Chronic Stress, Neurotransmitter Plasticity, and Body Weight

Transcription

1

2 Chronic Stress, Neurotransmitter Plasticity, and Body Weight A dissertation submitted to the Division of Research and Advanced Sciences of the University of Cincinnati In partial fulfillment of the requirement for the degree of Doctor of Philosophy In the Graduate Program in Neuroscience of the College of Medicine December 2011 By Jonathan N. Flak B.S. University of Michigan, 2004 Committee Chair: James P. Herman, Ph.D. 1

3 General Abstract Chronic stress exposure is associated with the manifestation of many diseases and disorders, which may be due in part to changes in neural structure and function via the activation/ recruitment of neural circuits altering synaptology, morphology, and gene expression. However, the neural dysfunction associated with chronic stress- related pathology is believed to be plastic and returns to normal following a combination of sufficient recovery, coping, pharmaceuticals, and/or therapy. Thus, the changes in neural organization by chronic stress and factors that alter how chronic stress can regulate neuronal structure/function are integral to our understanding of chronic stress- related pathology. Thus, the studies within this dissertation attempt to further our understanding of both the effects and regulatory factors in chronic stress- related pathology. The data in chapter 2 demonstrated that chronic stress increases the number of excitatory neurotransmitter inputs to mppvn CRH neurons, the cells that trigger HPA axis response to stress. The experiment in chapter 3 identified several regions known to project into the PVN as being recruited by unpredictable chronic stress, but not a predictable regimen. The study in chapter 4 indicated that the PVN- projecting noradrenergic neurons are necessary for appropriate responses to acute, but not chronic stress. Collectively, the studies included in chapter 5 demonstrated that changes in body weight homeostasis are sufficient to alter the physiological indices of chronic stress, perhaps via opposing regulation of the PVN. These investigations have laid the groundwork for subsequent analyses, progressing toward deciphering the neural circuits that regulate responses to chronic stress and how these changes contribute to the manifestation of disorders such as depression and PTSD. 2

4 3

5 Acknowledgements I would like to thank Jim for the opportunity to train within his lab. He afforded me the essential opportunities to grow as a scientist both within the laboratory and at meetings. Although he did not think it was the best decision at the time, I truly appreciate the freedom that he gave me to explore projects that sparked ideas I hope to investigate with future experiments. He also allowed me to learn from my own mistakes, which I truly believe is the best method to grow in science. I cannot thank the whole Herman lab enough. It was a tremendous, collaborative environment. Due to the existing teamwork, we were able to complete several extremely large and challenging experiments, that I am still amazed we were able to accomplish. I especially have to thank Susan and her skill of jello caking. Thanks Ben, Kenny, Mark, Jen, Annette, Ryan, Anne, Rong, Amy, Dennis, Eduardo, Brent, Jessica, Miyuki, Michelle, Helmer, Yve, Renu, Naomi, and Dayna! Of the numerous memories that I will look back to with nostalgia, I have to single out the mentoring I graciously received from both Eric and Tia. We commonly had stimulating, scientific discussions that have challenged and formed my scientific thinking. I have no doubt that they will both be tremendous mentors someday. I hope that I can someday repay a fraction of what I owe them. 4

6 I would also like to thank Randall for his support. I think of him as my unofficial second mentor. He provided me with many opportunities to meet investigators, even though I was a member of another lab and he did not have to. I have to also thank the Sakai lab. These days, it is really difficult to differentiate between the Herman and Sakai labs. Not only members, but equipment also wanders from lab to lab. Special thanks to Mike and Karen. I would like to thank the rest of my committee. Thanks Stephen, Steve, and Neil for taking time out of your busy schedules. I would also like to thank the whole Metabolic Diseases Institute (aka Genome Research Institute) and the many investigators and trainees I have had the opportunity to interact with through the years. It is an exceptional community to train as a scientist, especially in neuroendocrinology. In addition to scientific help, Karen and Amanda also provided me with moral support in times of need. Thanks! Lastly, I would like to thank the UC Neuroscience Program. 5

7 Table of Contents General Abstract... 2 Acknowledgements... 4 Table of Contents... 6 List of Figures and Tables... 8 Abbreviations Chapter 1: General Introduction and Background Stress History of Stress Research The Physiological Stress Responses Glucocorticoid Action Chronic Stress Exposure Stress Neurocircuitry Stress Neurocircuitry: Upstream Limbic Structures Stress Neurocircuitry: PVN Stress Neurocircuitry: PVN- projecting Regions Metabolic Modification of Chronic Stress Exposure Dissertation Chapter 2: Chronic Stress- Induced Neurotransmitter Plasticity Abstract Introduction Materials and Methods Results Discussion Tables Figures Chapter 3: Identification of Chronic Stress- Recruited Circuits Using ΔFosB Abstract Introduction Materials and Methods Results Discussion

8 Table Figures Chapter 4: PVN- Projecting Noradrenergic Neurons Are Not Necessary for Chronic Stress HPA Facilitation Abstract Introduction Materials and Methods Results Discussion Tables Figures Chapter 5: Opposing Effects of Chronic Stress and Weight Restriction on Cardiovascular, Metabolic and Neuroendocrine Function Abstract Introduction Materials and Methods Results Discussion Tables Figures Chapter 6: General Discussion References

9 List of Figures and Tables Chapter Table Table Figure Figure Figure Figure Figure Figure Chapter Table Figure Figure Figure Figure Figure Figure Chapter Table Table Figure Figure Figure Figure Figure Figure Chapter Table Table Figure Figure

10 Figure Figure Figure Figure

11 Abbreviations ACTH: Adrenocortropic Hormone ANOVA: Analysis of Variance ARC: Arcuate Nucleus of the Hypothalamus AUC: Area under the Curve AVP: Arginine Vasopressin BDNF: Brain Derived Neurotrophic Factor BSA: Bovine Serum Albumin BST: Bed Nucleus of the Stria Terminalis camp: cyclic Adenosine Monophosphate CeA: Central Nucleus of the Amygdala CGL: Corrected Gray Level CREB: camp Response CRH: Corticotropin Releasing Hormone CVS: Chronic Variable Stress DBH: Dopamine Beta Hydroxylase DMH: Dorsomedial Nucleus of the Hypothalamus dp: Dorsal Parvocellular DSAP: DBH-conjugated Saporin E: Epinephrine GABA: Gamma Amino Butyric Acid GAD: Glutamic Acid Decarboxylase GR: Glucocorticoid Receptor HPA: Hypothalamo Pituitary Adrenocortical HR: Heart Rate IPAC: Interstitial Nucleus of the Posterior Anterior Commissure KPBS: Potassium Phosphate Buffered Saline LC: Locus Coeruleus LH: Lateral Hypothalamic Nucleus LSD: Least Significant Difference lp: Lateral Parvocellular LTD: Long Term Depression LTP: Long Term Potentiation MAP: Mean Arterial Pressure MeA: Medial Nucleus of the Amygdala mg: Magnocellular mipsc: Miniature Inhibitory Post Synaptic Current mp: Medial Parvocellular mpfc: Medial Prefrontal Cortex mpoa: Medial Preoptic Area MR: Mineralocorticoid Receptor mrna: Messenger RNA NE: Norepinephrine NGF: Nerve Growth Factor 10

12 NPY: Neuropeptide Y NTS: Nucleus of the Solitary Tract OT: Oxytocin PH: Posterior Hypothalamic Nucleus PKC: Protein Kinase C POMC: Pro-Opio Melanocortin PPG: Pre-proglucagon PSA-NCAM: Polysialic Acid Neural Cell Adhesion Molecule PTSD: Post Traumatic Stress Disorder PVN: Paraventricular Nucleus of the Hypothalamus PVT: Paraventricular Nucleus of the Thalamus RNA: Ribonucleic Acid ROI: Region of Interest RR: Repeated Restraint RT: Room Temperature RVLM: Rostral Ventrolateral Medulla SAP: Saporin SEM: Standard Error of the Mean SNS: Sympathetic Nervous System SON: Supraoptic Nucleus TH: Tyrosine Hydroxylase VEGF: Vascular Endothelial Growth Factor vglut2: Vesicular Glutamate Transporter 2 VMH: Ventromedial Nucleus of the Hypothalamus WM: Weight-Matched 11

13 Chapter 1 General Introduction and Background 12

14 I. Stress Stress can be defined as a real or perceived threat to homeostasis. Individuals know when they are stressed, but the stimuli vary greatly between beings. Despite the lack of control associated with stressful experiences, organisms have evolved responses to stress in order to deal with the stress at hand and prevent future complications. However, these responses can become maladaptive over time, in that chronic experience of stress is associated with anxiety, depression, heart disease, diabetes, as well as a host of other disease states and disorders. Because stressors can vary so greatly among the population, deciphering the neurological consequences of stress becomes a difficult practice. Even though there are so many fundamental differences in the stress we all face and the types of experiences that incite physiological and behavioral change, we can deliver the same stressors chronically to experimental animals to model the human experience. Using these models, we can locate the effects of chronic stress and pin down which of these effects are essential for physiological and behavioral indices associated with disease manifestation. This dissertation will begin with an introduction to our understanding of the mechanisms of chronic stress regulation, leading into the experimental data, and finally with a conclusion of how these findings fit into the context of our understanding of chronic stress. II. History of Stress Research. The history of stress research can be traced back to Claude Bernard, a 19 th century French physiologist, who taught his students about the maintenance of the internal milieu (Bernard, 1865). Despite the ever changing external environment, physical and chemical properties of tissue and fluids remain stable. When this self- regulating power fails, disease sets in. He likened the nervous system to a conductor promoting harmony within the symphony of our 13

15 physiology (Bernard, 1865). Walter Cannon, a 1930s American Physiologist, expanded on the teachings of Bernard to coin the term homeostasis, defined as the physiological regulation to sustain stability in function and keeping variables within an acceptable range. When these variables deviated from equilibrium, homeostasis was restored through compensatory and anticipatory adjustments by activating the sympathoadrenomedullary system or the emergency reaction. The sympathoadrenomedullary system is an acute physiological response driving the sympathetic nervous system, increasing skeletal muscle activity and arousing the central nervous system (Cannon, 1914). In addition to noxious environmental or internal stimuli that disrupt homeostasis, he described that strong emotional states, such as fear and anger, stimulate epinephrine release from the adrenal gland that act on peripheral tissues to prepare for fighting or fleeing (Cannon, 1914; Cannon and de la Paz, 1911). Following subsequent studies, we now know that the sympathoadrenomedullary system is not just activated following emergencies but also daily life events, such as meal ingestion, posture change, and locomotion (Goldstein and Kopin, 2007; Lake et al., 1976; Patel et al., 2002). While Walter Cannon s work laid the groundwork for modern day stress research, Hans Selye popularized the concept of stress and redefined it from the engineering term into the nonspecific response of the body to any demand upon it (Selye, 1976). During physician training in Hungary, Selye noticed that patients exhibited a few similar symptoms dismissed as unimportant, despite an array of actual diagnoses. When beginning his own line of research, Selye found that animals developed a pathological triad (enlarged adrenal glands, appearance of gastrointestinal ulcers, and atrophy of lymphoid tissue of the thymus, spleen, and lymph nodes) whether he injected tissue extracts or vehicle (Selye, 1976). These as well as subsequent observations led Selye to propose the existence of a general adaptation syndrome, which 14

16 included three phases: an alarm reaction, followed by adaptation, and finally exhaustion. Selye s general adaptation syndrome included Cannon s fight or flight response as the alarm reaction, but expanded on this subject by including resistance to stress after the alarm ceases. If organisms can no longer deal with the stress, exhaustion will set in, resulting in illness. Subsequent research has supported a number of the proposed ideas of Selye and Cannon. Cannon s unitary sympathoadrenomedullary system and Selye s concept that prolonged stress precipitates physical disease and mental disorder are both widely accepted today. Selye s pursuit of an adrenal gland- secreted steroid hormone producing the disease triad lead to the discovery of glucocorticoids. However, both of these scientists believed that the responses to stress were completely nonspecific. They both thought that disparate stressors would elicit a similar response of both the emergency reaction and the general adaptation syndrome. In fact, it was later shown that different stressors initiate the secretion of differing amounts of hormones, at different rates (Deboer et al., 1990; Pacak et al., 1995; Pacak et al., 1998), which fail to confirm Selye s predictions. The geometry of responses to stress can be driven by stimulus intensity, length of time, and modality. Therefore, wear and tear may not be due to merely responses of stress, but also to the individual stressors that initiate them. Thus, I will refer to stress as the initiation of the responses. In addition to the non- specificity of the stress response, Selye believed that responses habituate to all stressors in the adaptation phase, but this theory of cross- resistance to stress was later proven false. In fact, responses are facilitated to acute novel challenges following chronic periods of stress (Akana et al., 1994). 15

17 III. The Physiological Stress Responses. (2000)Today, Cannon s sympathetic nervous system (SNS) and Selye s neuroendocrinemediated responses are accepted as the two main physiological responses to stress. Within seconds following the onset of stress, cells within the intermediolateral division of the spinal column and adrenal medulla release their catecholaminergic (epinephrine (E) and norepinephrine (NE)) contents. The sympathetic nerves innervate glands, smooth muscle, blood vessels, and myocardium, increasing salivation, sweating, vasoconstriction, and myocardial contraction. E/NE can also act as hormones released by the adrenal medulla. Via the bloodstream, catecholamines, primarily E, can have a multitude of effects including elevating metabolic rate, glycogenolysis, increasing renin- angiotensin activity, bronchodilation, lactate production, skeletal vasodilation, relaxation of the bladder, and papillary dilation. Selye s proposed neuroendocrine cascade became known as the hypothalamo- pituitaryadrenocortical (HPA) axis response. Following stress onset, corticotropin- releasing- hormone (CRH) expressing neurons within the medial parvocellular (mp) division of the paraventricular nucleus of the hypothalamus (PVN) release their contents from stores located in the external zone of the medial eminence into the portal hypopheseal- pituitary circulation. CRH acts via corticotropin releasing hormone receptor 1 receptors expressed in anterior pituitarycorticotropes to stimulate the release of adrenocorticotropic hormone (ACTH) through a cyclic adenosine monophosphate (camp) pathway. Arginine- vasopressin (AVP), a potent synergistic factor with CRH, drives phospholipase C to stimulate protein kinase C (PKC) and intracellular calcium release, which potentiate ACTH secretion. AVP can act in the anterior pituitary both from co- release within mppvn neurons and by magnocellular AVP- expressing neurons. ACTH travels through the general circulation, acting on melanocortin-2 receptors within the 16

