at an earlier age than in any other behavioural study to date, it is possible that the behavioural abnormalities they observed are prominent in young ZnT3 KO mice and diminish as the animals mature. Comparison of young and mature mice in the same study would help to determine if this is the case. If the results of Yoo et al. (2016) can be further verified and replicated, they would represent a remarkable advancement in our understanding of the importance of ZnT3 in the normal development and function of the brain.

1.1.6.3 Experience-dependent plasticity For the most part, the behavioural characterization of ZnT3 KO mice has showed a consistent pattern, wherein these mice behave normally under standard or “unchallenging” conditions but begin to exhibit cognitive and sensory deficits when the tasks are made CHAPTER 1: General Introduction 35 difficult. This pattern is further supported by a novel line of research emerging from Richard Dyck’s laboratory, indicating that ZnT3 KO mice are deficient in certain forms of experience-dependent plasticity. That is, their brains appear normal under standard laboratory conditions, but they fail to exhibit changes that normal mice would show in response to certain experiences or treatments. The first major finding was that ZnT3 KO mice, when housed in an enriched environment for several weeks, did not show an increase in adult neurogenesis (i.e., the generation of new neurons) in the hippocampal dentate gyrus (Chrusch, 2015). WT mice, on the other hand, did show increased neurogenesis, consistent with previous research (Kempermann et al., 1997). Moreover, whereas WT mice benefitted from the enriched environment in terms of their ability to perform a spatial object recognition task and the MWT, ZnT3 KO mice did not. These findings were followed by the observation that ZnT3 KO mice also fail to exhibit increased hippocampal neurogenesis in response to chronic treatment with the antidepressant drug fluoxetine (Boon, 2006), which, like environmental enrichment, normally increases neurogenesis (Malberg et al., 2000). Finally, in an entirely different region of the brain – the barrel field of the primary somatosensory cortex, which responds to sensory input mediated by the vibrissae, or whiskers – ZnT3 KO mice also fail to exhibit experience-dependent plasticity. In normal mice, when sensory experience is chronically deprived by trimming of a whisker, the sensory-evoked response in the corresponding region of the barrel cortex is diminished relative to the response evoked by stimulation of the surrounding, untrimmed whiskers. In ZnT3 KO mice, however, this phenomenon is not observed (Chrusch, 2015). These results raise interesting, new questions. Among them: if ZnT3 KO mice fail to show neural plasticity and to adapt their behaviour in response to seemingly positive manipulations (such as environmental enrichment and antidepressant treatment), then how might they respond to a negative experience, such as chronic stress? Would an inability to adapt their behaviour make them more vulnerable to the effects of stress? Or, perhaps, could an inability to modulate neural function in response to experience protect against some of the harmful effects of stress on the brain? Primarily, these are the research questions addressed in the present thesis.

1.2 STRESS It is a necessary condition of life – for humans as for all other animals – to be faced with a constantly changing environment. To survive and flourish, an organism must have CHAPTER 1: General Introduction 36 some capacity to adapt to these changes and to meet challenges. These simple observations form the basis for the concept of stress and the stress response. The pioneering endocrinologist Hans Selye (1936) was the first to adopt the term stress to describe a specifically biological phenomenon. His definition evolved over time, with stress eventually coming to mean “the nonspecific response of the body to any demand” (Selye, 1976, p. 1). He was also the first to characterize the biological stress response (or stress syndrome), which he referred to more precisely as the general adaptation syndrome, based on the observation that a plethora of differing treatments and experiences provoked a generalized response consisting of enlargement of the adrenal glands, atrophy of the thymus and related lymphatic structures, and the development of stomach ulcers. This eventually led to the characterization of the hypothalamic-pituitary-adrenal (HPA) axis, the neuroendocrine system through which the brain initiates the response to stress, ultimately triggering the release of the stress hormone cortisol (CORT; or corticosterone, in rodents) from the adrenal glands. Though stress generally carries a negative connotation, it must be noted that if stress is the nonspecific response of the body to any demand, then stress, so defined, must be associated with positive, as well as negative, experience. And even when stemming from a negative experience, such as the threat of a predator, the stress response is obviously a necessary and adaptive component of an animal’s physiology. But undeniably, the negative connotation of stress is not entirely unearned, for stress is not without its significant downsides. One notable example is the damaging effects that stress and the stress response can exert on mental health, particularly when the cause of the stress (i.e., the stressor) is recurring, uncontrollable, or inescapable. It is well established that an increased number of stressful life events – such as those involving loss, humiliation, entrapment, or danger – increases the likelihood of suffering from mental disorders such as depression and anxiety (Kendler et al. 1999; Kendler et al., 2003; Kendler & Gardner, 2016). Clearly, the stress response, and particularly the response to chronic stress, must precipitate biological changes that predispose individuals towards these undesirable mental health outcomes – outcomes which often degrade quality of life far beyond the proximate effects of the initial stressor. It is no surprise, then, that the effects of stress on the body – and particularly on the brain – have been extensively studied, both by clinical researchers and by basic scientists looking to understand the causes and effects of the stress response, often using rodents in CHAPTER 1: General Introduction 37 order to delve more deeply into the underlying mechanisms. To this end, many methods have been used to induce stress in rodents. These include restraining them in a confined space, forcing them to swim in an inescapable tank, housing them in isolation for extended periods of time, exposing them to the scent of a natural predator, subjecting them to unavoidable electric shocks, and many others. In an effort to reduce the predictability of chronic stress – as predictability can potentially lead to habituation and therefore a decreased behavioural and physiological response to the stressor – and to increase the validity of experimentally-induced stress as a model of the varying and unpredictable stressors that humans often face, it is not uncommon for researchers to combine numerous stressors and to vary them across a period of several weeks to create the experience of chronic, unpredictable stress (Katz, 1981; Willner et al., 1987).

1.2.1 Social Defeat Stress One of the more common stress paradigms used in rodents is social defeat stress. Social defeat methods harness the innate territoriality of (usually) male rodents to provoke agonistic encounters between unfamiliar pairs. This often results in one animal emerging as dominant, while the other suffers defeat and swiftly learns to adopt a submissive role. The procedure can consist of a single defeat episode to induce acute stress, or repeated social defeat (RSD) over the course of days or weeks to induce chronic stress. Like most experimental paradigms, a number of different procedures have been proposed for inducing RSD stress in rodents – as well as in some non-rodent species such as tree shrews (e.g., Tornatzky & Miczek, 1994; Avitsur et al., 2001; Bartolomucci et al., 2001; Fuchs & Flügge, 2002; Huhman et al., 2003; Dubreucq et al., 2012). One such variant for stressing mice was first described in the late 1980s by a group in Siberia (Kudryavtseva et al., 1991) but in recent years has become better associated with the work of Eric Nestler and colleagues. In this procedure, male C57BL/6 mice are subjected to daily defeats by a larger, more aggressive CD-1 mouse, usually over the course of 10 days (Berton et al., 2006; Golden et al., 2011). In between defeats, the mice are housed in close quarters but separated by a perforated barrier, allowing sensory contact but preventing fighting or injury. One key strength of social defeat stress is its ethological validity (Gray et al., 2015). Social stress is relevant to a wide variety of species, considering that many species form social hierarchies and territorial relationships that are essential for gaining access to resources such as food and mates. Furthermore, while physical conflict and defeat is rare in CHAPTER 1: General Introduction 38 many modern human societies, the stressors that are relevant to people are often still social in nature or involve coping with social environments. Another key feature is that – as in humans, where not all who experience stress go on to develop stress-related disorders – not all mice respond similarly to RSD stress, and a reliable method exists for grouping mice into distinct populations: one that is behaviourally susceptible and one that is more resilient (Krishnan et al., 2007; Golden et al., 2011). The susceptible mice show a syndrome of behavioural changes that, in many ways, reflects the symptoms of human depression. The changes include social avoidance, heightened anxiety, anhedonia, altered body weight, increased amplitude of the circadian body temperature rhythm, increased sensitivity to cocaine-conditioned place preference, and an increased CORT response to stress. In contrast, the resilient mice show some of the same effects, such as increased anxiety and a sensitized endocrine response, but are in many ways more similar to non-stressed controls, as they do not exhibit social avoidance, anhedonia, weight change, altered place preference, or altered circadian amplitude. Social defeat is not entirely without drawbacks as a stress model, and particularly as a model of psychological stress. For one, the aggressive nature of the defeats often results in injury, which adds a physical component to the stress – as well as potential confounding effects of inflammation. Though measures can be taken to reduce the extent of injury, such as limiting the duration of the defeats, it is very difficult – if not impossible – to eliminate injury completely while retaining the defeat component, which limits the procedure’s utility as a purely psychological stress model. Also, as far as the common varieties of laboratory rodent go, social defeat is much more practical in males then females, due to females being less aggressive towards intruders. This has led to almost all rodent social defeat research being conducted in males. A third drawback is that most protocols for RSD involve the isolated housing of the control animals as well as the defeated ones, which likely induces a degree of stress in the control animals if the experiments are conducted over an extended period of time.

1.2.2 Neurobiology of Stress and Stress Resilience Experiments using RSD, along with many other stress procedures, have revealed previously unknown mechanisms through which stress impacts the brain. Moreover, some of these mechanisms have been found to distinguish mice that are behaviourally resilient to stress from those that are susceptible, indicating that these mechanisms could be promising CHAPTER 1: General Introduction 39 targets for the development of new antidepressant therapies. These mechanisms involve many different regions of the brain – including the prefrontal cortex, BNST, amygdala, hypothalamus, periaqueductal gray, and others – but two that have attracted a great deal of attention are the hippocampus and, particularly over the last decade or so, the nucleus accumbens (NAc). The hippocampus has long been a major focus of research in stress neurobiology, in part because it is a key mediator of the negative-feedback response to HPA axis activation (Jacobson & Sapolsky, 1991). Glucocorticoid and mineralocorticoid receptors, which respond to the CORT, are abundant in the hippocampus (McEwen et al., 1968; Chao et al., 1989; McEwen et al., 2015). Accordingly, the structure and function of the hippocampus is highly responsive to stress. Chronic stress in rats, induced by repeated daily restraint, causes atrophy of the apical dendrites in CA3 pyramidal neurons (Watanabe et al., 1992), which can be prevented by inhibiting CORT secretion, blocking NMDA receptors (Magariños & McEwen, 1995) or eliminating corticotropin releasing factor receptor 1 (CRFR1) in the forebrain (Wang et al., 2011). Similar dendritic atrophy is also observed following chronic social stress in tree shrews (Magariños et al., 1996). Chronic stress also causes rats to perform poorly on spatial learning tasks, which are dependent on the hippocampus, and recovery from dendritic atrophy or treatments that block atrophy rescue, reduce or prevent the cognitive impairments (Luine et al., 1994; Sousa et al., 2000; Wang et al., 2011), implying a causal relationship. In addition to its morphological effects on existing neurons, stress may also impact the generation of new neurons in the hippocampus. The hippocampal dentate gyrus represents a rare case in which, unlike most of the brain, neurogenesis persists into adulthood, at least in non-human animals (Altman, 1962; Altman & Das, 1965; Kaplan & Hinds, 1977), and possibly in humans as well (Boldrini et al., 2018; but see Sorrells et al., 2017). Adult hippocampal neurogenesis can be assessed by experiments in which newborn neurons are birth-dated by injecting 5-bromo-2′- deoxyuridine (BrdU), a thymidine analog, which is taken up by cells in S-phase, labeling them at the time of injection. The survival of these newborn cells can then be examined immunohistochemically at a later timepoint. Such studies reveal that neurogenesis is generally suppressed by stress in rats and tree shrews (Gould et al., 1997; Czéh et al, 2001; Czéh et al, 2002; Pham et al., 2003). Adding relevance to these findings, studies in humans suggest that stress-induced structural changes may be involved in the development of depression, as depression is associated with a smaller volume of the hippocampus (Sheline CHAPTER 1: General Introduction 40 et al., 1996; McKinnon et al., 2009; Cole et al., 2011), which may be attributable to stress decreasing the size of existing neurons and, perhaps, the production of new neurons – though the evidence for a causal effect of stress and depression on hippocampal volume is not conclusive. The NAc has been a more recent focus of stress research, particularly research using the RSD model to understand factors that determine the behavioural outcomes of stress. The NAc is situated in the ventral portion of the striatum and is a key hub in the brain’s reward circuitry. One groundbreaking finding, which helped to spark interest in the NAc as a critical neurobiological mediator of stress susceptibility, was that BDNF protein levels in this region are increased in susceptible, but not resilient, mice following RSD stress (Berton et al., 2006; Krishnan et al., 2007). The source of this BDNF is not the NAc itself, as local knockdown of the BDNF gene has no effect, but rather is the ventral tegmental area (VTA); knocking down BDNF in this region blocks the increase in accumbal BDNF levels and promotes resilience. Activity-dependent anterograde transport and release of BDNF occurs via dopaminergic cells in the VTA that project to the NAc, and the phasic firing rate of these cells is chronically elevated by RSD stress in susceptible mice (Krishan et al., 2007). Furthermore, it has been established that it is BDNF signaling through TrkB, rather than dopamine signaling, this is the critical factor in the NAc for determining susceptibility (Koo et al., 2016). Downstream of TrkB activation, pERK is increased in D1-positive medium spiny neurons in the NAc shell of susceptible mice.

1.2.3 Interactions Between Stress and the Zincergic System Combining what is known about stress with current knowledge about vesicular zinc allows predictions to be made regarding how ZnT3 KO mice will be affected by chronic stress – though as will be seen, the matter is not entirely straightforward. From the standpoint of zinc neurobiology, several mechanisms by which RSD stress affects the brain are of particular interest. The role of glutamatergic neurotransmission is an obvious starting point. Glutamatergic signaling in the NAc, and the relative balance between GluA1 and GluA2 subunits, has been implicated in the response to RSD. In the NAc, the transcription factor ΔFosB, which is a positive mediator of stress resilience following RSD, increases the expression of the GluA2 subunit, reducing the excitability of medium spiny neurons in response to glutamate and making mice more resilient to stress (Vialou et al., 2010). Decreasing glutamatergic transmission in the NAc by pharmacologically blocking CHAPTER 1: General Introduction 41

AMPA receptors also promoted resilience. Zinc has a predominantly inhibitory effect on AMPA receptors and glutamatergic neurotransmission (as described in section 1.1.4.1), and vesicular zinc is abundant throughout the striatum (Slomianka et al., 1990; Sørensen et al., 1995); given this, the loss of zinc might be expected to potentiate glutamatergic neurotransmission in the NAc, making mice more susceptible to stress. The relative strength of the glutamatergic connections between various forebrain structures and the NAc also shapes the response to RSD. In particular, the prefrontal cortex and ventral hippocampus play key roles, with decreased glutamatergic transmission from the latter region promoting resilience and decreased glutamatergic transmission from the former tilting the balance toward susceptibility (Bagot et al., 2015). Though it is not known precisely which NAc afferents contain vesicular zinc, it must be some of them, because the NAc (and striatum more generally) does not contain zincergic cell bodies (Brown & Dyck, 2002), despite being abundant in vesicular zinc-containing terminals. Considering that the prefrontal cortex and hippocampus both contain zincergic neurons, both are candidates to be a source of this vesicular zinc. In either case (or both cases), the loss of vesicular zinc could, by disinhibiting glutamate receptors, potentiate glutamatergic neurotransmission, influencing the behavioural response to RSD. However, the direction of the effect is difficult to predict without knowing which of the pathways contains vesicular zinc. Adult hippocampal neurogenesis also provides a link between vesicular zinc and resilience to stress. Shortly after RSD stress, mice exhibit reduced proliferation of cells in the dentate gyrus of the hippocampus, consistent with the pattern that neurogenesis is suppressed by stress, as described above. There appear to be important exceptions to this rule, however, because when newborn cells are labeled with BrdU 1-day after the final episode of RSD, the 4-week survival of these cells is dramatically increased in stressed mice relative to non-stressed controls (Lagace et al., 2010). Moreover – contrary to what one might expect, given the link between antidepressant treatment and increased neurogenesis (Malberg et al., 2000) – a greater number of surviving cells is associated with behavioural susceptibility to RSD, and ablating neurogenesis promotes stress resilience (Lagace et al., 2010). In mice that lack vesicular zinc, there is a failure to upregulate neurogenesis in response to environmental enrichment, fluoxetine treatment, and hypoglycemia (Chrusch, 2015; Boon, 2016; Suh et al., 2009) If the survival of adult-born neurons is similarly unaffected by RSD in ZnT3 KO mice, then this might confer a pro-resilience phenotype. A third link between vesicular zinc and the outcome of stress is BDNF signaling. As CHAPTER 1: General Introduction 42 described in the preceding section, increased BDNF signaling in the NAc is associated with greater susceptibility to stress (Berton et al., 2006; Krishnan et al., 2007; Koo et al., 2016). And as described in section 1.1.4.4, there are several routes through which zinc can interact with BDNF and TrkB. If zinc transactivates TrkB (Huang et al., 2008; but see Helgager et al., 2014), then vesicular zinc signaling in the NAc should promote susceptibility to stress, and elimination of vesicular zinc should have the opposite effect. On the other hand, a lack of vesicular zinc appears to render mice incapable of upregulating BDNF in the hippocampus, at least in response to certain treatments (Chrusch, 2015; Boon, 2016). The mechanism behind this is unknown, but one possibility is that vesicular zinc release is required to activate extracellular enzymes that cleave proBDNF into its mature form (J. Hwang et al., 2005; I. Hwang et al. 2011). If this is the case, then elimination of vesicular zinc might be expected to prevent the increase in accumbal BDNF signaling following RSD stress, thereby promoting resilience.

1.3 SUMMARY OF EXPERIMENTS The experiments described in the present thesis were designed to investigate the outcomes of chronic stress in ZnT3 KO mice, as part of the broader project of understanding the role of vesicular zinc in neural function and, specifically, plasticity. Chapter 2 presents the behavioural outcomes, on a battery of tests designed to assess social, emotional, and cognitive behaviour, of ZnT3 KO mice subjected to RSD stress. The effects of stress on some neuroanatomical parameters, including microglial activation, hippocampal cell proliferation, and the volume of several brain regions, are also included. The results of Chapter 3 serve as a replication of the primary finding from Chapter 2 – that ZnT3 KO mice exhibit decreased susceptibility to social avoidance following RSD stress – and extend the initial results through an examination of how chronic fluoxetine treatment following RSD stress affects hippocampal neurogenesis and social and emotional behaviour. In Chapter 4, the effects of genetic elimination of ZnT3 and vesicular zinc on levels of BDNF and TrkB in the neocortex and hippocampus are characterized, as well as how these effects are modulated by differences in age and sex of the mice. The effects of RSD stress on BDNF and TrkB levels in the NAc and hippocampus are also described. Finally, Chapter 5 provides a general discussion, integrating and interpreting the results and laying out directions for future research. CHAPTER 1: General Introduction 43

1.4 STATEMENT OF CONTRIBUTION Section 1.1 of this chapter includes text from a review article by McAllister and Dyck (2017). The text was written by the present author. CHAPTER 2: Vesicular Zinc and Social Stress 44

CHAPTER TWO: ELIMINATION OF VESICULAR ZINC ALTERS THE BEHAVIOURAL AND NEUROANATOMICAL EFFECTS OF SOCIAL DEFEAT STRESS IN MICE 2.1 INTRODUCTION The physiological and psychological response to stress is often beneficial, helping an organism to meet environmental challenges it might otherwise fail to overcome. Yet the stress response can also be harmful, particularly in cases where the stressor is chronic or recurring. Stressful life events increase the likelihood of suffering from numerous illnesses, including mental disorders such as depression (Kendler et al. 1999; Kendler & Gardner, 2016) and anxiety (Kendler et al., 2003). For this reason, the effects of stress on the rodent brain have been extensively studied, with the ultimate goal of understanding the neural mechanisms by which stress contributes to mental illness in humans. One way to induce stress in rodents is by exposing them repeatedly to episodes of social defeat by a more aggressive animal of the same species. A common protocol, developed in the late 1980s by Kudryavtseva and colleagues (Kudryavtseva et al., 1991) and refined and extended by many researchers since, involves subjecting a male mouse to brief, daily episodes of defeat by a series of dominant, aggressive mice. Between defeats, the mouse and its aggressor are housed in close quarters but separated by a perforated barrier, allowing sensory contact but preventing fighting or injury. This procedure results in a syndrome of behavioural changes, aspects of which resemble human depression. These include avoidance of other mice (similar to social withdrawal), decreased preference for sucrose (similar to anhedonia), increased anxiety, a sensitized endocrine response to acute stress, and altered circadian rhythms (Berton et al., 2006; Krishnan et al., 2007). But a certain proportion are more resilient, exhibiting some of the same changes as susceptible mice but behaving in other respects more similarly to non-defeated controls (Krishnan et al., 2007). This variability in outcome reflects what is seen in humans – not all people who experience stress go on to develop mood disorders (Sheerin et al., 2018) – and provides a model with which to probe the biological factors that predispose an animal toward susceptibility or resilience. Many such mechanisms have been discovered (Han & Nestler, 2017), but one potential mechanism that has yet to be examined is vesicular zinc. Vesicular zinc refers to zinc ions that are sequestered in the synaptic vesicles of neurons (McAllister & Dyck, 2017) – in the forebrain, zinc is found in a subset of glutamatergic neurons (Beaulieu et al., 1992; CHAPTER 2: Vesicular Zinc and Social Stress 45

Sindreu et al., 2003). This zinc is released in an activity-dependent manner and can modulate a plethora of targets, including glutamate receptors (Paoletti et al., 2009; Vergnano et al., 2014; Anderson et al., 2015; Kalappa et al., 2015). Notably, vesicular zinc storage is the responsibility of a single zinc transporting protein called ZnT3 (Palmiter et al., 1996; Wenzel et al., 1997), encoded by the SLC30A3 gene. When this transporter is eliminated, as in the ZnT3 knockout (KO) mouse, vesicular zinc can no longer be detected (Cole et al., 1999). These mutant mice thus provide a useful tool with which to study the function of vesicular zinc in the CNS. Despite the prevalence of vesicular zinc in the forebrain, mice that lack ZnT3 and vesicular zinc do not show a strong behavioural phenotype (Cole et al., 2001; Thackray et al., 2017). However, mounting evidence indicates that these mice are subtly abnormal. They perform normally when tested using a standard fear conditioning protocol but show deficient learning in a “weaker” training paradigm (Martel et al., 2010). They can perform an object recognition task when the interval between training and testing is short (Wu & Dyck, 2018), but not when it is extended to longer times (Martel et al., 2011). And ZnT3 KO mice can discriminate between textures when the difference is pronounced, but they lack the ability to detect fine textural differences (Wu & Dyck, 2018). This emerging pattern of behavioural deficits under challenging or complex conditions raises the question of how ZnT3 KO mice might respond to the challenge of chronic stress. There is reason to believe that vesicular zinc signaling could modulate stress outcomes. It has not been confirmed whether co-release of zinc occurs in the glutamatergic pathways that modulate the behavioural response to social defeat stress – e.g., ventral hippocampus to nucleus accumbens (NAc), prefrontal cortex (PFC) to NAc, and PFC to amygdala (Kumar et al., 2014; Bagot et al., 2015). But vesicular zinc-containing axon terminals are abundant in these brain regions (Pérez-Clausell & Danscher, 1985; Frederickson et al., 1992), and zinc-containing neurons are known to form reciprocal connections between PFC and amygdala (Christensen & Frederickson, 1998; Cunningham et al., 2007). Given this, we sought to characterize how ZnT3 KO mice respond to repeated social defeat (RSD), to understand whether vesicular zinc influences the outcomes of chronic stress. CHAPTER 2: Vesicular Zinc and Social Stress 46

2.2 METHOD 2.2.1 Animals All protocols were approved by the Life and Environmental Sciences Animal Care Committee at the University of Calgary and followed the guidelines for the ethical use of animals provided by the Canadian Council on Animal Care. Mice were housed in temperature- and humidity-controlled rooms maintained on a 12:12 light/dark cycle (lights on during the day). Food and water were provided ad libitum except where otherwise noted. WT and ZnT3 KO mice, on a mixed C57BL/6×129/Sv genetic background, were bred from heterozygous pairs. Offspring were housed with both parents until P21, at which point they were weaned and housed in standard cages (28 × 17.5 × 12 cm with bedding, nesting material, and one enrichment object) in groups of 2-5 same-sex littermates. CD-1 mice used for the RSD procedure were retired breeders, 4-12 months old, acquired from the University of Calgary Transgenic Services Facility or from Charles River. CD-1 mice were single- housed in standard cages except as described below.

2.2.2 Experimental Design For a diagram depicting the experimental design, see Figure 2.1. At 8-10 weeks of age, WT and ZnT3 KO mice were assigned either to the stress or control condition. The stress condition consisted of 10 days of RSD (day 1 to 10), followed by isolated housing for the remainder of the experiment. The control mice remained in standard housing (described in the above section) throughout the experiment and were briefly handled daily from day 1 to day 10. Additional control groups of WT and ZnT3 KO mice that were housed under the same conditions as the stress groups (i.e., in pairs separated by a partition for 10 days, followed by isolated housing) were also generated and subjected to a subset of the behavioural tests (see Appendix A). One day post-RSD (day 11), all mice were subjected to social interaction testing (WT- control: n = 20; WT-stress: n = 22; KO-control: n = 22; KO-stress: n = 22). Following this test, one cohort of mice was killed, and their brains were extracted and processed for immunofluorescence analysis (n = 6 per group). The remaining mice were subjected to further behavioural testing (WT-control: n = 14; WT-stress: n = 16; KO-control: n = 16; KO- stress: n = 16). On day 25 (15 days post-stress), following the completion of behavioural testing, a subset of these mice was killed, and brains were extracted and processed for magnetic resonance imaging (MRI) volumetric analysis (WT-control: n = 8; WT-stress: n = CHAPTER 2: Vesicular Zinc and Social Stress 47

9; KO-control: n = 9; KO-stress: n = 10).

2.2.3 Repeated Social Defeat The RSD procedure was adapted from Golden et al. (2011). Mice were subjected to daily episodes of defeat over 10 days. For each defeat, the mouse was transferred to a novel CD-1 mouse's cage for a period of 5 min. During this time, the CD-1 resident would reliably attack the smaller intruder, whereas the intruder would attempt to escape or exhibit submissive postures. After three attacks (with an attack defined as an uninterrupted episode of physical interaction, almost always resulting in vocalizations from the intruder) the intruder was placed in a cylindrical mesh enclosure (8.5 cm ⌀) for the remainder of the 5 min, allowing the mice to interact in close proximity but restricting further fighting or injury1. Following this, the intruder was housed with the CD-1 resident that had just defeated it, but with the two mice separated by a perforated acrylic partition that divided the large cage (24 × 45.5 × 15 cm) lengthwise into two compartments, allowing for visual, auditory, and olfactory contact between the mice, but limiting physical interaction. Prior to each defeat session, the intruder mice were rotated between cages, in order to prevent them from habituating to a particular CD-1 resident. After the final defeat, the mice were returned to single-housing in standard cages. To ensure that the CD-1 residents would reliably engage with and defeat the intruders, CD-1 mice were screened for aggressiveness prior to the experiments. The screening procedure consisted of three trials (one per day for 3 days) in which an intruder was introduced into the cage of the CD-1 resident for 3 min. (The intruders used for screening were C57BL/6 mice and ZnT3 heterozygotes, 9-20 weeks old, retired from previous experiments or surplus from our breeding colony.) Only CD-1 mice that attacked the intruder within 30 s on two or more consecutive trials were used for the experiment. To further promote territoriality and aggression, the CD-1 mice were housed in large cages for at least 24 h prior to the introduction of the first intruder, and remained in the same cages throughout the 10-day procedure.

1 In pilot experiments the mice were allowed unrestricted contact over the entire 5 min defeat session, consistent with Golden et al. (2011). However we found that this resulted in excessive wounding of the mice, which – aside from raising concerns about animal welfare – too frequently necessitated the termination of the social defeat procedure prior to the 10th day. Therefore the procedure was altered as described above. After limiting the contact to three attacks wounding was greatly reduced, and no further mice had to be removed prematurely from the experiment. CHAPTER 2: Vesicular Zinc and Social Stress 48

2.2.4 Behavioural Assessment All testing was conducted during the light phase of the mice’s light/dark cycle, and the mice were allowed to habituate to the behavioural testing room for at least 30 min prior to the start of each test. The behavioural tests were conducted in the order presented below.

2.2.4.1 Social interaction The procedure for the social interaction test was adapted from Golden et al. (2011). The test was conducted under dim red light. The apparatus for the test was an open field (40 × 40 cm) constructed of white corrugated plastic. The test consisted of three 150 s phases, each separated by 60 s. For the first phase, a cylindrical mesh enclosure (10 cm ⌀, with a weighted glass flask fixed to the roof to prevent mice from climbing on top of the enclosure or pushing and moving the enclosure) was placed against a wall of the field; the mouse was then placed along the center of the opposing wall and allowed to explore freely. The second phase was the same, but with a novel, age-matched mouse of the same strain (novel conspecific) placed inside the enclosure. For the third phase, the conspecific was replaced by a novel, aggressive CD-1 mouse. In between trials, the mouse being tested was returned to its home cage. A different but identical mesh enclosure was used for the first phase versus the second and third phases, in order to reduce odours on the enclosure that might influence the exploratory behaviour of the mouse. Between testing each mouse, the enclosures and the field were cleaned with Virkon; the field was also cleaned of urine and feces between each phase. The test was recorded using a digital video camera with night- vision capability (Sony HDR-SR8), and scoring was automated using tracking software (ANY-maze, version 4.73). The following parameters were scored: “interaction time” (i.e., time spent in the interaction zone, defined as a 26 × 16 cm rectangle around the mesh enclosure); “corner time” (i.e., time spent in either of the two corners of the field opposing the enclosure, with each corner encompassing a 9 × 9 cm area); total distance traveled; and “immobility time” (with immobility defined as a period longer than 2 s during which the mouse did not move, with detection sensitivity set at 90%). Social interaction ratios were calculated by dividing interaction time with the CD-1 mouse during the third phase by interaction time with the empty enclosure during the first phase.

2.2.4.2 Elevated plus-maze The elevated plus-maze (EPM) test exploits the natural propensity of mice to explore CHAPTER 2: Vesicular Zinc and Social Stress 49

novel areas, as well as their tendency to prefer dark, enclosed spaces to brightly lit, exposed ones. This creates an approach-avoid conflict that allows for the assessment of anxiety, as more anxious mice should tend to spend less time exploring the exposed areas of the maze. The apparatus was a plus-shaped wooden structure, painted white, with two opposing open arms (5 × 28 cm), two opposing closed arms (5 × 28 cm, 12.5 cm high walls), and a central area where the arms intersected (5 × 6 cm). The entire structure was elevated 30 cm above the ground and was illuminated by very dim light (~3 lux). Mice were tested individually for a period of 5 min, starting in the center of the maze facing toward one of the closed arms. Between trials, the maze was cleaned with 70% ethanol. Activity was video recorded and automatically scored using ANY-maze. The following parameters were scored: open arm time (i.e., time spent on either of the open arms), center area time, and total distance traveled. Arm entries were defined as 95% or more of the mouse’s area within an arm.

2.2.4.3 Novelty-suppressed feeding Like the EPM, novelty-suppressed feeding (NSF) is a test of anxiety-like behaviour that makes use of two conflicting motivations, those being the drive to feed when hungry and the mouse’s natural inclination to avoid feeding in a novel, brightly lit, open environment. The latency to feed in this scenario provides an indicator of anxiety, with longer latencies assumed to reflect greater anxiety. The protocol for this test was adapted from Samuels and Hen (2011). Mice were food deprived for 16 h prior to the test. The test was conducted in an open field under bright lighting (~800 lux). The floor of the field was covered with wood-chip bedding. (The field was not entirely novel to the mice, as it was the same apparatus used for the social interaction test, but it did include some novel features, such as the bedding on the floor, the room in which the field was situated, and the lighting of the room.) A food pellet (standard mouse chow) was fixed to a small platform in the center of the field with an elastic band, preventing the mouse from moving the pellet. The latency to begin feeding was recorded, up to a maximum time of 10 min, at which point the test was terminated and the mouse was assigned a latency time of 600 s. The mouse was then returned to its home-cage (with its cage-mates temporarily removed), and transported immediately to an adjacent, dimly lit (~3 lux) room. A pre-weighed food pellet was placed in the food hopper, and the latency to begin feeding in the home cage was recorded; a maximum latency score of 180 s was given if the latency to feed exceeded that length. After the mouse began feeding, it was allowed 5 min to eat, after which the pellet was removed CHAPTER 2: Vesicular Zinc and Social Stress 50

and weighed to calculate food consumption. Body weight was also recorded both prior to food deprivation and after the test.

2.2.4.4 Spatial Y-maze The spatial Y-maze protocol was adapted from the one described for rats by Conrad et al. (2003). The apparatus was a three-armed wooden structure, painted black. Each arm measured 10.5 by 48 cm, with 15 cm high walls. The structure was elevated off the ground, and there were numerous objects and visual cues surrounding the maze to be used for spatial orientation. The floor of the maze was covered in bedding, which was mixed between trials to prevent odours being used as non-spatial cues. For the training phase of the test, mice were placed at the end of the “starting arm” (which was kept constant for all mice) and allowed to explore for 15 min, with one arm blocked off by a partition. The mice were returned to the maze 3 h later for a 5 min test phase. The mice were started in the same arm as in the first phase, but this time the partition was removed, providing a novel arm that had not previously been explored. The test phase was video recorded for scoring. Because mice have a natural propensity to explore novel areas, it was assumed that mice with intact spatial memory would make a greater percentage of their total entries into the novel arm in comparison to the “other arm” (i.e., the third arm that was not the starting arm or novel arm). The total number of arm entries was also used as an indicator of locomotor activity in the test.

2.2.4.5 Conditioned fear A 3-day conditioned fear test was conducted to assess both cued and contextual fear memory. The former involves learning an association between an initially neutral stimulus (e.g., a tone) and a noxious stimulus (e.g., an electric shock administered through the floor of the chamber), whereas the latter involves forming an association between a noxious stimulus and the context in which it is encountered. The test was performed using a conditioning box (Hamilton-Kinder LM1000-B). Between mice, the inside of the box was wiped with a disinfectant solution. On day 1, after a 2 min period of acclimatization to the box, a 20 s tone was presented. A single foot shock (2 s, 0.3 mA) was administered coinciding with the final 2 s of the tone. After an additional 30 s, the mouse was removed from the box. On day 2, the apparatus was altered to provide a novel context (black walls were replaced with white walls, a solid plastic insert was placed over the metal grid floor, CHAPTER 2: Vesicular Zinc and Social Stress 51

coconut scent was added to the chamber, and Virkon was used in place of 70% ethanol as a disinfectant). Mice were allowed to explore the box for 1 min, after which the tone was presented for 2 min to assess the fear response to the auditory cue (i.e., cued fear memory). On day 3, the apparatus was reverted to its original conditions, and the mice tested for 3 min to assess their fear response to the context (i.e., contextual fear memory). The activity of the mice was video-recorded for analysis of freezing (defined as total immobility, excluding minor movements associated with breathing), which is an indicator of fear in mice. Quantification of freezing was automated using ANY-maze (minimum freezing duration = 500 ms).

2.2.5 Anatomical Analyses 2.2.5.1 Immunofluorescence labeling Mice were deeply anaesthetized with an overdose of sodium pentobarbital, and transcardially perfused with phosphate buffered saline (PBS) until the blood was cleared, followed by perfusion with 4% paraformaldehyde (PFA) in PBS. Brains were extracted and post-fixed overnight in 4% PFA in PBS at 4 °C. The spleen and adrenal glands were also extracted and weighed. After post-fixing, the brains were transferred to a sucrose solution (30% sucrose, 0.02% sodium azide in PBS) and stored at 4 °C. Brains were cut coronally into six series of 40 µm sections using a freezing, sliding microtome (American Optical, Model #860). One series was labeled for the cellular proliferation marker Ki67. A second series was labeled for the microglia marker ionized calcium-binding adapter molecule 1 (Iba1). For immunofluorescence labeling, the procedure was as follows: 3 × 10 min wash in PBS; blocking for 1 h in PBS containing 0.3% Triton X-100 (PBSx) and 4% normal goat serum (NGS); incubation overnight at room temperature in PBSx containing 2% NGS and the primary antibody (1:1000 rabbit anti-Iba1, Wako #019-19741; or 1:2000 rabbit anti- Ki67, Leica NCL-Ki67p); 3 × 10 min wash in PBSx; incubation overnight at room temperature in PBSx containing the secondary antibody (1:1000, biotin-conjugated goat anti-rabbit, Jackson ImmunoResearch 111-065-144); 3 × 10 min wash in PBSx; incubation for 1 h in PBSx containing the tertiary antibody (1:1000, Alexa-Fluor 594-conjugated streptavidin, Jackson ImmunoResearch 016-580-084) with 4′,6-diamidino-2-phenylindole (DAPI; 1:1000) added for the final 15 min; 3 × 10 min wash in PBS. Sections were mounted on gelatin-coated slides, coverslipped with fluorescence mounting medium, and stored at 4 CHAPTER 2: Vesicular Zinc and Social Stress 52

°C.

2.2.5.2 Hippocampal cell counting Ki67+ cells were counted using an epi-fluorescence microscope (Zeiss Axioskop 2) with a 63×/1.40 objective. Cells were counted in the granule cell layer and the subgranular zone (SGZ; defined as three cell-widths from the hilar edge of the granule cell layer) in all sections containing the hippocampal dentate gyrus. The counts were then multiplied by six (since the brains were sectioned into six series) to estimate the total number of Ki67+ cells.

2.2.5.3 Microglial analysis Microglia were assessed in the PFC (prelimbic region), basolateral amygdala (BLA), dorsal hippocampus (dHPC; dentate gyrus region), and ventral hippocampus (vHPC; CA3 region). Images were captured bilaterally from three sections containing each region of interest (ROI), resulting in a total of six images per ROI. Changes in microglial morphology, such as increased soma size and changes in process length or number, occur when microglia become “activated.” A thresholding method was used to provide a gross assessment of such changes. This method involves binarizing an image into areas of positive and negative labeling (Beynon & Walker, 2012). Though the method does not give specific information about changes in morphology, changes in the percentage of positive labeling can indicate that changes in morphology have occurred. Images for analysis were generated by capturing z-stacks throughout the depth of the tissue section using a confocal microscope (Nikon C1si) with a 20×/0.75 objective. The “volume render” function of EZ-C1 software (Nikon) was used to collapse the stack into a single, focused image. The images were processed with the “subtract background” and “sharpen” functions of ImageJ (http://rsb.info.nih.gov/ij/), and the “threshold” function was used to binarize the images. The optimal threshold level was determined across several sections from different brains, and then applied uniformly to all images. The percentage of area containing Iba1+ labeling within the ROI was then measured. Measurements from the six images were averaged to give a single score for each ROI from each brain. To provide a finer analysis of microglial morphology, soma area and arborization area were measured directly. Using the images described above, ten cells per brain (five from each hemisphere) were analyzed from each ROI. To avoid selection bias, the five microglia with their somata closest to the center of the image were selected. Arborization CHAPTER 2: Vesicular Zinc and Social Stress 53

area was determined by using the “polygon selections” tool in ImageJ to connect the most distal point of each process (Alboni et al., 2016). The area of the resulting polygon was then measured. Soma area was determined by using the “freehand selections” tool to trace the area of the soma. Measurements were then averaged across the 10 cells to give average scores for each ROI from each brain. Finally, microglial density was also quantified. Images were captured using a microscope (Zeiss Axioskop 2) with a 10×/0.30 objective. The number of microglia within the ROI was counted using the “multi-point” tool in ImageJ, then divided by the area of the ROI. Measurements from the six images were averaged to give a single score for each ROI from each brain.

2.2.5.4 MRI acquisition and analysis Mice were perfused as described above. Brains were extracted, weighed, and stored in 4% PFA in PBS at 4 °C. MRI acquisition was conducted as previously described (Wright et al., 2017, 2018). Brains were washed overnight in PBS and embedded in 2-3% agar for ex vivo MRI using a 4.7 T Bruker MRI (Bruker, Ettlingen, Germany Biospin, USA). A 3D multiple gradient echo sequence was acquired using a cryogenically-cooled RF coil and the following imaging parameters: repetition time = 110 ms; echo times = 4, 8, 12…80 ms; matrix = 176 × 128 × 70; field of view = 17.6 × 12.8 × 7 mm3; and voxel size = 0.1 mm3. Echoes were averaged offline for ROI delineation. For analysis, six a priori ROIs – prefrontal cortex, hippocampus, corpus callosum, parietal cortex, lateral ventricles, and amygdala – were traced per hemisphere using ITK- SNAP (www.itksnap.org) as previously described (Shultz et al., 2013; Wright et al., 2017, 2018). “Prefrontal cortex” was defined as all cortex in the 12 slices anterior to the forceps minor. “Hippocampus” started at the anterior tip of the CA3 field and continued for 15 slices (encompassing most of dorsal, but not ventral, hippocampus). Analysis of the remaining structures was limited to these 15 slices. “Parietal cortex” was defined as all cortex dorsal and lateral to the corpus callosum, with the rhinal fissure serving as the ventral boundary. “Amygdala” was defined as everything ventral to the rhinal fissure and lateral to the external capsule and striatum (thus including structures surrounding the amygdala such as the entorhinal and piriform cortex). ROI volumes were determined using Fslutils, a component of FMRIB’s Software Library (FSL, www.fmrib.ox.ac.uk/fsl). MRI analysis was conducted by a researcher blind to the experimental conditions. CHAPTER 2: Vesicular Zinc and Social Stress 54

2.2.6 Statistical Analysis Statistical analyses were conducted using IBM SPSS Statistics (version 24). Unless otherwise stated, comparisons were conducted by two-way analysis of variance (ANOVA) with genotype (WT vs. ZnT3 KO) and stress (control vs. stress) as factors. Significant interactions were followed-up with Bonferroni-corrected simple effects tests using the pooled error term, unless equality of variances could not be assumed (Levene’s test: p < .05), in which case non-pooled error terms were used. All ANOVA results are reported in Table 2.4. Comparisons of control vs. resilient vs. susceptible mice were done by one-way ANOVA, with significant effects followed-up by Fisher’s Least Significant Difference (LSD) test unless homogeneity of variances could not be assumed (p < .05 for Levene’s test), in which case pairwise comparisons were conducted using t-tests with equal variances assumed or not assumed, as appropriate. Means are presented ± standard deviation.

2.3 RESULTS 2.3.1 Behavioural Assessment 2.3.1.1 Social interaction Time spent in the interaction zone and in the corners of the field was first examined for the three phases of the test (Figure 2.2A). In the first phase (empty cage; Table 2.1), stress decreased interaction time by 15% [F(1,82) = 4.92, p = .029], with no difference between genotypes. For corner time, there was no effect of stress or genotype. In the second phase (novel conspecific; Figure 2.2B), stress had differing effects on interaction time depending on the genotype of the mice [stress × genotype interaction: F(1,82) = 6.53, p = .012]. Stress decreased interaction time by 59% in the WT mice [F(1,82) = 26.53, p < .001, Bonferroni-corrected], but did not significantly affect the ZnT3 KO mice [F(1,82) = 2.62, p = .109]. Similarly, there was a significant interaction for corner time [F(1,82) = 6.36, p = .014]. Stress more than doubled the time spent in the corners by the WT mice [F(1,82) = 28.16, p < .001, Bonferroni-corrected], but did not significantly affect the ZnT3 KO mice [F(1,82) = 3.34, p = .071]. Together, these results indicate that stress caused WT mice to become socially avoidant of a novel conspecific, while ZnT3 KO mice were unaffected. For the third phase (CD-1 aggressor; Figure 2.2C), stress decreased interaction time by 60% [F(1,82) = 22.53, p < .001], and increased corner time by 118% [F(1,82) = 37.18, p < .001]. Interaction time with the CD-1 mouse did not differ between genotypes, but there CHAPTER 2: Vesicular Zinc and Social Stress 55

was a difference in corner time [F(1,82) = 4.12, p = .046], with the WT mice spending more time in the corners than the ZnT3 KO mice. To summarize, while only WT mice were avoidant of a novel conspecific following stress, stress caused both WT and ZnT3 KO mice to avoid a novel CD-1 mouse. There were no differences in the total distance traveled in the field during phase 1 of the test (Table 2.1), indicating that baseline differences in locomotion did not affect the results. During phase 2, stress more than doubled the amount of time spent immobile [F(1,82) = 38.72, p < .001; Table 2.1], with no difference between genotypes. The same was true for phase 3 [F(1,82) = 19.53, p < .001; Table 2.1]. The social interaction ratio with a CD-1 mouse is the standard measure used to define susceptibility to stress in the RSD model, so we also compared the groups on this measure. One mouse was excluded from this analysis (and from subsequent analyses that involve characterizing mice as resilient or susceptible) because it spent no time in the interaction zone during phase 1, which prevented us from calculating a ratio. Stress decreased the interaction ratio by 47% [F(1,81) = 8.32, p = .005; Figure 2.2D], with no difference between genotypes. Using the social interaction ratios, mice that underwent the RSD procedure were characterized as either susceptible or resilient to stress. Generally, the standard used to define susceptibility is an interaction ratio of less than 1, which indicates less time spent in the interaction zone when a social target is present than when no social target is present. After viewing the data, we noted that our control mice tended to have lower interaction ratios than are usually reported (e.g., by Golden et al., 2011), and few mice would be classified as resilient by the standard criteria. This could be due to the mixed C57BL/6×129/Sv genetic background of the mice. In comparison to the C57BL/6 mice used in most experiments, there is evidence that 129/SvEv mice are more vulnerable to the behavioural effects of RSD stress (Dadomo et al., 2011). We therefore decided to use a criterion of less than 0.5 to define susceptibility. By this definition, 11 mice (5 WT, 6 ZnT3 KO) were characterized as resilient and 32 were susceptible (26% of mice were resilient). Comparisons between these groups are displayed in Table 2.2. Unsurprisingly – considering that resilience was defined based on social interaction – the resilient mice, in comparison to the susceptible mice, spent more time interacting with a conspecific mouse (p = .046) and CD-1 mouse (p < .001), less time in the corners during exposure to the CD-1 mouse (p < .001), and less time immobile in the presence of a conspecific (p = .009) or a CD-1 mouse (p CHAPTER 2: Vesicular Zinc and Social Stress 56

< .001).

2.3.1.2 Elevated plus-maze Stress increased anxiety-like behaviour [F(1,58) = 8.74, p = .004; Figure 2.3A], as indicated by the stressed mice spending less time – indeed, almost no time at all – on the open arms. There was no difference between genotypes. Stress also decreased the amount of time spent in the center of the maze by 40% [F(1,58) = 10.99, p = .002; Figure 2.3B]. The ZnT3 KO mice tended to spend less time in the center of the maze than did the WT mice, though this effect was not significant [F(1,58) = 3.21, p = .078]. There was no effect of stress or difference between genotypes on total distance traveled (WT-control: 5.6 ± 1.5; WT- stress: 5.8 ± 2.0; KO-control: 6.2 ± 1.7; KO-stress: 5.2 ± 1.1), indicating that differences in locomotor activity did not influence the results of this test. Of the mice that were subjected to the behavioural test battery beyond the social interaction test, 9 mice (5 WT, 4 ZnT3 KO) were characterized as resilient and 22 were susceptible (Table 2.2). There were significant differences between groups in the amount of time spent in the open arms [F(2,58) = 3.97, p = .024] and in the amount of time spent in the center of the maze [F(2,58) = 5.53, p = .006]. Both susceptible (p = .019) and resilient mice (p = .002) spent significantly less time on the open arms than did control mice, with no significant difference from each other (p = .076). Susceptible mice also spent less time in the center than did control mice (p = .001), whereas resilient mice did not differ from the controls (p = .296) or susceptible mice (p = .131).

2.3.1.3 Novelty-suppressed feeding Providing further evidence of anxiety-like behaviour, stress increased the latency to begin feeding in the novel open field [F(1,58) = 40.43, p < .001; Figure 2.3C], with the stressed mice taking more than twice as long to begin feeding than controls. There was no difference between genotypes. In the home cage, there was no significant difference in the latency to feed between the stressed and control mice [F(1,58) = 0.56, p = .459; Table 2.1], suggesting that the difference in the novel field was indeed due to increased sensitivity to the anxiogenic environment, and not due to more general changes in feeding behaviour. Further supporting this, there was no effect of stress on the amount of food consumed in the home cage [F(1,58) = 1.06, p = .307; Table 2.1]. The WT mice did tend to consume more than the ZnT3 KO mice, though this difference was not significant [F(1,58) = 3.68, p = .060]. CHAPTER 2: Vesicular Zinc and Social Stress 57

Interestingly, there was a significant effect of stress on the change in body weight over the 16 h food restriction period [F(1,58) = 42.65, p < .001], with the stressed mice losing 33% more weight than the control mice. There was no difference between genotypes. This pattern of results was the same when weight change was calculated as a percentage of initial body weight (results not shown). In comparing susceptible, resilient and control mice (Table 2.2), there were significant differences in the latency to feed in the novel field [F(2,58) = 19.89, p < .001] and the amount of weight lost over the food deprivation period [F(2,58) = 21.46, p < .001]. Both susceptible (p < .001) and resilient mice (p < .001) had significantly longer latencies to feed than controls but did not differ from each other (p = .623). Similarly, both susceptible (p < .001) and resilient mice (p < .001) lost more weight than controls but did not differ from each other (p = .843).

2.3.1.4 Spatial Y-maze Stress did not alter the percentage of total arm entries made into the novel arm or the amount of time spent in the novel arm, nor was there an effect of genotype on either measure (Table 2.1). Thus, stress did not appear to have a deleterious effect on spatial memory in either genotype. There was a significant interaction between the effects of stress and genotype on the total number of arm entries [F(1,58) = 4.18, p = .045; Table 2.1], with stress tending to increase arm entries in the WT mice [F(1,58) = 1.39, p = .244] and decrease arm entries in the ZnT3 KO mice [F(1,58) = 2.97, p = .090], though neither effect was significant. Wilcoxon signed-rank tests were used to compare the percentage of entries into the novel arm to the percentage of entries into the “other” arm within each group. All four groups made a significantly greater percentage of arm entries into the novel arm than the “other” arm, suggesting that all groups could remember which arm had been inaccessible during the training phase 3 hours prior (WT-control: 40.3 ± 6.2 vs. 31.1 ± 5.2, Z = 2.50, p = .013; WT-stress: 42.0 ± 4.4 vs. 28.7 ± 5.9, Z = 3.30, p = .001; KO-control: 40.2 ± 6.3 vs. 31.7 ± 4.5, Z = 2.66, p = .008; KO-stress: 40.5 ± 5.0 vs. 32.9 ± 5.8, Z = 2.51, p = .012). There were no significant differences between susceptible, resilient, and control mice in this test (Table 2.2). CHAPTER 2: Vesicular Zinc and Social Stress 58

2.3.1.5. Conditioned fear There were no significant effects of genotype or stress on the percentage of time spent freezing before, during, or after the presentation of the tone/shock on day 1 of the test (p > .15 for all comparisons; data not shown), indicating that there were no baseline differences in fear or freezing behaviour, and no differences in the initial response to the tone/shock. Cued fear memory was assessed on day 2 in a novel context. There was no effect of stress or genotype on freezing time before cue presentation (Figure 2.4A), indicating no baseline differences in the fear response to the novel context. Stress had differing effects on cued fear memory depending on the genotype of the mice [stress × genotype interaction: F(1,58) = 5.89, p = .018; Figure 2.4A]. In the ZnT3 KO mice, stress enhanced fear memory, as indicated by a 42% increase in freezing time during the cue presentation [F(1,58) = 10.01, p = .004, Bonferroni-corrected]. In the WT mice, stress had no significant effect [F(1,58) = 0.10, p = .752]. However, it also appeared that control ZnT3 KO mice showed weaker fear memory than control WT mice, as was previously reported by Martel et al. (2010). Therefore, we also directly compared the control groups, and confirmed that ZnT3 KO mice froze less than WT mice during cue presentation [t-test: t(28) = 2.16, p = .040], indicating weaker cued fear memory. Contextual fear memory was assessed on day 3. There was no significant effect of stress on contextual fear memory [F(1,58) = 3.02, p = .087; Figure 2.4B], though stress did tend to increase freezing. There was, however, a significant difference between the genotypes [F(1,58) = 6.65, p = .012], with the ZnT3 KO mice showing weaker contextual fear memory, freezing 24% less than the WT mice. Finally, we examined whether cued or contextual fear memory differed between susceptible, resilient, and control mice (Table 2.2). There were no significant differences either during the cue presentation [F(2,58) = 2.04, p = .139] or during context re-exposure [F(2,58) = 1.73, p = .187].

2.3.2 Anatomical Analyses 2.3.2.1 Body and organ weights To assess whether stress affected body weight, we calculated the change in weight from 1-day pre-RSD to 1 day-post RSD. Five mice (2 WT-stress, 2 KO-control, 1 KO-stress) were excluded from this analysis because body weight data were missing from one of the CHAPTER 2: Vesicular Zinc and Social Stress 59

two time points. Neither stress nor genotype influenced the change in body weight. On average, mice gained a small amount of weight over the time period (WT-control: 0.4 ± 1.0 g; WT-stress: 0.2 ± 1.5 g; KO-control: 0.4 ± 1.0 g; KO-stress: 0.3 ± 1.0 g). There was no difference in weight change between control, susceptible, and resilient mice (Table 2). Spleen and adrenal weights were also examined in the mice killed 1 day following stress. Stress significantly increased spleen weight [F(1,20) = 5.76, p = .026], but there was no difference between genotypes (WT-control: 74.1 ± 6.8 mg; WT-stress: 92.6 ± 16.5 mg; KO- control: 81.1 ± 20.7 mg; KO-stress: 108.1 ± 37.6 mg). When spleen weights were analyzed as a percentage of total body weight, the same pattern of findings was observed (data not shown). Stress also significantly increased adrenal gland weight [F(1,20) = 15.86, p < .001], with no difference between genotypes (WT-control: 4.1 ± 0.7 mg; WT-stress: 4.8 ± 0.8 mg; KO-control: 3.2 ± 0.5 mg; KO-stress: 5.3 ± 1.3 mg). When adrenal gland weight was analyzed as a percentage of total body weight, however, the effect of stress was found to differ between genotypes [stress × genotype interaction: F(1,20) = 7.87, p = .011]. Stress increased adrenal weight (normalized to body weight) in the ZnT3 KO mice [KO-control: 0.13 ± 0.03%; KO-stress: 0.21 ± 0.04%; F(1,20) = 23.43, p < .001], but not in the WT mice [WT-control: 0.16 ± 0.02%; WT-stress: 0.18 ± 0.03%; F(1,20) = 0.76, p = .786]. There was no significant effect of stress or genotype on body weight in this sample (data not shown).

2.3.2.2 Hippocampal cell proliferation The number of cells in the dentate gyrus positively-labeled for the proliferation marker Ki67 was assessed in brains collected 24 h after the final episode of stress. Cell proliferation was not significantly affected by stress, nor did it differ between genotypes (Figure 2.5).

2.3.2.3 Microglial analysis Using a thresholding procedure, gross changes in microglial morphology were assessed in brains collected 1 day after the final episode of stress (Figure 2.6). In the PFC and dHPC, there was no effect of stress or difference between genotypes. Likewise, there was no significant effect of stress or difference between genotypes in the vHPC, though there was a trend toward an interaction between stress and genotype [F(1,20) = 4.30, p = .051]. Finally, in the BLA, there was no significant effect of stress [F(1,20) = 4.15, p = .055], though the stressed mice did tend to have more Iba1+ labeling than the controls. There was CHAPTER 2: Vesicular Zinc and Social Stress 60

no difference between genotypes [F(1,20) = 0.49, p = .490]. Next, finer changes in microglial morphology were assessed (Table 2.3). There was no effect of stress or genotype on the soma area or arborization area of microglia in the PFC, dHPC, vHPC, or BLA. Finally, the density of microglia was assessed (Table 2.3). There was no effect of stress or genotype on microglial density in the four regions examined. In summary, when microglial status was assessed, either by thresholding of Iba1 immunolabeling, by directly measuring morphological parameters, or by quantification of microglial density, there were no significant differences across several brain regions.

2.3.2.4 MRI volumetric analysis Volumes of several brain regions were assessed by ex vivo MRI of brains collected 15 days after the final episode of stress (Figure 2.7A). First, we verified that there was no effect of stress or genotype on body weight in the subset of mice from which brains were collected (WT-control: 26.0 ± 3.1 g; WT-stress: 26.3 ± 1.9 g; KO-control: 27.2 ± 3.1 g; KO- stress: 26.9 ± 2.9 g). For the corpus callosum, the effect of stress differed based on the genotype of the mice [stress × genotype interaction: F(1,20) = 7.66, p = .009; Figure 2.7B], with stress decreasing corpus callosum volume by 8.2% in the ZnT3 KO mice (p = .002; Bonferroni-corrected) while having no effect in the WT mice (p = .918). It also appeared that, in the control groups, ZnT3 KO mice had larger corpus callosum volumes than WT mice. This was confirmed using a post-hoc Tukey test (WT-control vs. KO-control: p < .001). For the parietal cortex, there was no significant effect of stress, but there was a difference between genotypes [F(1,20) = 7.66, p = .009; Figure 2.7C], with parietal cortex being 3.6% larger in the ZnT3 KO mice than in the WT mice. For the lateral ventricles, there was no significant effect of stress or difference between genotypes. However, the interpretation of these effects was complicated by a trend toward an interaction [F(1,20) = 3.99, p = .054; Figure 2.7D], suggesting that differing effects of stress could be obscuring a difference between genotypes. We therefore compared ventricular volume in the non- stressed controls. However, the difference was not significant (post-hoc Tukey test: WT- control vs. KO-control: p = .114). There was no effect of stress or difference between genotypes in the volume of prefrontal cortex (WT-control: 11.4 ± 0.7 mm3; WT-stress: 11.7 ± 0.6 mm3; KO-control: 11.2 ± 0.8 mm3; KO-stress: 11.8 ± 1.3 mm3), amygdala (WT-control: 9.9 ± 0.7 mm3; WT-stress: 10.2 ± 0.5 mm3; KO-control: 10.2 ± 0.6 mm3; KO-stress: 10.5 ± 0.9 mm3), or hippocampus CHAPTER 2: Vesicular Zinc and Social Stress 61

(WT-control: 4.5 ± 0.3 mm3; WT-stress: 4.5 ± 0.4 mm3; KO-control: 4.8 ± 0.5 mm3; KO-stress: 4.7 ± 0.5 mm3).

2.4 DISCUSSION Social avoidance is a well-characterized outcome of RSD stress (Berton et al., 2006; Krishnan et al., 2007; Golden et al. 2011). We observed that both WT and ZnT3 KO mice became avoidant of an aggressive CD-1 mouse following stress. Only WT mice avoided a novel same-strain conspecific, however; stressed ZnT3 KO mice did not. Given previous reports that male ZnT3 KO mice show increased (Martel et al., 2011) and decreased (Yoo et al., 2016) social interaction with a novel mouse, it is worth highlighting that we found no baseline differences in sociability; interaction time with either a conspecific or a CD-1 mouse did not differ between genotypes in non-stressed controls. This is also consistent with our previous finding that female ZnT3 KO mice exhibit normal amounts of social interaction time with a novel conspecific in the three-chambered social interaction test (Thackray et al., 2017). Given that WT mice in our study became avoidant of both CD-1 mice and same- strain conspecifics, the lack of conspecific avoidance by ZnT3 KO mice could be interpreted as a cognitive deficit, such that these mice successfully learn to avoid CD-1 mice – a response that is relatively “easy” to learn, considering these are the mice by which they were defeated – but fail to generalize the avoidance response to mice of a different strain. This interpretation fits with previous findings of mild impairments in ZnT3 KO mice, including in fear memory (Martel et al., 2010), texture discrimination (Wu & Dyck, 2018), object recognition memory (Martel et al., 2011) and spatial reversal or working memory (Cole et al., 2001; Martel et al., 2011; Sindreu et al., 2011). On the other hand, it is arguable whether the lack of conspecific avoidance should be considered a “cognitive deficit.” At least from an anthropomorphic perspective, the ideal response to social stress would be to avoid the cause of the stress while not generalizing this response into depression-like withdrawal from all social interactions. Another interpretation of our findings, then, is that the lack of conspecific avoidance indicates resilience to stress. Defining a lack of stress-induced avoidance as a desirable trait – and, conversely, avoidance as an indicator of susceptibility to depression-like effects – is a common interpretation, for a number of reasons: 1) avoidance has obvious parallels to social withdrawal, a symptom of depression; 2) avoidance is associated with a number of other CHAPTER 2: Vesicular Zinc and Social Stress 62

depression-like outcomes (Krishnan et al., 2007), including reduced sucrose preference, altered circadian function, and decreased body weight; 3) avoidance can be reversed by chronic treatment with antidepressant drugs (Tsankova et al., 2006). Because resilience is usually defined by interaction with a CD-1 mouse – rather than a conspecific – it is somewhat difficult to directly compare the “resilience” observed by ZnT3 KO mice in our study to the “resilience” reported in many other studies. One way to further address this issue would be to examine whether ZnT3 KO mice are resilient to other depression-like behaviours following RSD. Sucrose preference or circadian function would be good candidates; unfortunately, neither was assessed in the current study. We examined anxiety, but this does little to clarify the matter, because increased anxiety can occur independently from social avoidance (Krishnan et al., 2007); that is, mice can become exhibit anxiety-like behaviour without being susceptible to depression-like behaviours. Our own results support this; when we grouped mice as susceptible or resilient based on interaction with a CD-1 mouse, both groups showed increased anxiety-like behaviour relative to non-stressed controls. We did assess body weight, which provides information about susceptibility to depression-like effects, as both weight gain and weight loss are symptoms of depression. However, we did not observe an effect of stress. This is perhaps not surprising, as the effect of RSD on body weight is quite variable between studies, with reports of weight loss or attenuated weight gain (Kudryavtseva et al., 1991; Krishnan et al., 2007; Venzala et al., 2012) as well as increased weight gain (Bartolomucci et al., 2004; Dubreucq et al., 2012). Interestingly, while body weight was not affected by stress under standard feeding conditions, we did observe that stressed mice lost more weight than controls when they were deprived of food for a 16 h period, likely indicating metabolic changes induced by RSD. This is supported by the previous finding that RSD increases levels of ghrelin and the amount of food consumed without increasing body weight (Lutter et al., 2008), indicating that defeated mice require a greater caloric intake to maintain the same bodyweight as controls. We also examined how stress affects cognition in ZnT3 KO mice, using the fear conditioning paradigm. RSD had no significant effect on contextual fear memory, but the results of the cued fear memory test were more interesting. Stress enhanced cued fear memory in ZnT3 KO mice, but WT mice were not affected. Interpreting this effect is complicated, however, because the two genotypes did not start out from the same baseline; CHAPTER 2: Vesicular Zinc and Social Stress 63

under control conditions, ZnT3 KO mice showed weaker fear memory than WT mice. Thus, rather than enhancing fear memory beyond “normal” levels, RSD only brought fear memory in the ZnT3 KO mice up to the level normally exhibited by WT mice. We also observed ZnT3 KO mice to have weaker fear memory in the contextual fear test. Put together, these findings support the results of Martel et al. (2010), who found that fear memory is impaired in ZnT3 KO mice when they are trained using a single tone-shock pairing. While the cause of this impairment is not known, vesicular zinc release is required for spike-timing dependent long-term potentiation in cortico-amygdalar projections (Kodirov et al., 2006), which may be involved in fear learning. The lack of a strong effect of stress on fear memory was somewhat surprising, considering that others have observed enhanced fear memory after chronic restraint in rats (Conrad et al., 1999; Suvrathan et al., 2014) and RSD in mice (Fuertig et al., 2016; Lisboa et al., 2018). It is possible that the effect did not persist over the 10-day gap between RSD and fear memory testing in the present study. We also failed to detect an effect of RSD on memory in a spatial Y-maze test, despite the well-documented deleterious effect of chronic stress on spatial memory (Conrad et al., 1996; Conrad, 2010; Wang et al., 2011). It might be the case that a longer period of stress is required; at 10 days, the duration of stress in our experiment was relatively short. And the period of time between RSD and testing is, again, a possible factor. The hippocampal atrophy that occurs in response to 21 days of restraint stress recovers within 5-10 days (Conrad et al., 1999). It is possible that any memory deficits produced by stress in our experiment recovered over the 8 days between RSD and the Y-maze test. It seems somewhat paradoxical that, relative to WT mice, stress had less of an effect on ZnT3 KO mice in terms of their social interaction behaviour, but more of an effect when it came to fear memory. However, this is consistent with some previous findings. Following a condensed RSD procedure, in which eight defeats were conducted over 2 days, mice that did not become socially avoidant, relative to those that did, showed stronger cued fear memory, impaired fear extinction, and greater generalization of the fear response to a neutral cue (Meduri et al., 2013; Dulka et al., 2015). This suggests that the trade-off for greater resilience to stress-induced social avoidance might be greater sensitivity to stress- induced potentiation of fear memory, which has obvious parallels to post-traumatic stress disorder. On the other hand, Chou et al. (2014) reported that greater resilience to social avoidance following defeat stress is associated with stronger – not weaker – cued fear CHAPTER 2: Vesicular Zinc and Social Stress 64

memory. It is likely that the relationship between these behavioural phenotypes depends on methodological details, such as the stress protocol used. If ZnT3 KO mice are resilient to stress, one might predict that other means of lowering zinc levels in the brain would make mice more resilient, and that raising levels would have the opposite effect. This appears not to be the case. Zinc deficiency increases the endocrine response to acute stress (Watanabe et al., 2010) and increases anxiety- and depression-like behaviours in rats (Takeda et al., 2007, 2012; Whittle et al., 2009). And administering zinc as an adjunctive treatment to antidepressant drugs increases their effectiveness in humans (Siwek et al., 2009; Ranjbar et al., 2014) and decreases depression- like behaviour in mice (Ding et al., 2016). The matter is complex, however, because manipulating dietary zinc intake has much broader effects than just altering the vesicular zinc pool or even the brain zinc pool. Chronic stress can alter zinc levels in the brain (Tao et al., 2013; Dou et al., 2014), but, to our knowledge, beyond the present results it has not been shown whether brain zinc levels – vesicular or total – predict the behavioural outcomes of chronic stress. In rats, it has been shown that hippocampal total zinc levels are correlated with behavioural outcomes 1 week following acute stress, with greater zinc levels being associated with a greater behavioural response (Sela et al., 2017). However, this study did not demonstrate whether the relationship was causal. To uncover the neural mechanisms that cause the altered behavioural response to RSD in mice that lack vesicular zinc, we examined some aspects of neuroanatomy that are known to be affected by stress. The first was hippocampal neurogenesis. Chronic social stress decreases the proliferation, survival, differentiation, and maturation of adult-born cells in the hippocampal dentate gyrus, at least at certain time points (Czéh et al., 2001; Van Bokhoven et al., 2011; Chen et al., 2015; McKim et al., 2016). Further, some behavioural effects of RSD are mediated by changes in neurogenesis (Lagace et al., 2010; Lehmann et al., 2013). ZnT3 KO mice fail to show the increase in neurogenesis that is normally seen following hypoglycemia (Suh et al., 2009), and we have observed that these mice also do not show the increase in neurogenesis that normally results from enriched housing (Chrusch, 2015). We therefore speculated that the effect of stress on neurogenesis might also be abnormal in ZnT3 KO mice, and that this might account for the altered behavioural profile. Here, we only examined cell proliferation 24 h after the last episode of defeat stress, with the intention of examining longer-term cell survival in greater depth in future experiments. We found that proliferation was unaffected by stress, regardless of CHAPTER 2: Vesicular Zinc and Social Stress 65

genotype. Though unexpected, this is consistent with previous findings that the number of proliferating cells in S-phase is decreased immediately after the completion of 10 days of RSD, but not 24 h later (Lagace et al., 2010). Next, we examined the status of microglia, the brain’s endogenous immune cells. RSD increases the expression of inflammatory markers by microglia and, in certain stress- responsive brain regions, causes microglia to adopt an “activated” morphology with shorter, thicker processes; importantly, this microglial activation is associated with anxiety-like behaviour, and manipulations that prevent microglial activation also prevent the development of anxiety (Wohleb et al., 2011, 2014; McKim et al., 2018). While we are unaware of direct evidence that microglia are abnormal in ZnT3 KO mice, microglia do express receptors that are sensitive to modulation by zinc, such as P2X7 receptors (Liu et al., 2008). Further, pretreating cultured microglia with zinc prior to stimulation by lipopolysaccharide increases the release of pro-inflammatory cytokines (Higashi et al., 2017). We assessed microglial morphology using a thresholding procedure that has previously been effective at detecting RSD-induced changes (Wohleb et al., 2011, 2014; McKim et al., 2018). However, we were unable to detect an effect of stress on microglial morphology or density in the BLA, dorsal hippocampus, ventral hippocampus, or PFC. While it is possible that a more detailed morphological analysis would detect differences, our results do not support changes in microglial morphology following stress, despite the development of anxiety-like behaviour in our mice. The RSD protocol that has been found to induce microglial activation (e.g., by Wohleb et al., 2011) involves introducing a dominant CD-1 intruder into a cage of three mice for 2 h. It is possible that this is more stressful than the protocol used in the current study, and very likely results in greater wounding and peripheral inflammation, which may contribute to the microglial activation. Finally, we used MRI to conduct a volumetric analysis. Control ZnT3 KO mice had greater corpus callosum volumes than WT mice, and stress decreased corpus callosum volume only in the ZnT3 KO mice. To our knowledge, this is the first indication of white matter abnormalities in mice that lack vesicular zinc. Interestingly, it has previously been observed that corpus callosum volume is larger in mice that are resilient to RSD than in mice that are susceptible, though whether this is a cause or effect of stress susceptibility is unclear (Anacker et al., 2016). Assuming that a larger corpus callosum is a protective factor, this could explain why ZnT3 KO mice show diminished susceptibility to social avoidance, though it does not explain why corpus callosum volume was reduced by stress CHAPTER 2: Vesicular Zinc and Social Stress 66

only in the ZnT3 KO mice. We also found that parietal cortex was larger in ZnT3 KO mice than in WT mice, supporting the results of Yoo et al. (2016). Finally, it appeared that ZnT3 KO mice might have larger lateral ventricles than WT mice, though this difference was not significant, and so would need to be verified in additional studies. While these findings are interesting, there were limitations to our volumetric analysis. Our analysis of the hippocampus included most of dorsal hippocampus but excluded ventral hippocampus, and our definitions of parietal cortex, corpus callosum, and lateral ventricle included portions, but not the entirety, of these structures. Also, due to the challenges inherent in accurately delineating regions of interest on MRI, our definition of “amygdala” also included parts of surrounding structures. In future studies, it would be valuable to apply complementary methods of quantifying regional volumes, as well as other MRI methods, such as diffusion MRI, for assessing inter- and intra-structural connectivity.

2.5 CONCLUSION Our primary aim was to examine how chronic stress, in the form of RSD, impacts mice that lack vesicular zinc due to genetic deletion of ZnT3. We found that these mice, unlike WT mice, did not become avoidant of a novel conspecific, suggesting increased resilience to the depression-like effects of stress. ZnT3 KO mice were not entirely unaffected by stress, however; they did become avoidant of a CD-1 mouse and also exhibited stress- induced anxiety-like behaviour. Finally, cued fear memory was enhanced by stress in ZnT3 KO mice, but not in their WT counterparts. Thus, a lack of vesicular zinc modulates the outcomes of RSD, but not in a straightforward fashion. We were unable to account for these behavioural effects through differences in hippocampal neurogenesis or microglial activation. However, we did observe ZnT3 KO mice to have larger corpus callosum volumes than WT mice. Further study will be required to determine whether this neuroanatomical abnormality is protective against the depression-like effects of RSD.

2.6 ACKNOWLEDGEMENTS & STATEMENT OF CONTRIBUTION The author thanks Abril Valverde Rascón, Katy Sandoval, and Angela Pochakom for their assistance in quantifying microglial density. The author also acknowledges the facilities provided by the HBI Advanced Microscopy Platform (Calgary, AB, Canada) and Dr. Vincent Ebacher for his assistance with confocal imaging. Finally, the author acknowledges the facilities, and the scientific and technical assistance, of the National CHAPTER 2: Vesicular Zinc and Social Stress 67

Imaging Facility at the Florey Institute of Neuroscience and Mental Health (Parkville, VIC, Australia). This chapter is a slightly modified version of an article by McAllister et al. (2018a). The present author wrote the article and designed and conducted the experiments, with the exception of the MRI volumetric analysis, which was designed and conducted by Dr. David K. Wright, Ryan Wortman, and Dr. Sandy R. Shultz. Dr. Richard H. Dyck contributed to the study design. The author thanks them for their contribution. CHAPTER 2: Vesicular Zinc and Social Stress 68

FIGURES

Figure 2.1 Timeline depicting the experimental design. WT and ZnT3 KO mice were subjected to 10 days of stress, consisting of daily episodes of social defeat, followed by isolated housing for the remainder of the experiment. Control WT and ZnT3 KO mice remained in group housing with their same-sex littermates throughout the experiment.

CHAPTER 2: Vesicular Zinc and Social Stress 69

Figure 2.2 Social interaction behaviour of WT and ZnT3 KO mice following repeated social defeat stress. A. Diagram of the social interaction apparatus. Each phase lasted 2.5 min. B. Time spent in the interaction zone and corner zones with a novel conspecific in the holding cage (phase 2). Stressed WT mice spent less time in the interaction zone, and more time in the corners, than control mice, whereas stressed ZnT3 KO mice did not differ from controls. C. Regardless of genotype, stress decreased time in the interaction zone and increased time in the corner zones when a novel aggressive CD-1 mouse was in the holding cage (phase 3). D. Stress decreased the social interaction ratios of the mice. Error bars represent 95% CIs. #effect of stress, p < .05; *follow-up test to significant interaction, p < .05 CHAPTER 2: Vesicular Zinc and Social Stress 70

Figure 2.3 Anxiety-like behaviour of WT and ZnT3 KO mice following repeated social defeat stress. A. Stress decreased the time spent on the open arms during a 5 min test in an elevated plus-maze, indicating increased anxiety. B. Stress also decreased time spent in the center area of the elevated plus-maze. C. Stress increased the latency to begin feeding when food-deprived mice were placed in a novel environment, again indicating increased anxiety. The maximum allowed time was 10 min. D. Relative to controls, stressed mice lost more weight over the 16 h food deprivation period prior to the novelty-suppressed feeding test. Error bars represent 95% CIs. #effect of stress, p < .05 CHAPTER 2: Vesicular Zinc and Social Stress 71

Figure 2.4 Conditioned fear memory in WT and ZnT3 KO mice following repeated social defeat stress. Training was conducted on day 1 and consisted of a single tone-shock pairing. A. On day 2, the freezing response to the tone presented in a novel context (i.e., cued fear memory) was assessed. Cued fear memory was enhanced by stress in ZnT3 KO mice, but was unaffected by stress in WT mice. The blue bars indicate the percentage of time spent freezing in the novel context prior to cue presentation, showing that there were no baseline differences. B. On day 3, contextual fear memory was assessed by measuring the time spent freezing when mice were re-exposed to the context in which they were previously shocked. ZnT3 KO mice exhibited less freezing, indicating worse fear memory, than WT mice. Error bars represent 95% CIs. †effect of genotype, p < .05; *follow-up test to significant interaction, p < .05

CHAPTER 2: Vesicular Zinc and Social Stress 72

Figure 2.5 Hippocampal neurogenesis in WT and ZnT3 KO mice following repeated social defeat stress. A. Sample image, taken at the same magnification at which cell counts were conducted, of cells in the subgranular zone (SGZ) of the hippocampal dentate gyrus immunolabeled for the cell proliferation marker Ki67. B. Estimates of the total number of Ki67-positive cells in the granule cell layer and SGZ of the dentate gyrus, in brains collected 24 h after the last episode of defeat stress. There was no effect of stress or difference between genotypes. Error bars represent 95% CIs.

CHAPTER 2: Vesicular Zinc and Social Stress 73

CHAPTER 2: Vesicular Zinc and Social Stress 74

Figure 2.6 Microglial morphology in WT and ZnT3 KO mice following repeated social defeat stress. A. Image of microglia in the basolateral amygdala, immunolabeled for the microglial marker Iba1. B. The same image, after a thresholding procedure was applied to binarize the image into areas of Iba1-positive and Iba1-negative labeling. A greater percentage of positive labeling can reflect a change in microglial morphology indicative of microglial activation. C-F. Quantification of the percentage of total area positively labeled for Iba1 across several regions of interest (prefrontal cortex, dorsal hippocampus, ventral hippocampus, and basolateral amygdala, respectively), in brains collected 24 h after the last episode of defeat stress. There was no effect of stress or difference between genotypes. Error bars represent 95% CIs.