18 adrenal cortex to stimulate the synthesis and secretion of glucocorticoids (corticosterone in rodents and cortisol in humans) through a camp- dependent pathway. In non- stressful states, the HPA axis is activated in a pulsatile fashion, controlled by circadian factors. IV. Glucocorticoid Action. The lasting effects of stress exposure are believed to be mediated by the HPA axis and not the SNS, since glucocorticoid elevations persist for minutes or hours following stress onset, but SNS responses terminate long before then. Despite having a faster on- response in the SNS versus the HPA axis, SNS activation does not typically produce enduring changes on wholeorganism physiology. Thus, the temporal differences between the SNS and HPA axis utilize very different means by which the two responses to stress can influence present and future stress reactivity. In addition to a prolonged period of HPA axis activation versus the SNS, glucocorticoids bind to steroid receptors (glucocorticoid (GR) and mineralocorticoid receptors (MR)) throughout the body, which can have delayed and persistent effects on gene expression (De Nicola et al., 1998) and cellular plasticity (McEwen et al., 1991). Both of these receptors act as transcription factors that bind to glucocorticoid response elements within sensitive genes, thereby both directly regulating gene expression (Zilliacus et al., 1995) and indirectly via protein- protein interactions (De Bosscher et al., 2003; Gottlicher et al., 1998). Glucocorticoids have a fold higher affinity for MR than GR (Reul and de Kloet, 1985), and are primary ligands for MR action within the brain, due to the absence of enzymatic degradation processes by 11-beta hydroxysteroid dehydrogenase seen in the kidney, colon, and exocrine glands (Funder et al., 1988). On the other hand, GR is expressed in most cell types throughout the body. Similar to 17

19 expression within peripheral tissue, GR is widely expressed in brain, whereas MR is confined primarily to the cortex, hippocampus, hypothalamus, septum, and amygdala (Chao et al., 1989; de Kloet et al., 1986). In these brain areas, glucocorticoids are believed to act on MR when released in low concentrations, but on both MR and GR in high concentrations. Thus, stressinduced levels of glucocorticoids are believed to regulate physiological and behavioral function through the GR- mediated effects and not necessarily MR. In addition to the classical effects of GR activation on energy homeostasis and immune function, glucocorticoids bind to GR to shut off its own synthesis and secretion through negative feedback inhibition of the HPA axis. Negative feedback is believed to occur in three phases: rapid (minutes), intermediate (hours), and delayed (days) (Kellerwood and Dallman, 1984). Rapid feedback is rate- sensitive (Kellerwood and Dallman, 1984) and occurs in response to rapidly rising levels of glucocorticoids, mediating these effects by non- genomic mechanisms. However, genomic actions of acute glucocorticoid release likely govern intermediate and delayed feedback (Kellerwood and Dallman, 1984). Glucocorticoids mediate feedback of the HPA axis through the activation of GR. GR is highly expressed within the both the pituitary corticotropes and mppvn CRH- expressing cells, providing direct inhibition of two segments of the HPA axis. In support for these sites as being critical to glucocorticoid negative feedback, adrenalectomized animals exhibit profound elevations in CRH expression and ACTH release, indicating that glucocorticoids are necessary for proper control of the HPA axis. GR is expressed within extra- hypothalamic sites as well, especially within upstream limbic sites known to be critical in stress regulation. In fact, deletion of GR within the hippocampus, medial prefrontal cortex (mpfc), and parts of the amygdala dysregulate responses to stress and dexamethasone challenge, indicating that glucocorticoid 18

20 action within these upstream limbic sites is essential for proper HPA axis negative feedback (Boyle et al., 2005). Despite evidence for the necessity of neural action within glucocorticoid negative feedback, the specific circuitry mediating these three types of feedback has not been discovered. V. Chronic Stress exposure Following acute stress exposure, glucocorticoids produce relatively small changes in gene expression, but the relatively small changes accumulate when exposure becomes prolonged or chronic. As Selye proposed, disease and mental disorder set in when the organism can no longer adapt to and cope with stress. In support of this connection between the chronicity of stress and manifestation of disease/ disorder, chronic stress is associated with periods of depression and anxiety (Kendler et al., 1999). This connection is potentially due to cumulative glucocorticoid exposure, since elevated glucocorticoids are associated with both depressed mood and cognitive impairment. Furthermore, melancholic depressive patients exhibit facilitated HPA axis responses to stress, hypercortisolemia, and attenuated glucocorticoid negative feedback (Modell et al., 1997), which provide further evidence connecting cumulative glucocorticoid exposure and mental disorder. However, it is unclear whether glucocorticoid exposure precipitates mental disorder or mental disorder produces hypercortisolemia. Regardless of the direction of connection, Cushing s disorder patients, characterized by chronically elevated glucocorticoid levels, commonly become depressed (McEwen, 2005), indicating that cumulative glucocorticoid exposure can precipitate mental disorder. Clearly, there is an important connection between chronic stress, glucocorticoid exposure, and mental health. 19

21 Previous studies in rodents have furthered our understanding of the consequences of chronic stress on HPA axis regulation and how these effects can negatively influence mood. Following chronic restraint stress, the glucocorticoid response is cut in half after the fifth exposure (Cole et al., 2000), indicating that animals can adapt to stressful situations and limit their glucocorticoid exposure. However, this does not occur in immobilized or unpredictably stressed animals (Dobrakovova et al., 1982; Herman et al., 1995a), suggesting that prolonged or chronic drive of the HPA axis is not easily adaptable. In fact, chronically stressed rodents display anhedonia, anorexia, hypercortisolemia, facilitated HPA axis responses, and reduced glucocorticoid negative feedback (Pariante and Lightman, 2008; Thomson and Craighead, 2008; Willner, 1997), which are consistent with a depressive- like state. In addition to these negative consequences of stress, chronic stress induces hypertension, memory impairments, immune suppression, and cellular atrophy, which together provide the wear and tear on physiological function proposed by Selye within his general adaptation syndrome. VI. Stress Neurocircuitry In addition to these physiological effects, chronic stress can profoundly alter neuronal function, which may be necessary for the behavioral and physiological consequences of chronic drive of the HPA axis. However, these effects are believed to be plastic and return to normal following a combination of sufficient recovery, coping, pharmaceuticals, and/or therapy. Thus, neuroplasticity is a critical mediator of physiological and behavioral dysfunction in mental disorders, since patients commonly display disruptions in brain activation and organization that recover following successful treatment (Krishnan and Nestler, 2008; Pittenger and Duman, 2008). Chronic stress likely induces these changes by recruiting and dysregulating stress- 20

22 sensitive circuits. At this point, the chronic stress- sensitive circuits are unknown. Through the last twenty years, many individual structures that regulate stress responses have been discovered, but it is unclear how these regions connect in order to manifest behavioral and physiological dysfunction. This section will review the current state of known stress neurocircuitry, both in terms of chronic and acute stress. Before critical stress- activated/regulated regions are discussed, we can speculate the integral areas and connectivity, since many neural structures appear responsive to either psychogenic or systemic stressors. Psychogenic stressors, which include fear or restraint in rodents, stimulate the HPA axis and SNS during anticipation and appraisal of danger, while systemic stressors, for example, hypoxia or hypovolemia, initiate the stress responses before the organism realizes the situation is threatening. Psychogenic stress primarily triggers stress responses through limbic efferents, but systemic stimuli primarily activates the stress responses via brainstem pathways (Emmert and Herman, 1999; Herman and Cullinan, 1997). These terms will be used in order to differentiate between potentially connected stress- sensitive nuclei. VII. Stress Neurocircuitry: Upstream Limbic Regions Limbic regions upstream from the hypothalamus, such as the mpfc, amygdala, and hippocampus, are heavily implicated in stress regulation, since reduced size, integrity, and activity in these regions is seen in numerous psychiatric disorders (Pittenger and Duman, 2008; Shin and Liberzon, 2011; Shin et al., 2006). Furthermore, some of these characteristics of psychiatric disorders are alleviated by antidepressant therapy, underscoring the importance of proper neurochemical balance within these regions. Each of these structures has the potential to regulate physiological and behavioral function through their known effects on both emotional 21

23 memory and HPA axis regulation. As a result, research has focused on elucidating the connections between stress, psychosis, and these limbic regions, with particular emphasis on understanding how malfunction within these brain regions manifests disease. To this point, results indicate that the hippocampus inhibits stress responding. In humans, there is a positive correlation of hippocampal volume and self- esteem (Pruessner et al., 2005), an inverse correlation of self- esteem and cortisol response (Kirschbaum et al., 1995), and an inverse correlation of hippocampal volume and cortisol responses to stress tasks (Pruessner et al., 2005), indicating that hippocampus size, HPA axis responses, and mood are connected. In support of this association, hippocampal volume and neuronal integrity are reduced in depressed (MacQueen et al., 2003; Stockmeier et al., 2004) and PTSD patients (Bremner et al., 2003; Schuff et al., 2001). Hippocampal lesion studies note basal glucocorticoid hypersecretion (Fendler et al., 1961; Knigge, 1961), enhanced mppvn CRH and AVP mrna (Herman et al., 1995c; Herman et al., 1989), and elevated basal AVP (Sapolsky et al., 1989), all suggestive for inhibition of the PVN. In addition to indirect inhibition of the PVN, the hippocampal lesions prolong HPA axis responses to psychogenic (Herman et al., 1995c; Herman et al., 1998; Kant et al., 1984), but not systemic stressors (Bradbury et al., 1993; Magarinos et al., 1987), suggesting that the hippocampus plays a role in the termination of psychogenic stress responses. Feedback studies indicate that this stress termination is likely due to hippocampus- mediated glucocorticoid negative feedback (Jacobson and Sapolsky, 1991), which is supported by the fact that this region densely expresses both GR and MR (McEwen et al., 1968; Reul and de Kloet, 1985). Moreover, hippocampal GR and MR antagonist implants attenuate HPA axis responses to acute psychogenic stimuli (Feldman and Weidenfeld, 1999), providing further evidence of direct action of glucocorticoids on the hippocampus to alter stress responses. The principal 22

24 hypothalamus- projecting hippocampal neurons lie in the ventral subiculum and ventral CA1 (Kohler, 1990; Meibach and Siegel, 1977). These neurons project to areas that contain rich populations of gamma amino butyric acid (GABA) neurons (peri- PVN, bed nucleus of the stria terminalis (BST), dorsomedial hypothalamic nucleus (DMH), and medial preoptic area (mpoa) (Cullinan et al., 1993; Okamura et al., 1990), indicating that the hippocampal control of PVN reactivity is mediated indirectly. In support of the indirect action of the hippocampus stimulating PVN GABA, subicular afferents are primarily glutamatergic, and lesions of this area reduce hypothalamic glutamate (Walaas and Fonnum, 1980). In opposition to the hippocampus, the amygdala plays an excitatory role in responding to stress. In addition to its role in processing threatening stimuli, the amygdala also influences HPA axis activity. In humans, stimulation of the amygdala activates the HPA axis (Gallagher et al., 1987), and cortisol levels are correlated with amygdala response to emotional stimuli (van Stegeren et al., 2008), indicating that activation of both the amygdala and HPA axis are intertwined. The main outputs from the amygdala are the central nucleus of the amygdala (CeA) and the medial nucleus of the amygdala (MeA). The MeA is activated to a greater degree by psychogenic stress (Cullinan et al., 1995; Emmert and Herman, 1999; Pezzone et al., 1992), and lesion studies confirm that this area is required for proper HPA axis responsiveness to psychogenic, not systemic stimuli (Feldman et al., 1994). Stimulation of the MeA facilitates corticosterone secretion and CRH depletion (Dunn and Whitener, 1986), providing further support of the HPA excitatory effects of the MeA. However, the MeA is not necessary for chronic stress responses (Solomon et al., 2010), and therefore may only play a role in acute responses to stress. In contrast to the MeA, the CeA does not regulate acute HPA axis responses to psychogenic stress (Dayas et al., 1999; Xu et al., 1999) and is more responsive to systemic 23

25 stress (Cullinan et al., 1995; Emmert and Herman, 1999; Yamamoto et al., 1997). Both the CeA and MeA may regulate PVN activity via direct contacts (Canteras et al., 1995; Marcilhac and Siaud, 1997; Prewitt and Herman, 1998), but these projections are fairly limited. Alternatively, the amygdalar outputs may act through projections to the BST (Canteras et al., 1995; Prewitt and Herman, 1998). In addition to the PVN and BST, the MeA also projects to the mpoa, and the CeA projects to the nucleus of the solitary tract (NTS), providing alternative means for stress activation. The basolateral amygdala (BLA) is an additional amygdalar nucleus responsive to acute psychogenic stress (Cullinan et al., 1995), but is not necessary for acute responses to stress (Seggie, 1987). However, inactivation of the BLA enhances chronic stress- induced HPA facilitation (Bhatnagar and Dallman, 1998), indicating that this site is important for chronic, but not acute stress. Unlike the amygdala and the hippocampus, the mpfc does not merely have excitatory or inhibitory role in stress responding, but a combination of the two. Similar to the hippocampal findings, PTSD and depressed patients exhibit reduced size and activity (Baxter et al., 1989; De Bellis et al., 2000; Fennema-Notestine et al., 2002; Rajkowska et al., 1999), suggesting that dysfunction in this region may be important to the manifestation of mental disorder. Furthermore, increased mpfc activity is correlated with decreased cortisol secretion (Kern et al., 2008), indicative for a primarily stress inhibitory role of the mpfc. Additionally, lesions of the rodent mpfc reduce HPA axis activity in response to psychogenic, but not systemic stress (Figueiredo et al., 2003). Like the hippocampus, corticosterone implants attenuate HPA axis responses to stress, indicating that this is an additional site of glucocorticoid negative feedback (Akana et al., 2001). The stress relieving effects of the mpfc may dependent on side, as lesions of the right, but not the left, inhibit HPA axis activation (Sullivan and Gratton, 1999). In 24