CHAPTER 2: Vesicular Zinc and Social Stress 75

Figure 2.7 MRI volumetric analysis in WT and ZnT3 KO mice following repeated social defeat stress. Brains were collected 15 days after the final episode of stress. A. Depiction of the various regions of interest. There was no effect of stress or difference between genotypes in the volume of the prefrontal cortex, amygdala, or hippocampus B. Stress decreased the volume of the corpus callosum in the ZnT3 KO mice but had no effect on the WT mice. C. ZnT3 KO mice had larger parietal cortex volumes than did WT mice. D. While there was no main effect of stress or genotype, the volume of the lateral ventricles was larger in the ZnT3 KO controls than in the WT controls. Error bars represent 95% CIs. †effect of genotype, p < .05; *follow-up test to significant interaction, p < .05 CHAPTER 2: Vesicular Zinc and Social Stress 76

TABLES Table 2.1 Additional behavioural measures. Statistics are reported as mean ± standard deviation. #Main effect of stress, p < .05 *Stress × genotype interaction, p < .05

ZnT3 KO ZnT3 KO WT control WT stress control defeated Social interaction (n = 20) (n = 22) (n = 22) (n = 22) test Distance (m) – 8.9 ± 1.7 8.5 ± 1.9 8.7 ± 1.8 8.7 ± 2.6 empty cage Interaction time (s) 61.3 ± 13.2 53.9 ± 18.7 59.7 ± 21.1 49.4 ± 19.2 – empty cage# Corner time (s) – 25.4 ± 7.9 27.0 ± 13.0 25.3 ± 13.0 32.3 ± 23.4 empty cage Immobility time (s) – 14.1 ± 11.4 39.8 ± 17.4 18.5 ± 16.3 33.9 ± 15.0 conspecific# Immobility time (s) – 17.9 ± 8.82 42.2 ± 28.2 23.9 ± 25.2 50.6 ± 36.0 CD-1#

Novelty- suppressed (n = 14) (n = 16) (n = 16) (n = 16) feeding Latency to feed (s) 31.4 ± 32.9 25.9 ± 27.6 38.6 ± 39.4 31.3 ± 33.1 – home cage Food consumption 0.12 ± 0.09 0.09 ± 0.06 0.08 ± 0.06 0.07 ± 0.06 (g)

Y-maze spatial (n = 14) (n = 16) (n = 16) (n = 16) memory Novel arm entries 40.3 ± 6.2 42.0 ± 4.4 40.2 ± 6.3 40.5 ± 5.0 (% of total entries)

Novel arm time (s) 100.8 ± 31.3 94.9 ± 32.6 101.9 ± 28.4 92.6 ± 33.5

Total arm entries* 14.4 ± 6.8 16.8 ± 5.4 17.9 ± 5.1 14.5 ± 4.8

CHAPTER 2: Vesicular Zinc and Social Stress 77

Table 2.2 Comparison of control, susceptible, and resilient mice. Statistics are reported as mean ± standard deviation. *Different from control, p < .05 †Different from susceptible, p < .05

Control Susceptible Resilient

Social interaction (n = 42) (n = 32) (n = 11) test Distance traveled 8.8 ± 1.8 8.5 ± 1.9 9.5 ± 1.9 (m) – empty cage Immobile time (s) – 16.4 ± 14.2 40.0 ±15.3* 26.1 ± 14.8† conspecific Immobile time (s) – 21.0 ± 19.2 54.3 ± 32.7* 22.1 ± 16.8† CD1 Interaction time (s) – 65.3 ± 24.5 35.0 ± 28.4* 53.3 ± 23.4† conspecific Interaction time (s) – 49.9 ± 26.6 6.1 ± 7.0* 62.8 ± 35.1† CD-1 Corner time (s) – 30.1 ± 22.0 60.9 ± 32.1* 49.4 ± 30.0* conspecific Corner time (s) – 38.5 ± 27.3 96.7 ± 36.6* 45.5 ± 33.0† CD-1

Elevated plus-maze (n = 30) (n = 22) (n = 9)

Open arm time (s) 13.8 ± 20.7 3.8 ± 7.2* 0.8 ± 1.5*

Center time (s) 31.2 ± 18.6 16.7 ± 9.9* 23.9 ± 15.7

Distance traveled 6.0 ± 1.6 5.5 ± 1.4 5.9 ± 1.9 (m)

Novelty- suppressed (n = 30) (n = 22) (n = 9) feeding Latency to feed (s) – 207.0 ± 134.1 432.7 ± 151.2* 406.2 ± 87.5* novel cage

Weight loss (g) 2.7 ± 0.6 3.6 ± 0.5* 3.5 ± 0.5*

CHAPTER 2: Vesicular Zinc and Social Stress 78

Y-maze spatial (n = 30) (n = 22) (n = 9) memory Novel arm entries 40.2 ± 6.1 41.4 ± 4.4 40.6 ± 5.6 (% of total entries)

Novel arm time (s) 101.4 ± 29.3 85.6 ± 27.9 105.8 ± 32.6

Total arm entries 16.3 ± 6.1 15.1 ± 4.7 17.9 ± 5.5

Conditioned Fear (n = 30) (n = 22) (n = 9)

Freezing time (s) – 55.9 ± 21.2 64.3 ± 18.9 68.2 ± 8.5 cued fear Freezing time (s) – 49.1 ± 24.9 56.9 ± 23.8 64.4 ± 14.7 contextual fear

Body weight (n = 40) (n = 30) (n = 10)

Change from pre- to 0.4 ± 1.0 0.1 ± 1.3 0.8 ± 1.1 post-RSD (g)

CHAPTER 2: Vesicular Zinc and Social Stress 79

Table 2.3. Microglial analysis. Measurement of morphological parameters and microglial density. Statistics are reported as mean ± standard deviation.

Wild type Wild type ZnT3 KO ZnT3 KO control stress control defeated Microglial (n = 6) (n = 6) (n = 6) (n = 6) arborization area 2625.5 ± 2879.1 ± 2444.9 ± 2499.0 ± PFC (μm2) 278.0 411.4 630.6 379.5 3656.5 ± 3273.2 ± 3327.1 ± 3251.1 ± dHPC (μm2) 410.0 421.9 353.5 372.7 5657.4 ± 5072.7 ± 5539.5 ± 5832.6 ± vHPC (μm2) 653.6 981.6 834.0 774.4 3048.5 ± 3167.5 ± 3313.4 ± 3387.7 ± BLA (μm2) 200.7 522.6 394.0 337.6

Microglial soma (n = 6) (n = 6) (n = 6) (n = 6) area

PFC (μm2) 127.2 ± 11.0 134.6 ± 16.5 118.6 ± 23.9 121.2 ± 9.7

dHPC (μm2) 105.8 ± 17.0 99.5 ± 11.0 109.6 ± 23.6 102.2 ± 8.9

vHPC (μm2) 109.5 ± 9.5 105.2 ± 5.7 109.3 ± 5.4 105.6 ± 9.6

BLA (μm2) 111.3 ± 9.3 107.5 ± 11.1 106.4 ± 7.4 103.2 ± 7.9

Microglial density (n = 6) (n = 6) (n = 6) (n = 6)

PFC 439.9 ± 43.7 444.1 ± 37.8 469.4 ± 45.4 460.0 ± 38.5 (microglia/mm2) dHPC 294.6 ± 41.2 274.6 ± 21.7 305.2 ± 36.8 290.3 ± 54.3 (microglia/mm2) vHPC 199.6 ± 15.2 195.4 ± 21.8 201.6 ± 20.2 196.0 ± 18.4 (microglia/mm2) BLA 435.8 ± 39.8 416.7 ± 13.1 436.0 ± 30.1 441.2 ± 17.0 (microglia/mm2)

CHAPTER 2: Vesicular Zinc and Social Stress 80

Table 2.4 ANOVA results (effects of genotype and stress).

D.F. Stress Genotype Stress × Genotype

Social interaction

test Interaction time – 1, 82 F = 4.92, p = .029 F = 0.58, p = .450 F = 0.13, p = .719 empty cage Interaction time – 1, 82 F = 23.20, p < .001 F = 1.39, p = .243 F = 6.53, p = .012 conspecific Interaction time – 1, 82 F = 22.53, p < .001 F = 0.04, p = .835 F = 0.26, p = .608 CD-1 Corner time – 1, 82 F = 1.66, p = .201 F = 0.60, p = .440 F = 0.63, p = .430 empty cage Corner time – 1, 82 F = 25.74, p < .001 F = 1.81, p = .182 F = 6.36, p = .014 conspecific Corner time – CD- 1, 82 F = 37.18, p < .001 F = 4.12, p = .046 F = 2.27, p = .136 1 Interaction ratio – 1, 81 F = 8.32, p = .005 F = 0.09, p = .762 F < 0.01, p = .948 CD-1 Distance traveled 1, 82 F = 0.33, p = .566 F < .01, p = .953 F = 0.24, p = .624 – empty cage Immobility time – 1, 82 F = 38.72, p < .001 F = 0.05, p = .826 F = 2.45, p = .121 conspecific Immobility time – 1, 82 F = 19.53, p < .001 F = 1.54, p = .219 F = 0.04, p = .840 CD-1

Elevated plus-

maze

Open arm time 1, 58 F = 8.74, p = .004 F = 2.46, p = .122 F = 0.54, p = .465

Center time 1, 58 F = 10.99, p = .002 F = 3.21, p = .078 F = 0.77, p = .384

Distance traveled 1, 58 F = 0.97, p = .328 F < 0.01, p = .983 F = 2.23, p = .141

CHAPTER 2: Vesicular Zinc and Social Stress 81

Novelty- suppressed feeding Latency to feed (s) 1, 58 F = 40.43, p < .001 F = 0.27, p = .606 F = 0.09, p = .764 – novel cage Latency to feed (s) 1, 58 F = 0.56, p = .459 F = 0.54, p = .467 F = 0.01, p = .918 – home cage Food consumption 1, 58 F = 1.06, p = .307 F = 3.68, p = .060 F = 0.66, p = .419 (g)

Weight loss (g) 1, 58 F = 42.65, p < .001 F = 2.47, p = .121 F = 0.05, p = .828

Y-maze spatial

memory Novel arm entries 1, 58 F = 0.56, p = .458 F = 0.35, p = .554 F = 0.27, p = .603 (% of total entries)

Novel arm time (s) 1, 58 F = 0.89, p = .348 F = 0.01, p = .943 F = 0.05, p = .830

Total arm entries 1, 58 F = 0.12, p = .726 F = 0.16, p = .689 F = 4.18, p = .045

Conditioned fear

Freezing time – 1, 58 F < 0.01, p = .969 F = 0.10, p = .760 F = 0.41, p = .523 training pre-cue Freezing time – 1, 58 F = 1.77, p = .189 F = 0.70, p = .406 F = 0.23, p = .632 training during cue Freezing time – 1, 58 F = 0.14, p = .705 F = 0.05, p = .825 F = 0.01, p = .918 training post-cue Freezing time – novel context pre- 1, 58 F = 0.82, p = .369 F = 0.05, p = .824 F = 0.32, p = .577 cue Freezing time – 1, 58 F = 3.88, p = .054 F = 0.99, p = .324 F = 5.89, p = .018 cued fear Freezing time – 1, 58 F = 3.02, p = .087 F = 6.65, p = .012 F = 0.55, p = .463 contextual fear

CHAPTER 2: Vesicular Zinc and Social Stress 82

Body and organ weights

Weight change – 1, 77 F = 0.24, p = .627 F = 0.04, p = .835 F = 0.01, p = .933 pre- to post-RSD

Spleen weight 1, 20 F = 5.76, p = .026 F = 1.40, p = .250 F = 0.20, p = .660

Spleen weight (% 1, 20 F = 5.40, p = .031 F = 2.71, p = .116 F = 0.76, p = .394 of body weight)

Adrenal weight 1, 20 F = 15.86, p < .001 F = 0.25, p = .662 F = 3.92, p = .062

Adrenal weight (% 1, 20 F = 16.32, p < .001 F < 0.01, p = .985 F = 7.87, p = .011 of body weight)

Body weight 1, 20 F = 1.00, p = .328 F = 1.16, p = .294 F = 0.56, p = .463

Hippocampal cell

proliferation

Ki67+ cells 1, 20 F = 1.23, p = .281 F = 2.62, p = .121 F = 0.77, p = .391

Microglial

activation

Iba1+ area – PFC 1, 20 F = 1.34, p = .261 F = 0.53, p = .476 F = 0.62, p = .440

Iba1+ area – dorsal 1, 20 F = 1.05, p = .319 F = 0.12, p = .728 F = 0.14, p = .718 HPC Iba1+ area – 1, 20 F = 2.32, p = .143 F = 0.05, p = .831 F = 4.30, p = .051 ventral HPC

Iba1+ area – BLA 1, 20 F = 4.15, p = .055 F = 0.49, p = .490 F = 0.16, p = .692

Iba1+ cell density – 1, 20 F = 0.02, p = .881 F = 1.79, p = .196 F = 0.16, p = .690 PFC Iba1+ cell density – 1, 20 F = 1.13, p = .300 F = 0.64, p = .433 F = 0.03, p = .877 dorsal HPC Iba1+ cell density – 1, 20 F = 0.39, p = .537 F = 0.03, p = .872 F = 0.01, p = .926 ventral HPC Iba1+ cell density – 1, 20 F = 0.39, p = .539 F = 1.24, p = .279 F = 1.22, p = .283 BLA CHAPTER 2: Vesicular Zinc and Social Stress 83

MRI volumetric

analysis

PFC volume 1, 32 F = 1.88, p = .180 F = 0.02, p = .897 F = 0.34, p = .563

Hippocampus 1, 32 F = 0.28, p = .603 F = 2.39, p = .132 F = 0.57, p = .457 volume

Amygdala volume 1, 32 F = 1.39, p = .251 F = 1.25, p = .271 F = 0.01, p = .943

Lateral ventricle 1, 32 F = 1.31, p = .261 F = 1.90, p = .177 F = 3.99, p = .054 volume Corpus callosum 1, 32 F = 5.66, p = .024 F = 16.07, p < .001 F = 6.40, p = .017 volume Parietal cortex 1, 32 F = 1.37, p = .251 F = 7.66, p = .009 F = 1.06, p = .311 volume

Body weight 1, 32 F < 0.01, p = .986 F = 0.94, p = .339 F = 0.09, p = .768

CHAPTER 3: Vesicular Zinc and Fluoxetine 84

CHAPTER THREE: SURVIVAL OF ADULT-BORN CELLS IN THE HIPPOCAMPUS IS ENHANCED BY SOCIAL DEFEAT STRESS AND BY CHRONIC FLUOXETINE TREATMENT 3.1 INTRODUCTION Depression is a leading cause of disability worldwide (World Health Organization, 2017) and, as such, has been the subject of intensive scientific research. Yet much remains unknown about how depression develops, as well as how experiences such as chronic stress, which is a risk factor for depression (Kendler et al. 1999; Kendler & Gardner, 2016), contribute to this process. Much is also unknown about how pharmacological treatments for depression – the most common being selective serotonin reuptake inhibitor (SSRI) drugs – work on the brain to ameliorate the disorder, and why certain individuals respond to these drugs while others do not. One key to unravelling the etiology of depression is to understand the long-term structural changes that occur in the depressed brain, as well as how these changes are reversed, or compensated for, by antidepressant treatments. A component of structural neuroplasticity that has relevance to both depression and the effects of antidepressant drugs is adult neurogenesis. This refers to the process of generating new neurons in the mature brain (Altman, 1962; Altman & Das, 1965). Predominantly, adult neurogenesis takes place at two sites: the subventricular zone and the hippocampal dentate gyrus. The latter has been linked to the etiology of depression. In the dentate gyrus – specifically in the subgranular zone (SGZ) of the granule cell layer, along the border of the hilus – new cells are generated by asymmetrically-dividing radial glial-like neural progenitor cells (Encinas et al., 2006). The newly generated cells, called amplifying progenitor cells, themselves divide symmetrically to produce mostly immature neurons and a smaller number of glia (Tanapat et al., 2001; Pham et al., 2003). For the most part, the neurons migrate into the granule cell layer (Cameron et al., 1993). Some exist only transiently, undergoing apoptosis within days or weeks of being born (Tanapat et al., 2001; Dayer et al., 2003; Sairanen et al., 2005). Others survive, mature and become integrated into existing neural circuits (van Praag et al., 2002; Kee et al., 2007), affecting hippocampal function and, ultimately, behaviour (Snyder et al., 2001; Aimone et al., 2009; Clelland et al., 2009; Sahay et al., 2011). The effects of stress, depression, and antidepressant treatment on hippocampal neurogenesis have been well-described, at least in non-human animals. While it is difficult, CHAPTER 3: Vesicular Zinc and Fluoxetine 85 if not impossible, to truly replicate human depression in rodents, procedures that induce stress and model depression-like behaviour have been used to study these effects. Stress has a generally suppressive effect on neurogenesis, whether it be acute or chronic social stress in tree shrews (Gould et al., 1997; Czéh et al, 2001), acute predator odor stress in rats (Tanapat et al., 2001), chronic social stress or chronic restraint stress in rats (Czéh et al, 2002; Pham et al., 2003), or chronic mild stress or chronic social stress in mice (Mitra et al., 2006; Surget et al., 2011). Chronic treatment with SSRIs, such as fluoxetine, has the opposite effect, boosting neurogenesis in rats and mice (Malberg et al., 2000; Sairanen et al., 2005; Encinas et al., 2006). Importantly, this increase in neurogenesis seems to be required for some, though not all, of the behavioural effects of SSRIs in mice that have been chronically stressed (Santarelli et al., 2003; Surget et al., 2008, 2011; Wang et al., 2008; David et al., 2009). How these findings apply to human depression is not entirely clear, but there is evidence that treatment with antidepressant drugs, either tricyclics or SSRIs, increases cell proliferation in the SGZ in depressed people (Boldrini et al., 2012; but see Lucassen et al., 2010), and that depression is associated with smaller hippocampal volume (Sheline et al., 1996; McKinnon et al., 2009; Cole et al., 2011) and fewer dentate granule cells (Boldrini et al., 2013) – though whether impaired neurogenesis contributes to the smaller size of the hippocampus is unknown. Past research, both from our laboratory and from others (Suh et al., 2009), has implicated vesicular zinc as being involved in modulating adult hippocampal neurogenesis. Normally, neurogenesis can be enhanced in rodents by transferring them from standard laboratory housing (i.e., “shoebox” cages) to more complex, stimulating environments (Kempermann et al., 1997, 1998), by allowing them to exercise on running wheels (van Praag et al., 1999), or, as mentioned above, by treating them with SSRIs. In experiments conducted in our laboratory, we observed that ZnT3 KO mice, which lack the ability to store zinc in synaptic vesicles and release zinc in response to neural activity, did not show enhanced neurogenesis in response to enriched housing (Chrusch, 2015) or treatment with fluoxetine (Boon, 2016). Furthermore, these mice did not show the behavioural benefits (i.e., enhanced spatial cognition) of enriched housing, suggesting that hippocampal neurogenesis may be required for these benefits. The necessity of neurogenesis for some, though not all, of the cognitive benefits of environmental enrichment is supported by other findings (Bruel-Jungerman et al., 2005; Meshi et al., 2006; Garthe et al., 2015). Previously, we studied the effects of subjecting ZnT3 KO mice to repeated social CHAPTER 3: Vesicular Zinc and Fluoxetine 86 defeat (RSD) stress (see Chapter 2), which results in behavioural changes that reflect aspects of human depression (Krishnan et al., 2007). We found that ZnT3 KO mice exhibit some of the same effects of stress as WT mice, including increased anxiety-like behaviour and social avoidance of an aggressive CD-1 mouse. However, unlike WT mice, ZnT3 KO mice did not become socially avoidant of a same-strain conspecific, suggesting that ZnT3 KO mice are less susceptible to social avoidance following RSD stress. The present study sought to extend these results, with two main objectives. Given that vesicular zinc is required for modulation of neurogenesis in response to certain experiences, we first sought to assess if it is also required for modulation of neurogenesis in response to RSD stress. If so, then this might help to explain our finding that ZnT3 KO mice are less susceptible to stress-induced social avoidance. Second, given that neurogenesis has been implicated in the antidepressant effects of SSRIs, we sought to test whether ZnT3 KO mice would show behavioural benefits from chronic treatment with the SSRI fluoxetine. Our hypothesis, based on our previous findings, was that fluoxetine would increase neurogenesis in WT mice, and that this would be associated with reduced social avoidance of a CD-1 mouse and reduced anxiety-like behaviour, whereas ZnT3 KO mice would not exhibit increased neurogenesis, and would show no behavioural benefits from fluoxetine treatment.

3.2 METHOD 3.2.1 Animals See section 2.2.1.

3.2.2. Experimental Design For a diagram depicting the experimental design, see Figure 3.1. At 8-10 weeks of age, WT and ZnT3 KO mice were assigned to one of four treatment conditions: control + vehicle (WT: n = 11; KO: n = 9); control + fluoxetine (WT: n = 10; KO: n = 10); stress + vehicle (WT: n = 10; KO: n = 9); stress + fluoxetine (WT: n = 10; KO: n = 10). The stress consisted of 10 days of RSD (day 1 to 10), followed by isolated housing for the remainder of the experiment. The control mice remained in standard housing (described in section 2.2.1) throughout the experiment. One-day post-stress (day 11), mice were subjected to the first social interaction test. Immediately following this test, 5-bromo-2′-deoxyuridine (BrdU; Sigma) was administered to label cells in S-phase of the cell cycle. Three injections (100 mg/kg, IP) of BrdU dissolved CHAPTER 3: Vesicular Zinc and Fluoxetine 87 in sterile-filtered PBS (15 mg/ml) were administered at 8-h intervals. Also immediately following the social interaction test, the mice began 4 weeks of fluoxetine treatment. Four- days post-stress (day 14), the mice were subjected to the novelty-suppressed feeding (NSF) test. The NSF test was re-administered 3 weeks later (day 35), after 24 d of fluoxetine treatment. Finally, the social interaction test was re-administered after the 4 weeks of fluoxetine treatment was completed (day 39). Following this test, the mice were killed, and their brains were extracted and processed for BrdU immunofluorescence

3.2.3 Repeated Social Defeat See section 2.2.3.

3.2.4. Fluoxetine Treatment Fluoxetine hydrochloride (Santa Cruz) was administered orally (approximately 25mg/kg/day) via the drinking water, as previously described (McAllister et al., 2012). Fluoxetine was dissolved in distilled water (1 mg/ml), and then diluted to the appropriate concentration in tap water. To achieve the correct dosage, mice were weighed, and the amount of water consumed over the previous 3-4 days was measured using calibrated water bottles. In cages with multiple mice (i.e., group-housed controls), the average weight of the mice was used to calculate dosages. The dosage was adjusted, and fresh fluoxetine solution provided, two times per week. Vehicle-treated mice were weighed and handled on the same schedule but received only tap water.

3.2.5. Behavioural Assessment All testing was conducted during the light phase of the mice’s light/dark cycle, and mice were habituated to the testing room for at least 30 min prior to the start of each test.

3.2.5.1 Social interaction See section 2.2.4.1.

3.2.5.2 Novelty-suppressed feeding See section 2.2.4.3.

3.2.6 Quantification of Cell Survival Mice were perfused, brains were extracted, and tissue sections were prepared for immunofluorescence as described in section 2.2.5.1. One series of tissue sections was used CHAPTER 3: Vesicular Zinc and Fluoxetine 88 for BrdU immunofluorescence labeling. The procedure was as follows: 1 h in 2 M hydrochloric acid with gentle agitation every 15 min; 6 × 15 min wash in PBS; overnight incubation at room temperature in PBS containing 0.3% Triton X-100 and 2% normal goat serum with the primary antibody (1:200 rat anti-BrdU, Bio-Rad MCA2060); 3 × 10 min wash in PBS; overnight incubation at room temperature in PBS containing the secondary antibody (1:500, biotin-conjugated goat anti-rat, Jackson ImmunoResearch 112-065-167); 3 × 10 min wash in PBS; 1 h incubation in PBS containing the tertiary antibody (1:500, Alexa-Fluor 594-conjugated streptavidin, Jackson ImmunoResearch 016-580-084) with 0.1 µg/ml of 4′,6-diamidino-2-phenylindole (DAPI; Sigma) added for the final 15 min; 3 × 10 min wash in PBS. Sections were mounted on gelatin-coated slides, coverslipped with fluorescence mounting medium, and stored at 4 °C until analysis.

3.2.7 Statistical Analysis Statistical analyses were conducted using IBM SPSS Statistics (version 24). Unless otherwise stated, comparisons were conducted by three-way analysis of variance (ANOVA) with genotype (WT vs. ZnT3 KO), stress (control vs. stress), and drug (vehicle vs. fluoxetine) as factors. Significant interactions were followed-up with Bonferroni-corrected simple effects tests using the pooled error term, unless equality of variances could not be assumed (Levene’s test: p < .05), in which case non-pooled error terms were used. All ANOVA results are reported in Table 3.3 and Table 3.4. Means are presented ± standard deviation.

3.3 RESULTS 3.3.1 Behavioural Assessment 3.3.1.1 Social interaction 3.3.1.1.1 Replication of previous results Previously, we found that RSD stress caused WT mice to become socially avoidant of both novel conspecifics and aggressive CD-1 mice, whereas stressed ZnT3 KO mice avoided only the CD-1 aggressors (see Chapter 2). Here, we examined whether we could replicate this effect in a different sample of mice. We examined behaviour in the initial social interaction test, 1 d after the final episode of RSD. Because this test was conducted before the start of fluoxetine treatment, we were able to combine the vehicle- and fluoxetine- treated groups for analysis. We conducted planned contrasts to test the a priori hypotheses, CHAPTER 3: Vesicular Zinc and Fluoxetine 89 based on our previous results, that in the interaction test with a novel conspecific 1) interaction time and corner time would not differ between control and stressed ZnT3 KO mice, and 2) interaction time would be significantly decreased, and corner time significantly increased, in stressed WT mice relative to controls. We further hypothesized that in the social interaction test with an aggressive CD-1 mouse, there would be significant main effects of stress on interaction time and corner time, with stress decreasing interaction time and decreasing corner time. We first examined interaction with a novel conspecific (Figure 3.2A). Consistent with our hypotheses, stress decreased interaction time in the WT mice by 32% [F(1,39) = 5.15, p = .024], while having no significant effect on the ZnT3 KO mice [F(1,36) = 0.70, p = .410; Welch’s test for unequal variances]. It should be noted, though, that a two-way ANOVA showed no significant interaction between the effects of genotype and stress [F(1,75) = 1.36, p = .248]. Time spent in the corners of the field exhibited a similar pattern of effects (Figure 3.2A). Planned contrasts showed that stress had a significant effect on the WT mice, more than doubling corner time [F(1,39) = 5.26, p = .034; Welch’s test], while having no significant effect on the ZnT3 KO mice [F(1,36) = 3.72, p = .062; Welch’s test], though stress did tend to increase corner time in the ZnT3 KO mice. Again, a two-way ANOVA showed no significant interaction between the effects of genotype and stress [F(1,75) = 0.33, p = .568]. Next, we examined behaviour during the social interaction test with a novel, aggressive CD-1 mouse (Figure 3.2B). For interaction time, a two-way ANOVA confirmed our hypothesis that there would be a significant main effect of stress [F(1,75) = 34.70, p < .001] but no effect of genotype [F(1,75) = 1.42, p = .238] or interaction [F(1,75) = 1.59, p = .212]. Stress decreased interaction time with the CD-1 mouse by 61%. For corner time, a two-way ANOVA also confirmed our hypothesis that there would be a significant main effect of stress [F(1,75) = 30.18, p < .001], but no effect of genotype [F(1,75) = 0.10, p = .750], or interaction [F(1,75) = 1.93, p = .169]. In the presence of a CD-1 mouse, stress more than doubled corner time. Finally, stress decreased the interaction ratio with a CD-1 mouse [F(1,75) = 14.71, p < .001], and there was no significant difference between genotypes [F(1,75) = 3.23, p = .074], though WT mice tended to have lower interaction ratios than ZnT3 KO mice (Table 3.1). In the first phase of the social interaction test, with an empty holding cage, stress had no effect on total distance travelled [F(1,75) = 1.56, p = .216], nor was there a difference CHAPTER 3: Vesicular Zinc and Fluoxetine 90 between genotypes (Table 3.1). There was, however, a significant effect of stress on time spent in the interaction zone, consistent with our previous results (see Chapter 2). Stress decreased interaction zone time by 19% [F(1,75) = 8.28, p = .005]. There was no significant difference between genotypes [F(1,75) = 3.07, p = .084], though interaction time with the empty holding cage tended to be lower in ZnT3 KO mice than in WT mice.

3.3.1.1.2 Effect of fluoxetine treatment Four weeks after the first test, we examined social interaction behaviour a second time to assess how stress-induced social avoidance was affected by chronic fluoxetine treatment. Because the effects of stress on social interaction with a novel conspecific differed between genotypes on the first social interaction test, we focused here on interaction with an aggressive CD-1 mouse, as this was decreased in both genotypes on the first test, providing a similar baseline from which to assess the effects of fluoxetine treatment. Our hypotheses were that 1) in stressed mice that received vehicle treatment, social avoidance would persist to the second test (i.e., the mice would exhibit decreased interaction time and increased corner time relative to controls); and 2) fluoxetine treatment would reduce social avoidance (i.e., increase interaction time and decrease corner time) in stressed WT mice, but not in stressed ZnT3 KO mice. We first examined time spent in the interaction zone (Figure 3.3A). In the second social interaction test, there was no longer a significant main effect of stress on interaction time [F(1,71) = 2.80, p = .098]. This did not appear to be due to a reversal in social avoidance specific to the fluoxetine-treated mice, however [stress × drug: F(1,71) = 0.13, p = .716]; both the fluoxetine-treated and vehicle-treated mice showed similar amounts of interaction time, and simple-effects tests showed that there was no effect of stress even when the comparison was limited to vehicle-treated mice [F(1,71) = 0.84, p = .362]. This suggests that, even without drug treatment, social avoidance diminished over the 4 weeks between the first and second test. Our first hypothesis was thus not confirmed. Because there was no persistent effect of stress in the vehicle-treated mice, these results could not provide an adequate test of our second hypothesis. There was no significant main effect of fluoxetine treatment [F(1,71) = 0.71, p = .403] or interaction between fluoxetine treatment and any of the other factors. We next examined time spent in the corners of the field (Figure 3.3B). Unlike interaction time, there was a significant main effect of stress on corner time [F(1,71) = CHAPTER 3: Vesicular Zinc and Fluoxetine 91

29.27, p < .001], with the stressed mice spending more time in the corners than the control mice. Simple effects tests showed that this was also the case when the comparison was limited to vehicle-treated mice [F(1,37) = 17.34, p < .001; Welch’s test] and when it was further broken down by genotype [WT: F(1,19) = 7.12, p = .021, Welch’s test; KO: F(1,16) = 12.71, p = .003], supporting our first hypothesis. Regarding our second hypothesis, simple effects tests showed that there was no effect of fluoxetine treatment on the stressed WT mice [F(1,35) = 1.57, p = .218] or the stressed ZnT3 KO mice [F(1,35) = 1.17, p = .286]. Thus, our hypothesis that fluoxetine treatment would decrease corner time in WT mice but not in ZnT3 KO mice was not confirmed. There was, however, a significant main effect of fluoxetine treatment [F(1,71) = 5.59, p = .021], with fluoxetine decreasing the amount of time spent in the corners. This was not specific to the stressed mice [stress × drug: F(1,71) = 0.28, p = .600], and simple-effects tests showed that, when control and stressed mice were compared separately, there was no effect of fluoxetine treatment on either the control [F(1,36) = 3.42, p = .073] or stressed mice [F(1,35) = 2.73, p = .108]. To summarize, fluoxetine reduced corner time globally, rather than having an antidepressant-like effect specific to the stressed mice. One possible explanation for the lack of effect of fluoxetine on the stressed mice was that we included both mice that were susceptible to stress (as defined by an interaction ratio < 0.5 on the first social interaction test) and resilient to stress (interaction ratio ≥ 0.5) in our analyses. Therefore, we repeated these analyses excluding mice that were resilient to stress on the first test (4 WT-stress-vehicle, 4 KO-stress-vehicle, 4 KO-stress-fluoxetine). We first examined time spent in the interaction zone (Figure 3.4A). In this case, there was a significant main effect of stress [F(1,59) = 4.40, p = .040], with the stressed mice spending less time in the interaction zone. This was not specific to the vehicle-treated mice, however; simple effects tests showed that, when vehicle- and fluoxetine-treated mice were compared separately, there was no effect of stress on vehicle-treated [F(1,59) = 2.04, p = .159] or fluoxetine-treated mice [F(1,59) = 2.40, p = .127]. Once again, our first hypothesis was not confirmed, and therefore our results could not provide a test of our second hypothesis. There was no significant effect of fluoxetine treatment [F(1,59) = 1.22, p = .275] or interaction between fluoxetine treatment and any of the other factors. We next examined time spent in the corners, again excluding resilient mice from the analysis (Figure 3.4B). Once again, there was a significant main effect of stress on corner CHAPTER 3: Vesicular Zinc and Fluoxetine 92 time [F(1,59) = 32.20, p < .001], with the stressed mice spending more time in the corners than the control mice. Simple effects tests showed that this was also the case when the comparison was limited to vehicle-treated mice [F(1,27) = 20.90, p < .001] and when it was further broken down by genotype [WT: F(1,27) = 17.93, p < .001; KO: F(1,27) = 5.42, p = .028], supporting our first hypothesis. Regarding our second hypothesis, simple effects tests showed that there was no effect of fluoxetine treatment on the stressed WT mice [F(1,23) = 3.86, p = .062] or on the stressed ZnT3 KO mice [F(1,23) = 0.23, p = .633]. Thus, our hypothesis that fluoxetine treatment would decrease corner time in WT mice but not in ZnT3 KO mice was not confirmed. There was a significant main effect of fluoxetine treatment [F(1,59) = 6.41, p = .014], with fluoxetine decreasing the amount of time spent in the corners. As in our broader analysis including all mice, this effect was not specific to the stressed animals [stress × drug: F(1,59) = 0.62, p = .435]. Again, fluoxetine reduced corner time globally, rather than having an antidepressant-like effect specific to the stressed mice. Finally, to assess within-subjects changes in social interaction behaviour across the 4 weeks of fluoxetine treatment, we used the social interaction ratios with a CD-1 mouse to calculate difference scores for each mouse (Figure 3.5), subtracting the observed ratio on the first social interaction test (t1) from the observed ratio on the second test (t2) (i.e., t2 - t1). A positive number thus represents an increase in the social interaction ratio across the 4 weeks of treatment. For this analysis, we again excluded mice that were found to be resilient to stress (interaction ratio ≥ .5) at t1. Another mouse (KO-control-fluoxetine) was also excluded, due to a very small amount of time spent investigating the empty cage (3.9 s), which would have resulted in a very large interaction ratio. First, we assessed how social interaction behaviour changed across time in the non-stressed control groups, using one- sample t-tests to determine whether the observed mean difference score for each group differed from zero (a score of zero indicates no change over time). Surprisingly, in the WT- control mice that received vehicle treatment, social interaction ratios increased significantly from t1 to t2 [t(10) = 2.92, p = .015]. With this exception, however, interaction ratios were fairly stable in the control groups, remaining consistent over time [WT-control-fluoxetine: t(9) = 0.93, p = .376; KO-control-vehicle: t(8) = 0.29, p = .782; KO-control-fluoxetine: t(8) = 0.20, p = .845]. We next assessed how social interaction behaviour changed over time in the stressed mice. Consistent with our hypothesis, in stressed mice that received vehicle treatment, interaction ratios did not change over time in either WT [t(5) = 1.05, p = .340] or CHAPTER 3: Vesicular Zinc and Fluoxetine 93

ZnT3 KO mice [t(4) = 1.79, p = .148]. But, contrary to our hypothesis, treating stressed WT mice with fluoxetine did not increase interaction [t(9) = 1.57, p = .151]. Fluoxetine treatment also had no significant effect on stressed ZnT3 KO mice [t(5) = 2.06, p = .095], though it tended to increase interaction ratios. Thus, our hypothesis that fluoxetine treatment would decrease social avoidance in WT mice, but not in ZnT3 KO mice, was not confirmed; if anything, the trend was toward the opposite pattern of an effect in the ZnT3 KO mice but not in the WT mice.

3.3.1.2 Novelty-suppressed feeding Anxiety-like behaviour in the NSF test was first assessed on day 4 post-RSD (Figure 3.6A). Three mice (1 KO-stress-vehicle, 2 KO-stress-fluoxetine) were excluded from this analysis (and analysis of the results from the second NSF test) because they did not respond well to the overnight food deprivation. These mice were cold to the touch and moved very little, or not at all, when placed in the novel open field. The mice were immediately withdrawn from the NSF test, placed on a heating pad, and provided with food. Consistent with our previous findings (see Chapter 2), stress increased the latency to feed in the novel field by 87% [F(1,68) = 16.25, p < .001], indicating increased anxiety-like behavior. There was no difference between genotypes. On the first NSF test, after only 3 days of fluoxetine treatment, we predicted that the drug would have no effect, which was confirmed [main effect of fluoxetine: F(1,68) = 0.01, p = .943]. There was no significant effect of stress on latency to feed in the home cage [F(1,68) = 3.12, p = .082], though there was a trend toward an increase in the stressed mice (Table 3.2). The effect in the novel field, but not in the home cage, suggests that the increased latency to feed was caused by the anxiogenic environment of the novel field, rather than by more general effects of stress on feeding behaviour. In further support of this, there was also no effect of stress on the amount of food consumed during the 5 min test in the home cage [F(1,68) = 1.16, p = .286; Table 3.2]. Interestingly, there was significant interaction between fluoxetine treatment and genotype of the mice [F(1,68) = 4.63, p = .035], with fluoxetine tending to increase food consumption in the WT mice and decrease consumption in the ZnT3 KO mice. However, follow-up simple effects tests did not show a significant effect of fluoxetine treatment on either the WT [F(1,68) = 5.09, p = .054; Bonferroni-corrected] or ZnT3 KO mice [F(1,68) = 0.72, p = .400]. CHAPTER 3: Vesicular Zinc and Fluoxetine 94

The effect of stress on anxiety-like behaviour persisted at least 3 weeks to the second

NSF test. Difference scores, subtracting the latency to feed on the first test (t1) from the latency to feed on the second test (t2), showed that all groups had shorter latencies at t2, and that the decrease in latency was greater in the stressed mice [F(1,68) = 7.94, p = .006;

Table 3.2], likely because they started from a much higher baseline at t1. Despite this, there was still a significant effect of stress on latency to feed at t2 [F(1,68) = 10.41, p = .002; Figure 3.6B], with the stressed mice exhibiting an 118% increase relative to non-stressed controls. There was also a trend toward an interaction between the effect of stress and the genotype of the mice [F(1,68) = 3.54, p = .064]. Follow-up simple effects tests showed that there was no effect of stress on the WT mice [F(1,39) = 1.29, p = .264; Welch’s test] but that stress increased latency to feed in the ZnT3 KO mice [F(1,33) = 8.18, p = .022; Welch’s test; Bonferroni-corrected]. This suggests that the effect of stress on anxiety-like behaviour may be more persistent in ZnT3 KO mice than in WT mice. However, because the interaction effect was not significant, this interpretation would require further experimental validation. As in the first NSF test, there was no significant effect of stress on the latency to feed in the home cage test [F(1,68) = 2.46, p = .121; Table 3.2]. On the second test, after more than 3 weeks of drug treatment, we anticipated that the stressed, fluoxetine-treated mice would exhibit less anxiety-like behaviour than the stressed, vehicle-treated mice, and that this effect might be specific to the WT mice. However, we did not observe a significant effect of fluoxetine treatment on latency to feed in the novel field [F(1,68) = 0.23, p = .632], and the effect of fluoxetine treatment did not differ based on genotype [stress × genotype interaction: F(1,68) = 0.41, p = .527]. Although fluoxetine treatment did not affect anxiety-like behaviour, 24 days of fluoxetine treatment did significantly increase the amount of food consumed in the home cage test [F(1,68) = 39.93, p < .001; Table 3.2]. There was also a trend toward a three-way interaction between stress, genotype, and fluoxetine treatment on food consumption, though this was not significant [F(1,68) = 3.69, p = .059]. Finally, we examined differences in the amount of weight lost over the 16 h food restriction period prior to both the first and second NSF tests. Previously (see Chapter 2), we found that RSD stress increased the amount of weight lost, suggesting stress-induced metabolic changes. Here, we also observed the same effect on the first NSF test [main effect of stress: F(1,68) = 15.75, p < .001; Figure 3.7A], and this effect persisted to the second NSF CHAPTER 3: Vesicular Zinc and Fluoxetine 95 test [F(1,68) = 69.00, p < .001; Figure 3.7B]. These effects were also significant when weight loss was calculated as a percentage of body weight (data not shown). Interestingly, 3 days of fluoxetine treatment had no effect on weight loss in the first test [F(1,68) = 0.83, p < .365], but on the second test, after 24 days of treatment, there was a significant effect [F(1,68) = 17.99, p < .001], with fluoxetine decreasing the amount of weight lost over the food restriction period. Again, this effect was also significant when weight loss was calculated as a percentage of body weight (data not shown).

3.3.2 Quantification of Cell Survival One-day post-RSD, we injected mice with BrdU to label newborn cells. Brains were collected 4 weeks later, and cell survival was assessed by counting the number of BrdU+ cells in the granule cell layer and SGZ of the hippocampal dentate gyrus (Figure 3.8A). Stress had a significant effect on cell survival, but, contrary to our hypothesis, the effect was to increase survival [F(1,71) = 21.88, p < .001; Figure 3.8B], with stress increasing the number of BrdU+ cells by 37%. Further contradicting our hypothesis, modulation of cell survival by stress was not limited to the WT mice [stress × genotype interaction: F(1,71) = 0.84, p = .362]; follow-up simple effects tests showed that stress increased cell survival in both WT [F(1,71) = 16.28, p < .001; Bonferroni-corrected] and ZnT3 KO mice [F(1,71) = 6.81, p = .022; Bonferroni-corrected]. We used a priori planned contrasts to test our second hypothesis, that fluoxetine would increase cell survival in WT mice but have no effect on ZnT3 KO mice. Consistent with our hypothesis, we found that fluoxetine increased cell survival in WT mice [F(1,71) = 5.00, p = .028] while having no significant effect on ZnT3 KO mice [F(1,71) = 1.31, p = .256]. However, an ANOVA did not show that the effect of fluoxetine treatment differed significantly between genotypes [genotype × drug interaction: F(1,71) = 0.53, p = .470]. Instead, there was a significant main effect of fluoxetine treatment [F(1,71) = 5.64, p = .020], with fluoxetine increasing cell survival by 18%. Thus, our results clearly support an effect of fluoxetine treatment on neurogenesis, but only moderately support our previous finding that fluoxetine has no effect on neurogenesis in ZnT3 KO mice. Finally, we examined whether the effect of stress on cell survival differed between mice that were susceptible or resilient to stress (based on the results of the first social interaction test, as defined in section 3.3.1.1.2). A one-way ANOVA [F(2,73) = 12.14, p < .001] with Fisher’s LSD post-hoc test showed that, relative to controls, cell survival was CHAPTER 3: Vesicular Zinc and Fluoxetine 96 increased both in susceptible (p < .001) and resilient mice (p = .001). Survival in susceptible and resilient mice did not differ (p = .871).

3.4 DISCUSSION Stress is commonly thought of as an inhibitor of hippocampal neurogenesis; in part, this fact has underlain the neurogenesis theory of depression since its inception (Jacobs et al., 2000). Focusing specifically on chronic social stress in mice, most previous reports indicate that stress inhibits neurogenesis, decreasing cell proliferation, survival, or neuronal differentiation (Mitra et al., 2006; Ferragud et al., 2010; Schloesser et al., 2010; Chen et al., 2015; Walker et al., 2015; McKim et al., 2016). It is interesting, then, that we observed a strong positive effect of stress on neurogenesis. Newly-dividing cells were labeled by administering BrdU to mice 1 day after the final episode of RSD stress. When brains were collected 4 weeks later, and the number of BrdU+ cells in the SGZ and granule cell layer of the hippocampal dentate gyrus was counted, a substantial stress-induced increase was observed. Though uncommon, the observation of increased neurogenesis in response to stress is not unprecedented (Lagace et al., 2010; De Miguel et al., 2018). In particular, the present results are in alignment with those of Lagace et al., who – using a very similar, though more severe, RSD protocol – demonstrated an increase in the 4-week survival of cells born 1-day post-RSD. Lagace et al. also found that the increase in cell survival was limited to mice that were susceptible to social avoidance following stress, and that ablating neurogenesis by cranial x-ray irradiation made mice more resilient. We could not replicate the finding that enhanced survival is associated with susceptibility; in our experiment both susceptible and resilient mice showed greater cell survival than non- stressed controls, though we note that our definition of susceptible (social interaction ratio < 0.5) differed from the definition used by Lagace et al. (interaction ratio < 1). The finding that blocking neurogenesis promotes resilience is also at odds with a recent report that the activity of adult-born granule cells, specifically in the ventral dentate gyrus, promotes resilience to social avoidance following RSD stress (Anacker et al., 2018). Previously (see Chapter 2), we found that cell proliferation, as measured by immunolabeling for the endogenous proliferation marker Ki67, was not increased in the dentate gyrus 1 day after the final episode of RSD stress. This provides evidence that the increased number of BrdU+ cells observed in the present study was due to an increase in the rate of cell survival, rather than an increase in proliferation at the time of the BrdU CHAPTER 3: Vesicular Zinc and Fluoxetine 97 injections. It should be noted that, based solely on our own results, we cannot rule out the possibility that RSD stress increased blood-brain barrier (BBB) permeability, resulting in a greater concentration of BrdU reaching the hippocampus. This, rather than increased survival, could explain the increase in the number of BrdU+ cells observed 4 weeks later. However, Lagace et al. (2010) injected mice with BrdU 1-day post-RSD and killed them 2 h later to examine proliferation, and they found no effect of RSD stress on the number of BrdU+ cells (or Ki67+ cells) at this timepoint. This indicates that increased BrdU labeling due to increased BBB permeability is likely not a factor in these results, and it also supports our observation that proliferation is not increased 1-day post-RSD. One limitation of the present experiment is that neuronal differentiation of BrdU+ cells was not examined. Adult-born cells in the dentate gyrus mostly develop into neurons (Encinas et al., 2006), but some instead become glial cells or do not express standard neuronal or glial markers. We did not quantify the percentage of BrdU+ cells that expressed a neuronal marker, so we do not know the number or proportion of surviving cells that became neurons. However, Lagace et al. (2010) found that the percentage of surviving BrdU+ cells that expressed the neuronal marker NeuN was consistent between stressed and control mice at approximately 75%. Others have shown that neuronal differentiation is not affected by chronic fluoxetine treatment (Malberg et al., 2000; Santarelli et al., 2003; Encinas et al., 2006). And previous results from our own lab show that neuronal differentiation is not affected by a lack of vesicular zinc in naïve, environmentally enriched, or fluoxetine-treated mice (Chrusch, 2015; Boon, 2016). Together, these findings suggest that neuronal differentiation is unlikely to have been affected by stress or fluoxetine treatment in the present experiment. Based on previous findings that ZnT3 KO mice do not show modulation of adult hippocampal neurogenesis in response to environmental enrichment or fluoxetine treatment (Chrusch, 2015; Boon, 2016), one goal of the present experiment was to assess whether neurogenesis is modulated by stress in these mice. The results clearly show that it can be. This establishes that hippocampal neurogenesis is not completely incapable of being upregulated by experience in mice that lack vesicular zinc. But what might differentiate the experiences that affect neurogenesis from those that do not? The survival of adult-born neurons following fluoxetine treatment or environmental enrichment is dependent on brain-derived neurotrophic factor (BDNF) signaling (Sairanen et al., 2005; Rossi et al., 2006). In our previous work, we found that WT mice exhibit increased hippocampal BDNF CHAPTER 3: Vesicular Zinc and Fluoxetine 98 in response to environmental enrichment and fluoxetine treatment, but ZnT3 KO mice do not (Chrusch, 2015; Boon, 2016) – this quite possibly accounts for the failure to modulate neurogenesis in response to these experiences. RSD stress, on the other hand, does not appear to affect hippocampal BDNF levels either 1-day or 4-weeks post RSD, based on our own results (see Chapter 4) and the results of others (Lagace et al., 2010). Thus, it appears likely that RSD stress can increase neurogenesis in ZnT3 KO mice because it exerts its neurogenic effects through a pathway other than BDNF signaling. A second goal of the present experiment was to verify the earlier finding from our laboratory that vesicular zinc is required for the neurogenic effect of chronic fluoxetine treatment (Boon, 2016), and to extend this finding from female ZnT3 KO mice – which were used in the previous experiment – to male ZnT3 KO mice. The results provided partial support for this previous finding. A priori planned comparisons did show that fluoxetine treatment increased neurogenesis (specifically cell survival, as described below) in WT mice but not in ZnT3 KO mice. However, an ANOVA revealed no significant interaction between the effect of fluoxetine treatment and the genotype of the mice, indicating that fluoxetine treatment did not have a significantly greater effect on WT mice than on ZnT3 KO mice. Thus, while the present results do not disconfirm our previous findings, further validation will be required to more conclusively determine whether this effect extends to males. More straightforwardly, there was a clear positive effect of chronic fluoxetine treatment on neurogenesis in WT mice, consistent with many previous reports (Malberg et al., 2000; Santarelli et al., 2003; Sairanen et al., 2005; Encinas et al., 2006). Because BrdU was administered on the first day of the 4-week fluoxetine treatment, and because there is no acute effect of fluoxetine on cell proliferation (Malberg et al., 2000; Santarelli et al., 2003), it can be concluded that the effect of fluoxetine treatment was to increase cell survival. We also assessed the effects of RSD stress, and subsequent treatment with fluoxetine, on the behaviour of WT and ZnT3 KO mice. Consistent with our previous findings (see Chapter 2), RSD stress led to avoidance of an aggressive CD-1 mouse in the social interaction test and increased latency to feed in the NSF test, which can be interpreted as depression-like (i.e., social withdrawal) and anxiety-like behaviours, respectively. Furthermore, the anxiety-like behaviour persisted for at least 3 weeks, and, among mice that showed social avoidance (i.e., were defined as being susceptible to stress) on the first test, the social avoidance behaviour persisted for at least 4 weeks. The persistent nature of these effects allowed us to examine whether they could be reversed by CHAPTER 3: Vesicular Zinc and Fluoxetine 99 chronic treatment with fluoxetine. Contrary to several previous findings (Santarelli et al., 2003; Surget et al., 2008; Wang et al., 2008; David et al., 2009), we could not detect an effect of fluoxetine on anxiety-like behaviour in the NSF test. In the social interaction test with an aggressive CD-1 mouse, fluoxetine had a moderate effect, decreasing the amount of time spent in the corners (i.e., decreasing social avoidance), but not significantly increasing the time spent interacting (i.e., in close proximity) with the CD-1 target. This is broadly consistent with findings that chronic fluoxetine treatment decreases social avoidance following RSD stress in mice (Berton et al., 2006; Tsankova et al., 2006; Vialou et al., 2015; but see Venzala et al., 2012), though the effect may differ across strains (Razzoli et al., 2011). It should be noted, though, that the effect of fluoxetine in the present experiment was a general decrease in social avoidance across all groups, rather than an antidepressant- like effect that was specific to the stressed mice. Overall, the behavioural results provided no support for our hypothesis that ZnT3 KO mice would fail to benefit from the anti-anxiety and antidepressant-like effects of fluoxetine – though the lack of robust effects of fluoxetine treatment, even in WT mice, make it difficult to draw any strong conclusions regarding this hypothesis. In the future, this hypothesis might be better tested by using a higher dosage of fluoxetine, or perhaps a different antidepressant drug that exerts stronger behavioural effects. Interestingly, the most robust effect of fluoxetine treatment in the present experiment was to attenuate the amount of weight lost over the 16 h food deprivation period prior to the NSF test. This was the opposite of the effect of RSD stress, suggesting that chronic fluoxetine treatment may help to decrease the metabolic abnormality caused by RSD, though the exact nature of this abnormality remains to be uncovered. Notably, increased caloric intake following RSD stress has been linked to increased ghrelin signaling (Lutter et al., 2008), and fluoxetine treatment has been found to reverse the stress-induced increase in food intake, while also altering the circadian profile of ghrelin levels (Kumar et al., 2013). Finally, this experiment also served as a replication of our previous finding (see Chapter 2) that RSD stress causes social avoidance of a same-strain conspecific in WT mice, but not in ZnT3 KO mice, suggesting decreased susceptibility to stress in mice that lack vesicular zinc. A priori planned contrasts did show that, as expected, RSD stress increased social avoidance (i.e., decreased interaction time and increased corner time) in WT mice, but not in ZnT3 KO mice. However, an ANOVA failed to show a significant interaction between the effect of RSD stress and the genotype of the mice; that is, the effect of RSD CHAPTER 3: Vesicular Zinc and Fluoxetine 100 stress on social avoidance was not significantly greater in the WT mice than in the ZnT3 KO mice. Thus, the present results provide moderate, but not strong, support for our previous findings.

3.5 CONCLUSION The objective of the present experiment was to address two main questions. The first was whether mice that lack vesicular zinc would exhibit modulation of adult hippocampal neurogenesis by RSD stress. The second was whether hippocampal neurogenesis in these mice could be modulated by chronic fluoxetine treatment, and whether neurogenesis would be required for any anti-anxiety or antidepressant-like effects of the drug. The answer to the first question was clear. In both WT and ZnT3 KO mice, RSD stress clearly enhanced the rate of survival of cells born 1 day after the final episode of defeat, indicating that neurogenesis can be modulated by stress in ZnT3 KO mice. Regarding the second question, we found no evidence that the behavioural effects of fluoxetine were absent, or in any way altered, in mice that lack vesicular zinc – though the lack of strong behavioural effects of fluoxetine even in WT mice did not provide an ideal test of this hypothesis. However, the results did provide moderate support for our previous finding that vesicular zinc is required for the neurogenic effect of chronic fluoxetine treatment.

3.6 ACKNOWLEDGEMENTS The author thanks Nicoline Bihelek for her assistance with running the NSF test.

CHAPTER 3: Vesicular Zinc and Fluoxetine 101

FIGURES

Figure 3.1 Timeline depicting the experimental design. WT and ZnT3 KO mice were subjected to 10 days of repeated social defeat stress, consisting of daily episodes of social defeat, followed by isolated housing for the remainder of the experiment. Control WT and ZnT3 KO mice remained in group housing with their same-sex littermates throughout the experiment. Beginning 1-day post-stress, and lasting for 28 days, some mice received fluoxetine (25 mg/kg/day) in their drinking water. This procedure resulted in eight treatment groups (n = 9-11). CHAPTER 3: Vesicular Zinc and Fluoxetine 102

Figure 3.2 Replication of social interaction results from Chapter 2. The social interaction test was conducted 1-day post RSD. Because this test was conducted prior to the onset of fluoxetine treatment, groups were collapsed across this variable. A. Interaction with a novel same-strain conspecific. Stress decreased interaction time and increased corner time in WT mice but had no significant effect on ZnT3 KO mice. B. Interaction with a novel, aggressive CD-1 mouse. Stress decreased time spent in the interaction zone and increased time spent in the corner zones. Error bars represent 95% CIs. #effect of stress, p < .05; *simple effects test, p < .05 CHAPTER 3: Vesicular Zinc and Fluoxetine 103

Figure 3.3 Results from the second social interaction test, after 4 weeks of fluoxetine. A. Interaction time with a novel, aggressive CD-1 mouse. Contrary to our hypothesis, there was no persistent effect of stress on interaction time. Fluoxetine treatment also had no effect on stressed WT or ZnT3 KO mice. B. Corner time during the interaction test with a novel, aggressive CD-1 mouse. The stressed mice exhibited more corner time, indicating that the effect of stress on social avoidance persisted at least 4 weeks. Fluoxetine treatment decreased corner time, indicating decreased social avoidance, though this effect was not limited to the stressed mice. Error bars represent 95% CIs. #effect of stress, p < .05; ‡effect of fluoxetine, p < .05; n.s. indicates not significant. CHAPTER 3: Vesicular Zinc and Fluoxetine 104

Figure 3.4 Results from the second social interaction test, excluding susceptible mice. A. Interaction time with a novel, aggressive CD-1 mouse. Stress decreased interaction time, but fluoxetine did not have a counteracting effect, either overall or specifically in stressed WT or ZnT3 KO mice. B. Corner time during the social interaction test with a novel, aggressive CD-1 mouse. Stress increased corner time; together with decreased interaction time, this indicates that social avoidance persisted at least 4 weeks. Fluoxetine treatment decreased corner time, indicating decreased social avoidance, though this effect was not limited to the stressed mice. Error bars represent 95% CIs. #effect of stress, p < .05; ‡effect of fluoxetine, p < .05; n.s. indicates not significant. CHAPTER 3: Vesicular Zinc and Fluoxetine 105

Figure 3.5 Comparison of social interaction behaviour in the first and second tests. Difference scores were calculated, with a positive score indicating an increase in social interaction from the first to the second test. Social interaction was defined using the interaction ratio, which was calculated by dividing the time spent interacting with a novel, aggressive CD-1 mouse (phase 3 of the test) by the time spent interacting with an empty holding cage (phase 1 of the test). With one exception, interaction ratios were stable over time at the group level, as indicated by a mean score not differing from zero. Interaction ratios did increase significantly over time in the WT-control-vehicle treated mice. There was no effect of stress, fluoxetine treatment, or genotype of the mice on this measure. Error bars represent 95% CIs. *difference from zero, p < .05

CHAPTER 3: Vesicular Zinc and Fluoxetine 106

Figure 3.6 Anxiety-like behaviour in the novelty-suppressed feeding (NSF) tests. A. Results from the first NSF test, after 3 days of fluoxetine treatment. Stress increased the latency to feed in a novel open field, indicating increased anxiety-like behaviour. As anticipated, fluoxetine had no effect after only 3 days of treatment. There was also no difference between genotypes. B. Results from the second NSF test, after 24 days of fluoxetine treatment. The effect of stress on anxiety-like behaviour persisted to the second test. Contrary to our hypothesis, chronic fluoxetine did not decrease anxiety-like behaviour. There was also no difference between genotypes. Error bars represent 95% CIs. #effect of stress, p < .05 CHAPTER 3: Vesicular Zinc and Fluoxetine 107

Figure 3.7 Weight loss over the 16 h food restriction period prior to the novelty-suppressed feeding (NSF) tests. A. Results from the first NSF test, after 3 days of fluoxetine treatment. Stress increased the amount of weight lost, suggesting stress-induced changes in metabolism. B. Results from the second NSF test, after 24 days of fluoxetine treatment. The effect of stress persisted to the second NSF test, with stress increasing weight loss. Chronic fluoxetine had the opposite effect, decreasing the amount of weight lost, though this effect was not specific to the stressed mice. There were no significant differences between genotypes. Error bars represent 95% CIs. #effect of stress, p < .05; ‡effect of fluoxetine, p < .05 CHAPTER 3: Vesicular Zinc and Fluoxetine 108

Figure 3.8 Hippocampal cell survival in WT and ZnT3 KO mice following repeated social defeat stress and chronic fluoxetine treatment. BrdU was administered (3 injections × 100 mg/kg/injection over 16 h) 1-day post-stress, and brains were collected 4 weeks later to examine cell survival. A. Sample image of the hippocampal dentate gyrus immunolabeled for BrdU. BrdU+ cells were counted in the granule cell layer and subgranular zone. A′. Image showing the magnification at which cells were counted. B. Stress increased the 4- week survival of cells born 1-day after the final episode of stress. Fluoxetine treatment during the 4-week period also increased cell survival. Simple effects tests indicated that, as hypothesized, fluoxetine increased survival in WT mice but not in ZnT3 KO mice. Error bars represent 95% CIs. #effect of stress, p < .05; ‡effect of fluoxetine, p < .05 CHAPTER 3: Vesicular Zinc and Fluoxetine 109

TABLES Table 3.1 Additional behavioural measures from the first social interaction test. The test was conducted prior to the start of fluoxetine treatment. Statistics are reported as mean ± standard deviation. #Main effect of stress, p < .05 Wild type Wild type ZnT3 KO ZnT3 KO control stress control defeated Social interaction (n = 21) (n = 20) (n = 19) (n = 19) test Distance (m) – 8.7 ± 1.8 8.1 ± 2.1 9.1 ± 2.0 8.6 ± 1.4 empty cage Interaction time (s) 72.5 ± 18.4 62.0 ± 24.2 67.2 ± 19.0 51.1 ± 20.0 – empty cage#

CHAPTER 3: Vesicular Zinc and Fluoxetine 110

Table 3.2 Additional behavioural measures from the novelty-suppressed feeding tests. Statistics are reported as mean ± standard deviation. #main effect of stress, p < .05 ‡main effect of drug, p < .05 *genotype × drug interaction, p < .05 Vehicle Wild type ZnT3 KO Control Stress Control Stress

First test (n = 11) (n = 10) (n = 9) (n = 8)

Latency to feed 31.1 ± 22.4 47.2 ± 31.1 33.9 ± 29.2 51.9 ± 57.3 (s) – home cage Consumption 0.14 ± 0.06 0.14 ± 0.05 0.17 ± 0.06 0.15 ± 0.05 (g)*

Second test (n = 11) (n = 10) (n = 9) (n = 8)

Latency to feed 23.5 ± 21.6 23.6 ± 16.2 10.3 ± 6.8 41.5 ± 47.6 (s) – home cage Consumption 0.12 ± 0.04 0.16 ± 0.04 0.13 ± 0.07 0.11 ± 0.04 (g)‡ Latency to feed -105.8 ± 73.7 -195.2 ± 132.7 -86.0 ± 66.8 -151.9 ± 78.1 (s) – diff. score#

CHAPTER 3: Vesicular Zinc and Fluoxetine 111

Table 3.2 (continued)

Fluoxetine Wild type ZnT3 KO Control Stress Control Stress

First test (n = 10) (n = 10) (n = 10) (n = 8)

Latency to feed 37.3 ± 36.4 42.7 ± 33.0 26.6 ± 22.1 41.1 ± 26.7 (s) – home cage Consumption 0.18 ± 0.05 0.17 ± 0.05 0.16 ± 0.07 0.13 ± 0.04 (g)*

Second test (n = 10) (n = 10) (n = 10) (n = 8)

Latency to feed 26.3 ± 20.5 26.2 ± 24.6 15.4 ± 8.6 16.5 ± 9.9 (s) – home cage Consumption 0.22 ± 0.05 0.17 ± 0.04 0.22 ± 0.07 0.22 ± 0.06 (g)‡ Latency to feed -95.2 ± 113.8 -113.1 ± 91.0 -90.8 ± 127.5 -194.0 ± 140.8 (s) – diff. score#

CHAPTER 3: Vesicular Zinc and Fluoxetine 112

Table 3.3 ANOVA results from the first social interaction test. The test was conducted prior to the start of fluoxetine treatment (effects of genotype and stress).

D.F. Genotype Stress Genotype × stress

First social

interaction test Interaction time - 1, 75 F = 1.00, p = .321 F = 5.23, p = .025 F = 1.36, p = .248 conspecific Corner time - 1, 75 F = 0.51, p = .476 F = 8.80, p = .004 F = 0.33, p = .568 conspecific Interaction time - 1, 75 F = 1.42, p = .238 F = 34.70, p < .001 F = 1.59, p = .212 CD-1 Corner time - 1, 75 F = 0.10, p = .750 F = 30.18, p < .001 F = 1.93, p = .169 CD-1 Interaction ratio – 1, 75 F = 3.28, p = .074 F = 14.71, p < .001 F = 0.52, p = .472 CD-1 Interaction time – 1, 75 F = 3.07, p = .084 F = 8.28, p = .005 F = 0.36, p = .549 empty cage Distance – empty cage 1, 75 F = 0.99, p = .323 F = 1.56, p = .216 F < 0.01, p = .997

CHAPTER 3: Vesicular Zinc and Fluoxetine 113

Table 3.4 ANOVA results (effects of genotype, stress, and drug). Results are from the second social interaction test, after 4 weeks of fluoxetine treatment; from the first and second novelty-suppressed feeding (NSF) tests, after 3 days and 24 days of fluoxetine treatment, respectively; and from the analysis of cell survival.

D.F. Genotype Stress Drug

Social

interaction test

All mice

Interaction time – 1, 71 F = 0.13, p = .722 F = 2.80, p = .098 F = 0.71, p = .403 CD-1 Corner time – 1, 71 F = 1.51, p = .223 F = 29.27, p < .001 F = 5.59, p = .021 CD-1 Susceptible

only Interaction time – 1, 59 F = 0.40, p = .529 F = 4.40, p = .040 F = 1.22, p = .275 CD-1 Corner time – 1, 59 F = 3.09, p = .084 F = 32.20, p < .001 F = 6.41, p = .014 CD-1 CD-1 Interaction 1, 58 F = 0.74, p = .393 F = 4.12, p = .047 F = 0.08, p = .779 Ratio – diff. score

NSF test

First test

Latency to feed – 1, 68 F = 0.07, p = .787 F = 16.25, p < .001 F = 0.01, p = .943 novel field Latency to feed – 1, 68 F = 0.03, p = .876 F = 3.12, p = .082 F = 0.29, p = .595 home cage

Weight loss 1, 68 F = 0.12, p = .735 F = 15.75, p < .001 F = 0.83, p = .365

Food 1, 68 F = 0.22, p = .643 F = 1.16, p = .286 F = 0.82, p = .369 consumption

Second test

Latency to feed – 1, 68 F = 0.07, p = .787 F = 16.25, p < .001 F = 0.01, p = .943 novel field CHAPTER 3: Vesicular Zinc and Fluoxetine 114

Latency to feed – 1, 68 F = 0.59, p = .445 F = 2.46, p = .121 F = 0.50, p = .484 home cage

Weight loss 1, 68 F = 0.07, p = .790 F = 69.00, p < .001 F = 17.99, p < .001

Food 1, 68 F = 0.24, p = .624 F = 0.38, p = .542 F = 39.93, p < .001 consumption Latency to feed – 1, 68 F = 0.02, p = .892 F = 7.94, p = .006 F = 0.22, p = .642 difference score

Cell Survival

BrdU+ cell count 1, 71 F = 0.38, p = .540 F = 21.88, p < .001 F = 5.64, p = .020

CHAPTER 3: Vesicular Zinc and Fluoxetine 115

Table 3.4 (continued)

Genotype × Genotype × Genotype × drug Stress × drug stress stress × drug

All mice

Interaction time – F = 0.25, p = .622 F = 0.27, p = .604 F = 0.13, p = .716 F = 0.13, p = .721 CD-1

Corner time – CD- F = 0.01, p = .905 F = 0.01, p = .912 F = 0.28, p = .600 F < 0.01, p = .951 1

Susceptible only

Interaction time – F = 0.57, p = .452 F = 0.01, p = .927 F < 0.01, p = .986 F = 0.06, p = .813 CD-1

Corner time – CD- F = 0.49, p = .486 F = 0.88, p = .351 F = 0.62, p = .435 F = 0.80, p = .376 1

CD-1 Interaction F = 7.18, p = .010 F = 6.53, p = .013 F = 6.35, p = .014 F = 0.05, p = .826 Ratio – diff. score

NSF test

First test

Latency to feed – F = 2.50, p = .118 F = 0.50, p = .482 F = 0.16, p = .692 F = 0.22, p = .644 novel field

Latency to feed – F = 0.13, p = .720 F = 0.42, p = .520 F = 0.22, p = .644 F = 0.06, p = .813 home cage

Weight loss F = 2.43, p = .124 F = 0.05, p = .831 F = 0.01, p = .922 F = 1.05, p = .310

Food consumption F = 0.45, p = .503 F = 4.63, p = .035 F = 0.15, p = .699 F = 0.07, p = .788

Second test

Latency to feed – F = 3.54, p = .064 F = 0.41, p = .527 F = 1.27, p = .263 F = 0.42, p = .522 novel field

Latency to feed – F = 2.45, p = .122 F = 1.52, p = .222 F = 2.17, p = .146 F = 2.10, p = .152 home cage CHAPTER 3: Vesicular Zinc and Fluoxetine 116

Weight loss F = 0.50, p = .482 F = 0.42, p = .520 F = 0.10, p = .751 F = 2.71, p = .104

Food consumption F = 0.35, p = .555 F = 2.50, p = .118 F = 2.30, p = .134 F = 3.69, p = .059

Latency to feed – F = 0.40, p = .531 F = 2.03, p = .159 F = 0.12, p = .729 F = 1.23, p = .271 difference score

Cell Survival

BrdU+ cell count F = 0.84, p = .362 F = 0.53, p = .470 F = 0.21, p = .649 F = 0.42, p = .520

CHAPTER 4: Vesicular Zinc and BDNF 117

CHAPTER FOUR: BRAIN-DERIVED NEUROTROPHIC FACTOR AND TRKB LEVELS IN MICE THAT LACK VESICULAR ZINC: EFFECTS OF AGE, SEX, AND STRESS 4.1 INTRODUCTION The divalent cation zinc has essential biological functions throughout the body, including in the brain. Though most zinc in the brain is tightly-bound in protein structures, a portion of the brain’s zinc exists in a “free” (unbound or loosely-bound) state, making it readily available to participate in signaling functions (reviewed by McAllister & Dyck, 2017). The concentration of extracellular free zinc is relatively low, in the low nanomolar range (Frederickson et al., 2006a), and the cytosolic concentration even lower, in the picomolar range (Colvin et al., 2010). However, in certain regions of the brain and spinal cord, a considerable pool of free zinc can be found stored in the synaptic vesicles of neurons (Pérez-Clausell & Danscher, 1985). This vesicular zinc can be released synaptically in an activity-dependent manner (Assaf & Chung, 1984; Howell et al., 1984; Aniksztejn et al., 1987), elevating the extracellular free zinc concentration to – according to most estimates – the low micromolar range (Frederickson et al., 2006b), though hundreds of micromolar might be achievable with intense stimulation. One intriguing target of zinc is brain-derived neurotrophic factor (BDNF), a member of the neurotrophin family (Barde et al., 1982). Like other excreted peptides, BDNF is produced in the cell soma as a larger precursor protein. Pre-proBDNF is cleaved intracellularly into proBDNF (32 kDa). ProBDNF is then cleaved, by furin or other proprotein convertases, to produce mature BDNF (14 kDa), which forms an active homodimer (Mowla et al., 2001). BDNF is anterogradely transported to axon terminals in dense core vesicles (Conner et al., 1997; Michael et al., 1997; Fawcett et al., 1997; Dieni et al., 2012), from where it can be released along with the cleaved pro-domain (Kohara et al., 2001; Matsumoto et al., 2008). Uncleaved proBDNF can also be released and processed extracellularly, by plasmin or matrix metalloproteinases (MMPs), into mature BDNF (Lee et al., 2001; Gray & Ellis, 2008; Nagappan et al., 2009; but see Matsumoto et al., 2008). In addition to pre-synaptic release, BDNF is also stored postsynaptically in dendrites and spines, and it can be released to act as a retrograde or autocrine signal (Wong et al., 2015; Harward et al., 2016; Choo et al., 2007). Once released, BDNF can exert effects through tropomyosin receptor kinase B (TrkB), allosterically dimerizing these receptors and inducing their kinase function (Klein CHAPTER 4: Vesicular Zinc and BDNF 118 et al., 1991). This activates several signaling cascades, including the Ras, Rac, PI3-kinase, and PLC-γ1 pathways (Reichardt, 2006). Truncated variants of TrkB are also expressed, with unique cytoplasmic domains that lack catalytic kinase function (Klein et al., 1990), though the T1 variant has its own BDNF-dependent signaling pathway that results in intracellular calcium release (Rose et al., 2003). When expressed in the same cells, truncated TrkB forms heterodimers with full-length TrkB and inhibits its function (Eide et al., 1996). There is in vitro evidence that zinc can increase BDNF mRNA expression (I. Hwang et al., 2011), act directly on the BDNF protein (Ross et al., 1997; Post et al., 2008; Travaglia et al., 2013), activate extracellular enzymes that cleave BDNF from its precursor (J. Hwang et al., 2005; I. Hwang et al., 2011; Poddar et al., 2016), and transactivate TrkB through BDNF-independent mechanisms (Huang et al., 2008; Huang & McNamara, 2012). However, uncertainty remains about how vesicular zinc actually interacts with BDNF in the intact brain. One way of studying this is to examine how the BDNF-TrkB pathway is affected in mice that lack the dedicated vesicular zinc transporter, ZnT3 (Palmiter et al., 1996; Wenzel et al., 1997). Elimination of this protein results in a total loss of vesicular zinc (Cole et al., 1999). Several studies have used this approach, but they have produced mixed results. Adlard et al. (2010) found that hippocampal proBDNF levels, but not BDNF or TrkB levels, are reduced in ZnT3 KO mice at 3 months of age. By 6 months, both proBDNF and TrkB levels are reduced. Similarly, Nakashima et al. (2011) showed that TrkB mRNA levels in barrel cortex are reduced in 2-month-old male ZnT3 KO mice. On the other hand, Helgager et al. (2014) showed that ZnT3 KO mice, aged 3-6 months, have normal levels of hippocampal TrkB, but increased BDNF levels and increased TrkB phosphorylation. And Yoo et al. (2016) found that 5-week-old male ZnT3 KO mice have increased levels of TrkB in the hippocampus and cortex, increased mature BDNF and proBDNF in the cortex, and increased mature BDNF but decreased proBDNF in the hippocampus. One obstacle to synthesizing the results of this literature are differences in the age and sex of the mice examined. Our goal in the present experiment was to address this limitation by comparing WT and ZnT3 KO mice of both sexes at two different ages within the same study. In addition, based on our previous finding that ZnT3 KO mice are less susceptible to social avoidance following repeated social defeat (RSD) stress, and on findings that BDNF-TrkB signaling in the nucleus accumbens (NAc) is a critical determinant of stress susceptibility (Berton et al., 2006; Krishnan et al., 2007; Koo et al., CHAPTER 4: Vesicular Zinc and BDNF 119

2016), we also sought to examine whether ZnT3 KO mice would show abnormalities in BDNF and TrkB levels following RSD stress.

4.2 METHOD 4.2.1 Animals See section 2.2.1.

4.2.2 Experimental Design 4.2.2.1 Experiment one Brain tissue was collected from 40 mice. These included WT and ZnT3 KO, male and female, and young (5-week-old) and mature (12-week-old) mice, resulting in eight experimental groups (n = 5 for each).

4.2.2.2 Experiment two Tissue was collected from 38 mice. Male WT and ZnT3 KO mice at 8-10 weeks of age were assigned to the control condition or the stress condition. Stress consisted of 10 days of RSD (see sections 2.2.2 and 2.2.3 for details), resulting in four experimental groups (WT- control: n = 6; KO-control: n = 6; WT-stress: n = 6, KO-stress: n = 6). Mice were tested in the social interaction test on day 11, and brain tissue was collected on day 12. The effects of stress on female mice were not tested, due to limitations of the RSD method; however, we did collect brain tissue from non-stressed, female WT (n = 7) and ZnT3 KO (n = 7) mice at 8- 10 weeks of age.

4.2.3 Sample Preparation Mice were briefly anaesthetized with isoflurane and killed by decapitation. The brain was rapidly extracted, and the neocortices and hippocampi were dissected. The neocortical samples contained primarily posterior cortex, to avoid including striatum in the sample. For NAc samples, a 1 mm coronal section (from approximately 0.5-1.5 mm anterior to bregma) was prepared using a mouse brain matrix (ASI Instruments, #RBM-2000C). Tissue samples (containing NAc core and shell) were dissected bilaterally using an 11- gauge sample corer (Fine Science Tools). The extracted tissue was frozen on dry ice and stored at -80 °C. For protein analysis, tissue samples were placed in chilled RIPA buffer (Millipore- Sigma, 50 mM Tris-HCl, 150 mM NaCl, 0.25% deoxycholic acid, 1% NP-40, 1 mM EDTA; CHAPTER 4: Vesicular Zinc and BDNF 120

NAc: 100 µl; hippocampus: 200 µl; neocortex: 400 µl) containing protease and phosphatase inhibitors (Thermo Scientific Halt Protease and Phosphatase Inhibitor Cocktail, EDTA- Free) and homogenized using a bead lyser (5 min, 50 Hz). The lysates were placed on ice for 1 h and then centrifuged at 4 °C (15 min, 12,000 g). The supernatants were collected, protein concentrations were determined by Bradford assay (Bio-Rad), and the supernatants were stored at 20 °C until further analysis.

4.2.4 Western Blotting Protein samples were heated at 95 °C for 3 min in sample buffer containing 2% 2- mercaptoethanol. Samples (30 µg of protein per lane) were separated by electrophoresis on 12% SDS-PAGE gels (Experiment One) or on 4-20% gradient gels (TGX precast gels, Bio- Rad; Experiment Two) and transferred to PVDF membranes (Bio-Rad). The membranes were blocked for 1 h in Odyssey tris-buffered saline (TBS) blocking buffer (LI-COR Biosciences) and incubated overnight at 4 °C in Odyssey blocking buffer containing one or more of the following primary antibodies: rabbit anti-BDNF (1:1000; Santa Cruz Biotechnology, Inc.; sc-546); rabbit anti-TrkB (1:1000, Cell Signaling Technology; #4603); mouse anti-beta actin (1:2000; Abcam; ab8224). Blots were washed 3 × 10 min in TBS containing 0.1% Tween-20 (TBST), incubated for 1 h in Odyssey blocking buffer containing the secondary antibodies (1:10,000 anti-rabbit IRDye 800, 1:15,000 anti-mouse IRDye 680; LI-COR Biosciences), and again washed 3 × 10 min in TBST. Blots were imaged using an Odyssey infrared imaging system (LI-COR Biosciences) and quantified by densitometry using ImageJ (https://imagej.nih.gov/ij/index.html). Levels of BDNF or TrkB were normalized to the level of beta-actin.