26 addition to the evidence of stress inhibitory action of the mpfc, there is also an hyper-activated region of the mpfc in depressed patients (area 25), and deep brain stimulation directed toward area 25 improves mood in patients with treatment resistant depression (Lozano et al., 2008), indicating that this is an important site of dysfunction in depression. Hyper- activation of a subregion of the mpfc may contribute to depression via SNS activation, since pharamacological blockade of the mpfc inhibits SNS responses to psychogenic stress (Resstel et al., 2006). The heterogeneity in responses is similar in rodents, due to differing control of the infralimbic and prelimbic mpfc on PVN activation. The prelimbic, but not infralimbic, mpfc attenuates HPA axis activity, via projections to GABAergic BST neurons (Radley et al., 2006; Radley et al., 2009). The prelimbic and infralimbic mpfc may regulate responses to stress through their differential projections. The infralimbic and prelimbic mpfc both innervate the amygdala, raphe, and DMH, but the infralimbic preferentially innervates the NTS and posterior hypothalamic nucleus (PH) (Hurley et al., 1991; Sesack et al., 1989; Terreberry and Neafsey, 1987), regions important for control of both cardiovascular and HPA axis regulation. Chronic stress structurally re- organizes the hippocampus, mpfc, and amygdala. A key feature of structural re- organization is altered dendritic architecture. Chronic stress reduces length of the apical dendrites in both the hippocampus and mpfc (Cook and Wellman, 2004; Watanabe et al., 1992), both of which can be also be produced by corticosterone administration alone (Wellman, 2001; Woolley et al., 1990). Another stress- sensitive feature of these neural structures is hippocampal neurogenesis, which is compromised by chronic stress (Dranovsky and Hen, 2006). Structural regulation of dendritic length and neurogenesis are presumed mechanisms for hippocampal and cortical atrophy associated with depression and PTSD. In contrast to the hippocampus and mpfc, dendrites within the amygdala hypertrophy following 25

27 chronic stress (Vyas et al., 2003; Vyas et al., 2002), indicating reciprocal regulation of the hippocampus/mpfc and amygdala. These morphological changes within limbic structures are associated with electrophysiological impairments in long term potentiation (LTP) (Foy et al., 1987) and enhancements in long term depression (LTD) (Xu et al., 1997) within the hippocampus, but much less is known about the electrophysiological consequences within the amygdala and the mpfc. Behavioral deficits also mirror the morphological and electrophysiological changes, with impairments in hippocampal and mpfc- dependent learning (Conrad et al., 1996; Liston et al., 2006) and enhancements in amygdala dependent learning (Conrad et al., 1999). These changes in limbic structures may be due to chronic stress- induced elevations in extracellular glutamate levels (Lowy et al., 1993) or regulation plasticityassociated signaling pathways. Chronic stress increases the phosphorylation of hippocampal map kinase (MAPK) and camp response element binding (CREB) (Pardon et al., 2005), two pathways known to be important in LTD/LTP. In addition, chronic stress reduces hippocampal brain derived neurotrophic factor (BDNF) (Smith et al., 1995) and vascular endothelial growth factor (VEGF) (Heine et al., 2005) (predicting loss of function), but elevates nerve growth factor (NGF) (Alfonso et al., 2006) expression, indicating that stress does not have universally negative effects on neurotrophic factors. The above neuroplastic changes within the hippocampus, mpfc, and amygdala function may be critical to chronic stress- related pathology, since antidepressants reverse a number of these changes. In fact, antidepressants are a component within the treatment of both depression and PTSD. Using electron microscopy, antidepressants elevate the number of synapses within the CA1 pyramidal cell layer (Hajszan et al., 2005), indicating the strong neuroplastic action by antidepressants. Furthermore, tianeptine, an atypical antidepressant, reverses CA1 dendritic 26

28 atrophy (Magarinos et al., 1999). In addition, antidepressants reverse stress- induced impairment of LTP and enhancement of LTD (Holderbach et al., 2007; Rocher et al., 2004; Vouimba et al., 2006), suggesting that antidepressants have the potential to restore pre- stress neuronal function. Antidepressants also increase the rate of proliferation and survival of newborn cells, BDNF expression, CREB phosphorylation, and VEGF expression, indicating that antidepressants provide opposing actions on hippocampal function to chronic stress. In addition to these clear opposing effects of antidepressants to chronic stress, there are a large number of effects on processes that chronic stress does not appear to regulate, indicating that antidepressants do not merely act by mitigating the effects of chronic stress. Together, the data suggests that antidepressants have profound neuroplastic effects, but vary by the antidepressant used, which provide insight into the array of deficits in depression and PTSD. VIII. Stress Neurocircuitry: PVN While both acute and chronic responses to stress involve the recruitment of complex neurocircuitry, the PVN is the final site of integration of sensory and visceral stimuli. In support of the integrative capabilities of this region, the PVN, a cell- dense heart- shaped region surrounding the third ventricle, receives direct projections from a range of neural sites, including areas known to be cortical relays (thalamus and BST), hypothalamic mediators of physiological function (DMH, ventromedial hypothalamic nucleus (VMH), and arcuate nucleus (ARC), PH, lateral hypothalamic nucleus (LH)), and brainstem regions (NTS and rostral ventrolateral medulla (RVLM)). The parvocellular neurons within this nucleus primarily contain peptide transmitters (CRH, AVP, and oxytocin). CRH- positive neurons are largely located within the mppvn division and project to the external zone of the median eminence, initiating the HPA 27

29 axis response to stress. In addition to these classical neuroendocrine neurons, a subset of neurons expressing oxytocin as well as a host of other transmitters, located within the dorsal cap of the PVN and lateral parvocellular divisions, project to both the NTS and spinal column (Swanson and Kuypers, 1980), indicating the capacity for this nucleus to initiate both the HPA and SNS responses to stress. However, sympathetic and parasympathetic pre-autonomic neurons are sprinkled within the PVN (Kreier et al., 2006). The magnocellular neurons are positioned lateral to the parvocellular component and express AVP and oxytocin in separate neurons. These neurons are integral to responding to dehydration and lactation. Interestingly, PVN dendrites do not extend outside the nucleus (van den Pol, 1982), affording the possibility that the neurons within this region also regulate the function within the other neurons within the PVN. The PVN is primarily innervated by glutamatergic, noradrenergic, and GABAergic axons, and there is evidence that these fibers directly contact CRH mppvn neurons (Liposits et al., 1986; Miklos and Kovacs, 2002; van den Pol, 1991). The PVN expresses NMDA, AMPA, and metabotropic glutamate receptors (Eyigor et al., 2005; Herman et al., 2000; Scaccianoce et al., 2003). Furthermore, glutamate agonists stimulate the release of corticosterone (Makara and Stark, 1975), and PVN- directed antagonist delivery attenuates stress- induced elevations in corticosterone (Feldman and Weidenfeld, 1997; Ziegler and Herman, 2000), indicating that glutamate stimulates HPA axis activity at the level of the mppvn. Glutamate is not the only stimulatory neurotransmitter in the mppvn. For instance, the PVN expresses α adrenoceptors (Day et al., 1999). In support for the stimulatory of PVN norepinephrine (NE) on the HPA axis, local NE administration activates CRH and AVP gene transcription and PVN cfos (Cole and Sawchenko, 2002; Szafarczyk et al., 1987). GABA is the primary inhibitory PVN 28

30 neurotransmitter, as GABA A antagonism induces corticosterone secretion and PVN cfos (Cole and Sawchenko, 2002). The paraventricular nucleus is a highly plastic region, in that the magnocellular neurons are known to expand and increase the number of connections following dehydration and lactation (Hatton and Walters, 1973; Theodosis and Poulain, 1984). In addition to the known neuroplastic properties of neurons within the PVN, this region is one of the most highly vascularized regions in the brain (McGinty et al., 1983), leading to the possibility of immediate regulation from the periphery including glucocorticoids. Following tonic activation by stress, the parvocellular neurons can be profoundly altered as well. Following chronic periods of stress, mppvn CRH and AVP mrna is increased (Herman et al., 1995b; Imaki et al., 1991; Kiss and Aguilera, 1993; Makino et al., 1995), which may lead to facilitation in HPA axis activation. This is similar to what is found following adrenalectomy (Kovacs et al., 1986; Swanson and Simmons, 1989) and opposite to following chronic glucocorticoid administration (Kovacs et al., 1986; Kovacs and Mezey, 1987), suggesting reduced inhibitory glucocorticoid regulation of mppvn CRH activation. This is supported by a reduction in GR mrna within the mppvn following chronic stress (Herman et al., 1995a). These neurons also begin to co- express AVP following chronic stress (Volpi et al., 2004), which further facilitates ACTH release. The peptides expressed within these neurons are not the only area of adaptation following chronic stress. In addition to peptide these changes in peptide regulation, chronic stress alters expression of ionotropic glutamate and GABA A subunits (Cullinan and Wolfe, 2000; Ziegler et al., 2005) in a manner consistent with enhanced excitability. Postmortem studies connect PVN dysregulation with disease, since depressed and Alzheimer disease patients display elevated CRH mrna and 29

31 CRH and AVP peptide (Raadsheer et al., 1994a; Raadsheer et al., 1994b; Raadsheer et al., 1994c; Wang et al., 2008). IX. Stress Neurocircuitry: PVN- projecting regions Even though the integrative capabilities of the PVN are accepted, the upstream targets initiating these changes have yet to be determined. Very little is known about how the neurons within these regions can be regulated by chronic stress and the role they play in responses to chronic stress. However, tract tracing and lesion studies have been performed within a number of these areas to suggest that they are important stress regulatory regions. The noradrenergic supply to the PVN is better defined than that of GABA and glutamate, arising from the NTS and RVLM. A2/C2 catecholaminergic axons primarily innervate the parvocellular PVN neurons, whereas A1/C1 innervate the magnocellular PVN (Cunningham and Sawchenko, 1988). Lesions of ascending medullary PVN inputs (Li et al., 1996) and PVN- directed injection of saporin conjugated to a dopamine- beta- hydroxylase antibody (Ritter et al., 2003) blunt HPA axis responses to glucoprivation and Il-1beta injection, but not swim and shock, indicating that this PVN connection is only necessary for systemic stress. However, both psychogenic and systemic stress activate the brainstem neurons (Cullinan et al., 1995; Teppema et al., 1997) and release PVN NE (Pacak et al., 1993; Palkovits et al., 1999), indicating that these areas are responsive to all types of stress. Furthermore, unpredictable chronic stress elevates NTS tyrosine- hydoxylase mrna independent of glucocorticoid levels (Zhang et al., 2010), indicating that the experience of stress dysregulates this area, which may contribute to associated physiological and behavioral dysfunction. 30

32 A significant proportion of PVN tonic inhibition arises from the GABAergic surround. These neurons are innervated by the ventral subiculum, lateral septum, and medial amygdala (Canteras et al., 1995; Canteras and Swanson, 1992; Risold and Swanson, 1997), indicating that forebrain regions can influence HPA axis reactivity through intermediate regulation of the PVNsurrounding GABAergic neurons. Delivery of kynurenic acid, an ionotropic glutamate receptor antagonist, exacerbates corticosterone responses to restraint (Cole and Sawchenko, 2002; Ziegler and Herman, 2000), indicating that glutamatergic innervation of the peri- PVN inhibits PVN activation. While acute stress activates this region (Cullinan et al., 1996), both chronic and acute stress elevate local glutamic acid decarboxylase (GAD) expression (Bowers et al., 1998), suggesting its involvement in acute and chronic stress regulation. Together, the data suggest that peri- PVN GABAergic neurons are an important control on the HPA axis. The BST is one of the primary relay sites from the forebrain to the PVN. It connects upstream forebrain sites with both the hypothalamus and brainstem. Primarily, it is known to be essential for conditioned fear, as an extension of the central amygdala. The BST is responsive to both acute and chronic stress, suggesting the involvement of the BST in stress regulation. Even though the anterior and posterior BST both project directly to the PVN (Dong and Swanson, 2004, 2006), the former, expressing CRH (Phelix and Paull, 1990), excites (Choi et al., 2007) and the latter, expressing GAD (Ju et al., 1989), inhibits (Choi et al., 2007) acute stress responding. In support of this dichotomy in function, BST stimulation can excite and inhibit corticosterone secretion (Casada and Dafny, 1991; Dunn, 1987). Despite the clear evidence of the BST in acute stress regulation, neither subregion is necessary for chronic stress responses (Choi et al., 2008a; Choi et al., 2008b). 31

33 The paraventricular thalamus (PVT) is a region that supplies a considerable amount of glutamate to both the rostral and caudal PVN (Ulrich-Lai et al., 2011b). In addition to the PVN, the PVT projects throughout the forebrain, known for its roles in both arousal and reward (Parsons et al., 2007; Van der Werf et al., 2002; Vertes and Hoover, 2008). Evidence indicates that it plays a role in the habituation and sensitization of the HPA axis following chronic stress (Bhatnagar and Dallman, 1998; Bhatnagar et al., 2002), but does not regulate acute stress responding. The PVT receives projections from a host of stress- activated regions (Vertes and Hoover, 2008), including the mpfc and ventral subiculum, suggesting that it may be a chronic stress integrative region to control stress responding. Additionally, PVT fibers innervate the BLA (Vertes and Hoover, 2008), suggesting linkage between the two sites in a chronic stressrecruited neurocircuit. In addition to regions that project to the PVN expressing either excitatory or inhibitory neurotransmitters, the DMH and mpoa contain both glutamatergic and GABAergic neurons that innervate the mppvn, possibly providing complex regulation of stress responses via PVN afferents. The DMH is primarily known to regulate the SNS, in that DMH inactivation attenuates (Stotz-Potter et al., 1996b) and stimulation exacerbates cardiovascular responses to psychogenic (Bailey and Dimicco, 2001), but not systemic stress. The mpoa is critical to body temperature homeostasis and also communicates gonadal hormone regulation of hypothalamic function (Stotz-Potter et al., 1996b). While lesion studies suggest a predominant inhibitory action of the mpoa and DMH on HPA axis drive (Bealer, 1986; Viau and Meaney, 1996), the dorsal DMH and lateral mpoa contain glutamatergic neurons that activate the PVN following stimulation (Bailey and Dimicco, 2001; Saphier and Feldman, 1986; Ulrich-Lai et al., 2011b). 32

34 The GAD67- positive and vesicular glutamate transporter 2 (vglut2)- positive PVN- projecting neurons are wholly separate (Herman et al., 2003). The previously mentioned nuclei are not the only regions known to supply the PVN with GABA and glutamate. The ARC and VMH are additional regions that contain glutamatergic and GABAergic neurons that project to the PVN. They also are known to play critical roles in metabolic regulation and thus will be discussed within the following section. X. Metabolic modification of chronic stress exposure While considerable study has focused on the role of chronic stress- induced neuroplasticity, physiological status can also produce changes in known stress regulatory regions independent of experience. While changes in physiological state can be considered stressors themselves, physiological challenges elicit specific counter- regulatory responses promoting the re- establishment of homeostasis. Thus physiological changes could either mitigate or exacerbate effects of chronic stress on associated pathologies. One such change in physiological state that clearly alters stress regulation is metabolic dyshomeostasis. Since glucocorticoids mobilize glucose, it is logical that energy storage would in turn regulate circulating glucocorticoid levels. However, there is evidence of both stress- induced hyperphagia (Mathes et al., 2009) and anorexia (Kaye, 2008), indicating that the regulation of stress on metabolic state appears to be dependent on the individual or qualities of the experience. Furthermore, both weight gain and weight loss are criteria for the diagnosis of depression, noting that hypo and hyper compensation of metabolic regulatory circuits may be involved in the pathophysiology of mood disorders. However, the changes in weight in depressed patients are assumed to be 33