4.2.5 Enzyme-Linked Immunosorbent Assay (ELISA) BDNF levels were also quantified using a sandwich ELISA kit (Aviscera Bioscience, SK00752-01) with good selectivity for mature BDNF over proBDNF (Polacchini et al., 2015). Samples were assayed in duplicate (protein loaded per well: 350 µg for NAc, 500 µg for hippocampus, 800 µg for neocortex) following the provided instructions. Briefly, samples were incubated for 2 h in a 96-well plate precoated with a monoclonal antibody against BDNF. The plate was then incubated for 2 h with the biotinylated detection antibody, followed by 1 h incubation with streptavidin-HRP conjugate. Tetramethylbenzidine (TMB) substrate solution was added, and colour was developed for 18 min before addition of the CHAPTER 4: Vesicular Zinc and BDNF 121 stop solution (0.5 M HCl). All incubations were conducted at room temperature on an orbital shaker. Absorbance was measured at 450 nm for 0.1 s using a microplate reader (Wallac 1420 Victor2; Perkin Elmer Life Sciences). BDNF concentrations were determined based on a standard curve and converted to pg of BDNF per mg of total protein. Where multiple plates were required to assay all samples (i.e., Experiment One), the plates used were from the same lot, and were run simultaneously.

4.2.6 Social Interaction Test See section 2.2.4.1.

4.2.7 Statistical Analysis Statistical analyses were conducted using IBM SPSS Statistics (Version 24). Unless otherwise specified, data were analyzed by three-way analysis of variance (ANOVA), with sex (male vs. female), age (5 week vs. 12 week), and genotype (WT vs. ZnT3 KO) as factors (Experiment One) or by two-way ANOVA with stress (control vs. stress) and genotype (WT vs. ZnT3 KO) as factors. Significant interactions were followed-up with Bonferroni- corrected simple effects tests using the pooled error term, unless equality of variances could not be assumed (Levene’s test: p < .05), in which case non-pooled error terms were used. All ANOVA results, including non-significant interactions, are summarized in Table 4.3 (Experiment One) and Table 4.4 (Experiment Two).

4.3 RESULTS 4.3.1 Experiment One 4.3.1.1 Hippocampus Levels of BDNF in the hippocampus were first quantified by densitometry on Western immunoblots (Figure 4.1A). The antibody against BDNF detected three bands between 11-kDa and 18-kDa; the lower molecular weight band was assumed to be mature BDNF, as its position most closely resembled that of recombinant human BDNF (Alomone Labs, #B-250). The other bands were presumed to be intermediate cleavage products of proBDNF, as suggested by Chacón-Fernández et al. (2016). Two more bands were also apparent, around 25-kDa and 28-kDa. These bands could also represent intermediate products, or they could have resulted from non-specific binding of the antibody. Previous work has shown a 28-kDa truncated form of proBDNF (Mowla et al., 2001). There was, at most, a very faint band around 32-kDa, where proBDNF would be expected. Given previous CHAPTER 4: Vesicular Zinc and BDNF 122 estimates that mature BDNF levels are over 10 times greater than proBDNF levels in the hippocampus (Matsumoto et al., 2008; Dieni et al., 2012), it is possible that proBDNF levels were too low to detect robustly. Mature BDNF levels, measured by densitometry (Figure 4.2A), did not differ based on sex [F(1,32) = 0.01, p = .924], age [F(1,32) = 0.68, p = .416], or genotype [F(1,32) = 2.69, p = .111]. To quantitatively assess BDNF, we also measured BDNF concentrations by ELISA (Figure 4.2B), using a kit that is selective against mature BDNF. Here, we found that age had differing effects on male and female mice [age × sex interaction: F(1,32) = 7.95, p = .008]. Specifically, BDNF levels did not differ between 5-week-old and 12-week-old males (p = .151), but in females, BDNF levels were significantly greater in the older mice than in the younger mice (p = .034; Bonferroni-corrected). Overall, no significant effect of genotype was observed [F(1,32) = 0.36, p = .552]. Though this pattern of results differed somewhat from the results obtained by Western blotting, the BDNF levels quantified by the two methods were significantly correlated [r(38) = .49, p = .001]. Levels of TrkB were also assessed by Western blotting (Figure 4.1A). The antibody against TrkB detected two bands; one around 135 kDa, assumed to be full-length TrkB, and one around 90 kDa, assumed to be truncated TrkB (TrkB.T). TrkB.T appeared to be more abundant than the full-length version, consistent with previous findings (Fryer et al., 1996). TrkB levels (Figure 4.3A) did not differ based on sex [F(1,32) = 0.47, p = .496], age [F(1,32) = 0.34, p = .566], or genotype [F(1,32) = 1.60, p = .215]. Likewise, TrkB.T levels (Figure 4.3B) did not differ based on sex [F(1,32) < 0.01, p = .958], age [F(1,32) < 0.01, p = .959], or genotype [F(1,32) = 1.49, p = .231].

4.3.1.2 Neocortex Levels of BDNF in the neocortex, as assessed by Western blotting, were higher in the 12-week-old mice compared to the 5-week-old mice [main effect of age: F(1,32) = 11.56, p = .002; Figure 4.4A]. No significant effect of sex [F(1,32) = 2.85, p = .101] or genotype [F(1,32) = 0.92, p = .345] was observed. BDNF concentrations in the neocortex were also assessed by ELISA (Figure 4.4B). This produced a somewhat different pattern of results than did Western blotting, though a similar pattern to the ELISA results observed in the hippocampus. Age had differing effects on male and female mice [age × sex interaction: F(1,32) = 6.60, p = .015]. BDNF levels did not differ between 5-week-old and 12-week-old males (p = .355), but in females, BDNF CHAPTER 4: Vesicular Zinc and BDNF 123 levels were significantly greater in the older mice than in the younger mice (p = .022; Bonferroni-corrected). Overall, no significant effect of genotype was observed [F(1,32) = 0.35, p = .557]. Unlike for the hippocampus, the neocortical BDNF levels measured by ELISA and by Western blotting were not significantly correlated [r(38) = .18, p = .266]. TrkB levels in the neocortex (Figure 4.5A) were significantly higher in male mice than female mice [F(1,32) = 5.13, p = .030]. There was no effect of age [F(1,32) = 0.89, p = .353] or genotype [F(1,32) = 1.28, p = .267]. Levels of TrkB.T (Figure 4.5B) did not differ based on sex [F(1,32) = 0.10, p = .757], age [F(1,32) = 0.08, p = .777], or genotype [F(1,32) = 1.29, p = .265].

4.3.2 Experiment Two 4.3.2.1 Nucleus accumbens BDNF levels in the NAc were first assessed by Western blotting (Figure 4.6A). Contrary to what was anticipated, there was no effect of stress on accumbal BDNF levels [F(1,20) = 2.47, p = .132]. There was a difference between genotypes, however, with the ZnT3 KO mice having significantly lower BDNF levels than the WT mice [F(1,20) = 9.43, p = .006]. We next sought to verify the effect of ZnT3 status on accumbal BDNF levels by ELISA (Figure 4.6B). Two data points were excluded as outliers (concentration > 50 pg/mg; 1 WT-control, 1 KO-control). Unexpectedly, there was no correlation between BDNF levels quantified by Western blotting and levels quantified by ELISA [r(20) = .05, p = .817], and the ELISA results showed no significant difference between genotypes [F(1,18) = 0.92, p = .355]. There was also no effect of stress [F(1,18) = 1.74, p = .204]. We also examined whether female ZnT3 KO mice show abnormalities in accumbal BDNF levels. Comparisons by one-way ANOVA detected no difference between WT and ZnT3 KO mice when BDNF levels were assessed either by Western blotting [F(1,12) = 2.80, p = .120; Figure 4.6C] or by ELISA [F(1,12) = 0.08, p = .788; Figure 4.6D]. BDNF levels detected by the two methods were, again, not significantly correlated [r(12) = -.29, p = .324], though when the male and female data were combined, there was a significant correlation [r(34) = .38, p = .024]. Levels of TrkB and TrkB.T in the NAc were also assessed by Western blotting (Table 4.1). In male mice, there was no effect of stress on TrkB levels [F(1,20) = 0.66, p = .425] or difference between genotypes [F(1,20) = 0.57, p = .458]. For TrkB.T, there was also no effect CHAPTER 4: Vesicular Zinc and BDNF 124 of stress [F(1,20) = 0.17, p = .689] or difference between genotypes [F(1,20) = 0.05, p = .826]. For female mice, comparisons by one-way ANOVA showed that there was no difference between genotypes in levels of TrkB [F(1,12) = 0.12, p = .738] or TrkB.T [F(1,12) = 0.55, p = .473].

4.3.2.2 Hippocampus BDNF levels, as assessed by Western blotting (Table 4.1), did not differ between genotypes in the hippocampus [F(1,20) = 0.89, p = .358], nor was there an effect of stress [F(1,20) = 1.35, p = .260]. For levels of TrkB (Table 4.1), there was no effect of stress [F(1,20) = 0.37, p = .553] or genotype [F(1,20) = 1.01, p = .328]. Finally, for TrkB.T levels, there was no effect of stress [F(1,20) = 0.04, p = .836] or genotype [F(1,20) = 0.22, p = .646].

4.3.2.3 Social interaction behaviour For Experiment Two, we assessed social interaction behaviour 1 day after the final episode of defeat stress. We measured time spent in the interaction zone and in the corner zones across the three phases of the test. In the first phase (empty holding cage; Table 4.2), there was no effect of stress [F(1,20) = 0.09, p = .772] or genotype [F(1,20) = 0.51, p = .484] on interaction time, nor was there an effect of stress [F(1,20) = 1.28, p = .272] or genotype [F(1,20) = 1.69, p = .209] on corner time. For total distance traveled, there was also no effect of stress [F(1,20) = 0.09, p = .774] or genotype [F(1,20) = 1.55, p = .227]. For the second phase of the test (novel conspecific; Figure 4.7A), there was an interaction between the effects of stress and genotype on interaction time [F(1,20) = 6.99, p = .016] and on corner time [F(1,20) = 8.96, p = .007]. Contrary to our previous findings (see Chapter 2), it was the ZnT3 KO mice that spent more time in the corners [F(1,10) = 8.48, p = .032, Bonferroni-corrected], whereas stress has no effect on corner time in WT mice [F(1,10) = 1.23, p = .293]. Similarly, stress decreased interaction time in ZnT3 KO mice [F(1,10) = 9.15, p = .026, Bonferroni-corrected], while having no effect on WT mice [F(1,10) = 0.91, p = .364]. For the third phase of the test (novel CD-1 mouse; Figure 4.7B), there was no effect of stress [F(1,20) = 2.04, p = .168] or genotype [F(1,20) = 0.10, p = .756] on interaction time. Similarly, for corner time there was no effect of stress [F(1,20) = 2.09, p = .164] or genotype [F(1,20) = 2.49, p = .130]. Finally, interaction ratios (Table 4.2) did not differ based on the genotype of the mice [F(1,20) < 0.01, p = .954] or on stress exposure [F(1,20) = 3.65, p = CHAPTER 4: Vesicular Zinc and BDNF 125

.071], though stress did tend to decrease interaction ratios. Previously, it has been shown that RSD stress increases BDNF levels in the NAc, and that this is associated with the development of social avoidance behaviour (Berton et al., 2006; Krishnan et al., 2007). We were unable to detect an effect of stress on accumbal BDNF levels, but there was considerable variability in social interaction within each group, which could potentially mask such an effect at the group level. Therefore, we also examined the correlations between NAc BDNF levels, as measured by Western blotting, and indicators of social avoidance. There was no significant correlation between BDNF levels and the interaction ratio with a CD-1 mouse [r(22) = .19, p = .367] or the raw amount of time spent interacting with a CD-1 mouse [r(22) = .21, p = .309]. The same was true for BDNF levels as determined by ELISA [respectively: r(20) = -.20, p = .369; r(20) = -.20, p = .364], and also if the control mice were excluded from the analysis, limiting the sample just to mice that had experienced RSD stress (data not shown). This provides further evidence that RSD stress did not increase accumbal BDNF levels in the present experiment.

4.4 DISCUSSION Previous examinations of BDNF and TrkB levels in the brains of ZnT3 KO mice have produced seemingly discrepant results. One of the objectives of the present experiment was to determine whether differences in age or sex of the mice tested might account for the discrepancy. The primary finding was that elimination of vesicular zinc, by genetic inactivation of ZnT3, did not affect hippocampal or neocortical levels of BDNF or TrkB, regardless of whether mice were male or female, young or mature. Thus, our results do not explain the discrepancy in the literature – to some extent, they add to it. Our results are in closest agreement with those of Adlard et al. (2010), who found that hippocampal TrkB levels do not decline in ZnT3 KO mice until somewhere between 3-6 months; older than the mice examined in our study. They also found that mature BDNF levels are normal in ZnT3 KO mice at least up to 6 months of age. Our results are in partial agreement with Helgager et al. (2014), who reported that overall hippocampal TrkB levels are normal in mature ZnT3 KO mice – though TrkB phosphorylation is enhanced. However, they also reported that hippocampal BDNF levels are elevated, for which we found no evidence. Our results correspond least closely with those of Yoo et al. (2016), who found elevated levels of both mature BDNF and TrkB (most likely truncated TrkB, based on their figure 4.3B) in 5-week-old ZnT3 KO mice. CHAPTER 4: Vesicular Zinc and BDNF 126

It is worth noting that the present results are consistent with our own previous findings. Specifically, we have found that ZnT3 KO mice show abnormal hippocampal BDNF levels following enriched housing (Chrusch, 2015) or chronic fluoxetine treatment (Boon, 2016), but in both experiments there was no difference between control WT and control ZnT3 KO mice. Together with the present results, this suggests that under “normal” conditions (e.g., standard housing, no drug treatment), vesicular zinc does not contribute to BDNF production or processing – at least in the hippocampus or neocortex. But vesicular zinc is necessary for the BDNF “boost” induced by certain experiences. How might this work? First, assume that mature BDNF is the dominant BDNF product in the adult brain, and proBDNF levels are relatively minimal, for which there is some evidence (Matsumoto et al, 2008; Dieni et al., 2012). Apparently, under normal conditions the brain has adequate enzymatic capacity to process most proBDNF into mature BDNF intracellularly, and this is not dependent on vesicular zinc. Under certain conditions that increase BDNF gene expression – such as environmental enrichment or chronic antidepressant treatment (Nibuya et al., 1996; Zajac et al., 2010; Tsankova et al., 2006) – proBDNF production will be increased, which may exceed the enzymatic capacity to process proBDNF, leading to its build-up within vesicles. It is unlikely that zinc interacts with BDNF within vesicles, as BDNF is localized to dense core vesicles (Michael et al., 1997; Dieni et al., 2012) and zinc is found in clear, round vesicles (Pérez-Clausell & Danscher, 1985). More likely is that zinc interacts with BDNF in the extracellular space, since both are secreted in response to neuronal activity (Frederickson et al., 2006b; Matsumoto et al., 2008; Nagappan et al., 2009; Wong et al., 2015). A build-up of proBDNF within vesicles would lead to increased proBDNF secretion. ProBDNF can be cleaved by extracellular MMPs (Lee et al., 2001), protease enzymes that require zinc to perform their catalytic function. Zinc, which promotes the activity of these enzymes at a (probably) physiologically- relevant concentration of 10 µM (J. Hwang et al., 2005), might increase the processing of proBDNF into mature BDNF within the synaptic cleft. Interestingly, MMP activity also appears to promote the expression of tissue plasminogen activator (I. Hwang et al., 2011), which, by activating plasmin, also increases the extracellular capacity to cleave proBDNF into mature BDNF (Pang et al., 2004). Alternatively, it is possible that vesicular zinc acts upstream of BDNF gene expression; this could be tested by examining whether ZnT3 KO mice show increased BDNF mRNA expression in response to environmental enrichment or CHAPTER 4: Vesicular Zinc and BDNF 127 antidepressant treatment. There is in vitro evidence that exposure to zinc can increase BDNF mRNA expression in cortical neurons, through an unknown mechanism that is dependent on MMPs (I. Hwang et al., 2011). Vesicular zinc release has also been shown to activate the mitogen-activated protein kinase (MAPK) pathway presynaptically in the hippocampal mossy fibers (Sindreu et al., 2011). One downstream target of MAPK is cyclic AMP response element-binding protein (CREB) (Impey et al., 1998), which can activate BDNF gene transcription (Tao et al., 1998). Although we found no effects of eliminating vesicular zinc on BDNF or TrkB levels in the present experiment, there were unexpected effects of age and sex. For BDNF levels, the results obtained by Western blotting and by ELISA were somewhat divergent; we tend to favour the ELISA results, for two reasons. One is that ELISA provides quantitative estimates of protein levels, based on a standard curve of known BDNF concentrations, versus estimates of relative differences from the Western blots. The other is that samples were assayed in duplicate for the ELISA, versus only once for the Western blots. The clearest result, as it was observed in both the hippocampus and neocortex, was that BDNF levels increased with age – from 5 weeks to 12 weeks – in female mice, but not in male mice. Other researchers have previously noted that BDNF levels increase over the first few postnatal weeks in mice and rats (Katoh-Semba et al., 1997; Kolbeck et al., 1999; Silhol et al., 2005; Yang et al., 2009). It is possible that BDNF expression peaks slightly later in females, which could explain why levels continue to increase after 5 weeks, whereas in male mice they apparently do not. We also observed that full-length TrkB levels were higher in male mice than female mice, though only in the neocortex. In our second experiment, we examined whether levels of BDNF and TrkB differ between WT and ZnT3 KO mice following exposure to RSD stress. Previously, we found that ZnT3 KO mice show reduced susceptibility to social avoidance following RSD (see Chapter 2), though we were unable to determine the mechanisms behind this effect. Other researchers have implicated BDNF signaling as a key factor in the behavioural outcomes of RSD stress. Specifically, mice that become socially avoidant following RSD have increased BDNF levels in the NAc, whereas mice that are resilient to social avoidance do not (Berton et al., 2006; Krishnan et al., 2007). Furthermore, blocking BDNF-TrkB signaling in the NAc prevents social avoidance following RSD stress (Koo et al., 2016). From experiments described above, we know that ZnT3 KO mice do not show normal upregulation of hippocampal BDNF levels in response to certain experiences. If the same held true for CHAPTER 4: Vesicular Zinc and BDNF 128 accumbal BDNF levels in response to the experience of stress, it could explain why these mice are less susceptible to social avoidance. In addition, we examined BDNF and TrkB levels in the hippocampus, because RSD has been shown to decrease hippocampal BDNF mRNA expression (Tsankova et al., 2006) and protein levels (Haenisch et al., 2009; but see Lagace et al., 2010). Several results are worth highlighting. First, we were unable to replicate our earlier finding that ZnT3 KO mice show reduced susceptibility to social avoidance following stress. Previously, we found that stressed ZnT3 KO mice did not avoid a novel, same-strain conspecific, whereas stressed WT mice did. Here, we found the opposite pattern – stressed ZnT3 KO mice, but not stressed WT mice, were avoidant of a novel conspecific, with the stressed ZnT3 KO mice spending an increased amount of time in the corners and a decreased amount of time in the interaction zone when the conspecific was present. Furthermore, we did not find a significant effect of stress on the amount of time spent interacting with an aggressive CD-1 mouse (though stress did tend to decrease interaction in both genotypes), whereas in our earlier study we found that both genotypes became avoidant of CD-1 mice. Given the small sample size in the present experiment, the results obtained from our earlier study are more likely indicative of the true effect of eliminating vesicular zinc. The effects of RSD stress on social avoidance are highly variable, with some mice showing no effect at all (i.e., resilience) and other mice becoming completely avoidant and showing no interaction over the entire testing period. Thus, a few highly resilient or highly susceptible mice could dramatically skew the results one way or the other, particularly with a small sample. Still, due to the divergent behavioural results, it is worth keeping in mind that the biochemical findings obtained from the small sample in the present experiment may not be applicable to the larger sample described in our previous study. Second, we were unable to replicate the finding that RSD stress increases accumbal BDNF levels – if anything, there was a trend toward a decrease. Elevated BDNF levels in the NAc are associated with social avoidance following RSD stress (Ma et al., 2016; Krishnan et al., 2007). We did not find a significant effect of stress on social avoidance of a CD-1 mouse, so one could argue that the social defeat procedure was not sufficiently stressful to induce social avoidance or an increase in BDNF. Indeed, our RSD procedure was modified to include less direct physical interaction than is commonly allowed (e.g., by Berton et al., 2006; Krishnan et al., 2007). However, in our previous study, the same RSD CHAPTER 4: Vesicular Zinc and BDNF 129 procedure produced robust avoidance of a CD-1 mouse, so the lack of an effect on avoidance in the present experiment was more likely the result of the small sample size. We also found no correlation between social avoidance and BDNF levels, indicating that BDNF levels were not elevated even in mice that did show social avoidance. This suggests that, contrary to previous findings, an increase in accumbal BDNF is not required for the development of social avoidance. Third, Western immunoblots showed that, regardless of stress, BDNF levels in the NAc were lower in male ZnT3 KO mice than in male WT mice. However, this effect was not supported by our ELISA results, and so should be considered – at best – preliminary and in need of further verification. We found no difference in accumbal BDNF levels between female WT and ZnT3 KO mice, suggesting that this effect could be specific to males. If the finding of decreased accumbal BDNF levels in male ZnT3 KO mice can be verified by additional experiments, it might begin to explain why these mice are less susceptible to social avoidance following stress, considering previous findings that elevated BDNF-TrkB signaling in the NAc is associated with avoidance (Ma et al., 2016; Koo et al., 2017). However, this explanation is called into question by our own failure to replicate the effect of RSD stress on BDNF levels. As far as we are aware, the present experiment was the first examination of BDNF levels in the NAc of ZnT3 KO mice. Finally, we found no differences in hippocampal BDNF levels, either between genotypes or between stressed and non- stressed mice. The lack of an effect of stress on hippocampal BDNF levels is consistent with Lagace et al. (2010).

4.5 CONCLUSION The results of the present experiments support the conclusion that – contrary to previous reports (Helgager et al., 2014; Yoo et al., 2016) – BDNF protein levels in the hippocampus and neocortex are not affected by the absence of vesicular zinc in naïve mice housed under standard laboratory conditions. This conclusion is consistent with our own previous findings, and it supports a broader model in which vesicular zinc modulates hippocampal and neocortical BDNF only under certain conditions, such as environmental enrichment. The absence of vesicular zinc may, however, decrease accumbal BDNF levels even in naïve mice, though support for this finding was mixed, and further verification is therefore required. We found no evidence that differences in age or sex of the mice tested could explain discrepant findings in the literature on how BDNF and TrkB levels are CHAPTER 4: Vesicular Zinc and BDNF 130 affected in ZnT3 KO mice; this discrepancy is, therefore, likely the result of other unidentified methodological factors. Finally, our results did not support previous findings that upregulation of BDNF-TrkB signaling in the NAc is required for the development of social avoidance following RSD stress (Berton et al., 2006; Krishnan et al., 2007; Koo et al., 2016), and we did not observe NAc BDNF levels to be abnormally affected by stress in ZnT3 KO mice. Thus, it remains unknown why mice that lack vesicular zinc signaling exhibit reduced susceptibility to social avoidance following stress.

4.6 ACKNOWLEDGEMENTS & STATEMENT OF CONTRIBUTION The author thanks Dr. Frank Visser and the HBI Molecular Core facility for use of their equipment and for consultation. This chapter is an extended version of an article by McAllister et al. (2018b). The present author wrote the article and designed and conducted the experiments. Nicoline Bihelek assisted with conducting the Western blotting experiments, and Dr. Richard H. Dyck assisted with tissue collection and contributed to the study design. The author thanks them for their contributions.

CHAPTER 4: Vesicular Zinc and BDNF 131

FIGURES

135 TrkB

80 TrkB.T

25 22

17 BDNF 11 Actin

Figure 4.1 Example Western blots for BDNF and TrkB. The antibody against BDNF detected numerous bands, assumed to be intermediate cleavage products of BDNF or nonspecific labeling. A band around 13-kDa was assumed to represent mature BDNF. The antibody against TrkB detected two variants; full-length TrkB and a truncated form of TrkB (TrkB.T).

CHAPTER 4: Vesicular Zinc and BDNF 132

Figure 4.2 Effects of age, sex, and ZnT3 status on BDNF protein levels in the hippocampus. A. BDNF levels, quantified by densitometry from Western blots and normalized to actin levels, were not affected by age, sex, or genotype. B. BDNF levels, quantified by ELISA, were greater in mature (12-week-old) female mice than in young (5- week-old) female mice, but did not differ between young and mature male mice. There was no difference between genotypes. Error bars represent SEM. ‡effect of age, p < .05 CHAPTER 4: Vesicular Zinc and BDNF 133

Figure 4.3 Effects of age, sex, and ZnT3 status on TrkB protein levels in the hippocampus. A. TrkB levels, quantified by densitometry from Western blots and normalized to actin levels, were not affected by age, sex, or genotype. B. TrkB.T levels were similarly unaffected by age, sex, or genotype. Error bars represent SEM. CHAPTER 4: Vesicular Zinc and BDNF 134

Figure 4.4 Effects of age, sex, and ZnT3 status on BDNF protein levels in the neocortex. A. BDNF levels, quantified by densitometry from Western blots and normalized to actin levels, were greater in mature (12-week-old) mice than in young (5-week-old) mice. There was no effect of sex or difference between genotypes. B. BDNF levels, quantified by ELISA, were greater in mature female mice than in young female mice, but did not significantly differ between young and mature male mice. There was no difference between genotypes. Error bars represent SEM. ‡effect of age, p < .05 CHAPTER 4: Vesicular Zinc and BDNF 135

Figure 4.5 Effects of age, sex, and ZnT3 status on TrkB protein levels in the neocortex. A. TrkB levels, quantified from Western blots and normalized to actin levels, were greater in male mice than in female mice. There was no effect of age, or difference between WT and ZnT3 KO mice. B. TrkB.T levels were not affected by age, sex, or genotype. Error bars represent SEM. *effect of sex, p < .05 CHAPTER 4: Vesicular Zinc and BDNF 136

Figure 4.6 Effects of stress and ZnT3 status on BDNF protein levels in the nucleus accumbens. A. In males, BDNF levels, quantified from Western blots and normalized to actin levels, were greater in WT mice than in ZnT3 KO mice. There was no effect of stress. B. In males, BDNF levels, quantified by ELISA, did not differ between genotypes, nor was there an effect of stress. C. In females, BDNF levels, quantified from Western blots and normalized to actin levels, did not differ between genotypes. D. Similarly, BDNF levels, quantified by ELISA, also did not differ between genotypes in female mice. Error bars represent SEM. †effect of age, p < .05 CHAPTER 4: Vesicular Zinc and BDNF 137

Figure 4.7 Social interaction behaviour of WT and ZnT3 KO mice following repeated social defeat stress. A. Time spent in the interaction zone and corner zones with a novel conspecific in the holding cage (phase 2). Stressed ZnT3 KO mice spent less time in the interaction zone, and more time in the corners, than control mice, whereas stressed WT mice did not differ from controls. B. Stress had no effect on time in the interaction zone or in the corner zones when a novel aggressive CD-1 mouse was in the holding cage (phase 3). Error bars represent SEM. *follow-up test to significant interaction, p < .05 CHAPTER 4: Vesicular Zinc and BDNF 138

TABLES Table 4.1 Experiment two: protein levels. Protein levels were quantified densitometrically from Western blots and expressed in arbitrary units (A.U.). BDNF, TrkB, and truncated TrkB (TrkB.T) levels were normalized to the level of actin in each sample. Statistics are reported as mean ± standard deviation. Wild type Wild type ZnT3 KO ZnT3 KO control stress control defeated Nucleus (n = 6) (n = 6) (n = 6) (n = 6) Accumbens

TrkB (A.U.) 0.76 ± 0.19 0.78 ± 0.25 0.64 ± 0.23 0.77 ± 0.19

TrkB.T (A.U.) 1.71 ± 0.26 1.81 ± 0.49 1.71 ± 0.47 1.74 ± 0.34

Nucleus (n = 7) (n = 7) Accumbens (♀)

TrkB (A.U.) 1.11 ± 0.20 N/A 1.14 ± 0.16 N/A

TrkB.T (A.U.) 2.72 ± 0.29 N/A 2.58 ± 0.37 N/A

Hippocampus (n = 6) (n = 6) (n = 6) (n = 6)

BDNF (A.U.) 1.68 ± 0.85 1.22 ± 0.36 1.28 ± 0.69 1.17 ± 0.29

TrkB (A.U.) 0.72 ± 0.09 0.75 ± 0.13 0.77 ± 0.08 0.77 ± 0.06

TrkB.T (A.U.) 2.30 ± 0.30 2.38 ± 0.19 2.31 ± 0.17 2.28 ± 0.17

CHAPTER 4: Vesicular Zinc and BDNF 139

Table 4.2 Experiment two: results of the social interaction test. The test was conducted 1 day after the final episode of stress. Statistics are reported as mean ± standard deviation.

Wild type Wild type ZnT3 KO ZnT3 KO control stress control defeated Social Interaction (n = 6) (n = 6) (n = 6) (n = 6) Test Distance traveled – 9.4 ± 2.0 8.3 ± 1.5 7.5 ± 2.9 8.1 ± 1.6 empty cage Interaction time – 62.3 ± 19.0 57.9 ± 25.8 54.6 ± 19.3 54.4 ± 8.8 empty cage Corner time – empty 21.1 ± 6.5 21.8 ± 16.0 36.8 ± 22.4 22.2 ± 10.9 cage Interaction ratio – 0.75 ± 0.64 0.45 ± 0.52 0.93 ± 0.65 0.30 ± 0.58 CD-1

CHAPTER 4: Vesicular Zinc and BDNF 140

Table 4.3 ANOVA results (effects of age, sex, and genotype).

D.F. Age Sex Genotype

Neocortex

BDNF 1, 32 F = 11.56, p = .002 F = 2.85, p = .101 F = 0.92, p = .345 densitometry BDNF Concentration 1, 32 F = 1.54, p = .224 F = 1.92, p = .176 F = 0.35, p = .557 (ELISA)

TrkB densitometry 1, 32 F = 0.89, p = .353 F = 5.13, p = .030 F = 1.28, p = .267

TrkB.T 1, 32 F = 0.08, p = .777 F = 0.10, p = .757 F = 1.29, p = .265 densitometry

Actin densitometry 1, 32 F = 1.75, p = .195 F = 2.00, p = .167 F < 0.01, p = .980

Hippocampus

BDNF 1, 32 F = 0.68, p = .416 F = 0.01, p = .924 F = 2.69, p = .111 densitometry BDNF Concentration 1, 32 F = 0.55, p = .362 F = 0.15, p = .697 F = 0.36, p = .552 (ELISA)

TrkB densitometry 1, 32 F = 0.34, p = .566 F = 0.47, p = .496 F = 1.60, p = .215

TrkB.T 1, 32 F < 0.01, p = .959 F < 0.01, p = .958 F = 1.49, p = .231 densitometry

Actin densitometry 1, 32 F = 0.17, p = .680 F = 0.03, p = .869 F = 0.03, p = .860

CHAPTER 4: Vesicular Zinc and BDNF 141

Table 4.3 (continued)

Age × sex × Age × sex Sex × genotype Age × genotype genotype

Neocortex

BDNF F = 0.99, p = .327 F = 2.07, p = .160 F = 0.03, p = .858 F = 1.04, p = .316 densitometry BDNF Concentration F = 6.60, p = .015 F = 0.09, p = .761 F = 0.80, p = .377 F = 0.42, p = .523 (ELISA)

TrkB densitometry F = 0.60, p = .446 F = 0.21, p = .649 F = 0.54, p = .470 F = 1.16, p = .290

TrkB.T F = 0.52, p = .477 F = 1.09, p = .304 F = 0.06, p = .816 F = 0.02, p = .904 densitometry

Actin densitometry F = 0.08, p = .776 F = 0.04, p = .840 F = 0.25, p = .619 F = 0.08, p = .777

Hippocampus

BDNF F = 1.27, p = .268 F = 0.01, p = .917 F = 0.73, p = .401 F < 0.01, p = .958 densitometry BDNF Concentration F = 7.95, p = .008 F = 1.26, p = .270 F = 0.62, p = .438 F = 0.45, p = .508 (ELISA)

TrkB densitometry F = 1.01, p = .322 F = 0.46, p = .504 F = 0.07, p = .795 F = 1.51, p = .228

Truncated TrkB F = 1.07, p = .309 F = 0.97, p = .332 F = 0.06, p = .814 F = 1.22, p = .278 densitometry

Actin densitometry F = 1.10, p = .302 F = 0.15, p = .705 F = 0.01, p = .921 F = 1.35, p = .254

CHAPTER 4: Vesicular Zinc and BDNF 142

Table 4.4 ANOVA results for experiment two (effects of stress and genotype).

D.F. Stress Genotype Stress × Genotype

Nucleus

Accumbens Actin 1, 20 F = 0.04, p = .846 F = 0.19, p = .666 F = 0.10, p = .759 densitometry BDNF/actin 1, 20 F = 2.47, p = .132 F = 9.43, p = .006 F = 0.10, p = .759 densitometry BDNF concentration 1, 18 F = 1.74, p = .204 F = 0.90, p = .355 F = 0.05, p = .832 (ELISA) TrkB/actin 1, 20 F = 0.66, p = .425 F = 0.57, p = .458 F = 0.37, p = .553 densitometry TrkB.T/actin 1, 20 F = 0.17, p = .689 F = 0.05, p = .826 F = 0.06, p = .810 densitometry

Nucleus

Accumbens (♀) Actin 1, 12 N/A F = 0.11, p = .743 N/A densitometry BDNF/actin 1, 12 N/A F = 2.80, p = .120 N/A densitometry BDNF concentration 1, 12 N/A F = 0.08, p = .788 N/A (ELISA) TrkB/actin 1, 12 N/A F = 0.12, p = .738 N/A densitometry TrkB.T/actin 1, 12 N/A F = 0.55, p = .473 N/A densitometry

Hippocampus

Actin 1, 20 F = 0.01, p = .908 F = 0.40, p = .534 F = 0.47, p = .502 densitometry BDNF/actin 1, 20 F = 1.35, p = .260 F = 0.89, p = .358 F = 0.50, p = .488 densitometry CHAPTER 4: Vesicular Zinc and BDNF 143

TrkB/actin 1, 20 F = 0.37, p = .553 F = 1.01, p = .328 F = 0.19, p = .670 densitometry TrkB.T/actin 1, 20 F = 0.04, p = .836 F = 0.22, p = .646 F = 0.40, p = .537 densitometry

Social

Interaction Test Distance traveled 1, 20 F = 0.09, p = .774 F = 1.55, p = .227 F = 1.03, p = .323 – empty cage Interaction time – 1, 20 F = 0.09, p = .772 F = 0.51, p = .484 F = 0.07, p = .795 empty cage Interaction time – 1, 20 F = 1.36, p = .257 F = 2.36, p = .140 F = 6.99, p = .016 conspecific Interaction time – 1, 20 F = 2.04, p = .168 F = 0.10, p = .756 F = 0.76, p = .394 CD-1 Interaction ratio – 1, 20 F = 3.65, p = .071 F < 0.01, p = .954 F = 0.45, p = .510 CD-1 Corner time – 1, 20 F = 1.28, p = .272 F = 1.69, p = .209 F = 1.54, p = .230 empty cage Corner time – 1, 20 F = 2.76, p = .113 F = 1.81, p = .193 F = 8.96, p = .007 conspecific Corner time – 1, 20 F = 2.09, p = .164 F = 2.49, p = .130 F = 0.57, p = .459 CD-1

CHAPTER 5: General Discussion 144

CHAPTER FIVE: GENERAL DISCUSSION The experiments described in this thesis were designed to characterize how mice that lack vesicular zinc respond to the experience of chronic stress, as induced by repeated social defeat (RSD), and – following stress – how they would respond to treatment with the antidepressant drug fluoxetine. Previous findings (Chrusch, 2015; Boon, 2016) have demonstrated a consistent pattern wherein ZnT3 KO mice fail to show neural plasticity in response to experience, be it barrel cortex plasticity in response to sensory deprivation by whisker plucking, or hippocampal neurogenesis in response to enriched housing or chronic fluoxetine treatment. Furthermore, the lack of hippocampal plasticity is associated with an inability to exhibit the cognitive benefits that are usually induced by environmental enrichment (Chrusch, 2015). From these findings followed the predictions that underlay the present experiments: 1) that ZnT3 KO mice would exhibit altered behavioural susceptibility to chronic stress, and 2) that ZnT3 KO mice would not show the neurogenic or, as a result, the behavioural effects of treatment with fluoxetine. As described in Chapter 2, examination of the social, emotional, and cognitive behaviour of ZnT3 KO mice following RSD stress confirmed the first prediction. The key finding was that – in contrast to WT mice, which became socially avoidant of a novel, aggressive CD-1 mouse following stress, and generalized this avoidance to a novel, same- strain conspecific – ZnT3 KO mice avoided CD-1 mice but not same-strain conspecifics, indicating reduced generalization of the avoidance response. This suggests that a lack of vesicular zinc protects against the depression-like outcome of social withdrawal following RSD stress. The results described in Chapter 3 provided some further support for this finding. Again, RSD stress caused social avoidance of a same-strain conspecific in WT mice but not in ZnT3 KO mice – though the previous finding that the effect of stress differed significantly between genotypes was not replicated (i.e., there was no significant interaction effect). In contrast, the social interaction results described in Chapter 4 showed the opposite pattern as in the previous two chapters, with the ZnT3 KO mice, but not the WT mice, avoiding a same-strain conspecific following stress. This experiment had, by far, the smallest sample size of the three. There is considerable inter-animal variability in the behavioural outcome of RSD stress, at least in terms of social interaction behaviour. This, coupled with the small sample size, may have led to the seemingly aberrant result in CHAPTER 5: General Discussion 145

Chapter 4. Despite this, the totality of the evidence from the three chapters supports the conclusion that the impact of RSD stress on social avoidance is reduced in ZnT3 KO mice. To understand the neuroanatomical basis by which ZnT3 KO mice are protected against the development of social avoidance following RSD stress, several potential mechanisms were examined. No evidence was found that microglial activation, hippocampal neurogenesis (either cell proliferation or survival), or hippocampal BDNF signaling differentiated WT from ZnT3 KO mice following stress – indeed, of these measures, only cell survival was affected by stress, and it was affected similarly in both genotypes. Two mechanisms did, however, stand out as warranting further investigation: volume of the corpus callosum and BDNF signaling in the nucleus accumbens (NAc). Regarding the corpus callosum, magnetic resonance imaging (MRI) volumetric analysis demonstrated that the volume of this structure is greater in ZnT3 KO mice than in WT mice. Assuming that stress decreases corpus callosum volume – as was observed in ZnT3 KO mice, though not in WT mice – and that smaller callosal volume is associated with social avoidance, then having a larger corpus callosum prior to stress may be protective against social avoidance and depression-like behaviour. Notably, smaller corpus callosum size has been observed to predict the development of late-life depression in women, though not in men (Cyprien et al., 2014), and is associated with early-onset adolescent depression (McMaster et al., 2013). Fractional anisotropy is also reduced in the corpus callosum of depressed people, suggesting abnormalities in structural integrity (Miyata et al., 2016; Dillon et al., 2018). And previous work in mice has shown that corpus callosum volume is negatively correlated with social avoidance following RSD stress; thus, a larger corpus callosum is associated with less social avoidance (Anacker et al., 2015). It is unclear whether the difference in callosal volume precedes stress or is the result of stress, however. In addition, Anacker et al. identified many other structures in which volume also correlates with social avoidance, either positively – as in the case of the ventral tegmental area (VTA), hypothalamus, and hippocampal CA3 region – or negatively – as in the case of the NAc, cingulate cortex, thalamus, bed nucleus of the stria terminalis, and raphe nuclei. It is possible that one or more of these represents the true causal factor, and the association with callosal volume is merely epiphenomenal. Still, the potential of a direct causal relationship between corpus callosum volume and stress resilience is interesting, though there is not much evidence to support what the underlying mechanism might be. One possibility is the loss of myelin; chronic restraint stress has been observed to reduce myelin CHAPTER 5: General Discussion 146 thickness and the number of oligodendrocytes in the corpus callosum of mice (Choi et al., 2017), though others failed to detect a difference in corpus callosum thickness following RSD stress (Laine et al., 2018). Another study found no difference in myelin thickness following chronic stress, but did note narrower nodes of Ranvier, as well as morphological abnormalities – including longer, thicker processes – in myelinating oligodendrocytes (Miyata et al., 2016). Taken together, these results suggest that an in-depth examination of myelination and oligodendrocyte number and morphology in ZnT3 KO mice – both stressed and stress- naïve – is warranted. However, another explanation for the enlarged callosal volume in ZnT3 KO mice is the larger volume of parietal cortex, which the MRI volumetric analysis also revealed. This is consistent with the results of Yoo et al. (2016), who also observed increased cortical size in male ZnT3 KO mice, as well as a greater number of neurons. If ZnT3 KO mice have more neocortical neurons, and specifically more callosal projection neurons, then this could also account for the increased corpus callosum size. It is worth highlighting that the present findings, along with those of Yoo et al., are the first indication of large-scale structural abnormalities in the brains of ZnT3 KO mice. A more in-depth morphological assessment of these mice, including both volumetric analysis and white- matter analysis, and including both females and males, would be highly beneficial for verifying and extending the present results. The other mechanism warranting further investigation as an explanation for the resilience to stress of ZnT3 KO mice is brain-derived neurotrophic factor (BDNF) signaling through tropomyosin receptor kinase B (TrkB) in the NAc. Based on previous findings, the potential link between BDNF, stress susceptibility, and vesicular zinc is compelling. RSD stress increases BDNF-TrkB signaling in the NAc, resulting from increased BDNF release from VTA neurons, but only in mice that are susceptible to depression-like behaviours following stress (Berton et al., 2006; Krishnan et al., 2007; Koo et al., 2016). Furthermore, local deletion of BDNF from VTA neurons confers resilience to stress (Berton et al., 2006; Krishnan et al., 2007). ZnT3 KO mice exhibit impairments at upregulating BDNF levels, at least in the hippocampus in response to certain treatments (Chrusch, 2015, Boon, 2016). If vesicular zinc, which is prevalent in the striatum and NAc (Slomianka et al., 1990; Sørensen et al., 1995), is also required for the upregulation of BDNF signaling in response to RSD stress, then this would provide an elegant explanation for why ZnT3 KO mice are less susceptible to stress-induced social avoidance. However, the support provided for this CHAPTER 5: General Discussion 147 idea by the present results was mixed, at best. There was no significant effect – positive or negative – of RSD stress on NAc BDNF levels, measured either by Western blotting or ELISA. Thus, an increase in NAc BDNF did not seem to be required for the development of social avoidance. On the other hand, NAc BNDF levels were lower in ZnT3 KO mice than in WT mice, as determined by Western blotting (though not confirmed by ELISA), which is consistent with low BDNF levels conferring resilience. The reason for the discrepancy between the present results and previous results (e.g., of Krishnan et al., 2007) regarding the effect of RSD on NAc BDNF levels is unclear. The RSD protocol used in the present experiments was less intense (a maximum of three attacks per defeat session vs. 10 min of uninterrupted physical interaction per session), so it is possible that the more intense protocol is required to affect BDNF release into the NAc. However, the less intense protocol was still sufficient to induce robust social avoidance, so it seems unlikely for this to be the case. Another possibility is a strain difference; the mice used in the present experiment were a mixed C57BL/6×129/Sv strain, which seem to be generally more susceptible to stress than the pure C57BL/6 mice used in the previous experiments; perhaps these mice exhibit relatively high levels of BNDF release into the NAc even at baseline, and do not exhibit a further increase in response to RSD stress. Whatever the explanation, replication of the decreased NAc BDNF levels observed in ZnT3 KO mice would be warranted, as well as further examination into the role this may play in conferring resilience to RSD stress. To this end, examining markers downstream of BDNF, such as phosphorylation of TrkB and ERK, would be helpful. Interestingly, phosphorylated

ERK (pERK) is increased following RSD stress in D2-receptor-negative cells (which are very likely D1-receptor-positive medium spiny neurons) in the NAc, but only in mice that are susceptible to stress, consistent with NAc BDNF levels being increased specifically in these mice (Koo et al., 2016). In the hippocampus, elimination of vesicular zinc reduces pERK, both at baseline and in response to activity, possibly because zinc slows ERK dephosphorylation by inhibiting MAPK tyrosine phosphate (Sindreu et al., 2011). This may provide a pathway by which vesicular zinc can modulate stress susceptibility, independently of a direct effect on BDNF levels – although the effect in the hippocampus is presynaptic in the mossy fibers, and so the mechanism would have to act somewhat differently to postsynaptically affect the medium spiny neurons in the NAc. As for the second major prediction of this thesis – that ZnT3 KO mice would not show the neurogenic or behavioural effects of treatment with fluoxetine – the results were CHAPTER 5: General Discussion 148 mixed. Regarding the effect on neurogenesis, the prediction was partially confirmed. Specifically, the 4-week survival of adult-born cells in the dentate gyrus was enhanced by chronic fluoxetine in WT mice but not in ZnT3 KO mice. However, the effect of fluoxetine treatment on cell survival was not significantly greater in the WT mice than in the ZnT3 KO mice (i.e., there was no significant interaction effect). Thus, while not conclusive, the present results provide some support for the previous finding that female ZnT3 KO mice fail to exhibit increased hippocampal neurogenesis – both in terms of proliferation and survival – in response to chronic fluoxetine treatment (Boon, 2016). The present results also extend this finding by demonstrating that ZnT3 KO mice are not entirely incapable of showing positive upregulation of neurogenesis; there was a robust increase in cell survival in response RSD stress in both WT and ZnT3 KO mice. Rather, the failure to upregulate neurogenesis only seems to apply to certain conditions, such as environmental enrichment (Chrusch et al., 2015), chronic fluoxetine (Boon et al., 2016), and hypoglycemia (Suh et al., 2009). Regarding the impact of chronic fluoxetine on behaviour, the prediction that ZnT3 KO mice would fail to exhibit any effects was not confirmed, though in some part this was due to methodological issues that prevented the prediction from being adequately tested, rather than outright refutation of the prediction itself. In the NSF test, it was predicted – based on past research (Santarelli et al., 2003; Surget et al., 2008; Wang et al., 2008; David et al., 2009) – that chronic fluoxetine would have an anxiolytic effect in the stressed mice. However, it did not. Similarly, in the social interaction test, it was predicted that fluoxetine would reduce the depression-like behaviour of social avoidance, again based on previous findings (Berton et al., 2006; Tsankova et al., 2006; Vialou et al., 2015). In this case, fluoxetine did have an effect, reducing the amount of time spent in the corners in the presence of a CD-1 mouse, though not significantly increasing the amount of time spent in close proximity interacting with it. However, this effect was not specific to the stressed animals; rather, fluoxetine decreased corner time in all groups. Since chronic fluoxetine did not have any clear anxiolytic or antidepressant effects, even in WT mice, it cannot be concluded whether elimination of vesicular zinc modulated these effects. Given that fluoxetine was administered orally in the drinking water, in comparison to the injections used in most previous studies, it is possible that a higher dosage is required to exert any behavioural effects – though the dosage was sufficient to affect neurogenesis. Thus, it appears that the neurogenic and behavioural effects of fluoxetine can be disassociated, at CHAPTER 5: General Discussion 149 least to some extent; this is also supported by the observation that fluoxetine treatment decreased social avoidance (i.e., corner time) in WT and ZnT3 KO mice, but only increased hippocampal cell survival in WT mice. One direction for future research would be to more conclusively delineate the pathway by which zinc is involved in upregulation of hippocampal neurogenesis following fluoxetine treatment or environmental enrichment. This is very likely to involve BDNF. Both chronic antidepressant treatment and environmental enrichment depend on increasing BDNF levels to increase the survival, and perhaps the proliferation, of new hippocampal neurons (Sairanen et al., 2005; Rossi et al., 2006; Choi et al., 2009; Pinnock et al., 2010), and both chronic antidepressant treatment and environmental enrichment fail to increase hippocampal BDNF levels or neurogenesis in ZnT3 KO mice (Chrusch, 2015; Boon, 2016). In contrast, RSD stress, which also increases neurogenesis but without any lasting effects on hippocampal BDNF levels (Lagace et al., 2010), can modulate neurogenesis in ZnT3 KO mice, as demonstrated by the present experiments. Put together, these results suggest that treatments specifically requiring BDNF to induce neurogenesis are not effective at doing so in ZnT3 KO mice, which suggests that vesicular zinc might interact directly with BDNF. The potential mechanisms behind this interaction are discussed extensively in Chapter 4. But an additional, and not mutually exclusive, possibility is that zinc acts on components of the signaling cascade downstream of TrkB activation that eventually lead to upregulation of neurogenesis. As highlighted above, ERK is a possible candidate here, as it is downstream of TrkB (Reichardt, 2006), and its phosphorylation is reduced in the hippocampus of ZnT3 KO mice (Sindreu et al., 2011). One of the main limitations of the present experiments was the nature of the mutant mice that were used. These mice are germ-line knockouts, meaning that they lack ZnT3 in all cells and tissues throughout all of life, including during development. The lack of tissue specificity is probably not a major problem, though it cannot be ruled out as a cofounding factor. Outside of the nervous system, ZnT3 mRNA is only strongly expressed in the testis, but it appears not to be translated into protein (Palmiter et al., 1996). ZnT3 mRNA has also been detected in a few other cell types (Smidt & Rungby, 2012). Of these, ZnT3 protein has been detected in the pancreatic islets, where it seems to be involved in glucose metabolism under conditions of beta-cell stress (Smidt et al., 2009). The lack of temporal specificity of the KO is likely a greater issue, as mechanisms almost certainly emerge to compensate for the loss of vesicular zinc. This would explain the CHAPTER 5: General Discussion 150 still somewhat curious observation that, despite the many established effects of zinc on a variety of neurotransmitter receptors and synaptic targets, mice that lack vesicular zinc exhibit only subtle abnormalities. Some evidence for this is provided by abnormalities in the expression of proteins related to synaptic function. In mature ZnT3 KO mice, dramatically reduced GluN2A expression and greater expression of AMPA receptors has been noted in the hippocampus (Adlard et al., 2010). The reduced GluN2A protein expression is also mirrored by reduced mRNA expression in the barrel cortex (Nakashima et al., 2011). Zinc has a pronounced inhibitory effect on GluN2A-containing NMDA receptors, so a decrease in GluN2A expression could reasonably be interpreted as a compensatory change to overcome excess excitation caused by a lack of the inhibition normally provided by vesicular zinc. It is also possible that the absence of ZnT3 causes neurodevelopment to be altered in these mice, since ZnT3 is detectable from late embryonic development onward, including in regions where it is not found in the mature brain (Valente & Auladell, 2002). The effects are probably not substantial, however, given the lack of gross behavioural or anatomical abnormalities in ZnT3 KO mice – though this is challenged somewhat by the present findings of larger parietal cortex and corpus callosum volume in ZnT3 KO mice. Whatever the case, in the future this methodological limitation will be surmountable by using conditional ZnT3 KO mice, in which ZnT3 can be eliminated in specific cell types and at specific time points. This will provide a better understanding of the true function of vesicular zinc in the normal brain. It is worth noting, though, that germline KO of ZnT3 likely provides a better model of humans with polymorphisms of the ZnT3 gene – assuming that such polymorphisms render the gene non-functional, for which there is at least some evidence (Hildebrand et al., 2015). However, given the rarity of such polymorphisms, and therefore the unlikeliness of inheriting two non-functioning alleles, the heterozygous ZnT3 KO is likely a better model of this than the homozygous KO used in the present experiments. Another limitation of these experiments was the use of only a single method of inducing chronic stress, and one that is only relevant to males1. This prevented us from

1 Recently, methods have been developed to overcome the challenge of getting male CD-1 mice to reliably attack female mice, allowing the RSD method to be applied to females. The “high-tech” option involves chemogenetic stimulation of the ventromedial hypothalamus to induce the CD-1 mouse to attack (Takahashi et al., 2017), whereas the “low-tech” option is to apply male urine to the CHAPTER 5: General Discussion 151 studying the effects of chronic stress on female ZnT3 KO mice. Though there is no strong evidence to believe that these effects would differ from the effects in male mice, it is entirely possible that they would. Indeed, there are documented sex differences in the regulation of vesicular zinc in the barrel cortex (Nakashima et al., 2010), the interaction between vesicular zinc and amyloid plaque load (Lee et al., 2002), the regulation of ZnT3 levels by estrogen (Lee et al., 2004) and the contribution of SLC30A3 polymorphisms to schizophrenia risk (Perez-Becerril et al., 2013, 2016). To extend the generalizability of the present results, the best option might be to use an entirely different chronic stress model that is effective in both male and female mice, such as chronic restraint stress or chronic unpredictable stress. Altering some other parameters of the stress protocol might also be worthwhile; for example, it could be tested whether some of the well-documented effects of stress that were not replicated in the present experiments, such as spatial memory impairment and microglial activation, could be induced by a longer duration of chronic stress – and, if so, whether ZnT3 KO mice would show altered susceptibility. If a lack of vesicular zinc promotes resilience to depression-like outcomes following chronic stress, then a reasonable assumption would be that targeted depletion of vesicular zinc would be a good therapeutic option for depression, or at least a good prophylactic against it. This strategy may be problematic, however, for at least two reasons. For one, while elimination of vesicular zinc does not cause any major abnormalities, it does cause minor cognitive and sensory impairments (Martel et al., 2010; 2011; Sindreu et al., 2011; Wu & Dyck, 2018), and reduces the threshold for drug-induced seizures (Cole et al., 2000), which would obviously be undesirable side-effects. Furthermore, given that these effects were observed in constitutive ZnT3 KO mice, in which there is very likely some compensation for the chronic loss of vesicular zinc, it is likely that these undesirable outcomes would be more pronounced following acute depletion of vesicular zinc. The other issue is that there is no straightforward way to specifically target vesicular zinc levels, barring the identification of a molecule that selectively blocks ZnT3 without affecting other, non-vesicular ZnTs. Vesicular zinc levels can be partially depleted by dietary zinc restriction (Takeda et al., 2003; Grabrucker et al., 2014), but dietary restriction would

female mouse (Harris et al., 2018). A third option is to use vicarious social defeat stress, a purely psychological stress model in which a mouse witnesses social defeat (Sial et al., 2015), which can be used to induce depression-like behaviours in female mice (Iñiguez et al., 2018) CHAPTER 5: General Discussion 152 affect zinc levels throughout the body – not just in the vesicular pool – and it is by no means clear that this would have a net antidepressant effect. Indeed, there is evidence that zinc deficiency increases the endocrine response to acute stress (Watanabe et al., 2010) and increases anxiety- and depression-like behaviour in rats (Takeda et al., 2007, 2012; Whittle et al., 2009). There is also evidence that dietary zinc supplementation is antidepressant in humans, at least when administered as an adjunctive treatment to other drugs (Siwek et al., 2009; Ranjbar et al., 2014). A better strategy, then, might be to identify the downstream effects of vesicular zinc signaling that are depressogenic, and to target these processes instead. One potential pitfall, though, is that if these factors include BDNF, TrkB, and pERK in the NAc – as previous findings suggest they might (Berton et al., 2006; Krishnan et al., 2007; Koo et al., 2016) – then any pharmacological treatment intended to produce an antidepressant effect by reducing or inhibited these factors would run up against the potentially pro-depressive effects of inhibiting these same factors in the hippocampus (Nibuya et al., 1995; Chen et al., 2001; Saarelainen et al., 2003). Though the results of the present thesis provide no clear implications for how to treat depression or reduce the deleterious effects of stress on the brain, they do make a significant contribution to our understanding of the role of ZnT3 and vesicular zinc in neural plasticity and behaviour. They reveal that the behavioural response to chronic stress is altered in mice that lack vesicular zinc, with these mice exhibiting reduced social avoidance. This strongly implies that vesicular zinc is involved in one or more of the mechanisms by which stress impacts the brain, as eliminating vesicular zinc confers resilience to one of the depression-like outcomes of stress. It is notable that, at least in this regard, the consequences of vesicular zinc loss appear not to be entirely negative. The present results also provide some evidence that the underlying mechanisms by which zinc influences the brain’s response to stress may involve the structure of the corpus callosum or BDNF signaling in the NAc. But the exact nature of the mechanisms remains to be uncovered. This and other challenges lie ahead on the road to more fully understanding the biological function of the vesicular storage, and synaptic release, of zinc. This question has intrigued anatomists and neurobiologists since the striking distribution of vesicular zinc in the hippocampus was first visualized over half a century ago, and will continue to be a source of intrigue for many years to come. REFERENCES 153

REFERENCES Adlard PA, Parncutt JM, Finkelstein DI, Bush AI (2010) Cognitive loss in zinc transporter- 3 knock-out mice: a phenocopy for the synaptic and memory deficits of Alzheimer’s disease? J Neurosci 30: 1631-1636. Adlard PA, Parncutt J, Lal V, James S, Hare D, Doble P, Finkelstein DI, Bush AI (2015) Metal chaperones prevent zinc-mediated cognitive decline. Neurobiol Dis 81: 196- 202. Aiba I, Shuttleworth CW (2013) Hypoxia limits inhibitory effects of Zn2+ on spreading depolarizations. PLoS One 8: e75739. Aiba I, Carlson AP, Sheline CT, Shuttleworth CW (2012) Synaptic release and extracellular actions of Zn2+ limit propagation of spreading depression and related events in vitro and in vivo. J Neurophysiol 107: 1032-1041. Aiba I, West AK, Sheline CT, Shuttleworth CW (2013) Intracellular dialysis disrupts Zn2+ dynamics and enables selective detection of Zn2+ influx in brain slice preparations. J Neurochem 125: 822-831. Aimone JB, Wiles J, Gage FH (2009) Computational influence of adult neurogenesis on memory encoding. Neuron 61: 187-202. Alboni S, Poggini S, Garofalo S, Milior G, El Hajj H, Lecours C, Girard I, Gagnon S, Boisjoly-Villeneuve S, Brunello N, Wolfer DP, Limatola C, Tremblay MÈ, Maggi L, Branchi I (2016) Fluoxetine treatment affects the inflammatory response and microglial function according to the quality of the living environment. Brain Behav Immun 58: 261-271. Altman J (1962) Are new neurons formed in the brains of adult mammals? Science 135: 1127-1128. Altman J, Das GD (1965) Autoradiographic and histological evidence of postnatal hippocampal neurogenesis in rats. J Comp Neurol 124: 319-336. Anacker C, Scholz J, O’Donnell KJ, Allemang-Grand R, Diorio J, Bagot RC, Nestler EJ, Hen R, Lerch JP, Meaney MJ (2016) Neuroanatomic differences associated with stress susceptibility and resilience. Biol Psychiatry 79: 840-849. Anacker C, Luna VM, Stevens GS, Millette A, Shores R, Jimenez JC, Chen B, Hen R (2018) Hippocampal neurogenesis confers stress resilience by inhibiting the ventral dentate gyrus. Nature 559: 98-102. REFERENCES 154

Anderson CT, Radford RJ, Zastrow ML, Zhang DY, Apfel UP, Lippard SJ, Tzounopoulos T (2015) Modulation of extrasynaptic NMDA receptors by synaptic and tonic zinc. Proc Natl Acad Sci U S A 112: E2705-2714. Andreini C, Banci L, Bertini I, Rosato A (2006) Counting the zinc-proteins encoded in the human genome. J Proteome Res 5: 196-201. Andrews GK (2000) Regulation of metallothionein gene expression by oxidative stress and metal ions. Biochem Pharmacol 59: 95-104. Andrews GK, Wang H, Dey SK, Palmiter RD (2004) Mouse zinc transporter 1 gene provides an essential function during early embryonic development. Genesis 40: 74-81. Aniksztejn L, Charton G, Ben-Ari Y (1987) Selective release of endogenous zinc from the hippocampal mossy fibers in situ. Brain Res 404: 58-64. Aras MA, Aizenman E (2011) Redox regulation of intracellular zinc: molecular signaling in the life and death of neurons. Antioxid Redox Signal 15: 2249-2263. Assaf SY, Chung SH (1984) Release of endogenous Zn2+ from brain tissue during activity. Nature 308: 734-736. Avitsur R, Stark JL, Sheridan JF (2001) Social stress induces glucocorticoid resistance in subordinate animals. Horm Behav: 247-257. Bagot RC, Parise EM, Peña C, Zhang HX, Maze I, Chaudhury D, Persaud B, Cachope R, Bolaños-Guzmán CA, Cheer J, Deisseroth K, Han MH, Nestler EJ (2015) Ventral hippocampal afferents to the nucleus accumbens regulate susceptibility to depression. Nat Commun 6: 7062. Bai S, Sheline CR, Zhou Y, Sheline CT (2013) A reduced zinc diet or zinc transporter 3 knockout attenuate light induced zinc accumulation and retinal degeneration. Exp Eye Res 108: 59-67. Barde YA, Edgar D, Thoenen H (1982) Purification of a new neurotrophic factor from mammalian brain. EMBO J 1: 549-553. Bartolomucci A, Palanza P, Gaspani L, Limiroli E, Panerai AE, Ceresini G, Poli MD, Parmigiani S (2001) Social status in mice: behavioral, endocrine and immune changes are context dependent. Physiol Behav 73: 401-410. Bartolomucci A, Pederzani T, Sacerdote P, Panerai AE, Parmigiani S, Palanza P (2004) Behavioral and physiological characterization of male mice under chronic psychosocial stress. Psychoneuroendocrinology 29: 899-910. Beaulieu C, Dyck R, Cynader M (1992) Enrichment of glutamate in zinc-containing REFERENCES 155

terminals of the cat visual cortex. Neuroreport 3: 861-864. Belloni-Olivi L, Marshall C, Laal B, Andrews GK, Bressler J (2009) Localization of Zip1 and Zip4 mRNA in the adult rat brain. J Neurosci Res 87: 3221-3230. Berton O, McClung CA, DiLeone RJ, Krishnan V, Renthal W, Russo SJ, Graham D, Tsankova NM, Bolanos CA, Rios M, Monteggia LM, Self DW, Nestler EJ (2006) Essential role of BDNF in the mesolimbic dopamine pathway in social defeat stress. Science 311: 864-868. Besser L, Chorin E, Sekler I, Silverman WF, Atkin S, Russel JT, Hershfinkel M (2009) Synaptically released zinc triggers metabotropic signaling via a zinc-sensing receptor in the hippocampus. J Neurosci 29: 2890-2901. Beyer N, Coulson DT, Heggarty S, Ravid R, Irvine GB, Hellemans J, Johnston JA (2009) ZnT3 mRNA levels are reduced in Alzheimer’s disease post-mortem brain. Mol Neurodegener 4: 53. Beynon SB, Walker FR (2012) Microglial activation in the injured and healthy brain: what are we really talking about? Practical and theoretical issues associated with the measurement of changes in microglial morphology. Neuroscience 225: 162-171. Bjorklund NL, Reese LC, Sadagoparamanujam VM, Ghirardi V, Woltjer RL, Taglialatela G (2012) Absence of amyloid β oligomers at the postsynapse and regulated synaptic Zn2+ in cognitively intact aged individuals with Alzheimer’s disease neuropathology. Mol Neurodegener 7: 23. Blakemore LJ, Trombley PQ (2004) Diverse modulation of olfactory bulb AMPA receptors by zinc. Neuroreport 15: 919-923. Boldrini M, Hen R, Underwood MD, Rosoklija GB, Dwork AJ, Mann JJ, Arango V (2012) Hippocampal angiogenesis and progenitor cell proliferation are increased with antidepressant use in major depression. Biol Psychiatry 72: 562-571. Boldrini M, Santiago AN, Hen R, Dwork AJ, Rosoklija GB, Tamir H, Arango V, Mann JJ (2013) Hippocampal granule neuron number and dentate gyrus volume in antidepressant-treated and untreated major depression. Neuropsychopharmacology 38: 1068-1077. Boldrini M, Fulmore CA, Tartt AN, Simeon LR, Pavlova I, Poposka V, Rosoklija GB, Stankov A, Arango V, Dwork AJ, Hen R, Mann JJ (2018) Human hippocampal neurogenesis persists throughout aging. Cell Stem Cell 22: 589-599. Boon JM (2016) Experience-dependent modulation of adult hippocampal neurogenesis in REFERENCES 156

female mice (Master’s thesis). Retrieved from The Vault: Electronic Theses and Dissertations. http://hdl.handle.net/11023/3041 Brown CE, Dyck RH (2002) Rapid, experience-dependent changes in levels of synaptic zinc in primary somatosensory cortex of the adult mouse. J Neurosci 22: 2617-2625. Brown CE, Dyck RH (2003) An improved method for visualizing the cell bodies of zincergic neurons. J Neurosci Methods 129: 41-47. Brown CE, Dyck RH (2004) Distribution of zincergic neurons in the mouse forebrain. J Comp Neurol 479: 156-167. Brown CE, Dyck RH (2005b) Retrograde tracing of the subset of afferent connections in mouse barrel cortex provided by zincergic neurons. J Comp Neurol 486: 48-60. Bruel-Jungerman E, Laroche S, Rampon C (2005) New neurons in the dentate gyrus are involved in the expression of enhanced long-term memory following environmental enrichment. Eur J Neurosci 21: 513-521. Bush AI, Pettingell WH, Multhaup G, d Paradis M, Vonsattel JP, Gusella JF, Beyreuther K, Masters CL, Tanzi RE (1994) Rapid induction of Alzheimer A beta amyloid formation by zinc. Science 265: 1464-1467. Cameron HA, Woolley CS, McEwen BS, Gould E (1993) Differentiation of newly born neurons and glia in the dentate gyrus of the adult rat. Neuroscience 56: 337-344. Carter RE, Aiba I, Dietz RM, Sheline C, Shuttleworth CW (2011) Spreading depression and related events are significant sources of neuronal Zn2+ release and accumulation. J Cereb Blood Flow Metab 31: 1073-1084. Casanovas-Aguilar C, Christensen MK, Reblet C, Martínez-García F, Pérez-Clausell J, Bueno-López JL (1995) Callosal neurones give rise to zinc-rich boutons in the rat visual cortex. Neuroreport 6: 497-500. Casanovas-Aguilar C, Reblet C, Pérez-Clausell J, Bueno-López JL (1998) Zinc-rich afferents to the rat neocortex: projections to the visual cortex traced with sodium selenite injections. J Chem Neuroanat 72: 97-108. Chácon-Fernández P, Säuberli K, Colzani M, Moreau T, Ghevaert C, Barde YA (2016) Brain-derived neurotrophic factor in megakaryocytes. J Biol Chem 291: 9872-9881. Chao HM, Choo PH, McEwen BS (1989) Glucocorticoid and mineralocorticoid receptor mRNA expression in rat brain. Neuroendocrinology 50: 365-371. Chen B, Dowlatshahi D, MacQueen GM, Wang JF, Young LT (2001) Increased hippocampal BDNF immunoreactivity in subjects treated with antidepressant medication. Biol REFERENCES 157

Psychiatry 50: 260-265. Chen CC, Huang CC, Hsu KS (2015) Chronic social stress affects synaptic maturation of newly generated neurons in the adult mouse dentate gyrus. Int J Neuropsychopharmacol 19: pyv097. Choo M, Miyazaki T, Yamazaki M, Kawamura M, Nakazawa T, Zhang J, Tanimura A, Uesaka N, Watanabe M, Sakimura K, Kano M (2017) Retrograde BDNF to TrkB signaling promotes synapse elimination in the developing cerebellum. Nat Commun 8: 195. Chorin E, Vinograd O, Fleidervish I, Gilad D, Herrmann S, Sekler I, Aizenman E, Hershfinkel M (2011) Upregulation of KCC2 activity by zinc-mediated neurotransmission via the mZnR/GPR39 receptor. J Neurosci 31: 12916-12926. Choi BY, Kim IY, Kim JH, Kho AR, Lee SH, Lee BE, Sohn M, Koh JY, Suh SW (2016) Zinc transporter 3 (ZnT3) gene deletion reduces spinal cord white matter damage and motor deficits in a murine MOG-induced multiple sclerosis model. Neurobiol Dis 94: 205-212. Choi MH, Na JE, Yoon YR, Lee HJ, Yoon S, Rhyu IJ, Baik JH (2017) Role of dopamine D2 receptor in stress-induced myelin loss. Sci Rep 7: 11654. Chou D, Huang CC, Hsu KS (2014) Brain-derived neurotrophic factor in the amygdala mediates susceptibility to fear conditioning. Exp Neurol 255: 19-29. Christensen MK, Frederickson CJ (1998) Zinc-containing afferent projections to the rat corticomedial amygdaloid complex: a retrograde tracing study. J Comp Neurol 400: 375-390. Christensen MK, Geneser FA (1995) Distrubtion of neurons of origin of zinc-containing projections in the amygdala of the rat. Anat Embryol 191: 227-237. Chrusch MJ (2015) The role of synaptic zinc in experience-dependent plasticity (Doctoral dissertation). Retrieved from The Vault: Electronic Theses and Dissertations. http://hdl.handle.net/11023/2369 Clelland CD, Choi M, Romberg C, Clemenson GD Jr, Fragniere A, Tyers P, Jessberger S, Saksida LM, Barker RA, Gage FH, Bussey TJ (2009) A functional role for adult hippocampal neurogenesis in pattern separation. Science 325: 210-213. Coddou C, Acuña-Castillo C, Bull P, Huidobro-Toro JP (2007) Dissecting the facilitator and inhibitor allosteric metal sites of the P2X4 receptor channel: critical roles of CYS132 for zinc potentiation and ASP138 for copper inhibition. J Biol Chem 282: 36879- REFERENCES 158

36886. Cole J, Costafreda SG, McGuffin P, Fu CH (2011) Hippocampal atrophy in first episode depression: a meta-analysis of magnetic resonance imaging studies. J Affect Disord 134: 483-487. Cole TB, Wenzel HJ, Kafer KE, Schwartzkroin PA, Palmiter RD (1999) Elimination of zinc from synaptic vesicles in the intact mouse brain by disruption of the ZnT3 gene. Proc Natl Acad Sci U S A 96: 1716-1721. Cole TB, Robbins CA, Wenzel HJ, Schwartzkroin PA, Palmiter RD (2000) Seizures and neuronal damage in mice lacking vesicular zinc. Epilepsy Research 39: 153-169. Cole TB, Martyanova A, Palmiter RD (2001) Removing zinc from synaptic vesicles does not impair spatial learning, memory, or sensorimotor functions in the mouse. Brain Res 891: 253-265. Colvin RA, Bush AI, Volitakis I, Fontaine CP, Thomas D, Kikuchi K, Holmes WR (2008) Insights into Zn2+ homeostasis in neurons from experimental and modeling studies. Am J Physiol Cell Physiol 294: C726-742. Colvin RA, Holmes WR, Fontaine CP, Maret W (2010) Cytosolic zinc buffering and muffling: their role in intracellular zinc homeostasis. Metallomics 2: 306-317. Conner JM, Lauterborn JC, Yan Q, Gall CM, Varon S (1997) Distribution of brain-derived neurotrophic factor (BDNF) protein and mRNA in the normal adult rat CNS: evidence for anterograde axonal transport. J Neurosci 17: 2295-2313. Conrad CD (2010) A critical review of chronic stress effects on spatial learning and memory. Prog Neuropsychopharmacol Biol Psychiatry 34: 742-755. Conrad CD, Galea LA, Kuroda Y, McEwen BS (1996) Chronic stress impairs rat spatial memory on the Y-maze, and this effect is blocked by tianeptine pretreatment. Behav Neurosci 110: 1321-1334. Conrad CD, Magariños AM, LeDoux JE, McEwen BS (1999) Repeated restraint stress facilitates fear conditioning independently of causing hippocampal CA3 dendritic atrophy. Behav Neurosci 113: 902-913. Conrad CD, Grote KA, Hobbs RJ, Ferayorni A (2003) Sex differences in spatial and non- spatial Y-maze performance after chronic stress. Neurobiol Learn Mem 79: 32-40. Cunningham MG, Ames HM, Christensen MK, Sorensen JC (2007) Zincergic innervation of medial prefrontal cortex by basolateral projection neurons. Neuroreport 18: 531-535. Cyprien F, Courtet P, Poulain V, Maller J, Meslin C, Bonafé A, Le Bars E, Ancelin ML, REFERENCES 159

Ritchie K, Artero S (2014) Corpus callosum size may predict late-life depression in women: a 10-year follow-up study. J Affect Disord 165: 16-23. Czéh B, Michaelis T, Watanabe T, Frahm J, de Biurrun G, van Kampen M, Bartolomucci A, Fuchs E (2001) Stress-induced changes in cerebral metabolites, hippocampal volume, and cell proliferation are prevented by antidepressant treatment with tianeptine. Proc Natl Acad Sci U S A 98: 12796-12801. Czéh B, Welt T, Fischer AK, Erhardt A, Schmitt W, Müller MB, Toschi N, Fuchs E, Keck ME (2002) Chronic psychosocial stress and concomitant repetitive transcranial magnetic stimulation: Effects on stress hormone levels and adult hippocampal neurogenesis. Biol Psychiatry 52: 1057-1064. Czupryn A, Skangiel-Kramska J (1997) Distribution of synaptic zinc in the developing mouse somatosensory barrel cortex. J Comp Neurol 386: 652-660. Dadomo H, Sanghez V, Di Cristo L, Lori A, Ceresini G, Malinge I, Parmigiani S, Palanza P, Sheardown M, Bartolomucci A (2011) Vulnerability to chronic subordination stress- induced depression-like disorders in adult 129SvEv male mice. Prog Neuropsychopharmacol Biol Psychiatry 35: 1461-1471. Danscher G (1981) Histochemical demonstration of heavy metals. A revised version of the sulphide silver method suitable for both light and electron microscopy. Histochemistry 71: 1-16. Danscher G (1982) Exogenous selenium in the brain. A histochemical technique for light and electron microscopical localization of catalytic bonds. Histochemistry 76: 281- 293. Danscher G, Jo SM, Varea E, Wang Z, Cole TB, Schrøder HD (2001) Inhibitory zinc- enriched terminals in mouse spinal cord. Neuroscience 105: 941-947. David DJ, Samuels BA, Rainer Q, Wang JW, Marsteller D, Mendez I, Drew M, Craig DA, Guiard BP, Guilloux JP, Artymyshyn RP, Gardier AM, Gerald C, Antonijevic IA, Leonardo ED, Hen R (2009) Neurogenesis-dependent and -independent effects of fluoxetine in an animal model of anxiety/depression. Neuron 62: 479-493. Dayer AG, Ford AA, Cleaver KM, Yassaee M, Cameron HA (2003) Short-term and long- term survival of new neurons in the rat dentate gyrus. J Comp Neurol 460: 563-572. De Miguel Z, Haditsch U, Palmer TD, Azpiroz A, Sapolsky RM (2018) Adult-generated neurons born during chronic social stress are uniquely adapted to respond to subsequent chronic social stress. Mol Psychiatry: in press. REFERENCES 160

Deshpande A, Kawai H, Metherate R, Glabe CG, Busciglio J (2009) A role for synaptic zinc in activity-dependent Aβ oligomer formation and accumulation at excitatory synapses. J Neurosci 29: 4004-4015. Dieni S, Matsumoto T, Dekkers M, Rauskolb S, Ionescu MS, Deogracias R, Gundelfinger ED, Kojima M, Nestel S, Frotscher M, Barde YA (2012) BDNF and its pro-peptide are stored in presynaptic dense core vesicles in brain neurons. J Cell Biol 196: 775- 788. Dillon DG, Gonenc A, Belleau E, Pizzagalli DA (2018) Depression is associated with dimensional and categorical effects on white matter pathways. Depress Anxiety 35: 440-447. Ding Q, Li H, Tian X, Shen Z, Wang X, Mo F, Huang J, Shen H (2016) Zinc and imipramine reverse the depression-like behavior in mice induced by chronic restraint stress. J Affect Disord 197: 100-106. Doering P, Stoltenberg M, Penkowa M, Rungby J, Larsen A, Danscher G (2010) Chemical blocking of zinc ions in CNS increases neuronal damage following traumatic brain injury (TBI) in mice. PLoS One 5: e0131. Dou X, Tian X, Zheng Y, Huang J, Shen Z, Li H, Wang X, Mo F, Wang W, Wang S, Shen H (2014) Psychological stress induced hippocampus zinc dyshomeostasis and depression-like behavior in rats. Behav Brain Res 273: 133-138. Draguhn A, Verdorn TA, Ewert M, Seeburg PH, Sakmann B (1990) Functional and

molecular distinction between recombinant rat GABAA receptor subtypes by Zn2+. Neuron 781-788. Dreixler JC, Leonard JP (1994) Subunit-specific enhancement of glutamate receptor responses by zinc. Mol Brain Res 22: 144-150. Dubreucq S, Kambire S, Conforzi M, Metna-Laurent M, Cannich A, Soria-Gomez E, Richard E, Marsicano G, Chaouloff F (2012) Cannabinoid type 1 receptors located on single-minded 1–expressing neurons control emotional behaviors. Neuroscience 204: 230-244. Dulka BN, Lynch JF 3rd, Latsko MS, Mulvany JL, Jasnow AM (2015) Phenotypic responses to social defeat are associated with differences in cued and contextual fear discrimination. Behav Processes 118: 115-122. Dreixler JC, Leonard JP (1994) Subunit-specific enhancement of glutamate receptor responses by zinc. Mol Brain Res 22: 144-150. REFERENCES 161