35 byproducts of either mental disorder or pharmaceutical treatment, but not contributing factors. Thus, there is very little data available on how changes in metabolic state can influence mood. Currently, laboratory animals do not voluntary over- consume food without completely changing their diet in experimental animals relative to controls, making it difficult to design experiments to test the effects of body weight gain on mood. However, stress regulation has been assessed in chronically food restricted or food deprived animals, conditions both of which induce hypercortisolemia (Akana et al., 1994; Kiss et al., 1994; Wan et al., 2003). However, food deprivation reduces central HPA axis drive (Akana et al., 1994; Kiss et al., 1994), while also reducing glucocorticoid clearance (Kiss et al., 1994), keeping glucocorticoids at an elevated state but conserving energy by restraining their excursion. In addition to reducing central drive of the HPA axis, the therapeutic properties of weight loss on hypertension and cardiovascular disease are well- accepted, providing an avenue for changes in weight to modulate sympathetic tone. In fact, long term weight loss is accompanied by reductions in cardiovascular responses to psychogenic stress in both rodents (Flak et al., 2011; Wan et al., 2003) and humans (Ashida et al., 2007; Torres and Nowson, 2007). Behaviorally, there are studies that report pro- and antidepressive /anxiety like behavior effects of food restriction (Chandler-Laney et al., 2007; Inoue et al., 2004; Jahng et al., 2007; Lutter et al., 2008a), which may be due to the degree of whether the paradigm reduced central stress drive or glucocorticoid clearance. The energy storage- mediated stress regulation may be driven by neurocircuits that are involved in both stress and metabolic homeostasis. Two separate mechanisms appear likely. Food restriction reduces PVN CRH mrna (Flak et al., 2011), which is known to have anorectic properties in addition to triggering the HPA axis. CRH administration reduces food intake (Britton et al., 1982; Gosnell et al., 1983), whereas lessoning the PVN induces hyperphagia and 34

36 associated obesity (Leibowitz et al., 1981; Touzani and Velley, 1992), suggesting that PVN CRH reduces food intake. Thus, attenuating PVN CRH output would additionally attenuate central drive of stress responses. In addition to known effects on PVN CRH, food restriction sensitizes reward circuits (Carr and Wolinsky, 1993; Carroll et al., 1979; Stamp et al., 2008) presumably to promote food intake and recover pre- restriction metabolic state. Chronic activation of reward circuits attenuates HPA and SNS responses to psychogenic stress (Ulrich-Lai et al., 2011a), which provides another mechanism of possible neuroplastic regulation of central stress drive. Changes in metabolic homeostasis may communicate changes to stress neurocircuitry by regulating circulating hormones. Secreted into the bloodstream relative to adiposity level, these hormones travel through the blood brain barrier, activating centrally expressed receptors, primarily located within the hypothalamus, to drive anabolism/catabolism. One of these hormones is leptin, a hormone primarily released from adipocytes and known for its anorectic actions. The circadian peaks in leptin corresponds with the glucocorticoid nadir and vice versa (Licinio et al., 1997), suggesting that there is reciprocal regulation between these two hormones. However, the role of leptin in stress regulation is unclear. Previous studies indicate leptin blunts HPA axis and behavioral responses to stress (Ahima et al., 1996; Heiman et al., 1997; Lu et al., 2006), potentially due to inhibited CRH release by hippocampal activation. Furthermore, both leptin and leptin receptor knockouts display elevated glucocorticoid levels (Chen et al., 1996; Chua et al., 1996), that can be alleviated by leptin administration in leptin knockouts. Yet, leptin also elevates sympathetic activity, glucocorticoids during the active cycle, number of cfos immunoreactive PVN neurons, CRH release, and PVN CRH mrna (Dunbar et al., 1997; Elmquist et al., 1997; Raber et al., 1997; Schwartz et al., 1996; Shek et al., 1998; van Dijk et al., 1997), providing evidence for an enhancement in the drive of stress responses. In addition to 35

37 acute stress, there is evidence that leptin may play a role in chronic stress. Chronic stress reduces circulating leptin (Lu et al., 2006), which may contribute to associated hypercortisolemia and HPA facilitation by relieving tonic inhibition. The role of leptin in mood may be complicated, since leptin receptors are expressed throughout the body including the adrenal medulla and cortex (Cao et al., 1997; Glasow et al., 1998). Therefore, leptin may act at different sites to activate and inhibit stress responding. In addition to leptin, insulin is also secreted relative to adiposity level and can act within the brain. While insulin is primarily known for its actions in the hypothalamus to regulate energy homeostasis, it can also alter cortical and hippocampal function. Insulin administration, while keeping glucose levels constant, facilitates memory formation (Craft et al., 1996), possibly due to enhancing synaptic strength. Insulin has been associated with both LTP and LTD by regulation of specific subunits of the AMPA receptor (Passafaro et al., 2001) and increasing post synaptic density (PSD)- 95 expression (Lee et al., 2005). Thus, altered insulin levels and degrees of sensitivity could modulate stress effects on neuroplasticity. As previously mentioned, leptin and insulin both act at the hypothalamus to regulate energy homeostasis, mainly via the ARC and VMH. Both leptin and insulin receptors are expressed within these nuclei (Havrankova et al., 1978; Mercer et al., 1996). Lesions of each of these hypothalamic nuclei result in hyperphagia and associated obesity, indicating the necessity of proper regulation of these regions on body weight homeostasis (Choi et al., 1999). The ARC contains neuropeptide y (NPY)/agouti related peptide (AGRP) expressing neurons that are orexigenic, whereas pro opio melanocortin (POMC) neurons are anorexigenic. These neurons all project to the PVN, and can drive/inhibit PVN activation. In addition to peptides, PVNprojecting neurons of the ARC and VMH also express glutamate (Ulrich-Lai et al., 2011b), 36

38 providing direct activation of stress responses regulated by adiposity levels. Indeed, chronic stress regulates the ARC in a manner predictive by their altered body weight (Solomon et al., 2010), suggesting that adiposity regulates neuronal function independent of chronic stress level that could have ramifications on mood. XI. Dissertation To this point we do not completely understand the central drive of stress responses and how they can be modulated, culminating in disease. Although, we do understand some of the essential brain regions in the initiation/termination of stress responses. We know a great deal about the PVN, hippocampus, amygdala, and mpfc, and how these areas are regulated by chronic stress. However, it is unclear how chronic stress affects the intermediary regions that alter stress responding and how these effects can influence mood. The following four chapters will extend upon previous studies deciphering the mechanisms of chronic stress regulation and potentially locate new targets for intervention in the treatment of chronic stress- related pathology. Chapter 2 assesses morphological changes to afferents to the PVN due to chronic stress exposure. Chapter 3 locates known stress regulatory brain regions tonically activated by chronic stress. Chapter 4 tests the necessity of PVN-projecting noradrenergic neurons on the physiological consequences of chronic stress. Chapter 5 investigates whether the reductions in weight gain associated with chronic stress can sufficiently modulate stress responding. 37

39 Chapter 2 Chronic Stress- Induced Neurotransmitter Plasticity in the PVN 38

40 I. Abstract Chronic stress precipitates pronounced enhancement of central stress excitability, marked by sensitization of hypothalamic- pituitary- adrenocortical (HPA) axis responses and increased adrenocorticotropic hormone (ACTH) secretagogue biosynthesis in the paraventricular nucleus of the hypothalamus (PVN). Chronic stress- induced enhancement of HPA axis excitability predicts increased excitatory and/or decreased inhibitory innervation of the parvocellular PVN. We tested this hypothesis by evaluating chronic variable stress (CVS) - induced changes in total (synaptophysin), glutamatergic (VGluT2), GABAergic (GAD65), and noradrenergic (DBH) terminal immunoreactivity on PVN parvocellular neurons using immunofluorescence confocal microscopy. CVS increased the total PVN bouton immunoreactivity as well as the number of glutamatergic and noradrenergic immunoreactive boutons in apposition to both the corticotropinreleasing- hormone (CRH) - immunoreactive cell bodies and dendrites within the parvocellular PVN. However, the number of GABAergic- immunoreactive boutons in the PVN was unchanged. CVS did not alter CRH median eminence immunoreactivity, indicating that CVS does not enhance CRH storage within the median eminence. Taken together, the data are consistent with a role for both glutamate and norepinephrine in chronic stress enhancement of HPA axis excitability. These changes could lead to an enhanced capacity for excitation in these neurons, contributing to chronic stress- induced hyper- reactivity of stress effector systems in the brain. II. Introduction The paraventricular nucleus of the hypothalamus (PVN) is an important site of integration in the regulatory control of the hypothalamo- pituitary- adrenocortical (HPA) axis 39

41 (Herman et al., 2003). Afferents from limbic, brainstem, and hypothalamic regions converge upon the medial parvocellular division of the PVN to both precipitate and terminate HPA axis responses to stressors, triggering the neuroendocrine cascade culminating in the release of glucocorticoids (Herman et al., 2003). The HPA axis can be activated by glutamatergic or noradrenergic afferent stimulation of corticotropin- releasing- hormone (CRH) neurons of the PVN. Intraventricular or local glutamate (Makara and Stark, 1975) or norepinephrine (Cole and Sawchenko, 2002; Szafarczyk et al., 1987) infusion stimulate adrenocorticotropic hormone (ACTH) secretion, corticosterone release, and PVN cfos activation. Local injections of alpha adrenergic and ionotropic glutamate receptor antagonists inhibit stress- induced corticosterone release and PVN cfos induction (Feldman and Weidenfeld, 1997; Itoi et al., 1994; Leibowitz et al., 1989; Ziegler and Herman, 2000). In contrast, PVN CRH neurons are inhibited by GABA inputs. Local blockade of GABA A receptors with bicuculline methiodide initiates PVN cfos activation in the absence of a stressor (Cole and Sawchenko, 2002), and local application of GABA A receptor agonists can inhibit stress- induced HPA axis responses (Stotz-Potter et al., 1996a). There is evidence for direct innervation of CRH neurons by noradrenergic, glutamatergic, and GABAergic terminals (Liposits et al., 1986; Miklos and Kovacs, 2002; van den Pol, 1991), and all of these transmitters regulate the electrophysiological activity of parvocellular PVN neurons (Boudaba et al., 1997; Boudaba et al., 1996; Daftary et al., 2000). Ionotropic glutamate receptors (Ziegler et al., 2005), alpha-1 adrenergic receptors (Day et al., 1999), and GABA A receptors (Cullinan, 2000) are all expressed in PVN CRH neurons, providing a means through which these transmitters can control activation and inhibition of HPA axis responses to stress. 40

42 Repeated or chronic stress in rodents produces numerous changes in the function and regulation of the HPA axis, including hypersecretion of corticosterone during the circadian trough (Herman et al., 1995a) and facilitated HPA axis responses to novel stressors (Akana et al., 1992). At the level of the PVN, chronic stress upregulates parvocellular PVN CRH and AVP mrna expression (Herman et al., 1995b; Imaki et al., 1991; Kiss and Aguilera, 1993; Makino et al., 1995). Chronic stress also produces alterations in ionotropic glutamate and GABAA subunit expression (Cullinan and Wolfe, 2000; Ziegler et al., 2005) that are consistent with enhanced excitability. At the cellular level, chronic stress attenuates mipsc frequency in the medial parvocellular PVN (Verkuyl et al., 2004). Finally, interleukin 1-beta and amphetamine- induced sensitization of the HPA axis are correlated with reduced dopamine- beta- hydroxylase (DBH) immunoreactivity in the parvocellular PVN (Jansen et al., 2003), suggesting that activitydependent changes in synaptic actions may be related to altered neurotransmitter innervation. The current study tests the hypothesis that chronic stress alters neurotransmitter innervation of parvocellular PVN CRH neurons in a manner that favors excitation. To test this hypothesis we quantified the number of glutamate, norepinephrine, and GABA immunoreactive pre- synaptic boutons in apposition to CRH neurons within the parvocellular division of the PVN following chronic variable stress (CVS). Our results indicate that exposure to CVS induces striking alterations in excitatory innervation of the parvocellular PVN and support the hypothesis that altered neurotransmitter innervation underlies the enhanced excitatory drive on the HPA axis following chronic stress. 41

43 III. Materials and Methods Subjects. Male Sprague Dawley rats from Harlan (Indianapolis, IN) weighing g on arrival were group housed two per cage for the duration of the experiment in clear polycarbonate cages containing granulated corncob bedding, with food and water available ad libitum. The colony room was temperature and humidity- controlled with a 12 hour light cycle (lights on 6:00 AM; lights off 6:00 PM). Rats acclimated to the colony facility for 5 weeks prior to experimental manipulations. All experimental procedures were conducted in accordance with the National Institutes of Health Guidelines for the Care and Use of Animals and approved by the University of Cincinnati Institutional Animal Care and Use Committee. Chronic stress procedure. Subjects were randomly assigned to either 1 week CVS (n = 8) or non- handled control (n = 8) groups. The chronic stress protocol consisted of twice daily (morning and afternoon) exposure to randomly assigned stressors, with occasional overnight stressors, for 1 week. Morning stressors were conducted between 8:30 AM and 10:30 AM and afternoon stressors were administered between 2:30 PM and 4:30 PM. Overnight stressors commenced immediately after cessation of the afternoon stressor and ended with the initiation of the following morning s stressor. Stressors consisted of rotation stress (1 hour at 100 rpm on a platform orbital shaker), warm swim (20 minutes at 31 C); cold swim (10 minutes at 18 C), cold room stress (kept in 4 C for 1 hour), and hypoxia (8% O 2 92% N 2 ), overnight social isolation (1 rat/cage), and overnight social crowding (6 rats/cage). In the morning following the last afternoon stressor, rats received an overdose of sodium pentobarbital and were perfused with phosphate buffered saline (PBS), followed by 4% paraformaldehyde. Brains were post- fixed 42