Dufner-Beattie J, Wang F, Kuo YM, Gitschier J, Eide D, Andrews GK (2003) The acrodermatitis enteropathica gene ZIP4 encodes a tissue-specific, zinc-regulated zinc transporter in mice. J Biol Chem 278: 33474-33481. Dyck R, Beaulieu C, Cynader M (1993) Histochemical localization of synaptic zinc in the developing cat visual cortex. J Comp Neurol 329: 53-67. Dyck RH, Chaudhuri A, Cynader MS (2003) Experience-dependent regulation of the zincergic innervation of visual cortex in adult monkeys. Cereb Cortex 13: 1094-1109. Eide FF, Vining ER, Eide BL, Zang K, Wang XY, Reichardt LF (1996) Naturally occurring truncated trkB receptors have dominant inhibitory effects on brain-derived neurotrophic factor signaling. J Neurosci 16: 3123-3129. Encinas JM, Vaahtokari A, Enikolopov G (2006) Fluoxetine targets early progenitor cells in the adult brain. Proc Natl Acad Sci U S A 103: 8233-8238. Evstratova A, Tóth K (2011) Synaptically evoked Ca2+ release from intracellular stores is not influenced by vesicular zinc in CA3 hippocampal pyramidal neurones. J Physiol 589: 5677-5689. Fawcett JP, Aloyz R, McLean JH, Pareek S, Miller FD, McPherson PS, Murphy RA (1997) Detection of brain-derived neurotrophic factor in a vesicular fraction of brain synaptosomes. J Biol Chem 272: 8837-8840. Ferragud A, Haro A, Sylvain A, Velázquez-Sánchez C, Hernández-Rabaza V, Canales JJ (2010) Enhanced habit-based learning and decreased neurogenesis in the adult hippocampus in a murine model of chronic social stress. Behav Brain Res 210: 134- 139. Franklin PA, Pullen RG, Hall GH (1992) Blood-brain exchange routes and distribution of 65Zn in rat brain. Neurochem Res 17: 767-771. Frederickson CJ (1989) Neurobiology of zinc and zinc-containing neurons. Int Rev Neurobiol 31: 145-238. Frederickson CJ, Howell GA, Haigh MD, Danscher G (1988) Zinc-containing fiber systems in the cochlear nuclei of the rat and mouse. Hear Res 36: 203-211. Frederickson CJ, Rampy BA, Reamy-Rampy S, Howell GA (1992) Distribution of histochemically reactive zinc in the forebrain of the rat. J Chem Neuroanat 5: 521- 530. Frederickson CJ, Giblin LJ, Krezel A, McAdoo DJ, Mueller RN, Zeng Y, Balaji RV, Masalha R, Thompson RB, Fierke CA, Sarvey JM, de Valdenebro M, Prough DS, Zornow MH REFERENCES 162

(2006a) Concentrations of extracellular free zinc (pZn)e in the central nervous system during simple anesthetization, ischemia and reperfusion. Exp Neurol 198: 285-293. Frederickson CJ, Giblin LJ 3rd, Balaji RV, Masalha R, Frederickson CJ, Zeng Y, Lopez EV, Koh JY, Chorin U, Besser L, Hershfinkel M, Li Y, Thompson RB, Krezel A (2006b) Synaptic release of zinc from brain slices: factors governing release, imaging, and accurate calculation of concentration. J Neurosci Methods 154: 19-29. Friedman B, Price JL (1984) Fiber systems in the olfactory bulb and cortex: a study in adult and developing rats, using the Timm method with the light and electron microscope. J Comp Neurol 223: 88-109. Friedlich AL, Lee JY, van Groen T, Cherny RA, Volitakis I, Cole TB, Palmiter RD, Koh JY, Bush AI (2004) Neuronal zinc exchange with the blood vessel wall promotes cerebral amyloid angiopathy in an animal model of Alzheimer’s disease. J Neurosci 24: 3453- 3459. Fryer RH, Kaplan DR, Feinstein SC, Radeke MJ, Grayson DR, Kromer LF (1996) Developmental and mature expression of full-length and truncated TrkB receptors in the rat forebrain. J Comp Neurol 374: 21-40. Fuchs E, Flügge G (2002) Social stress in tree shrews: Effects on physiology, brain function, and behavior of subordinate individuals. Pharmacol Biochem Behav 73: 247-258. Fuertig R, Azzinnari D, Bergamini G, Cathomas F, Sigrist H, Seifritz E, Vavassori S, Luippold A, Hengerer B, Ceci A, Pryce CR (2016) Mouse chronic social stress increases blood and brain kynurine pathway activity and fear behaviour: both effects are reversed by inhibition of indoleamine 2,3-dioxygenase. Brain Behav Immun 54: 59-72. Fukunaka A, Suzuki T, Kurokawa Y, Yamazaki T, Fujiwara N, Ishihara K, Migaki H, Okumura K, Masuda S, Yamaguchi-Iwai Y, Nagao M, Kambe T (2009) Demonstration and characterization of the heterodimerization of ZnT5 and ZnT6 in the early secretory pathway. J Biol Chem 284: 30798-30806. Garrett B, Slomianka L (1992) Postnatal development of zinc-containing cells and neuropil in the visual cortex of the mouse. Anat Embryol 186: 487-496. Garrett B, Geneser FA, Slomianka L (1991) Distribution of acetylcholinesterase and zinc in the visual cortex of the mouse. Anat Embryol 184: 461-468. Garrett B, Sørensen JC, Slomianka L (1992) Fluoro-Gold tracing of zinc-containing afferent connections in the mouse visual cortices. Anat Embryol 185: 451-459. REFERENCES 163

Garthe A, Roeder I, Kempermann G (2015) Mice in an enriched environment learn more flexibly because of adult hippocampal neurogenesis. Hippocampus 26: 261-271. Geiser J, De Lisle RC, Andrews GK (2013) The zinc transporter Zip5 (Slc39a5) regulates intestinal zinc excretion and protects the pancreas against zinc toxicity. PLoS One 8: e82149. Geiser J, Venken KJ, De Lisle RC, Andrews GK (2012) A mouse model of acrodermatitis enteropathica: loss of intestine zinc transporter ZIP4 (Slc39a4) disrupts the stem cell niche and intestine integrity. PLoS Genet 8: e1002766. Gilad D, Shorer S, Ketzef M, Friedman A, Sekler I, Aizenmann E, Hershfinkel M (2015) Homeostatic regulation of KCC2 activity by the zinc receptor mZnR/GPR39 during seizures. Neurobiol Dis 81: 4-13. Golan Y, Berman B, Assaraf YG (2015) Heterodimerization, altered subcellular localization, and function of multiple zinc transporters in viable cells using biochemical fluorescence complementation. J Biol Chem 290: 9050-9063. Golden SA, Covington HE 3rd, Berton O, Russo SJ (2011) A standardized protocol for repeated social defeat stress in mice. Nat Protoc 6: 1183-1191. Gould E, McEwen BS, Tanapat P, Galea LA, Fuchs E (1997) Neurogenesis in the dentate gyrus of the adult tress shrew is regulated by psychosocial stress and NMDA receptor activation. J Neurosci 17: 2492-2498. Grabrucker S, Jannetti L, Eckert M, Gaub S, Chhabra R, Pfaender S, Mangus K, Reddy PP, Rankovic V, Schmeisser MJ, Kreutz MR, Ehret G, Boeckers TM, Grabrucker AM (2014) Zinc deficiency dysregulates the synaptic ProSAP/Shank scaffold and might contribute to autism spectrum disorders. Brain 137: 137-152. Gray JM, Chaouloff F, Hill MN (2015) To stress or not to stress: a question of models. Curr Protoc Neurosci 70: 8.33.1-8.33.22. Gray K, Ellis V (2008) Activation of pro-BDNF by the pericellular serine protease plasmin. FEBS Lett 582: 907-910. Haenisch B, Bilkei-Gorzo A, Caron MG, Bönisch H (2009) Knockout of the norepinephrine transporter and pharmacologically diverse antidepressants prevent behavioral and brain neurotrophin alterations in two chronic stress models of depression. J Neurochem 111: 403-416. Harris AZ, Atsak P, Bretton ZH, Holt ES, Alam R, Morton MP, Abbas AI, Leonardo ED, Bolkon SS, Hen R, Gordon JA (2018) A novel method for chronic social defeat stress REFERENCES 164

in female mice. Neuropsychopharmacology 43: 1276-1283. Harward SC, Hedrick NG, Hall CE, Parra-Bueno P, Milner TA, Pan E, Laviv T, Hempstead BL, Yasuda Ryohei, McNamara JO (2016) Autocrine BDNF-TrkB signalling within a single dendritic spine. Nature 538: 99-103. Han MH, Nestler EJ (2017) Neural substrates of depression and resilience. Neurotherapeutics 14: 677-686. Haug FM (1967) Electron microscopical localization of the zinc in hippocampal mossy fibre synapses by a modified sulfide silver procedure. Histochemie 8: 355-368. Helgager J, Huang YZ, McNamara JO (2014) Brain-derived neurotrophic factor but not vesicular zinc promotes TrkB activation within mossy fibers of mouse hippocampus in vivo. J Comp Neurol 522: 3885-3899. Hershfinkel M, Moran A, Grossman N, Sekler I (2001) A zinc-sensing receptor triggers the release of intracellular Ca2+ and regulates ion transport. Proc Natl Acad Sci U S A 98: 11749-11754. Hershfinkel M, Kandler K, Knoch ME, Dagan-Rabin M, Aras MA, Abramovitch-Dahan C, Sekler I, Aizenman E (2009) Intracellular zinc inhibits KCC2 transporter activity. Nat Neurosci 12:725–727. Hildebrand MS, Phillips AM, Mullen SA, Adlard PA, Hardies K, Damiano JA, Wimmer V, Bellows ST, McMahon JM, Burgess R, Hendrickx Rik, Weckhuysen S, Suls A, de Jonghe P, Scheffer IE, Petrou S, Berkovic SF, Reid CA (2015) Loss of synaptic Zn2+ transporter function increases risk of febrile seizures. Sci Rep 5: 17816. Higashi Y, Aratake T, Shimizu S, Shimizu T, Nakamura K, Tsuda M, Yawata T, Ueba T, Saito M (2017) Influence of extracellular zinc on M1 microglial activation. Sci Rep 7: 43778. Hirzel K, Müller U, Latal AT, Hülsmann S, Grudzinska J, Seeliger MW, Betz H, Laube B (2006) Hyperekplexia phenotype of glycine receptor α1 subunit mutant mice identifies Zn2+ as an essential endogenous modulator of glycinergic neurotransmission. Neuron 52: 679-690. Holst B, Egerod KL, Schlid E, Vickers SP, Cheetham S, Gerlach LO, Storjohann L, Stidsen CE, Jones R, Beck-Sickinger AG, Schwartz TW (2007) GPR39 signaling is stimulated by zinc ions but not by obestatin. Endocrinology 148: 13-20.

Hosie AM, Dunne EL, Harvey RJ, Smart TG (2003) Zinc-mediated inhibition of GABAA receptors: discrete binding sites underlie subtype specificity. Nat Neurosci 6: 362- REFERENCES 165

369. Howell GA, Frederickson CJ (1990) Retrograde transport method for mapping zinc- containing fiber systems in the brain. Brain Res 515: 277-286. Howell GA, Welch MG, Frederickson CJ (1984) Stimulation-induced uptake and release of zinc in hippocampal slices. Nature 308: 736-738. Huang YZ, McNamara JO (2012) Neuroprotective effects of reactive oxygen species mediated by BDNF-independent activation of TrkB. J Neurosci 32: 15521-15532. Huang YZ, Pan E, Xiong ZQ, McNamara JO (2008) Zinc-mediated transactivation of TrkB potentiates the hippocampal mossy fiber-CA3 pyramid synapse. Neuron 57: 546-558. Huhman KL, Solomon MB, Janicki M, Harmon AC, Lin SM, Israel JE, Jasnow AM (2003) Conditioned defeat in male and female Syrian hamsters. Horm Behav 44: 293-299. Hwang JJ, Park M-H, Choi S-Y, Koh J-Y (2005) Activation of the Trk signaling pathway by extracellular zinc. J Biol Chem 280: 11995-12001. Hwang IY, Sun ES, An JH, Im H, Lee SH, Lee JY, Han PL, Koh JY, Kim YH (2011) Zinc- triggered induction of tissue plasminogen activator by brain-derived neurotrophic factor and metalloproteinase. J Neurochem 118: 855-863. Ibata Y, Otsuka N (1969) Electron microscopic demonstration of zinc in the hippocampal formation using Timm’s sulfide-silver technique. J Histochem Cytochem 17: 171-175. Ichinohe N, Rockland KS (2005a) Distribution of synaptic zinc in the macaque monkey amygdala. J Comp Neurol 489: 135-147. Ichinohe N, Rockland KS (2005b) Zinc-enriched amygdalo- and hippocampo-cortical connections to the inferotemporal cortices in macaque monkey. Neurosci Res 53: 57- 68. Impey S, Smith DM, Obrietan K, Donahue R, Wade C, Storm DR (1998) Stimulation of cAMP response element (CRE)-mediated transcription during contextual learning. Nat Neurosci 1: 595-601. Iñiguez SD, Flores-Ramirez FJ, Riggs LM, Alipio JB, Garcia-Carachure I, Hernandez MA, Sanchez DO, Lobo MK, Serrano PA, Braren SH, Castillo SA (2018) Vicarious social defeat stress induces depression-related outcomes in female mice. Biol Psychiatry 83: 9-17. Izumi Y, Auberson YP, Zorumski CF (2006) Zinc modulates bidirectional hippocampal plasticity by effects on the NMDA receptor. J Neurosci 7181-7188. Jacobs BL, van Praag H, Gage F (2000) Adult brain neurogenesis and psychiatry: a novel REFERENCES 166

theory of depression. Mol Psychiatry 5: 262-269. Jacobson L, Sapolsky R (1991) The role of the hippocampus in feedback regulation of the hypothalamic-pituitary-adrenal axis. Endocr Rev 12: 118-134. Jayasooriya AP, Ackland ML, Mathai ML, Sinclair AJ, Weisinger HS, Weisinger RS, Halver JE, Kitajka K, Puskás LG (2005) Perinatal ω-3 polyunsaturated fatty acid supply modifies brain zinc homeostasis during adulthood. Proc Natl Acad Sci U S A 102: 7133-7138. Jo SM, Danscher G, Schrøder HD, Won MH, Cole TB (2000a) Zinc-enriched (ZEN) terminals in mouse spinal cord: immunohistochemistry and autometallography. Brain Res 870: 163-169. Jo SM, Won MH, Cole TB, Jensen MS, Palmiter RD, Danscher G (2000) Zinc-enriched (ZEN) terminals in mouse olfactory bulb. Brain Res 865: 227-236. Jo SM, Kim YK, Wang Z, Danscher G (2002) Retrograde tracing of zinc-enriched (ZEN) neuronal somata projecting to the olfactory bulb. Brain Res 956: 230-235. Kalappa BI, Tzounopoulos T (2017) Context-dependent modulation of excitatory synaptic strength by synaptically released zinc. eNeuro 4: e0011-17.2017. Kalappa BI, Anderson CT, Goldberg JM, Lippard SJ, Tzounopoulos T (2015) AMPA receptor inhibition by synaptically released zinc. Proc Natl Acad Sci U S A 112: 15749-15754. Kaneko M, Noguchi T, Ikegami S, Sakurai T, Kakita A, Toyoshima Y, Kambe T, Yamada M, Inden M, Hara H, Oyanagi K, Inuzuka T, Takahashi H, Hozumi I (2015) Zinc transporters ZnT3 and ZnT6 are downregulated in the spinal cords of patients with sporadic amyotrophic lateral sclerosis. J Neurosci Res 93: 370-379. Kantheti P, Qiao X, Diaz ME, Peden AA, Meyer GE, Carskadon SL, Kapfhamer D, Sufalko D, Robinson MS, Noebels JL, Burmeister M (1998) Mutation in AP-3 delta in the mocha mouse links endosomal transport to storage deficiency in platelets, melanosomes, and synaptic vesicles. Neuron 21: 111-122. Kambe T, Tsuji T, Hashimoto A, Itsumura N (2015) The physiological, biochemical, and molecular roles of zinc transporters in zinc homeostasis and metabolism. Physiol Rev 95: 749-784. Kaplan MS, Hinds JW (1977) Neurogenesis in the adult rat: electron microscope analysis of light autoradiographs. Science 197: 1092-1094. Katoh-Semba R, Takeuchi IK, Semba R, Kato K (1997) Distribution of brain-derived REFERENCES 167

neurotrophic factor in rats and its changes with development in the brain. J Neurochem 69: 34-42. Katz RJ (1981) Animal model of depression: effects of electroconvulsive shock therapy. Neurosci Biobehav Rev 5: 273-277. Kay AR (2003) Evidence for chelatable zinc in the extracellular space of the hippocampus, but little evidence for synaptic release of Zn. J Neurosci 23: 6847-6855. Kay AR (2006) Imaging synaptic zinc: promises and perils. Trends Neurosci 29: 200-206. Kay AR, Tóth K (2006) Influence of location of a fluorescent zinc probe in brain slices on its response to synaptic activation. J Neurophysiol 95: 1949-1956. Kee N, Teixeira CM, Wang AH, Frankland PW (2007) Preferential incorporation of adult- generated granule cells into spatial memory networks in the dentate gyrus. Nat Neurosci 10: 355-362. Kempermann G, Kuhn HG, Gage FH (1997) More hippocampal neurons in adult mice living in an enriched environment. Nature 386: 493-495. Kempermann G, Brandon EP, Gage FH (1998) Environmental stimulation of 129/SvJ mice causes increased cell proliferation and neurogenesis in the adult dentate gyrus. Curr Biol 8: 939-942. Kendler KS, Gardner CO (2016) Depressive vulnerability, stressful life events and major depression: a longitudinal model. Psychol Med 46: 1865-1874. Kendler KS, Karkowski LM, Prescott CA (1999) Causal relationships between stressful life events and the onset of major depression. Am J Psychiatry 156: 837-841. Kendler KS, Hettema JM, Butera F, Gardner CO, Prescott CA (2003) Life event dimensions of loss, humiliation, entrapment, and danger in the prediction of onsets of major depression and generalized anxiety. Arch Gen Psychiatry 60: 789-796. Khalil R, Levitt JB (2013) Zinc histochemistry reveals circuit refinement and distinguishes visual areas in the developing ferret cerebral cortex. Brain Struct Funct 218: 1293- 1306. Klein R, Conway D, Parada LF, Barbacid M (1990) The trkB tyrosine protein kinase gene codes for a second neurogenic receptor that lacks the catalytic kinase domain. Cell 61: 647-656. Klein R, Nanduri V, Jing S, Lamballe F, Tapley P, Bryant S, Cordon-Cardo C, Jones KR, Reichardt LF, Barbacid M (1991) The trkB tyrosine protein kinase is a receptor for brain-derived neurotrophic factor and neurotrophin-3. Cell 66: 395-403. REFERENCES 168

Kodirov SA, Takizawa S, Joseph J, Kandel ER, Shumyatsky GP, Bolshakov VY (2006) Synaptically released zinc gates long-term potentiation in fear conditioning pathways. Proc Natl Acad Sci U S A 103: 15218-15223. Koh JY, Suh SW, Gwag BJ, He YY, Hsu CY, Choi DW (1996) The role of zinc in selective neuronal death after transient global cerebral ischemia. Science 272: 1013-6. Kohara K, Kitamura A, Morishima M, Tsumoto T (2001) Activity-dependent transfer of brain-derived neurotrophic factor to postsynaptic neurons. Science 291: 2419-2423. Kolbeck R, Bartke I, Eberle W, Barde YA (1999) Brain-derived neurotrophic factor levels in the nervous system of wild-type and neurotrophin gene mutant mice. J Neurochem 72: 1930-1938. Koo JW, Labonté B, Engmann O, Calipari ES, Juarez B, Lorsch Z, Walsh JJ, Friedman AK, Yorgason JT, Han MH, Nestler EJ (2016) Essential role of mesolimbic brain-derived neurotrophic factor in chronic social stress-induced depressive behaviors. Biol Psychiatry 80: 469-478. Krezel A, Maret W (2006) Zinc-buffering capacity of a eukaryotic cell at physiological pZn. J Biol Inorg Chem 11: 1049-1062. Krezel A, Maret W (2007) Dual nanomolar and picomolar Zn(II) binding properties of metallothionein. J Am Chem Soc 129: 10911-10921. Krishnan V et al. (2007) Molecular adaptations underlying susceptibility and resistance to social defeat in brain reward regions. Cell 131: 391-404. Kudryavtseva NN, Bakshtanovskaya IV, Koryakina LA (1991) Social model of depression in mice of C57BL/6J strain. Pharmacol Biochem Behav 38: 315-320. Kumar J, Chuang JC, Na ES, Kuperman A, Gillman AG, Mukherjee A, Zigman JM, McClung CA, Lutter M (2013) Differential effects of chronic social stress and fluoxetine on meal patterns in mice. Appetite 64: 81-88. Kumar S, Hultman R, Hughes D, Michel N, Katz BM, Dzirasa K (2014) Prefrontal cortex reactivity underlies trait vulnerability to chronic social defeat stress. Nat Commun 5: 4537. Lagace DC, Donovan MH, DeCarolis NA, Farnbauch LA, Malhotra S, Berton O, Nestler EJ, Krishnan V, Eisch AJ (2010) Adult hippocampal neurogenesis is functionally important for stress-induced social avoidance. Proc Natl Acad Sci U S A: 4436-4441. Laine MA, Trontti K, Misiewicz Z, Sokolowska E, Kulesskaya N, Heikkinen A, Saarnio S, Balcells I, Ameslon P, Greco D, Mattila P, Ellonen P, Paulin L, Auvinen P, Jokitalo REFERENCES 169

E, Hovatta I (2018) Genetic control of myelin plasticity after chronic psychosocial stress. eNeuro 5: ENEURO.0166-18.2018. Langmade SJ, Ravindra R, Daniels PJ, Andrews GK (2000) The transcription factor MTF-1 mediates metal regulation of the mouse ZnT1 gene. J Biol Chem 275: 34803-34809. Lavoie N, Jeyaraju DV, Peralta MR 3rd, Seress L, Pellegrini L, Tóth K (2011) Vesicular zinc regulates the Ca2+ sensitivity of a subpopulation of presynaptic vesicles at hippocampal mossy fiber terminals. J Neurosci 31: 18251-18265. Lee R, Kermani P, Teng KK, Hempstead BL (2001) Regulation of cell survival by secreted proneurotrophins. Science 294: 1945-1948. Lee JY, Mook-Jung I, Koh JY (1999) Histochemically reactive zinc in plaques of the Swedish mutant beta-amyloid precursor protein transgenic mice. J Neurosci 19: RC10. Lee JY, Cole TB, Palmiter RD, Koh JY (2000) Accumulation of zinc in degenerating hippocampal neurons of ZnT3-null mice after seizures: evidence against vesicular origin. J Neurosci 20: RC79. Lee JY, Kim JH, Hong SH, Lee JY, Cherny RA, Bush AI, Palmiter RD, Koh JY (2004) Estrogen decreases zinc transporter 3 expression and synaptic vesicle zinc levels in mouse brain. J Biol Chem 279: 8602-8607. Lee JY, Kim YJ, Kim TY, Koh JY, Kim YH (2008) Essential role for zinc-triggered p75NTR activation in preconditioning neuroprotection. J Neurosci 28: 10919-10927. Lee JY, Cho E, Kim TY, Kim DK, Palmiter RD, Volitakis I, Kim JS, Bush AI, Koh JY (2010) Apolipoprotein E ablation decreases synaptic vesicular zinc in the brain. Biometals 23: 1085-1095. Lee JY, Kim JS, Byun HR, Palmiter RD, Koh JY (2011) Dependence of the histofluorescently reactive zinc pool on zinc transporter-3 in the normal brain. Brain Res 1418: 12-22. Lee JY, Cho E, Seo JW, Hwang JJ, Koh JY (2012) Alteration of the cerebral zinc pool in a mouse model of Alzheimer’s disease. J Neuropathol Exp Neurol 71: 211-222. Lehmann ML, Brachman RA, Martinowich K, Schloesser RJ, Herkenham M (2013) Glucocorticoids orchestrate divergent effects on mood through adult neurogenesis. J Neurosci 33: 2961-2972. Li Y, Maret W (2008) Human metallothionein metallomics. J Anal At Spectrom 23: 1055- 1062. REFERENCES 170

Li Y, Hough CJ, Frederickson CJ, Sarvey JM (2001a) Induction of mossy fiber→CA3 long- term potentiation requires translocation of synaptically released Zn2+. J Neurosci 21: 8015-8025. Li Y, Hough CJ, Suh SW, Sarvey JM, Frederickson CJ (2001b) Rapid translocation of Zn(2+) from presynaptic terminals into postsynaptic hippocampal neurons after physiological stimulation. J Neurophysiol 86: 2597-2604. Li Y, Andereggen L, Yuki K, Omura K, Yin Y, Gilbert HY, Erdogan B, Asdourian MS, Shrock C, de Lima S, Apfel UP, Zhuo Y, Hershfinkel M, Lippard SJ, Rosenberg PA, Benowitz L (2017) Mobile zinc increases rapidly in the retina after optic nerve injury and regulates ganglion cell survival and optic nerve regeneration. Proc Natl Acad Sci U S A 114: E209-218. Lin DD, Cohen AS, Coulter DA (2001) Zinc-induced augmentation of excitatory synaptic currents and glutamate receptor responses in hippocampal CA3 neurons. J Neurophysiol 85: 1185-1196. Linkous DH, Flinn JM, Koh JY, Lanzirotti A, Bertch PM, Jones BF, Giblin LJ, Frederickson CJ (2008) Evidence that the ZnT3 protein controls the total amount of elemental zinc in synaptic vesicles. J Histochem Cytochem 56: 3-6. Lisboa SF, Niraula A, Resstel LB, Guimaraes FS, Godbout JP, Sheridan JF (2018) Repeated social defeat-induced neuroinflammation, anxiety-like behavior and resistance to fear extinction were attenuated by the cannabinoid receptor agonist WIN55,212-2. Neuropsychopharmacology 43: 1924-1933. Liu X, Surprenant A, Mao H-J, Roger S, Xia R, Bradley H, Jiang L-H (2008) Identification

of key residues coordinating functional inhibition of P2X7 receptors by zinc and copper. Mol Pharmacol 73: 252-259. Long Y, Frederickson CJ (1994) A zinc-containing fiber system of thalamic origin. Neuroreport 5: 2026-2028. Long Y, Hardwick AL, Frederickson CJ (1995) Zinc-containing innervation of the subicular region in the rat. Neurochem Int 27: 95-103. Lopantsev V, Wenzel HJ, Cole TB, Palmiter RD, Schwartzkroin PA (2003) Lack of vesicular zinc in mossy fibers does not affect synaptic excitability of CA3 pyramidal cells in zinc transporter 3 knockout mice. Neuroscience 116: 237-248. Lorca RA, Rozas C, Loyola S, Moreira-Ramos S, Zeise ML, Kirkwood A, Huidobro-Toro JP, Morales B (2011) Zinc enhances long-term potentiation through P2X receptor REFERENCES 171

modulation in the hippocampal CA1 region. Eur J Neurosci 33: 1175-1185. Lu M, Fu D (2007) Structure of the zinc transporter YiiP. Science 317: 1746-1748. Lu M, Chai J, Fu D (2009) Structural basis for autoregulation of the zinc transporter YiiP. Nat Struct Mol Biol 16: 1063-1067. Lu YM, Taverna FA, Tu R, Ackerley CA, Wang YT, Roder J (2000) Endogenous Zn2+ is required for the induction of long-term potentiation at rat hippocampal mossy fiber- CA3 synapses. Synapse 38: 187-197. Lucassen PJ, Stumpel MW, Wang Q, Aronica E (2010) Decreased number of progenitor cells but no response to antidepressant drugs in the hippocampus of elderly depressed patients. Neuropharmacology 58: 940-949. Luine V, Villegas M, Martinez C, McEwen BS (1994) Repeated stress causes reversible impairments of spatial memory impairments. Brain Res 639: 167-170. Lutter M, Sakata I, Osborne-Lawrence SO, Rovinsky SA, Anderson JG, Jung S, Birnbaum S, Yanagisawa M, Elmquist JK, Nestler EJ, Zigman JM (2008) The orexigenic hormone ghrelin defends against depressive symptoms of chronic stress. Nat Neurosci 11: 752-753. Lynch JW (2004) Molecular structure and function of the glycine receptor chloride channel. Physiol Rev 84: 1051-1095. Ma M, Ren Q, Yang C, Zhang JC, Yao W, Dong C, Ohgi Y, Futamura T, Hashimoto K (2016) Adjunctive treatment of brexpiprazole with fluoxetine shows a rapid antidepressant effect in social defeat stress model: role of BDNF-TrkB signaling. Sci Rep 6: 39209. MacMaster FP, Carrey N, Langevin LM (2013) Corpus callosal morphology in early onset adolescent depression. J Affect Disord 145: 256-259. Magariños AM, McEwen BS (1995) Stress-induced atrophy of apical dendrites of hippocampal CA3c neurons: involvement of glucocorticoid secretion and excitatory amino acid receptors. Neuroscience 69: 89-98. Magariños AM, McEwen BS, Flügge G, Fuchs E (1996) Chronic psychosocial stress causes apical dendritic atrophy of hippocampal CA3 pyramidal neurons in subordinate tree shrews. J Neurosci 16: 3534-3540. Malberg JE, Eisch AJ, Nestler EJ, Duman RS (2000) Chronic antidepressant treatment increases neurogenesis in adult rat hippocampus. J Neurosci 20: 9104-9110. Malenka RC, Bear MF (2004) LTP and LTD: an embarrassment of riches. Neuron 44: 5-21. REFERENCES 172

Mandava P, Howell GA, Frederickson CJ (1993) Zinc-containing neuronal innervation of the septal nuclei. Brain Res 608: 115-122. Maret W (2014a) Molecular aspects of zinc signals. In: Zing signals in cellular functions and disorders (Fukada T, Kambe T, eds), pp7-26. Japan: Springer. Maret W (2014b) Analyzing free zinc(II) ion concentrations in cell biology with fluorescent chelating molecules. Metallomics 7: 202-211. Maret W, Sandstead HH (2006) Zinc requirements and the risks and benefits of zinc supplementation. J Trace Elem Med Biol 20: 3-18. Maret W, Vallee BL (1998) Thiolate ligands in metallothionein confer redox activity on zinc clusters. Proc Natl Acad Sci U S A 95: 3478-3482. Martel G, Hevi C, Friebely O, Baybutt T, Shumyatsky GP (2010) Zinc transporter 3 is involved in learned fear and extinction, but not in innate fear. Learn Mem 17: 582- 590. Martel G, Hevi C, Kane-Goldsmith N, Shumyatsky GP (2011) Zinc transporter ZnT3 is involved in memory dependent on the hippocampus and perirhinal cortex. Behav Brain Res 223: 233-238. Maske H (1955) Über den topochemischen nachweis von zink im Ammonshorn verschiedener säugetiere. Naturwissenschaften 42: 424. Masters BA, Quaife CJ, Erickson JC, Kelly EJ, Froelick GJ, Zambrowicz BP, Brinster RL, Palmiter RD (1994) Metallothionein III is expressed in neurons that sequester zinc in synaptic vesicles. J Neurosci 14: 5844-5857. Matsumoto T, Rauskolb S, Polack M, Klose J, Kolbeck R, Korte M, Barde YA (2008) Biosynthesis and processing of endogenous BDNF: CNS neurons store and secrete BDNF, not pro-BDNF. Nat Neurosci 11: 131-133. Maycox PR, Kelly F, Taylor A, Bates S, Reid J, Logendra R, Barnes MR, Larminie C, Jones N, Lennon M, Davies C, Hagan JJ, Scorer CA, Angelinetta C, Akbar MT, Hirsch S, Mortimer AM, Barnes TR, de Belleroche J (2009) Analysis of gene expression in two large schizophrenia cohorts identifies multiple changes associated with nerve terminal function. Mol Psychiatry 14:1083-94. McAllister BB, Kiryanova V, Dyck RH (2012) Behavioural outcomes of perinatal maternal fluoxetine treatment. Neuroscience 226: 356-366. McAllister BB, Spanswick SC, Patel PP, Barneto AA, Dyck RH (2015) The effects of chronic fluoxetine treatment following injury of medial frontal cortex in mice. Behav Brain REFERENCES 173

Res 290: 102-116. McAllister BB, Dyck RH (2017) Zinc transporter 3 (ZnT3) and vesicular zinc in central nervous system function. Neurosci Biobehav Rev 80: 329-350. McAllister BB, Bihelek N, Dyck RH (2018) Brain-derived neurotrophic factor and TrkB levels in mice that lack vesicular zinc: effects of age and sex. BioRxiv preprint. https://doi.org/10.1101/421453 McAllister BB, Wright DK, Wortman RC, Shultz SR, Dyck RH (2018) Elimination of vesicular zinc alters the behavioural and neuroanatomical effects of social defeat stress in mice. Neurobiol Stress 9: 199-213. McEwen BS, Weiss J, Schwartz L (1968) Selective retention of corticosterone by limbic structures in rat brain. Nature 220: 911-912. McEwen BS, Bowles NP, Gray JD, Hill MN, Hunter RG, Karatsoreos IN, Nasca C (2015) Mechanisms of stress in the brain. Nat Neurosci 18: 1353-1363. McKim DB, Niraula A, Tarr AJ, Wohleb ES, Sheridan JF, Godbout JP (2016) Neuroinflammatory dynamics underlie memory impairments after repeated social defeat. J Neurosci 36: 2590-2604. McKim DB, Weber MD, Niraula A, Sawicki CM, Liu X, Jarrett BL, Ramirez-Chan K, Wang Y, Roeth RM, Sucaldito AD, Sobol CG, Quan N, Sheridan JF, Godbout JP (2018) Microglial recruitment of IL-1β-producing monocytes to brain endothelium causes stress-induced anxiety. Mol Psychiatry 23: 1421-1431. McKinnon MC, Yucel K, Nazarov A, MacQueen G (2009) A meta-analysis examining clinical predictors of hippocampal volume in patients with major depressive disorder. J Psychiatry Neurosci 34: 41-54. McMahon RJ, Cousins RJ (1998) Mammalian zinc transporters. J Nutr 128: 667-670. Meduri JD, Farnbauch LA, Jasnow AM (2013) Paradoxical enhancement of fear expression and extinction deficits in mice resilient to social defeat. Behav Brain Res 256: 580- 590. Mengual E, Casanovas-Aguilar C, Pérez-Clausell J, Giménez-Amaya JM (1995) Heterogeneous and compartmental distribution of zinc in the striatum and globus pallidus of the rat. Neuroscience 66: 523-537. Mengual E, Casanovas-Aguilar C, Pérez-Clausell J, Giménez-Amaya JM (2001) Thalamic distribution of zinc-rich terminal fields and neurons of origin in the rat. Neuroscience 102: 863-844. REFERENCES 174

Mellone M, Pelucchi S, Alberti L, Genazzani AA, Di Luca M, Gardoni F (2015) Zinc transporter-1: a novel NMDA receptor-binding protein at the postsynaptic density. J Neurochem 132: 159-168. Meshi D, Drew MR, Saxe M, Ansorge MS, David D, Santarelli L, Malapani C, Moore H, Hen R (2006) Hippocampal neurogenesis is not required for behavioral effects of environmental enrichment. Nat Neurosci 6: 729-731. Michael GJ, Averill S, Nitkunan A, Rattray M, Bennett DL, Yan Q, Priestly JV (1997) Nerve growth factor treatment increases brain-derived neurotrophic factor selectively in TrkA-expressing dorsal root ganglion cells and in their central terminations within the spinal cord. J Neurosci 17: 8476-8490. Miró-Bernié N, Sancho-Bielsa FJ, López-Garcia C, Pérez-Clausell J (2003) Retrograde transport of sodium selenite and intracellular injection of micro-ruby: a combined method to describe the morphology of zinc-rich neurones. J Neurosci Methods 127: 199-209. Mitra R, Sundlass K, Parker KJ, Schatzberg AF, Lyons DM (2006) Social stress-related behavior affects hippocampal cell proliferation in mice. Physiol Behav 89: 123-127. Miyata S, Taniguchi M, Koyama Y, Shimizu S, Tanaka T, Yasuno F, Yamamoto A, Iida H, Kudo T, Katayama T, Tohyama M (2016) Association between chronic stress-induced structural abnormalities in Ranvier nodes and reduced oligodendrocyte activity in major depression. Sci Rep 6: 23084. Mizukawa K, Tseng IM, Otsuka N (1989) Quantitative electron microscopic analysis of postnatal development of zinc-positive nerve endings in the rat amygdala using Timm’s sulphide silver technique. Dev Brain Res 50: 197-203. Molnár P, Nadler JV (2001) Synaptically-released zinc inhibits N-methyl-D-aspartate receptor activation at recurrent mossy fiber synapses. Brain Res 910: 205-207. Mott DD, Benveniste M, Dingledine RJ (2008) pH-Dependent inhibition of kainate receptors by zinc. J Neurosci 28: 1659-1671. Mowla SJ, Farhadi HF, Pareek S, Atwal JK, Morris SJ, Seidah NG, Murphy RA (2001) Biosynthesis and post-translational processing of the precursor to brain-derived neurotrophic factor. J Biol Chem 276: 12660-12666. Nagappan G, Zaitsev E, Senatorov VV, Yang J, Hempstead BL, Lu B (2009) Control of extracellular cleavage of proBDNF by high frequency neuronal activity. Proc Natl Acad Sci U S A 106: 1267-1272. REFERENCES 175