44 overnight in 4% paraformaldehyde and transferred to 30% sucrose at 4 C until they were cut on a freezing microtome. Immunohistochemical procedures. Brains were cut at 25 µm on a sliding microtome through hypothalamic regions and the resulting sections were stored in cryoprotectant (0.1 M phosphate buffer, 30% sucrose, 1% polyvinylpyrrolidone, and 30% ethylene glycol) at -20 C until used for immunohistochemistry. Table 2-1 lists the primary antibodies used. The synaptophysin antibody (Zymed Laboratories, San Francisco, CA; ) recognizes a single band of 38 kda on immunoblot of total homogenate from rat cerebral cortex (Melone et al., 2005). This antibody was designed against amino acids of the c-terminus of human synaptophysin using a synthetic peptide fragment with the sequence YGPQGDYGQQGYGPQGAPTSFSNQ (information obtained from Invitrogen, La Jolla, CA). The density of this synaptophysin immunoreactivity labels to a similar density compared to stereologically corrected electron micrographs (Micheva and Beaulieu, 1996). The CRH antiserum (RC70, courtesy of Wylie Vale) recognizes rat CRH by radioimmunoassay (Vale et al., 1983) and fails to label after preabsorption with the peptide (Sawchenko, 1987). The vglut2 antibody (Synaptic Systems, Goettingen, Germany; ) recognizes a single 65-kDa band using brain tissue (Zhou et al., 2007) and does not label following pre- absorption with the immunizing protein (Graziano et al., 2008). vglut2 immunoreactivity was specifically localized in regions known to receive projections from areas containing dense vglut2 mrna expression. It did not correspond with patterns of staining seen using antibodies against VGlut1 (data not shown). The GAD6 antibody (Developmental Studies Hybridoma Bank, Iowa City, IA) recognizes a 59-kDa band by Western blot using brain tissue, but not the 63-kDa GAD (Chang and Gottlieb, 1988). The antibody also 43

45 specifically immuno- precipitates GAD65, but not GAD67 (Kaufman et al., 1992). The axonal and perikaryal labeling of the DBH antibody (Chemicon, Temecula, CA; MAB308) is absent following pre- absorption with a 10-fold excess of bovine adrenal DBH, but not bovine PNMT (Rinaman, 2001). Sections were transferred from cryoprotectant to 50 mm potassium PBS (KPBS, ph 7.2; 40 mm potassium phosphate dibasic, 10 mm potassium phosphate monobasic, and 0.9% sodium chloride) at room temperature (RT). Cryoprotectant was rinsed (5 x 5 minutes) in KPBS; the sections were transferred to KPBS + 1.0% H 2 O 2, and incubated for 10 minutes at RT. Sections were then washed (5 x 5 minutes) in KPBS at RT and placed in blocking solution (50 mm KPBS, 0.1% bovine serum albumin (BSA), and 0.2% Triton X-100) for 1 hour at RT. Sections were incubated overnight at 4 C in primary antibody diluted in blocking solution. The following morning sections were rinsed in KPBS (5 x 5 minutes) and incubated in biotinylated anti- rabbit secondary antibody (Vector Laboratories, Burlingame, CA), diluted 1:500 in KPBS + 0.1% BSA for 1 hour at RT. Sections were rinsed in KPBS (5 x 5 minutes) and then treated with avidinbiotin complex (ABC, Vector Laboratories) at 1:1,000 in KPBS + 0.1% BSA for 1 hour at RT. Following this incubation, sections were rinsed again in KPBS (5 x 5 minutes) and subsequently incubated in biotin- labeled tyramide (PerkinElmer Life Sciences, Boston, MA) 1:250 in KPBS with 0.3% H2O2 for 10 minutes at RT. Sections were rinsed in KPBS (5 x 5 minutes) and incubated in Cy3- conjugated streptavidin (Jackson ImmunoResearch, West Grove, PA) diluted 1:500 for 30 minutes at RT on a shaker in the dark. For double- labeling, sections were rinsed (5 x 5 minutes) in KPBS and then incubated in the second primary antibody diluted as indicated in KPBS + 0.1% BSA. Following KPBS rinses (5 x 5 minutes), slices were then incubated in Alexa 488- labeled secondary antibody (Molecular Probes, Eugene, OR) diluted 1:500 in KPBS + 0.1% 44

46 BSA at RT for 30 minutes covered. Sections were rinsed 5 x 5 minutes in KPBS at RT following the final antibody incubation, mounted onto Superfrost Plus slides, and coverslipped with Gelvatol. To determine specificity of primary antibodies, control reactions were performed in the absence of one or both primary antibodies. Image collection and processing. Confocal imaging was performed on a Zeiss 510 Meta microscope system in multichannel mode. All confocal images processed for analysis were collected using a 63x oil immersion lens with a numerical aperture of 1.0; z- step, 0.5 µm; and image size 1024 X 1024 pixels. Cy3 was excited using the yellow line (568 nm) of the krypton/argon laser to collect images of synaptophysin, while the green line (488 nm) was used to collect images of Alexa 488- labeled neurotransmitters. All image acquisition and quantification was performed by individuals blind to the treatment conditions. Images were consistently collected at but not past the threshold of overexposure to standardize analysis parameters across image stacks. After all quantification had taken place, contrast and brightness were adjusted on images presented in the article prior to publication using Axiovision 4.4. Four z-stacks of the parvocellular division of the PVN (0.5 µm interval; optical sections per animal) were collected from each animal at approximately -1.8 mm from bregma (Paxinos and Watson, 1986). For analysis of median eminence CRH content, three z-stacks were taken at the medial portion of the median eminence, using only the Cy3 channel, with the same specifications at approximately -2.8 mm from bregma (Paxinos and Watson, 1986). CRH- labeling of neurons within the PVN was the guideline to determine that z-stacks were taken from the parvocellular division of the PVN as displayed in Figure

47 All image processing was performed on an IBM compatible computer using Zeiss LSM 510 Image Browser software. Five projections from the middle of each z-stack were created and subsequently quantified for percentage of the field density occupied by synaptophysin immunoreactivity. To produce each projection, z-stacks were subdivided into five consecutive images to ensure separation of synaptic boutons. Single projections (first angle = 0, maximum transparency) were generated for each subdivision of the z-stack. Projections were analyzed using the measurement function of Axiovision 4.4 software to obtain the field area percent occupied by the labeled synaptophysin within each projection. The threshold for pixel inclusion was obtained by analysis of several random projection images and was held constant for all images analyzed. The occupied field area percent across the z-stack was determined by averaging across the projections for each z-stack. For each animal the occupied field area percent was determined by averaging across the z-stacks taken from that animal. Finally, the field area percent was averaged across animals by treatment group (control vs. CVS). To determine CRH cell size and neurotransmitter innervation of the CRH cells, four CRH- expressing cells were chosen from each z-stack. In order to meet our criteria for selection, each cell had to have: 1) definitive CRH immunoreactivity, 2) the total z-plane of the soma visible within the z-stack, 3) a defined nucleus, and 4) sufficient separation from other cells to clearly identify neurotransmitter appositions. The cells were identified by slowly scrolling through the z-plane. The first four neurons in the z-stack that satisfied the criteria for analysis were selected by observers blind to treatment condition. In order to quantify the number of cell body neurotransmitter appositions, we counted the total number of immunoreactive boutons in apposition to the soma through the complete z-axis of each of the four cells. These appositions were defined by visualizing absolutely no visible space between the neurotransmitter terminal 46

48 marker and CRH cell membrane. Figure 2-2 demonstrates our criteria for the determination of neurotransmitter appositions. To ensure that each bouton was only counted once, the appositions were quantified while scrolling back and forth through all of the optical sections containing a specified quadrant of the immunoreactive cell body. Following the quantification of a given quadrant, we would move clockwise to the next quadrant until all of the appositions were accounted for. The geometry of CRH neurons did not permit visualization of full dendritic trees. To obtain unbiased estimates of neurotransmitter innervation of CRH dendrites, 8 12 dendrites per z-stack were chosen in a similar manner to the CRH- positive cells. Measured dendrites were all clearly separated from the other cells and dendrites. Dendritic appositions were counted in a similar fashion to soma appositions. Since the number of dendritic appositions is dependent on dendrite length, the length of each dendrite was calculated using the measurement function on Axiovision 4.4 in order to express the data as the number of appositions per 10 µm of CRH dendrite. For the analysis of cell size we estimated cell volume using the unbiased nucleator method (Gundersen, 1988). After estimating the radius at a consistent point (the nucleolus), we estimated cellular volume using the equation for the volume of a sphere: v = 4/3*r 3. Using Axiovision 4.4, we focused the CRH cell at the level of the nucleolus and measured the distance from the center of the nucleolus to the edge of the cell along a randomly selected angle between 1 90 and again 90, 180, and 270 degrees away from the original angle (Morrow et al., 2005). The radii were averaged for each cell and used to determine cellular volume. Each of these values (cell body appositions, dendritic appositions, and cell volumes) were averaged across the 47

49 z-stacks taken from each animal and then across the animals by each treatment group (control vs. CVS). Statistics. Data are reported as mean ± SEM. All data were analyzed by Student s t-tests. When necessary, data underwent a log transformation to achieve homogeneity of variance and then reanalyzed. Significance for each experiment was set at P < IV. Results Following the termination of chronic stress, we determined the efficacy of the chronic stress paradigm by assessing alterations in body weight change and thymus weight. As observed previously, CVS attenuated body weight gain (and decreased thymus weights (as a function of its body weight) (Table 2-2), providing verification of the efficacy of the stress regimen. Synaptophysin immunoreactivity was used as a marker of pre- synaptic terminals to determine whether chronic stress increased the number of synaptic inputs to the PVN (Fig. 2-3). The area occupied by immunoreactive synaptic boutons in the medial parvocellular PVN was calculated to quantify transmitter- specific terminal density. The relative density of synaptophysin in the CRH- containing medial parvocellular region of the PVN was higher in CVS than in control animals. This qualitative observation is supported by quantitative data (Fig. 2-3C) demonstrating that CVS exposure increased the area occupied by synaptophysin- labeled terminals within the PVN by about 125% [t(10) = 3.72; P < 0.05], consistent with neuroplastic alterations in pre- synaptic inputs to the PVN (Marqueze-Pouey et al., 1991). Next we tested whether chronic stress induces specific neurotransmitter changes to the CRH cell bodies of the medial parvocellular PVN. Therefore, we predicted that CVS either 48

50 increased the number of glutamatergic and/or noradrenergic terminals in apposition to the CRH cell bodies or decreased the number of GABAergic terminals. One week exposure elicited marked changes in both glutamatergic [t(14) = 8.898, P < 0.05] and noradrenergic [t(14) = 3.553, P < 0.05] terminals in apposition to CRH neurons, as illustrated in Figure 2-4. In contrast, there was no change in the GABAergic innervation of CRH somata. It is well known that neuronal somata and dendrites are selectively targeted by afferents from different cell groups. Given that medial parvocellular PVN glutamate and GABA inputs may emanate from multiple sources, we assessed CVS- induced changes in dendritic innervation (Fig. 5). In agreement with the results above, CVS increased the number of VGluT2 [t(14) = 2.371, P < 0.05] and DBH [t(14) = 2.705, P < 0.05] boutons in apposition to CRHimmunoreactive dendrites within the PVN. There were no changes in GAD65- immunoreactive dendritic appositions. These results indicate that CVS alters excitatory innervation of dendrites as well as somata of medial parvocellular CRH neurons. Magnocellular neurons in the supraoptic nucleus (SON) exhibit cellular hypertrophy following dehydration (Hatton and Walters, 1973; Mueller et al., 2005), but it is unknown whether chronic stress similarly affects CRH neuronal size within the PVN. We therefore estimated CRH somatic volume. One week of chronic stress caused a 148% increase in PVN CRH soma volume [t(14) = 4.844, P < 0.05], indicating that CVS induces cellular hypertrophy of PVN CRH neurons (Fig. 2-4T). Due to the observation of CRH cellular hypertrophy following 1 week of chronic stress, we also investigated whether 1 week of chronic stress alters CRH fiber density in the median eminence, where the terminals of the PVN CRH cells are located. CVS did not alter the CRH 49

51 fiber density within the median eminence (Fig. 2-6), indicating that the chronic stress is unlikely to alter the resting pool of releasable CRH in the PVN. V. Discussion The results of this study support the hypothesis that chronic stress produces striking alterations in neurotransmitter innervation of CRH neurons in the parvocellular PVN. Presynaptic innervation of the PVN, as determined by synaptophysin immunoreactivity, was markedly enhanced by prior chronic stress. Further analysis revealed that at least part of this enhanced pre- synaptic innervation is due to increased glutamatergic and noradrenergic inputs onto somata and dendrites of CRH- expressing cells in the parvocellular PVN following chronic stress. Enhanced excitatory innervation following chronic stress was associated with hypertrophy of parvocellular PVN CRH neurons, but not with changes in CRH stores in fibers located in the median eminence. Chronic excitation of the HPA axis due to CVS exposure produced a robust increase in the density of synaptophysin, a synaptic vesicle glycoprotein that is an indirect marker of nerve terminal innervation (Masliah and Terry, 1993). Increased density of synaptophysin is suggestive of increased synaptic input, in line with our hypothesis of altered neurotransmitter innervation following chronic stress. These data can be interpreted to indicate that number and/or strength of PVN synaptic input is enhanced post stress. The current results cannot distinguish among these possibilities. Prior work indicates that chronic restraint stress attenuates synaptophysin protein and mrna levels in the hippocampus (Cunha et al., 2006; Thome et al., 2001; Xu et al., 2004), a limbic region that exerts negative regulatory control over the HPA axis (Herman et al., 2005). Diminished levels of synaptophysin are suggestive of a loss of synaptic 50

52 contacts, and electron microscopy studies confirm that chronic stress decreases the total synapse density in CA3 hippocampal subfield (Sandi et al., 2003; Sousa et al., 2000b), as well as simple asymmetric synapse density in CA3 (Sandi et al., 2003). Collectively, these findings suggest that altered synaptic density occurs in multiple brain regions following stress exposure. Notably, the loss of synapses is associated with functional deficits, which may result in impaired HPA axis inhibition and contribute to the enhancement of synaptic density seen in the PVN (Sousa et al., 2000a; 2000b). Glutamatergic innervation of the CRH parvocellular PVN neurons, as reflected by VGluT2 immunoreactivity, was markedly increased following exposure to chronic stress. VGluT1 and VGluT2 immunoreactive terminals are accepted markers of pre- synaptic glutamatergic terminals, given that these trans- membrane proteins specifically transport glutamate into pre- synaptic vesicles (Moechars et al., 2006). Notably, VGluT2 is the primary form within the parvocellular division of the PVN (Herzog et al., 2001; Ziegler et al., 2002). Enhanced density of VGluT2 innervation following chronic stress likely reflects an increased capacity for glutamatergic stimulation of CRH neurons within the parvocellular division of the PVN. In agreement with the hypothesis, recent work indicates that VGluT2- immunoreactive axons densely innervate all parvocellular divisions of the PVN, with VGluT2 boutons establishing close contacts to all CRH neurons in the medial parvocellular PVN (Wittmann et al., 2005). Moreover, ultra- structural analysis verified that VGluT2 terminals closely appose CRHir perikarya and dendrites (Wittmann et al., 2005). While the anatomical origin(s) of the enhanced glutamatergic innervation is unknown, there are several likely hypothalamic and limbic sources, including the suprachiasmatic nucleus, dorsomedial hypothalamus, anterior hypothalamic nucleus, perifornical area, as well as within the PVN (Boudaba et al., 1997; 51

53 Daftary et al., 1998; Hermes et al., 1996). There is preliminary evidence to suggest that the increased glutamatergic innervation of CRH neurons of the PVN results in an increase in the synaptic excitation of these cells (Franco et al., 2007). Norepinephrine terminals in apposition to CRH neurons were also increased after CVS. The PVN receives prominent Norepinephrine/Epinephrine input from brainstem regions (Cunningham and Sawchenko, 1988; Sawchenko and Swanson, 1982), and NE exerts an excitatory influence on the HPA axis (Plotsky, 1987). Previous studies indicate that sensitized HPA axis responses to immobilization stress after chronic cold stress are attributable to enhanced responsiveness of postsynaptic α1-adrenergic receptors in the PVN (Ma and Morilak, 2005) rather than enhanced NE release per se. It is possible that enhanced responsiveness may be associated with increased proximity of NE terminals onto CRH- containing somata. Chronic stress may induce any number of structural or functional alterations in norepinephrine neurons that would increase PVN excitation. In fact, electrophysiological evidence points to norepinephrine having an excitatory effect on PVN neurons by enhancing pre- synaptic glutamate release (Daftary et al., 2000). It is unclear how changes in excitatory innervation affect the excitability of CRH neurons of the parvocellular PVN. It is logical to predict that the increased number of excitatory synapses would lead to an increased capacity for excitation of the neurons, but this remains to be tested. Nonetheless, it is important to consider that similar increases in excitatory innervation are observed in magnocellular neurons following both lactation (El Majdoubi et al., 1996) and dehydration (Mueller et al., 2005), and these changes are accompanied by an increase in glutamatergic excitatory synaptic inputs to these cells (Di and Tasker, 2004; Stern et al., 2000). 52

54 We found that the GABAergic innervation of the parvocellular division of the PVN was not altered by chronic stress. A prior study reported that chronic intermittent stress for 21 days markedly attenuated basal inhibitory postsynaptic currents in the parvocellular division of the PVN via a putative reduction in the number of functional GABA synapses (Verkuyl et al., 2004). These findings suggest that stress may result in a decrease in physiologically active synapses without affecting gross synaptic number. In addition, postsynaptic alterations in the PVN GABAergic innervation may also occur, as mrna transcripts encoding the β1 and β2 subunits of the GABA A receptor are diminished in the PVN following CVS (Cullinan and Wolfe, 2000). Previous studies in the magnocellular system indicate that chronic stimulation affects cell size. For example, magnocellular neurons in the supraoptic nucleus exhibit hypertrophy during lactation (Gies and Theodosis, 1994), and following chronic dehydration (Hatton and Walters, 1973; Mueller et al., 2005), and chronic restraint (Miyata et al., 1994), which suggests that cellular hypertrophy may be a common neurobiological adaptation to increased activity. We observed a significant increase in CRH cell volume following chronic stress (48%) that was similar to that reported in the magnocellular system. Confocal microscopic quantification of pre- synaptic terminals in apposition to neurons has been used previously to assess chronic dehydration- induced changes to neurotransmitter afferents of SON vasopressin neurons (Mueller et al., 2005). In this previous study, chronic dehydration increased the number of GABA and glutamate terminals in apposition to vasopressin neurons, which replicated previous studies of synaptic plasticity using quantitative electron microscopy (El Majdoubi et al., 1996; Gies and Theodosis, 1994). Thus, while the confocal approach cannot definitively identify synaptic contacts, it provides a reliable estimate of neuroplastic events occurring within CRH- containing cell groups in the PVN. 53

55 While these types of morphological changes have been widely documented in hypothalamic magnocellular neurons of both the SON and PVN, the cellular mechanism of these alterations has not yet been elucidated. Importantly, cell adhesion molecules have been linked to hypothalamic neuroplasticity, with Polysialic Acid- Neural Cell Adhesion Molecule (PSA- NCAM) being the most widely studied. PSA- NCAM is robustly expressed throughout the SON and PVN (Bonfanti et al., 1992) and is regulated during lactation (Soares et al., 2000). In addition, polysialylation of NCAM in the SON is necessary for the lactation- and dehydrationinduced increase in synapses onto magnocellular neurons (Theodosis et al., 1999). It remains to be determined whether PSA- NCAM plays a causal role in PVN stress- related plasticity. We did not observe a CVS- induced change in CRH content in the median eminence. These data suggest that capacity of CRH storage in the median eminence remains stable despite marked enhancement of CRH biosynthesis (Herman et al., 1995b; Imaki et al., 1991; Kiss and Aguilera, 1993; Makino et al., 1995), CRH cell size, and excitatory inputs to CRH neurons. Together, the data are consistent with increased post stress release of CRH, although this prediction requires testing. Previous studies also report inconsistent effects of stress on median eminence CRH stores. Chronic stress paradigms have equivocal effects on the quantity of CRH immunoreactivity in the median eminence (Chappell et al., 1986; de Goeij et al., 1991; Inoue et al., 1993). Acute immobilization stress also has inconsistent effects on CRH stores in the median eminence, with some reports indicating decrements in CRH immunoreactivity and others reporting no significant changes (Chappell et al., 1986; Culman et al., 1991; Inoue et al., 1993). There are some interpretive caveats that bear consideration. First, since we did not include an acute stress group, we cannot definitively conclude that the observed changes are specific to chronic stress. It is possible that a single stressful event can produce rapid changes in 54

56 synaptic organization. For example, there is some evidence indicating that acute dehydration (24 hours) can alter soma- somatic appositions and enhances dendritic bundling in magnocellular neurons of the supraoptic nucleus (Hatton et al., 1984; Tweedle and Hatton, 1987). Importantly, chronic dehydration greatly augments these changes, as well as also producing double synapses (Tweedle and Hatton, 1984, 1987). Whereas there are no data indicating such rapid changes in axo- somatic and axo- dendritic appositions, as seen in the current study, we cannot definitively preclude this possibility. Second, our approach did not allow us to quantify neurotransmitter appositions within the full dendritic tree of CRH neurons, as 1) full arborizations could not be clearly demonstrated within the 30-µm sections used in this study, and 2) we cannot assume that CRH immunoreactivity completely fills parvocellular dendrites. Thus, it is possible that we missed changes in synaptic organization occurring on distal dendrites. In summary, our data are consistent with a marked neuroplasticity of excitatory neurotransmitter innervation of PVN CRH neurons induced by chronic stress. These results are consistent with abundant data documenting maintenance and, indeed, exacerbation of HPA axis responses following chronic stress, despite the existence of an enhanced feedback signal (i.e., elevated glucocorticoids). Together, the data suggest that a chronic activation of the HPA axis induces adaptations in PVN CRH neurons that allow the system to maintain the capacity to mount stress responses despite its prior stress history. Inappropriate regulation of this neuroplastic mechanism may contribute to glucocorticoid hypersecretion (and perhaps hyposecretion) seen in stress- related disease states and brain aging. 55

57 Table 2-1: Primary Antibodies used Antigen Immunogen Manufacturer, species, type, catalog number Synaptophysin C-terminus of human Zymed laboratories (San Francisco, synaptophysin (288- CA), rabbit antisera, # ) Dilution used 1:300 CRH human/rat CRF(1-41) Wylie Vale, rabbit antisera 1:25,000 Vesicular glutamate transporter 2 GST-fusion protein containing amino acid residues of rat Synaptic Systems, rabbit polyclonal, #135402, lot # :1,500 GAD-6 VGluT2/DNPI Adult rat glutamic acid decarboxylase purified Developmental Studies Hybridoma Bank (David Gottlieb), mouse monoclonal antibody, DBH Purified bovine DBH Chemicon, mouse monoclonal, MAB308, lot# :100 1:2,500 Table 2-2: Body and thymus weight Treatment Group Control (n=8) CVS (n=8) body weight change (g) (from original weight) * Thymus weight (% body weight) * Data are expressed as mean + SEM. * CVS is significantly different from control at P <

58 Figure 2-1 Figure 2-1: Images demonstrating location of Z-stacks taken within the medial parvocellular PVN. A: Image taken at -1.8 mm bregma using a 10X objective. A Cy3 fluorophore is attached to the CRH antibody. The box represents the approximate location of where z-stacks were taken, in the area with the highest CRH cellular density. B: Representative projection of 10 optical sections collected 0.5 µm apart using a confocal microscope with a 63x objective. Scale bars = 1 mm in A; 20 µm in B. 57

59 Figure 2-2 Figure 2-2: Demonstration of neurotransmitter apposition quantification. This figure contains a single optical section taken from a confocal z-stack. In this case, DBH is in green and CRH in red. The solid white arrows in the figure demonstrate boutons that we defined as appositions. However, boutons pointed out by the dashed white arrows did not meet our criteria for an apposition. Scale bar = 10 µm. Figure 2-3 Figure 2-3: Effects of CVS on densities of synaptic terminals. Images are taken from the PVN of control rats (A) and rats treated with CVS for 1 week (B), and represent the projections of five optical sections 0.5 µm apart using a 63x objective. Chronic stress exposure markedly increased the number of synaptic inputs (synaptophysin) within the PVN (C). Data are expressed as mean percentage of control ± SEM; n = 8 for control, n = 8 for CVS; *CVS is significantly different from control at P < Scale bars = 20 µm. 58

60 Figure 2-4 Figure 2-4: Effect of CVS on somatic neurotransmitter appositions on CRH neurons. Series of four optical sections 0.5 µm apart showing VGluT2- labeled (A D; green), DBH- labeled (K N; green), and GAD- labeled (F I; green) immunoreactive pre- synaptic terminals co- localized with CRH- labeled neurons (magenta) in the parvocellular portion of the PVN. Quantitative results are illustrated in E,J,O,T. Chronic stress increased the number of glutamatergic and noradrenergic inputs to CRH neurons of the parvocellular PVN, but had no effect on GABAergic innervation. Chronic stress also caused an increase in the cell volume of CRH- expressing cells, measured at the nucleolus (T). Neurotransmitter apposition and CRH estimated volume data are expressed as the average number of appositions per cell ± SEM; n = 8 for control, n = 8 for CVS; *CVS is significantly different from control at P < Scale bars = 10 µm. 59

61 Figure 2-5 Figure 2-5: Increase in terminal appositions to PVN CRH dendrites following chronic stress. Series of four optical sections 0.5 _m apart for VGluT2- labeled (A D; green), DBH- labeled (K N; green), and GAD- labeled (F I; green) immunoreactive pre- synaptic terminals apposed to CRH- labeled dendrites (magenta) in the medial parvocellular PVN. Quantitative results are illustrated in E,J,O. Chronic stress increased the number of glutamatergic and noradrenergic appositions on CRH dendrites in the parvocellular PVN, but had no effect on the GABAergic innervation. Neurotransmitter apposition data are expressed as the average number of appositions per cell ± SEM; n = 8 for control, n = 8 for CVS; *CVS is significantly different from control at P < Scale bars = 10 µm. 60

62 Figure 2-6 Figure 2-6: CRH fiber density within the median eminence is not altered by chronic stress. A: The approximate area of the median eminence where confocal z-stacks were compiled from. B: A representative image of a projection of median eminence CRH used for quantification. One week of CVS had no effect on the density of CRH immunostaining in the median eminence, expressed as the average mean percentage of control ± SEM; n =8 for control and CVS rats (C). Scale bars = 1 mm in A; 20 µm in B. 61

63 Chapter 3 Identification of Chronic Stress- Recruited Circuitry in Rat Brain 62

64 I. Abstract Chronic stress induces pre-synaptic and post-synaptic modifications in the paraventricular nucleus of the hypothalamus (PVN) that are consistent with enhanced excitatory hypothalamopituitary-adrenocortical (HPA) axis drive. The brain regions mediating these molecular modifications are not known. We hypothesized that chronic variable stress (CVS) tonically activates stress-excitatory regions that interact with the PVN, culminating in stress facilitation. In order to identify chronically activated brain regions, ΔFosB, a documented marker of tonic neuronal activation, was assessed in known stress regulatory limbic and brainstem regions. Four experimental groups were included: CVS, repeated restraint (RR) (control for HPA habituation), animals weight-matched (WM) to CVS animals (control for changes in circulating metabolic factors due to reduced weight gain), and non-handled controls. CVS, but not RR or WM, induced adrenal hypertrophy, indicating sustained HPA axis drive only by CVS. CVS (but not RR or WM) selectively increased the number of FosB/ΔFosB nuclei in the nucleus of the solitary tract, posterior hypothalamic nucleus, and both the infralimbic and prelimbic divisions of the medial prefrontal cortex, indicating an involvement of these regions in chronic drive of the HPA axis. Increases in FosB/ΔFosB-immunoreactive cells were observed following both RR and CVS in the other regions (e.g., the dorsomedial hypothalamus), suggesting activation by both habituating and non-habituating stress conditions. The data suggest that unpredictable stress uniquely activates interconnected cortical, hypothalamic, and brainstem nuclei, suggesting the existence of a recruited circuitry mediating chronic drive of brain stress effector systems. 63