Nakashima AS, Oddo S, Laferla FM, Dyck RH (2010) Experience-dependent regulation of vesicular zinc in male and female 3xTg-AD mice. Neurobiol Aging 31: 605-613. Nakashima AS, Butt RH, Dyck RH (2011) Alterations in protein and gene expression within the barrel cortices of ZnT3 knockout mice: experience-independent and dependent changes. Neurochem Int 59: 860-870. Nibuya M, Morinobu S, Duman RS (1995) Regulation of BDNF and trkB mRNA in rat brain by chronic electroconvulsive seizure and antidepressant treatments. J Neurosci 15: 7539-7547. Nibuya M, Nestler EJ, Duman RS (1996) Chronic antidepressant administration increases the expression of cAMP response element binding protein (CREB) in rat hippocampus. J Neurosci 16: 2365-2372. Nozaki C, Vergnano AM, Filliol D, Ouagazzal AM, Le Goff A, Carvalho S, Reiss D, Gaveriaux-Ruff C, Neyton J, Paoletti P, Kieffer BL (2011) Zinc alleviates pain through high-affinity binding to the NMDA receptor NR2A subunit. Nat Neurosci 14: 1017-1022. Nydegger I, Rumschik SM, Kay AR (2010) Zinc is externalized rather than released during synaptic transmission. ACS Chem Neurosci 1: 728-736. Nydegger I, Rumschik SM, Zhao J, Kay AR (2012) Evidence for an extracellular zinc-veneer in rodent brains from experiments with Zn-ionophores and ZnT3 knockouts. ACS Chem Neurosci 3: 761-766. Ohana E, Hoch E, Keasar C, Kambe T, Yifrach O, Hershfinkel M, Sekler I (2009) Identification of the Zn2+ binding site and mode of operation of a mammalian Zn2+ transporter. J Biol Chem 284: 17677-17686. Oleson RH, Hyde TM, Kleinman JE, Smidt K, Rungby J, Larsen A (2016) Obesity and age- related alterations in the gene expression of zinc-transporter proteins in the human brain. Transl Psychiatry 6: e838. Palmiter RD, Cole TB, Quaife CJ, Findley SD (1996) ZnT-3, a putative transporter of zinc into synaptic vesicles. Proc Natl Acad Sci U S A 93: 14934-14939. Pan E, Zhang XA, Huang Z, Krezel A, Zhao M, Tinberg CE, Lippard SJ, McNamara JO (2011) Vesicular zinc promotes presynaptic and inhibits postsynaptic long-term potentiation of mossy fiber-CA3 synapse. Neuron 71: 1116-1126. Pang PT, Teng HK, Zaitsev E, Woo NT, Sakata K, Zhen S, Teng KK, Yung WH, Hempstead BL, Lu B (2004) Cleavage of proBDNF by tPA/plasmin is essential for long-term REFERENCES 176

hippocampal plasticity. Science 306: 487-491. Paoletti P, Vergnano AM, Barbour B, Casado M (2009) Zinc at glutamatergic synapses. Neuroscience 158: 126-136. Perez-Becerril C, Morris AG, Mortimer A, McKenna PJ, de Belleroche J (2013) Allelic variants in the zinc transporter-3 gene, SLC30A3, a candidate gene identified from gene expression studies, show gender-specific association with schizophrenia. Eur Psychiatry 29: 172-178. Perez-Becerril C, Morris AG, Mortimer A, McKenna PJ, de Bellaroche J (2016) Common variants in the chromosome 2p23 region containing the SLC30A3 (ZnT3) gene are associated with schizophrenia in female but not male individuals in a large collection of European samples. Psychiatry Res 246: 335-340. Pérez-Clausell J (1996) Distribution of terminal fields stained for zinc in the neocortex of the rat. J Chem Neuroanat 11: 99-111. Pérez-Clausell J, Danscher G (1985) Intravesicular localization of zinc in rat telencephalic boutons. A histochemical study. Brain Res 337: 91-98. Pérez-Clausell J, Frederickson CJ, Danscher G (1989) Amygdaloid efferents through the stria terminalis in the rat give origin to zinc-containing boutons. J Comp Neurol 290: 201-212. Perez-Rosello T, Anderson CT, Schopher FJ, Zhao Y, Gilad D, Salvatore SR, Freeman BA, Hershfinkel M, Aizenman E, Tzounopoulos (2013) Synaptic Zn2+ inhibits neurotransmitter release by promoting endocannabinoid synthesis. J Neurosci 33: 9259-9272. Perez-Rosello T, Anderson CT, Ling C, Lippard SJ, Tzounopoulos T (2015) Tonic zinc inhibits spontaneous firing in dorsal cochlear nucleus principal neurons by enhancing glycinergic neurotransmission. Neurobiol Dis 81: 14-19. Peters S, Koh J, Choi DW (1987) Zinc selectively blocks the action of N-Methyl-D-Aspartate on cortical neurons. Science 236: 589-593. Pham K, Nacher J, Hof PR, McEwen BS (2003) Repeated restraint stress suppresses neurogenesis and induces biphasic PSA-NCAM expression in the adult rat dentate gyrus. Eur J Neurosci 17: 879-886. Poddar R, Rajagopal S, Shuttleworth CW, Paul S (2016) Zn2+-dependent activation of the Trk signaling pathway induces phosphorylation of the brain-enriched tyrosine phosphatase STEP. J Biol Chem 291: 813-825. REFERENCES 177

Prasad AS, Halsted JA, Nadimi M (1961) Syndrome of iron deficiency anemia, hepatosplenomegaly, hypogonadism, dwarfism and geophagia. Am J Med 31: 532- 546. Polacchini A, Metelli G, Francavilla R, Baj G, Florean M, Mascaretti LG, Tongiorgi E (2015) A method for reproducible measurements of serum BDNF: comparison of the performance of six commercial assays. Sci Rep 5: 17989. Post JI, Eibl JK, Ross GM (2008) Zinc induces motor neuron death via a selective inhibition of brain-derived neurotrophic factor activity. Amyotroph Lateral Scler 9: 149-155. Ranjbar E, Shams J, Sabetkasaei M, M-Shirazi M, Rashidkhani B, Mostafavi A, Bornak E, Nasrollahzadeh J (2014) Effects of zinc supplementation on efficacy of antidepressant therapy, inflammatory cytokines, and brain-derived neurotrophic factor in patients with major depression. Nutr Neurosci 17: 65-71. Rassendren FA, Lory P, Pin JP, Nargeot J (1990) Zinc has opposite effects on NMDA and non-NMDA receptors expressed in Xenopus oocytes. Neuron 4: 733-740. Razzoli M, Carboni L, Andreoli M, Michielin F, Ballottari A, Arban R (2011) Strain-specific outcomes of repeated social defeat and chronic fluoxetine treatment in the mouse. Pharmacol Biochem Behav 97: 566-576. Rebola N, Lujan R, Cunha RA, Mulle C (2008) Adenosine A2A receptors are essential for long-term potentiation of NMDA-EPSCs at hippocampal mossy fiber synapses. Neuron 57: 121-134. Redenti S, Chappell RL (2004) Localization of zinc transporter-3 (ZnT-3) in mouse retina. Vision Res 44: 3317-3321. Redenti S, Chappell RL (2005) Neuroimaging of zinc released by depolarization of rat retinal cells. Vision Res 45: 3520-3525. Reichardt LF (2006) Neurotrophin-regulated signalling pathways. Philos Trans R Soc Lond B Biol Sci 361: 1545-1564. Rose CR, Blum R, Pichler B, Lepier A, Kafitz KW, Konnerth A (2003) Truncated TrkB-T1 mediates neurotrophin-evoked calcium signaling in glia cells. Nature 426: 74-78. Ross GM, Shamovsky IL, Lawrance G, Solc M, Dostaler SM, Jimmo SL, Weaver DF, Riopelle RJ (1997) Zinc alters conformation and inhibits biological activities of nerve growth factor and related neurotrophins. Nat Med 3: 872-878. Rossi C, Angelucci A, Constantin L, Braschi C, Mazzantini M, Babbini F, Fabbri ME, Tessarollo L, Maffei L, Berardi N, Caleo M (2006) Brain-derived neurotrophic factor REFERENCES 178

(BDNF) is required for the enhancement of hippocampal neurogenesis following environmental enrichment. Eur J Neurosci 24: 1850-1856.

Ruiz A, Walker MC, Fabian-Fine R, Kullmann DM (2004) Endogenous zinc inhibits GABAA receptors in a hippocampal pathway. J Neurophysiol 91: 1091-1096. Saarelainen T, Hendolin P, Lucas G, Koponen E, Sairanen M, MacDonald E, Agerman K, Haapasalo A, Nawa H, Aloyz R, Ernfors P, Castrén E (2003) Activation of the TrkB receptor is induced by antidepressant drugs and is required for antidepressant- induced behavioral effects. J Neurosci 23: 349-357. Safiulina VF, Fattorini G, Conti F, Cherubini E (2006) GABAergic signaling at mossy fiber synapses in neonatal rat hippocampus. J Neurosci 26: 597-608.s Sahay A, Scobie KN, Hill AS, O’Carroll CM, Kheirbek MA, Burghardt NS, Fenton AA, Dranovsky A, Hen R (2011) Increasing adult hippocampal neurogenesis is sufficient to improve pattern separation. Nature 472: 466-470. Sairanen M, Lucas G, Ernfors P, Castrén M, Castrén E (2005) Brain-derived neurotrophic factor and antidepressant drugs have different but coordinated effects on neuronal turnover, proliferation, and survival in the adult dentate gyrus. J Neurosci 25: 1089- 1094. Salazar G, Love R, Werner E, Doucette MM, Cheng S, Levey A, Faundez V (2004a) The zinc transporter ZnT3 interacts with AP-3 and it is preferentially targeted to a distinct synaptic vesicle subpopulation. Mol Biol Cell 15: 575-597. Salazar G, Love R, Styers ML, Werner E, Peden A, Rodriguez S, Gearing M, Wainer B, Faundez V (2004b) AP-3-dependent mechanisms control the targeting of a chloride channel (ClC-3) in neuronal and non-neuronal cells. J Biol Chem 279: 25430-25439. Salazar G, Craige B, Love R, Kalman D, Faundez V (2005) Vglut1 and ZnT3 co-targeting mechanisms regulate vesicular zinc stores in PC12 cells. J Cell Sci 118: 1911-1921. Salazar G, Falcon-Perez JM, Harrison R, Faundez V (2009) SLC30A3 (ZnT3) oligomerization by dityrosine bonds regulates its subcellular localization and metal transport capacity. PloS One 4: e5896. Samuels BA, Hen R (2011) Novelty-suppressed feeding in the mouse, in: Gould TD (Ed.), Mood and Anxiety Related Phenotypes in Mice: Characterization Using Behavioral Tests, Volume II, Neuromethods, vol. 63. Springer Science+Business Media, New York, pp. 107-121. Sanchez-Andres JV, Palop JJ, Ramirez C, Nacher J, Molowny A, Lopez-Garcia C (1997) REFERENCES 179

Zinc-positive presynaptic boutons of the rabbit hippocampus during early postnatal development. Dev Brain Res 103: 171-183. Santarelli L, Saxe M, Gross C, Surget A, Battaglia F, Dulawa S, Weisstaub N, Lee J, Duman RS, Arancio O, Belzung C, Hen R (2003) Requirement of hippocampal neurogenesis for the behavioral effects of antidepressants. Science 301: 805-809. Schloesser RJ, Lehmann M, Martinowich K, Manji HK, Herkenham M (2010) Environmental enrichment requires adult neurogenesis to facilitate recovery from psychosocial stress. Mol Psychiatry 15: 1152-1163. Schrøder HD, Danscher G, Jo SM, Su H (2000) Zinc-enriched boutons in rat spinal cord. Brain Res 868: 119-122. Sela H, Cohen H, Karpas Z, Zeiri Y (2017) Distinctive hippocampal zinc distribution patterns following stress exposure in an animal model of PTSD. Metallomics 9: 323- 333. Selye H (1936) A syndrome produced by diverse nocuous agents. Nature 138: 32. Selye H (1976) The stress of life (revised ed). New York: McGraw-Hill. Sensi SL, Paoletti P, Koh JY, Aizenman E, Bush AI, Hershfinkel M (2011) The neurophysiology and pathology of brain zinc. J Neurosci 31: 16076-16085. Sheerin CM, Lind MJ, Brown EA, Gardner CO, Kendler KS, Amstadter AB (2018) The impact of resilience and subsequent stressful life events on MDD and GAD. Depress Anxiety 35: 140-147. Sheline YI, Wang PW, Gado MH, Csernansky JG, Vannier MW (1996) Hippocampal atrophy in recurrent major depression. Proc Natl Acad Sci U S A 93: 3908-3913. Shultz SR, Cardamone L, Liu YR, Hogan RE, Maccotta L, Wright DK, Zheng P, Koe A, Gregoire MC, Williams JP, Hicks RJ, Jones NC, Myers DE, O’Brien TJ, Bouilleret V (2013) Can structural or functional changes following traumatic brain injury in the rat predict epileptic outcome? Epilepsia 54: 1240-1250. Sial OK, Warren BL, Alcantara LF, Parise EM, Bolaños-Guzmán CA (2016) Vicarious social defeat stress: bridging the gap between physical and emotional stress. J Neurosci Meth 258: 94-103. Silhol M, Bonnichon V, Rage F, Tapia-Arancibia L (2005) Age-related changes in brain- derived neurotrophic factor and tyrosine kinase receptor isoforms in the hippocampus and hypothalamus in male rats. Neuroscience 132: 613-624. Sindreu CB, Varoqui H, Erickson JD, Pérez-Clausell JP (2003) Boutons containing REFERENCES 180

vesicular zinc define a subpopulation of synapses with low AMPAR content in rat hippocampus. Cereb Cortex 13: 823-829. Sindreu C, Palmiter RD, Storm DR (2011) Zinc transporter ZnT-3 regulates presynaptic Erk1/2 signaling and hippocampus-dependent memory. Proc Natl Acad Sci U S A 108: 3366-3370. Sindreu C, Bayés A, Altafaj X, Pérez-Clausell J (2014) Zinc transporter-1 concentrates at the postsynaptic density of hippocampal synapses. Mol Brain 7:16. Siwek M, Dudek D, Paul IA, Sowa-Kućma M, Zięba A, Popik P, Pilc A, Nowak G (2009) Zinc supplementation augments efficacy of imipramine in treatment resistant patients: a double-blind, placebo-controlled study. J Affect Disord 118: 187-195. Slomianka L, Danscher G, Frederickson CJ (1990) Labeling the neurons of origin of zinc- containing pathways by intraperitoneal injections of sodium selenite. Neuroscience 38: 843-854. Slomianka L (1992) Neurons of origin of zinc-containing pathways and the distribution of zinc-containing boutons in the hippocampal region of the rat. Neuroscience 48: 325- 352. Slomianka L, Geneser FA (1997) Postnatal development of zinc-containing cells and neuropil in the hippocampal region of the mouse. Hippocampus 7: 321-340. Smart TG, Constanti A (1983) Pre- and postsynaptic effects of zinc on in vitro prepyriform neurones. Neurosci Lett 40: 205-211.

Smart TG, Moss SJ, Xie X, Huganir RL (1991) GABAA receptors are differentially sensitive to zinc: dependence on subunit composition. Br J Pharmacol 103: 1837-1839. Smart TG, Hosie AM, Miller PS (2004) Zn2+ ions: modulators of excitatory and inhibitory synaptic activity. Neuroscientist 10: 432-442. Smidt K, Rungby J (2012) ZnT3: a zinc transporter active in several organs. Biometals 25: 1-8. Smidt K, Jessen N, Petersen AB, Larsen A, Magnusson N, Jeppesen JB, Stoltenberg M, Culvenor JG, Tsatsanis A, Brock B, Schmitz O, Wogensen L, Bush AI, Rungby J (2009) SLC30A3 responds to glucose- and zinc variations in beta-cells and is critical for insulin production and in vivo glucose-metabolism during beta-cell stress. PloS One 4: e5684. Snyder JS, Kee N, Wojtowicz JM (2001) Effects of adult neurogenesis on synaptic plasticity in the rat dentate gyrus. J Neurophysiol 85: 2423-2431. REFERENCES 181

Sørensen JC, Tønder N, Slomianka (1993) Zinc-positive afferents to the rat septum originate from distinct subpopulations of zinc-containing neurons in the hippocampal areas and layers. Anat Embryol 188: 107-115. Sørensen JC, Slomianka L, Christensen J, Zimmer J (1995) Zinc-containing telencephalic connections to the rat striatum: a combined Fluoro-Gold tracing and histochemical study. Exp Brain Res 105: 370-382. Sorrells SF, Paredes MF, Cebrian-Silla A, Sandoval K, Qi D, Kelley KW, James D, Mayer S, Chang J, Auguste KI, Chang EF, Gutierrez AJ, Kriegstein AR, Mathern GW, Oldham MC, Huang EJ, Garcia-Verdugo JM, Yang Z, Alvarez-Buylla A (2018) Human hippocampal neurogenesis drops sharply in children to undetectable levels in adults. Nature 555: 377-381. Sousa N, Lukoyanov NV, Madeira MD, Almeida OFX, Paula-Barbosa MM (2000) Reorganization of the morphology of hippocampal neurites and synapses after stress- induced damage correlates with behavioural improvement. Neuroscience 97: 253- 266. Stoltenberg M, Nejsum LN, Larsen A, Danscher G (2004) Abundance of zinc ions in synaptic terminals of mocha mutant mice: zinc transporter 3 immunohistochemistry and zinc sulphide autometallography. J Mol Histol 35: 141-145. Stoltenberg M, Bush AI, Bach G, Smidt K, Larsen A, Rungby J, Lund S, Doering P, Danscher G (2007) Amyloid plaques arise from zinc-enriched cortical layers in APP/PS1 trangenic mice and are paradoxically enlarged with dietary zinc deficiency. Neuroscience 150: 357-369. Suh SW, Hamby AM, Gum ET, Shin BS, Won SJ, Sheline CT, Chan PH, Swanson RA (2008) Sequential release of nitric oxide, zinc, and superoxide in hypoglycemic neuronal death. J Cereb Blood Flow Metab 28: 1697-1706. Suh SW, Won SJ, Mamby AM, Yoo BH, Fan Y, Sheline CT, Tamano H, Takeda A, Liu J (2009) Decreased brain zinc availability reduces hippocampal neurogenesis in mice and rats. J Cereb Blood Flow Metab 29: 1579-1588. Surget A, Saxe M, Leman S, Ibarguen-Vargas Y, Chalon S, Griebel G, Hen R, Belzung C (2008) Drug-dependent requirement of hippocampal neurogenesis in a model of depression and of antidepressant reversal. Biol Psychiatry 64: 293-301. Surget A, Tanti A, Leonardo ED, Laugeray A, Rainer Q, Touma C, Palme R, Griebel G, Ibarguen-Vargas Y, Hen R, Belzung C (2011) Antidepressants recruit new neurons REFERENCES 182

to improve stress response regulation. Mol Psychiatry 16: 1177-1188. Suvrathan A, Bennur S, Ghosh S, Tomar A, Anilkumar S, Chattarji Sumantra (2014) Stress enhances fear by forming new synapses with greater capacity for long-term potentiation in the amygdala. Phil Trans R Soc Lond B Biol Sci 369: 20130151. Takahashi A, Chung JR, Zhang S, Zhang H, Grossman Y, Aleyasin H, Flanigan ME, Pfau ML, Menard C, Dumitriu D, Hodes GE, McEwen BS, Nestler EJ, Han MH, Russo SJ (2017) Establishment of a repeated social defeat model in female mice. Sci Rep 7: 12838. Takeda A, Akiyama T, Sawashita J, Okada S (1994) Brain uptake of trace metals, zinc and manganese, in rats. Brain Res 640: 341-344. Takeda A, Kawai M, Okada S (1997) Zinc distribution in the brain of Nagase analbuminemic rat and enlargement of the ventricular system. Brain Res 769: 193- 195. Takeda A, Hirate M, Tamano H, Nisibaba D, Oku N (2003) Susceptibility to kainate- induced seizures under dietary zinc deficiency. J Neurochem 85: 1575-1580. Takeda A, Tamano H, Kan F, Itoh H, Oku N (2007) Anxiety-like behavior of young rats after 2-week zinc deprivation. Behav Brain Res 177: 1-6. Takeda A, Fuke S, Ando M, Oku N (2009) Positive modulation of long-term potentiation at hippocampal CA1 synapses by low micromolar concentrations of zinc. Neuroscience 158: 585-591. Takeda A, Tamano H, Ogawa T, Takada S, Ando M, Oku N, Watanabe M (2012) Significance of serum glucocorticoid and chelatable zinc in depression and cognition in zinc deficiency. Behav Brain Res 226: 259-264. Takeda A, Suzuki M, Tempaku M, Ohashi K, Tamano H (2015) Influx of extracellular Zn2+ into hippocampal CA1 neurons is required for cognitive performance via long-term potentiation. Neuroscience 304: 209-216. Tanapat P, Hastings NB, Rydel TA, Galea LA, Gould E (2001) Exposure to fox odor inhibits cells proliferation in the hippocampus of adult rats via an adrenal hormone- dependent mechanism. J Comp Neurol 437: 496-504. Tao X, West AE, Chen WG, Corfas G, Greenberg ME (2002) A calcium-responsive transcription factor, CaRF, that regulates neuronal activity-dependent expression of BDNF. Neuron 33: 383-395. Tao L, Zheng Y, Shen Z, Li Y, Tian X, Dou X, Qian J, Shen H (2013) Psychological stress- REFERENCES 183

induced lower serum zinc and zinc redistribution in rats. Biol Trace Elem Res 155: 65-71. Thackray SE, McAllister BB, Dyck RH (2017) Behavioral characterization of female zinc transporter 3 (ZnT3) knockout mice. Behav Brain Res 321: 36-49. Timm F (1958) Zur histochemie der schwermetalle. Das sulfid-silber-verfahren. Dtsch Z Ges Gerichtl Med 46: 706-711. Tornatzky W, Miczek KA (1994) Behavioral and autonomic responses to intermittent social stress: different protection by clonidine and metoprolol. Psychopharmacology 116: 346-356. Travaglia A, La Mendola D, Magrí A, Pietropaolo A, Nicoletti VG, Grasso G, Malgieri G, Fattorusso R, Isernia C, Rizzarelli E (2013) Zinc(II) interactions with brain-derived neurotrophic factor N-terminal peptide fragments: inorganic features and biological perspectives. Inorg Chem 52: 11075-11083. Tsankova NM, Berton O, Renthal W, Kumar A, Neve RL, Nestler EJ (2006) Sustained hippocampal chromatin regulation in a mouse model of depression and antidepressant action. Nat Neurosci 9: 519-525. Ueno S, Tsukamoto M, Hirano T, Kikuchi H, Yamada MK, Nishiyama N, Nagano T, Matsuki N, Ikegaya Y (2002) Mossy fiber Zn2+ spillover modulates heterosynaptic N- methyl-D-aspartate receptor activity in hippocampal CA3 circuits. J Cell Biol 158: 215-220. Ugarte M, Osborne NN (1998) The localization of endogenous zinc and the in vitro effect of exogenous zinc on the GABA immunoreactivity and formation of reactive oxygen species in the retina. Gen Pharmac 30: 297-303. Ugarte M, Osborne NN (1999) The localization of free zinc varies in rat photoreceptors during light and dark adaptation. Exp Eye Res 69: 459-461. Valente T, Auladell C (2002) Developmental expression of ZnT3 in mouse brain: correlation between the vesicular zinc transporter protein and chelatable vesicular zinc (CVZ) cells. Glial and neuronal CVZ cells interact. Mol Cell Neurosci 21: 189-204. Valente T, Auladell C, Pérez-Clausell J (2002) Postnatal development of zinc-rich terminal fields in the brain of the rat. Exp Neurol 174: 215-229. Vallee BL, Auld DS (1990) Zinc coordination, function, and structure of zinc enzymes and other proteins. Biochemistry 29: 5647-5659. Van Bokhoven P, Oomen CA, Hoogendijk WJ, Smit AB, Lucassen PJ, Spijker S (2011) REFERENCES 184

Reduction in hippocampal neurogenesis after social defeat is long-lasting and responsive to late antidepressant treatment. Eur J Neurosci 33: 1833-1840. van Praag H, Kempermann G, Gage FH (1999) Running increases cell proliferation and neurogenesis in the adult mouse dentate gyrus. Nat Neurosci 2: 266-270. van Praag H, Schinder AF, Christie BR, Toni N, Palmer TD, Gage FH (2002) Functional neurogenesis in the adult hippocampus. Nature 415: 1030-1034. Venzala E, García-García AL, Elizalde N, Delagrange P, Tordera RM (2012) Chronic social defeat stress model: behavioural features, antidepressant action, and interaction with biological risk factors. Psychopharmacology 224: 313-325. Veran J, Kumar J, Pinheiro PS, Athané A, Mayer ML, Perrais D, Mulle C (2012) Zinc potentiates GluK3 receptor function by stabilizing the ligand binding domain dimer interface. Neuron 76: 565-578. Vergnano AM, Rebola N, Savtchenko LP, Pinheiro PS, Casado M, Kieffer BL, Rusakov DA, Mulle C, Paoletti P (2014) Zinc dynamics and action at excitatory synapses. Neuron 82: 1104-1114. Vialou V, et al. (2010) ΔFosB in brain reward circuits mediates resilience to stress and antidepressant responses. Nat Neurosci 13: 745-752. Vialou V, Thibault M, Kaska S, Cooper S, Gajewski P, Eagle A, Mazei-Robison M, Nestler EJ, Robison AJ (2015) Differential induction of FosB isoforms throughout the brain by fluoxetine and chronic stress. Neuropharmacology 99: 28-37. Vincent SR, Semba K (1989) A heavy metal marker of the developing striatal mosaic. Dev Brain Res 45: 155-159. Vogt K, Mellor J, Tong G, Nicoll R (2000) The actions of synaptically released zinc at hippocampal mossy fiber synapses. Neuron 26: 187-196. Yasuda S, Miyazaki T, Munechika K, Yamashita M, Ikeda Y, Kamizono A (2007) Isolation of Zn2+ as an endogenous agonist of GPR39 from fetal bovine serum. J Recept Signal Transduct Res 27: 235-246. Walker AK, Rivera PD, Wang Q, Chuang JC, Tran S, Osborne-Lawrence S, Estill SJ, Starwalt R, Huntington P, Morlock L, Naidoo J, Williams NS, Ready JM, Eisch AJ, Pieper AA, Zigman JM (2015) The P7C3 class of neuroprotective compounds exerts antidepressant efficacy in mice by increasing hippocampal neurogenesis. Mol Psychiatry 20: 500-508. Wang XD, Chen Y, Wolf M, Wagner KV, Liebl C, Scharf SH, Harbich D, Mayer B, Wurst W, REFERENCES 185

Holsboer F, Deussing JM, Baram TZ, Müller MB, Schmidt MV (2011) Forebrain CRHR1 deficiency attenuates chronic stress-induced cognitive deficits and dendritic remodeling. Neurobiol Dis 42: 300-310. Wang XB, Bozdagi O, Nikitczuk JS, Zhai ZW, Zhou Q, Huntley GW (2008) Extracellular proteolysis by matrix metalloproteinase-9 drives dendritic spine enlargement and long-term potentiation coordinately. Proc Natl Acad Sci U S A 105: 19520-19525. Wang ZY, Dahlström A (2008) Axonal transport of zinc transporter 3 and zinc containing organelles in the rodent adrenergic system. Neurochem Res 33: 2472-2479. Wang Z, Li JY, Dahlstrӧm A, Danscher G (2001) Zinc-enriched GABAergic terminals in mouse spinal cord. Brain Res 91: 165-172. Wang Z, Danscher G, Kim YK, Dahlstrom A, Jo SM (2002) Inhibitory zinc-enriched terminals in the mouse cerebellum: double-immunohistochemistry for zinc transporter 3 and glutamate decarboxylase. Neurosci Lett 321: 37-40. Wang ZY, Stoltenberg M, Huang L, Danscher G, Dahlstrӧm A, Shi Y, Li JY (2005) Abundant expression of zinc transporters in Bergmann glia of mouse cerebellum. Brain Res Bull 64: 441-448. Wang JW, David DJ, Monckton JE, Battaglia F, Hen R (2008) Chronic fluoxetine stimulates maturation and synaptic plasticity of adult-born hippocampal granule cells. J Neurosci 28: 1374-1384. Wang X, Wang ZY, Gao HL, Danscher G, Huang L (2006) Localization of ZnT7 and zinc ions in mouse retina–immunohistochemistry and selenium autometallography. Brain Res Bull 71: 91-96. Watanabe Y, Gould E, McEwen BS (1992) Stress induces atrophy of apical dendrites of hippocampal CA3 pyramidal neurons. Brain Res 588: 341-345. Watanabe M, Tamano H, Kikuchi T, Takeda A (2010) Susceptibility to stress in young rats after 2-week zinc deprivation. Neurochem Int 56: 410-416. Wenzel HJ, Cole TB, Born DE, Schwartzkroin PA, Palmiter RD (1997) Ultrastructural localization of zinc transporter-3 (ZnT-3) to synaptic vesicle membranes within mossy fiber boutons in the hippocampus of mouse and monkey. Proc Natl Acad Sci U S A 94: 12676-12681. Westbrook GL, Mayer ML (1987) Micromolar concentrations of Zn2+ antagonize NMDA and GABA responses of hippocampal neurons. Nature 328: 640-643. Whitfield DR, Vallortigara J, Alghamdi A, Howlett D, Hortobágyi T, Johnson M, Attems J, REFERENCES 186

Newhouse S, Ballard C, Thomas AJ, O’Brien JT, Aarsland D, Francis PT (2014) Assessment of ZnT3 and PSD95 protein levels in Lewy body dementias and Alzheimer’s disease: association with cognitive impairment. Neurobiol Aging 35: 2836-2844. Whittle N, Lubec G, Singewald N (2009) Zinc deficiency induces enhanced depression-like behaviour and altered limbic activation reversed by antidepressant treatment in mice. Amino Acids 36: 147-158. Willner P, Towell A, Sampson D, Sophokleous S, Muscat R (1987) Reduction of the sucrose preference by chronic unpredictable mild stress, and its restoration by a tricyclic antidepressant. Psychopharmacology 93: 358-364. Wohleb ES, Hanke ML, Corona AW, Powell ND, Stiner LM, Bailey MT, Nelson RJ, Godbout JP, Sheridan JF (2011) β-Adrenergic receptor antagonism prevents anxiety-like behavior and microglial reactivity induced by repeated social defeat. J Neurosci 31: 6277-6288. Wohleb ES, McKim DB, Shea DT, Powell ND, Tarr AJ, Sheridan JF, Godbout JP (2014) Re- establishment of anxiety in stress-sensitized mice is caused by monocyte trafficking from the spleen to the brain. Biol Psychiatry 75: 970-981. World Health Organization (2017) Depression and other common mental health disorders: global health estimates. (WHO Reference No. WHO/MSD/MER/2017.2) Geneva: WHO Document Production Services. Wright DK, O’Brien TJ, Shultz SR, Mychasiuk R (2017) Sex matters: repetitive mild traumatic brain injury in adolescent rats. Ann Clin Transl Neurol 4: 640-654. Wright DK, O’Brien TJ, Mychasiuk R, Shultz SR (2018) Telomere length and advanced diffusion MRI as biomarkers for repetitive mild traumatic brain injury in adolescent rats. Neuroimage Clin 18: 315-324. Wong YH, Lee CM, Xie W, Cui B, Poo MM (2015) Activity-dependent BDNF release via endocytic pathways is regulated by synaptotagmin-6 and complexin. Proc Natl Acad Sci U S A 112: E4475-4484. Wu HP, Dyck RH (2018) Signaling by synaptic zinc is required for whisker-mediated, fine texture discrimination. Neuroscience 369: 242-247. Xie X, Smart TG (1994) Modulation of long-term potentiation in rat hippocampal pyramidal neurons by zinc. Plüfgers Arch 427: 481-486. Yang J, Siao CJ, Nagappan G, Marinic T, Jing D, McGrath K, Chen ZY, Mark W, Tessarollo REFERENCES 187

L, Lee FS, Lu B, Hempstead BL (2009) Neuronal release of proBDNF. Nat Neurosci 12: 113-115. Yoo MH, Kim TY, Yoon YH, Koh JY (2016) Autism phenotype in ZnT3 null mice: involvement of zinc dyshomeostasis, MMP-9 activation and BDNF upregulation. Sci Rep 6: 28548. Zajac MS, Pang TY, Wong N, Weinrich B, Leang LS, Craig JM, Saffery R, Hannan AJ (2010) Wheel running and environmental enrichment differentially modify exon- specific BDNF expression in the hippocampus of wild-type and pre-motor symptomatic male and female Huntington’s disease mice. Hippocampus 20: 621-636. Zhang DQ, Ribelayga C, Mangel SC, McMahon DG (2002) Suppression by zinc of AMPA receptor-mediated synaptic transmission in the retina. J Neurophysiol 88: 1245- 1251. Zhang LH, Wang X, Stoltenberg M, Danscher G, Huang L, Wang ZY (2008) Abundant expression of zinc transporters in the amyloid plaques of Alzheimer’s disease brain. Brain Res Bull 77: 55-60. Zhang LH, Wang X, Zheng ZH, Ren H, Stoltenberg M, Danscher G, Huang L, Rong M, Wang ZY (2010) Altered expression and distribution of zinc transporters in APP/PS1 transgenic mouse brain. Neurobiol Aging 31: 74-87. Zheng W, Wang T, Yu D, Feng WY, Nie YX, Stoltenberg M, Danscher G, Wang ZY (2010) Elevation of zinc transporter ZnT3 protein in the cerebellar cortex of the AβPP/PS1 transgenic mouse. J Alzheimers Dis 20: 323-331. Zhou N, Wang CH, Zhang S, Wu DC (2013) The GLRA1 missense mutation W170S associates lack of Zn2+ potentiation with human hyperekplexia. J Neurosci 33: 17675-17681. Zimmer J, Haug FM (1978) Laminar differentiation of the hippocampus, fascia dentata and subiculum in developing rats, observed with the Timm sulphide silver method. J Comp Neurol 179: 581-618. APPENDIX A 188

APPENDIX A: PARTITIONED-HOUSING A.1 Partitioned-Housing When using RSD as a stressor, it is recommended that control mice be housed under the same conditions as the defeated animals (Golden et al., 2011). That is, controls should be housed in pairs and separated by a partition, as are the defeated mice, but – unlike the defeated mice – without daily episodes of direct physical interaction with their cagemate. After 10 days, the control mice are single-housed for the remainder of the experiment. While this approach provides a more rigorous control group than does using standard- housed controls, one drawback is that partitioned housing followed by isolated housing can impact some of the same depression- and anxiety-like behaviours as social defeat itself (Venzala et al., 2012), possibly due to stress associated with partitioned/isolated housing. Because the purpose of the present thesis was to compare stressed mice to mice in which stress was minimized as much as possible, we chose to keep our control mice in group housing throughout the experiments. The downside to this approach is that it cannot be concluded whether any observed effects of the stress procedure are caused by social defeat itself, or by the housing conditions associated with social defeat. Therefore, to assess whether 10 days of partitioned-housing, followed by isolated housing, could affect behaviour in similar ways to social defeat, we also generated partitioned-housing (PH) groups. WT (n = 13) and ZnT3 KO mice (n = 12) were pair-housed with a same-strain, non-littermate conspecific for 10 days in large cages, with the two mice separated by a partition that prevented them from physically interacting with one another. The mice were briefly handled each day. After 10 days, the mice were single-housed for the remainder of the experiment. These mice were assessed in the social interaction and NSF tests (see sections 2.2.4.1 and 2.2.4.3 for methods), following the same schedule described in Figure 2.1. Their performance was then compared to group-housed control mice and stressed mice, using the data collected from our larger behavioural test battery described in Chapter 2. The results were analyzed by two-way ANOVA, with genotype (WT vs. ZnT3 KO) and experimental treatment (group-housing vs. partitioned-housing vs. stress) as factors. Significant interactions were followed-up using Bonferroni-corrected simple-effects tests.

A.2 Results and Discussion We first examined social interaction behaviour with a novel, same-strain conspecific APPENDIX A 189

(Figure A1). For time spent in the interaction zone, there was a significant stress × genotype interaction [F(2,105) = 3.70, p = .028]. Amongst the WT groups, the stressed mice exhibited less interaction time than the group-housed controls (p < .001; Bonferroni- corrected) and PH controls (p < .001; Bonferroni-corrected), but amongst the ZnT3 KO groups, the stressed mice did not differ from either the group-housed controls (p = .118) or PH controls (p = .375). For time spent in the corners, there was also a significant stress × genotype interaction [F(2,105) = 4.33, p = .016]. Amongst the WT groups, the stressed mice exhibited more corner time than the group-housed controls (p < .001; Bonferroni-corrected) and PH controls (p < .001; Bonferroni-corrected), but in the ZnT3 KO groups, the stressed mice did not differ from either the group-housed controls (p = .070) or PH controls (p = .500). Together, these results indicate that stressed WT mice were socially avoidant of a novel conspecific, relative to both control groups, whereas stressed ZnT3 KO mice were not socially avoidant of the conspecific. We next examined social interaction behaviour with a novel, aggressive CD-1 mouse (Figure A2). For time spent in the interaction zone, there was a significant main effect of stress [F(2,105) = 14.50, p < .001], with the stressed mice exhibiting less interaction time than the group-housed controls (p < .001) and PH controls (p < .001). For time spent in the corners, there was also a significant main effect of stress [F(2,105) = 24.33, p < .001], with the stressed mice exhibiting more corner time than the group-house controls (p < .001) and PH controls (p < .001). These results indicate that, regardless of genotype, the stressed mice exhibited social avoidance of the CD-1 mouse, relative to both control groups. Finally, we examined anxiety-like behaviour, as inferred by the latency to feed in the NSF test (Figure A3). There was a significant main effect of stress [F(2,81) = 21.55, p < .001], with the stressed mice taking longer to feed than the group-housed controls (p < .001) and PH controls (p < .001). This indicates that, regardless of genotype, the stressed mice exhibited increased anxiety-like behaviour relative to both control groups. In summary, for both the social interaction and NSF tests, the stressed mice differed from the PH controls in the exact same manner as they differed from the group-housed controls. This indicates that the behavioural effects observed in the stressed mice were indeed caused by the experience of social defeat, and not the partitioned-housing conditions that accompanied social defeat.

APPENDIX A 190

FIGURES

1 5 0 * c

i *

f i

o c

1 0 0

)

(

n 5 0

n I 0

l s l s o H s o H s tr P e tr P e n T tr n O tr o W s o K s c T c T O W O K W K

1 5 0

B c

i *

f i

c * e

p s

n 1 0 0

)

(

i t

5 0

o C

l s l s o H s o H s tr P e tr P e n T tr n O tr o W s o K s c T c T O W O K W K

Figure A1. Interaction with a novel conspecific exhibited by partition-housed controls (“PH”), relative to group-housed mice (“control”) and mice subjected to RSD stress. A. Stress decreased interaction time in WT mice relative to group-housed or PH controls but did not affect interaction time in ZnT3 KO mice. B. Stress increased corner time in WT mice relative to both control groups but did not affect corner time in ZnT3 KO mice. Error bars represent 95% CIs. *follow-up test to significant interaction, p < .05 APPENDIX A 191

A # 1 5 0 * *

* * 1

)

s 1 0 0

(

i t

c 5 0

n I

l s l s o H s o H s tr P e tr P e n T tr n O tr o W s o K s c T c T O W O K W K

B # 1 8 0 * * * *

- 1 2 0

)

(

e 6 0

o C

l s l s o H s o H s tr P e tr P e n T tr n O tr o W s o K s c T c T O W O K W K

Figure A2. Social interaction with a novel CD-1 mouse in partition-housed controls (“PH”), relative to group-housed controls (“control”) and mice stressed by repeated social defeat stress. A. In both WT and ZnT3 KO mice, stress decreased interaction time relative to PH and group-housed mice. B. Regardless of genotype, RSD stress increased corner time relative to both control groups. Error bars represent 95% CIs. #main effect of stress; *p < .05 APPENDIX A 192

# * * * *

6 0 0

)

(

d 4 0 0

n e

t 2 0 0

a L

0 l l H s s ro s ro H s t P e t P e n T tr n O tr o s o K s c W T c T O W O K W K

Figure A3. Behaviour in the novelty-suppressed feeding test exhibited by partition-housed controls (“PH”), relative to group-housed controls (“control”) and mice subjected to repeated social defeat stress. In both WT and ZnT3 KO mice, stress increased the latency to begin feeding relative to both group-housed controls and PH controls. Error bars represent 95% CIs. #main effect of stress; *p < .05 APPENDIX B 193

APPENDIX B: COPYRIGHT PERMISSIONS APPENDIX B 194

APPENDIX B 195

APPENDIX B 196