65 II. Introduction Appropriate stress regulation is essential for the maintenance of homeostasis, as both elevated and attenuated glucocorticoid levels are associated with physiological and behavioral dysfunction (de Kloet et al., 2008; de Quervain et al., 2009; Handwerger, 2009; Lupien et al., 2007). The ability to adapt to adverse circumstances is a dynamic process that can be perturbed by a number of physiological and behavioral states, including chronic periods of stress. Following chronic stress, rats exhibit exaggerated glucocorticoid responses to novel stress, develop anhedonia, reduced body weight, and adrenal hypertrophy, a constellation of physical and behavioral sequelae consistent with a depressive-like state (Akana et al., 1992; Herman et al., 1995a; Papp et al., 1991). These physiological and behavioral effects are accompanied by the re-organization of neural structures, characterized by changes in synaptology (Carvalho- Netto et al., 2011), dendritic morphology (Cook and Wellman, 2004), and gene expression (Andrus et al., 2010), all likely mediated by chronically driven stress-sensitive circuits. However, the neural circuits that drive maladaptive responses to chronic stress are unknown. Previous studies suggest that the paraventricular nucleus of the hypothalamus (PVN) is a site that is essential for the initiation of stress responses. Chronic stress induces changes in PVN peptide expression (Herman et al., 1995b; Imaki et al., 1991; Kiss and Aguilera, 1993; Makino et al., 1995), neurotransmitter receptor expression (Cullinan and Wolfe, 2000; Ziegler et al., 2005), and neural excitability (Verkuyl et al., 2004) that are all consistent with an enhanced post-stress drive this nucleus. These findings suggest that PVN neurons are hyper-reactive and hyperresponsive to stimuli following repeated stimulation, contributing to the facilitatory effects of chronic stress on acute hormone release. In addition to these largely postsynaptic effects, chronic stress elevates the number of pre-synaptic excitatory neurotransmitter boutons 64

66 (glutamate and norepinephrine) in apposition to the CRH cell bodies and dendrites (Flak et al., 2009), indicating that the excitatory PVN afferents are are also enhanced following chronic stress. In this study, we mapped brain regions tonically activated by chronic stress, using an indirect marker of chronic neuronal activation, ΔFosB. FosB, a member of the fos-related antigen family of immediate early genes, shows peak induction 4-6 hours post-stimulus (Nestler et al., 2001). Following translation, FosB is cleaved and phosphorylated into the more stable ΔFosB (McClung et al., 2004; Ulery et al., 2006), which accumulates up in cells providing lasting effects on gene transcription (Nestler et al., 2001). Therefore, habituating chronic stress exposure (e.g. repeated restraint and food restriction) will not induce FosB/ ΔFosB within relevant chronic stress circuitry, as their responses attenuate over time. We used FosB/ΔFosB immunohistochemistry to test the hypothesis that chronic stress (CVS)-selectively recruits circuitry involved in control of stress responses, including afferent inputs to the PVN. III. Materials and Methods Subjects. Male Sprague-Dawley rats from Harlan (Indianapolis, IN), weighing g upon arrival, were housed one in clear polycarbonate cages containing granulated corncob bedding, with food and water available ad libitum. The colony room was temperature- and humiditycontrolled with a 12 h light cycle (lights on 6:00 am; lights off 6:00 pm). Rats acclimated to the colony facility for one week prior to experimental manipulations. All experimental procedures were conducted in accordance with the National Institutes of Health Guidelines for the Care and Use of Animals and approved by the University of Cincinnati Institutional Animal Care and Use Committee. Prior to experimental manipulation, the animals were weighed and placed into 65

67 groups such that there was no difference in starting body weight between groups. Rats (Chronic Variable Stress (CVS) (n=8), Repeated Restraint (RR) (n=8), Weight-Matched (WM) (n=9), and Control (n=8)) were perfused at least 16 hours following the termination of the day 14 stressor without acute stimulation, to negate the contribution of acute FosB induction and therefore assess central ΔFosB expression as a measure of chronic drive. Weight Matching. Animals were fed a sufficient amount of chow to produce a similar reduction in weight gain compared to the chronically stressed animals. In order to provide a relatively low-stressful means of food restriction, animals were fed 5 grams of chow in the morning at a random time between 7am and 11am and the rest between 5:30 and 6pm. Previous studies indicate that feeding the animals just before lights off does not shift their circadian rhythm. We added the morning feeding in order to limit the total amount of time without food (since rats typically eat small amounts during the light period of the circadian cycle). Feeding them in this manner does not shift their circadian pattern of heart rate and blood pressure (Flak et al., 2011). WM, RR, and CVS animals did not differ in body weight at any point in the experiment. Chronic Stress Procedure. Subjects were randomly assigned to either CVS, repeated restraint, weight-matched, or control groups. The chronic stress protocol consisted of twice-daily (morning and afternoon) exposure to randomly assigned stressors for two weeks. Morning stressors were conducted between 8:00 am and 11:30 am and afternoon stressors were administered between 1:30 pm and 5:00 pm. Stressors consisted of rotation stress (1 h at 100 rpm on a platform orbital shaker), warm swim (20 min at 31 o C); cold swim (10 min at 18 o C), cold room stress (kept in 4 o C for one hour) and hypoxia (8% O 2 92% N 2, 30 min). Repeatedly 66

68 restrained animals were placed in a Plexiglas restraint tube for one hour per day in the AM. The morning following the last afternoon stressor, rats received an overdose of sodium pentobarbital and were perfused with phosphate buffered saline, followed by 4% paraformaldehyde. Brains were post-fixed overnight in 4% paraformaldehyde and transferred to 30% sucrose at 4 o C until they were cut on a freezing microtome. Immunohistochemistry. Sections were cut at 35 µm on a sliding microtome throughout the brain and resulting sections were stored in cryoprotectant (0.1 M phosphate buffer, 30% sucrose, 1% polyvinylpyrrolidone, and 30% ethylene glycol) at 20 C until used for immunohistochemistry. Sections were transferred from cryoprotectant to 50 mm potassium phosphate buffered saline (KPBS; 40 mm potassium phosphate dibasic, 10 mm potassium phosphate monobasic, and 0.9% sodium chloride) at room temperature (RT). Cryoprotectant was rinsed off (5 x 5 min) in KPBS, the sections were transferred to KPBS + 1.0% H 2 O 2, and incubated for 10 minutes at room temperature (RT). Sections were then washed (5 x 5 min) in KPBS at RT, and placed in blocking solution (50 mm KPBS, 0.1% bovine serum albumin (BSA), and 0.2% Triton X-100) for 1 hour at RT. Sections were incubated overnight at 4 C in rabbit anti-fosb/δfosb primary (H75, Santa Cruz Biotechnologies; Santa Cruz, CA) diluted 1:300 in blocking solution. This antibody was raised against amino acids of the human FosB and detects two bands, one at kda (ΔFosB) and another at 45 kda (FosB). (Marttila et al., 2006). The following morning sections were rinsed in KPBS (5 x 5 min) and incubated in biotinylated anti-rabbit secondary antibody (Vector Laboratories, Inc., Burlingame, CA), diluted 1:500 in KPBS + 0.1% BSA for 1 hour at RT. Sections were rinsed in KPBS (5 x 5 min) and then treated with avidin-biotin complex (ABC, Vector Laboratories, Inc., Burlingame, CA) at 67

69 1:1,000 in KPBS + 0.1% BSA for 1 hour at RT. Following this incubation, sections were rinsed again in KPBS (5 x 5 min) and reacted with.02% diaminobenzamidine (Sigma Aldrich, St. Louis, MO) with.05% hydrogen peroxide. Sections were rinsed a final time in KPBS and coverslipped in DPX (Sigma Aldrich, St. Louis, MO) following mounting and dehydration through ethanol rinses. Sections used for quantification of co-labeled for FosB/ΔFosB and dopamine-beta-hydroxylase (DBH) were placed through a similar procedure on the first day of immunohistochemistry, but additionally incubated with mouse anti-dbh (Millipore; Temecula, California) at 1:2500. Following overnight incubation with both primary antibodies, the sections were rinsed in KPBS (5 X 5min) and incubated in both Cy3 goat anti-rabbit (Jackson Immuno Research; West Grove, PA) and Alexa 488 donkey anti-mouse (Molecular Probes; Eugene, OR) for thirty minutes. Next, the sections were washed a final time in KPBS and coverslipped with Fluka mounting medium (Sigma Aldrich; St. Louis, MO). The slides were left to dry for a sufficient time before capturing images for quantification. It is important to note that the FosB antibody used will recognize both cleaved (ΔFosB) and uncleaved forms (FosB). Uncleaved Fos B generally diminishes to baseline within 6 hours) (Nestler et al., 2001) and thus we should only be detecting ΔFosB at the time of kill (16h after the last stressor). Nonetheless, to be technically correct, we will refer to the immunohistochemical staining as FosB/ΔFosB. Cell counts. The number of FosB/ΔFosB immunoreactive neurons within the brain regions was quantified with the Zeiss Axiovision Program. Image magnifications were chosen in order to assess the number of immunoreactive nuclei in one unilateral image from each section. If 68

70 possible, four images per animal per region of interest (ROI) were collected. To quantify the number of immunoreactive nuclei, both a threshold gray level and minimum pixel size were determined for each objective using a subset of images for each region with varying signal and background intensities. The program recorded the number of immunoreactive nuclei within the defined ROI. All cell counts were converted to the number of immunoreactive cells per unit area and analyzed. The regions were delineated using characteristics of each nucleus taken from the Paxinos and Watson atlas (Paxinos and Watson, 2005a). and quantified at a similar distance from bregma within all experimental animals. The number of co-labeled FosB/ ΔFosB and DBH were counted manually. Images were collected on 2 right and 2 left NTS at approximately Bregma -14mm. Immunoreactive nuclei were first selected in the Cy3 channel. Then, the Alexa channel was added to record the number of cells that also contain cytosolic DBH staining. Statistics were analyzed using Sigma Stat (Systat Software, San Jose, California). Data are expressed as mean ± standard error. Outliers were determined if the value exceeded both 1.96 times the standard deviation and 1.5 times the interquartile range (McClave, 1994). The data were analyzed by one way ANOVA with a Fisher s LSD post-hoc test with group (CVS, WM, RR, and control) as a between subject factor. If an ANOVA failed homogeneity of the variance, the data underwent log transformation prior to analysis. 69

71 IV. Results We used three groups to control for non- chronic stress- specific changes in FosB induction: non- handled animals were used as a general unstressed control, weight- matched animals controlled for the passive effects of reducing weight gain to CVS levels, and repeatedly restrained animals controlled for the generalized induction under habituating conditions. At the end of the two week experiment, animals were all killed without acute disturbance, in order to minimize FosB production. As previously reported (Herman et al., 1995a), CVS caused adrenal hypertrophy {F(3,32)= , p<.001} and attenuated body weight gain {F(3,32)= , p<.001}. We used FosB/ΔFosB immunohistochemistry to identify regions containing immunoreactive nuclei. Surprisingly, a number of areas known to be sensitive to behavioral and metabolic disturbance exhibited little or no FosB/ΔFosB immunoreactivity, including the PVN, the ventromedial hypothalamic nucleus, the arcuate nucleus, and the central nucleus of the amygdala. Of the sites that contained FosB/ΔFosB immunoreactivity, we selected regions that have been previously been shown to contain glutamate and/or norepinephrine neurons that project to the PVN for FosB/ΔFosB analysis. In addition, we also chose to analyze the amygdala, hippocampus, and medial prefrontal cortex (mpfc), as they contained robust FosB/ΔFosB immunoreactivity, exhibit neuroplastic responses to chronic stress, and are upstream regulators of HPA axis function (Ulrich-Lai and Herman, 2009). To control for widespread effects of our stress paradigm, we analyzed FosB/ΔFosB immunoreactivity within the interstitial nucleus of the posterior limb of the anterior commissure (IPAC) and M1 of the motor cortex. The groups did not differ in their FosB/ΔFosB expression (Figure 3-1) within these two sites, indicating that our reported changes are specific to stress- sensitive neurons and do not reflect global neural activation. 70

72 PVN- projecting regions Our qualitative analyses located several sites that had variable levels of FosB/ΔFosB immunoreactivity across groups, the most notable being the nucleus of the solitary tract (NTS). FosB/ΔFosB in the NTS stood out, since staining was generally sparse in the brainstem. Quantitative analyses indicated that CVS increased FosB/ΔFosB within the NTS {F(3,30)=47.568, p<.001} (Figure 3-2A). This area is particularly interesting, as it includes the A2/C2 catecholaminergic neurons that project into the PVN (Cunningham and Sawchenko, 1988). Double-label analysis revealed increased colocalization of FosB/ΔFosB and the norepinephrine/epinephrine marker DBH in the NTS of CVS rats {F(3,31)=33.084, p<.001} (Figure 3-2B), indicating that unpredictable stress tonically activates noradrenergic NTS neurons. Importantly, no increases in NTS FosB/ΔFosB were observed in RR or WM groups. We also quantified FosB/ΔFosB in select regions known to include glutamate- expressing PVN- projecting neurons. CVS increased FosB/ΔFosB within the dorsomedial hypothalamic nucleus (DMH) {F(3,32)=27.534,p<.001} (Figure 3-3A) and posterior hypothalamic nucleus (PH) {F(3,23)=12.755,p<.001} (Figure 3-3B), indicating that chronic unpredictable stress activates these regions. However, RR also increased DMH FosB/ΔFosB expression (Figure 3-3B), suggesting that this region is chronically activated during both unpredictable and predictable stress. Upstream Limbic Structures We also quantified FosB/ΔFosB in the amygdala, hippocampus and prefrontal cortex, upstream limbic structures critical for stress regulation. FosB/ΔFosB immunoreactivity was rather abundant in these structures compared to the hypothalamic and brainstem regions 71

73 previously analyzed. Staining was particularly robust within the mpfc. Because side (Sullivan and Gratton, 1999) and subregion (Radley et al., 2006) effects are reported following lesions of the mpfc, we divided the region of interest into both left and right infralimbic and prelimbic mpfc to assess specific chronic stress- activated areas of the mpfc. CVS elevated the FosB/ΔFosB expression within the left infralimbic {F(3,31)=5.823, p=.03} (Figure 3-4A), left prelimbic {F(3,32)=7.119, p=.001} (Figure 3-4B), right infralimbic {F(3,32)=7.808, p<.001} (Figure 3-4C), and right prelimbic {F(3,31)=9.063, p<.001} (Figure 3-4D) cortices. In all cases, increased FosB/ΔFosB staining was specific to the CVS group. We also analyzed FosB/ΔFosB within the basolateral amygdala (BLA), as this nucleus exhibited the most prominent staining (FosB/ΔFosB immunoreactivity in the medial and central nuclei was minimal). BLA FosB/ΔFosB did not differ between treatment groups (Figure 3-5A). Hippocampal FosB/ΔFosB was restricted to the dentate gyrus. Our analyses found that CVS and RR both increased FosB/ΔFosB within the dentate gyrus {F(3, 31)=4.806, p=.008} (Figure 3-5B), indicating that this region is responsive to unpredictable as well as predictable stress. Discussion The data from the current study suggests that unpredictable stress (i.e., CVS) chronically activates a number of known stress-regulatory regions, including areas that project directly into the PVN (i.e., PH) and upstream limbic structures (i.e., mpfc) that indirectly regulate HPA axis activity. These data support the existence of a recruited chronic stress pathway that involves prefrontal cortex, posterior hypothalamus and NTS, putatively responsible for sustaining and amplifying stress responsivity during prolonged stimulation. Other regions, such as the DMH and DG, increase FosB/ΔFosB staining in response to both habituating and non-habituating 72

74 stress regimens, suggesting that these regions are not required for chronic drive of the HPA axis by unpredictable stress. Our analyses revealed several sites activated solely by a history of unpredictable stress and known to provide excitatory neurotransmitter input to the PVN. The NTS contained the most pronounced FosB/ΔFosB induction, including the A2/C2 catecholamine cell group. This finding is particularly important, as this cell group provides the majority of norepinephrine afferents to the medial parvocellular PVN (Cunningham and Sawchenko, 1988). Previous evidence suggests that NE/E from medullary regions play an excitatory role in HPA axis drive, as lesions of ascending medullary PVN inputs attenuate CRH immunoreactivity and PVN cfos responses to stress (Li et al., 1996). Furthermore, CVS elevates the NTS TH mrna and hnrna (Zhang et al., 2010), suggesting that unpredictable stress exposure enhances the output of A2/C2. In addition, CVS increases the number of PVN DBH-positive boutons in apposition to CRH neurons (Flak et al., 2009), indicating that chronic stress exposure enhances the noradrenergic influence in the initiation of stress responses. Thus, this region appears to be critical for responding to chronic stress. However, the NTS may not provide its effects on stress regulation directly through the PVN, since these neurons project to additional regions important in stress regulation including the rostral ventrolateral medulla, bed nucleus of the stria terminalis, and the amygdala (Ricardo and Koh, 1978). As these neurons are tonically activated over time by chronic stress, the output from this region is continually enhanced, facilitating HPA axis activity and suggesting that these neuroplastic changes within the NTS A2 region may form a final common pathway for mediating chronic stress-related behavioral and physiological dysfunction. 73

75 Importantly, some of the FosB/ΔFosB immunoreactive nuclei within the NTS were not DBH-positive. It is likely that non-noradrenergic NTS FosB/ΔFosB neurons, (e.g., those that express glucagon-like-peptide-1 (GLP-1)), may also play a role in stress regulation. For example, NTS GLP-1 can contribute to both psychogenic and systemic responses to stress (Kinzig et al., 2003), and GLP-1 neurons innervate numerous stress-regulatory regions, notably including the PVN (Larsen et al., 1997; Llewellyn-Smith et al., 2011; Tauchi et al., 2008). Furthermore, chronic stress reduces GLP-1-positive PVN fiber density and NTS PPG mrna (Zhang et al., 2010), suggestive of the role of these neurons in chronic stress regulation. It should be noted that GLP-1 and NE are not the sole transmitters expressed in NTS, as the region contains a host of neurotransmitters and peptides (Maley, 1996), including GABA, glutamate, cholecystokinin, calcitonin gene-related peptide, galanin, neuropeptide Y, among others, that could also be modified by chronic stress exposure In addition to the evidence that PVN NE release is driven by unpredictable but not predictable chronic stress exposure, our analyses found that CVS elevates PH FosB/ΔFosB immunoreactivity. Given evidence for a strong glutamatergic innervations of the PVN by the PH (Ulrich-Lai et al., 2011b), it is possible that unpredictable stress may also drive PVN glutamate release and contribute to enhanced density of glutamatergic appositions onto CRH neurons (Flak et al., 2009). In support, excitation of the PH can initiate cardiovascular and behavioral stress responses (DiMicco et al., 1986; Shekhar and DiMicco, 1987), and pharmacological blockade can attenuate stress-induced tachycardia (Lisa et al., 1989), suggesting that the PH may also influence other types of stress responses. In addition to glutamatergic neurons, the PH expresses a host of transmitters, including orexin, that are implicated in vigilance and regulating 74

76 wakefulness (Abrahamson and Moore, 2001). Given its selective recruitment during stress exposure, the PH has the potential to mediation of responses to prolonged stress. As previously shown (Perrotti et al., 2004), we observed pronounced ΔFosB staining within the medial prefrontal cortex following CVS. Previous studies have indicated that the prelimbic mpfc inhibit HPA axis responses to stress, whereas the infralimbic mpfc elevates glucocorticoid release (Radley et al., 2006) and increases pressor responses following acute stress exposure(resstel et al., 2006). Our results indicate that CVS activates the prelimbic and infralimbic mpfc to a similar degree, suggesting that the net impact of prefrontal activation may be translated downstream of the cortex. Prior studies indicate that the right mpfc plays a more prominent role than the left side in controlling responses to stress (Sullivan and Gratton, 1999), but we did not find a lateralized effect of CVS on mpfc activation. Overall, CVS clearly activates the mpfc and likely provides lasting changes to gene transcription to this region. In addition to its role in the tuning of the stress responses, the mpfc is critical to short term memory formation (Barsegyan et al., 2010; Floresco et al., 1997), reward processing (Capriles et al., 2003), appraisal (Schmitz and Johnson, 2007), and cognitive control (Maier et al., 2006), all of which can be regulated by chronic stress. The accumulation of FosB/ΔFosB in prefrontal cortical regions may provide a means by which chronic stress promotes long-lasting behavioral as well as hormonal dysfunction. Both CVS and RR induced the expression of ΔFosB within the DMH. The DMH is primarily known to regulate the SNS, in that DMH inactivation attenuates (Stotz-Potter et al., 1996b) and stimulation exacerbates cardiovascular responses to psychogenic (Bailey and Dimicco, 2001), but not systemic stress. This region a rich number of GABAergic neurons that are wholly separate from the glutamatergic neurons, which also project into the PVN and likely 75

77 regulate responses to stress (Herman et al., 2003). Note that it is possible that CVS may preferentially activate DMH glutamatergic neurons, while RR may activate GABAergic neurons. No changes in accumulation of FosB/ΔFosB were observed in any region in the WM group. These data indicate that the effect of CVS is not accounted for by changes in metabolic status induced by lowered body weight. Caloric restriction sensitizes reward circuits (Carr and Wolinsky, 1993; Carroll et al., 1979; Stamp et al., 2008) in order to re-establish previous energy storage levels, which are believed to be partially mediated by FosB/ΔFosB within the nucleus accumbens (Vialou et al., 2011). Interestingly, RR, CVS, and WM animals all displayed slight, non-significant increases in FosB/ΔFosB within the BLA, another known brain region important for reward processing. Since chronic stress can produce anhedonia, the ΔFosB within reward regions induced by stress may dampen, while caloric restriction may sensitize reward activation. The use of ΔFosB as a marker for tonic neuronal activation does potentially introduce false negatives, since the necessary mechanisms of FosB transformation into ΔFosB are, to this point, poorly understood. Therefore, it is possible that some nuclei contain neurons that are unable to convert Fosb to ΔFosB. For example, the PVN is activated (as determined by c-fos mrna expression and Fos protein staining) in response to each of the stressor of CVS and RR, but this area does not exhibit significant ΔFosB staining, despite displaying numerous other markers of chronic activation (e.g., increased CRH and vasopressin gene transcription, cellular hypertrophy). Studies have claimed that the induction of ΔFosB within the PVN following drug administration (Chocyk et al., 2006; Das et al., 2009; Nunez et al., 2010). However, the temporal aspects of these previous studies make it unclear whether it is FosB or ΔFosB is being detected. Thus, our analyses may have missed some areas that are tonically activated by WM, RR, and/or CVS if they are unable to cleave and phosphorylate FosB. However, DARPP-32 76

78 deletion attenuates ΔFosB accumulation, indicating that phosphorylation is necessary for ΔFosB (Hiroi et al., 1999). In addition, the terminal 21 amino acids, not present in ΔFosB, are shared by the other fos-related antigens (McClung et al., 2004), suggesting that the cleavage of this segment of the FosB sequence is required for ΔFosB. Identification of enzymes for phosphorylation and cleavage are required to adequately address heterogeneity of neuronal subtypes in FosB processing. It is also possible that some neuronal subtypes simply do not engage fosb transcription in response to stress. Nonetheless, the lack of a significant FosB/ΔFosB response in known stress-activated regions, such as the PVN, suggests that negative findings using ΔFosB mapping need be interpreted with caution. In conclusion, our analyses have located several brain regions specifically recruited by unpredictable stress, including the NTS, PH, and mpfc. Importantly, tract tracing studies indicate significant interconnectivity among these brain regions and the PVN (Figure 3-6). For example, the infralimbic mpfc has direct projections to the NTS (Vertes, 2004), which sends projections that terminate within the PH (Ciriello et al., 2003), DMH, (Ricardo and Koh, 1978) and the PVN (Cunningham and Sawchenko, 1988). Additionally, the PH (Ulrich-Lai et al., 2011b) projects into the PVN and NTS (Fontes et al., 2001). Finally, the prelimbic mpfc projects to the DMH and attenuates DMH cfos responses to acute stress (Radley et al., 2009), which provides an additional influence on PVN, PH, and NTS activation. Future tract tracing studies accompanied with chronic stress exposure are required to reveal the specific connections among these recruited regions, which may include some of the above proposed circuits and/or additional relays implicated in chronic stress-induced physiological and behavioral dysfunction. 77

79 Table 3-1: Organ and Body Weight Measures Control WM RR CVS Adrenal Weight (% Body Weight) 18.4 ± ± ± ± 0.5* Thymus Weight (% Body Weight) ± ± ± 5.3 # 95.3 ± 3.0 Body Weight Gain (grams) 43.3 ± ± 2.4 # 20.1 ± 3.2 # 12.6 ± 2.7 # CVS exposure reduced body weight gain and induced adrenal hypertrophy. Despite undergoing chronic stress, RR and WM did not alter adrenal weight, but did reduce body weight gain similar to CVS. Surprisingly, thymus weight was slightly elevated by RR. * denotes group significantly different from all groups. # denotes group significantly different from control group. 78

80 Figure 3-1 Figure 3-1: Control Regions. To control for the possibility that the chronic stress regimens globally activate the brain, we quantified FosB/ ΔFosB within the interstitial nucleus of the posterior limb of the anterior commissure (IPAC) and motor cortex. The groups did not differ in the number of FosB/ ΔFosB immunoreactive nuclei within these two regions. 79

81 Figure 3-2 Figure 3-2: Structure Supplying PVN Norepinephrine. The A2 region of the nucleus of the solitary tract (NTS) supplies the majority of the medial parvocellular PVN with norepinephrine. Thus, we examined FosB/ ΔFosB within the NTS. CVS increased the number of FosB/ ΔFosB immunoreactive neurons within the NTS and especially within DBH- positive neurons. Since WM and RR animals did not alter FosB/ ΔFosB, the data suggests that unpredictable stress recruits the NTS. * denotes group significantly different from all groups. The scale bar refers to 200 µm. 80

82 Figure 3-3 Figure 3-3: Structures Supplying PVN Glutamate. Since the dorsomedial (DMH) and posterior hypothalamic (PH) nuclei are two regions that supply the PVN with significant amounts of glutamate, we analyzed FosB/ ΔFosB within these regions. CVS increased the number of FosB/ ΔFosB immunoreactive neurons within the DMH and PH, suggesting that unpredictable recruits the DMH and PH. * denotes group significantly different from all groups. # denotes group significantly different from control group. The scale bar refers to 200 µm. 81

83 Figure 3-4 Figure 3-4: Medial Prefrontal Cortex. Because side and subregion of the mpfc have both been shown to differentially regulate responses to stress, we divided the mpfc into both left and right infralimbic and prelimbic mpfc to locate the specific chronic stress- activated areas of the mpfc. However, CVS elevated the FosB/ ΔFosB expression within the right infralimbic, right prelimbic, left infralimbic, and left prelimbic, indicating that unpredictable stress recruits all subregions of the mpfc. * denotes group significantly different from all groups. The scale bar refers to.5 mm. 82

84 Figure 3-5 Figure 3-5: Upstream Limbic Structures. Since chronic stress exposure classically re- wires these stress regulatory regions, we analyzed FosB/ ΔFosB within the Basolateral Amygdala and Dentate Gyrus. Chronic stress did not alter the number of FosB/ ΔFosB immunoreactive neurons within the Basolateral Amygdala. However, both RR and CVS increased the number of FosB/ ΔFosB immunoreactive neurons in the dentate gyrus, suggesting that both unpredictable and predictable stress recruits the dentate gyrus. # denotes group significantly different from control group. 83

85 Figure 3-6 Figure 3-6: Chronic stress recruited circuitry: Our data suggest the recruitment of a neural circuit underlying chronic drive of the HPA axis. This circuit begins with the activation of the prefrontal cortex (PFC) projecting to the nucleus of the solitary tract (NTS), which drive neurons within the posterior hypothalamic nucleus (PH). The PH activates the PVN via direct glutamatergic projections to the paraventricular nucleus of the hypothalamus (PVN), known to be a player in endocrine, behavioral, and metabolic homeostasis. Via this pathway, chronic stress may produce endocrine, behavioral, and metabolic dysfunction. 84

86 Chapter 4 PVN- Projecting Noradrenergic Neurons Are Not Necessary for Chronic Stress HPA Facilitation 85

87 I. Abstract Chronic variable stress (CVS) exposure modifies the paraventricular nucleus of the hypothalamus (PVN) in a manner consistent with enhanced central HPA drive. There is evidence that CVS enhances norepinephrine (NE) action within the PVN, suggesting a role in chronic stress regulation. Therefore, we hypothesized that PVN- projecting NE neurons were necessary for the stress facilitatory effects of CVS. We locally injected saporin toxinconjugated to a dopamine- beta- hydroxylase (DBH) antibody (DSAP) to selectively eliminate NE innervation of the PVN and controlled for non- specific effects of the toxin using an unconjugated form (SAP). As anticipated, DSAP reduced the density of PVN DBH immunoreactive terminals, but did not completely ablate DBH immunoreactivity throughout the brain (e.g. fibers densities were unaffected in the central nucleus of the amygdala). Reductions in DBH- positive fiber density were associated with less DBH- immunoreactive neurons in the nucleus of the solitary tract (NTS) and locus coeruleus (LC), as previously described. Following two weeks of CVS, PVN- targeted DSAP injection did not alter stress- induced adrenal hypertrophy and reductions in body weight gain, indicating that PVN- projecting NE neurons are not essential for these physiological effects of chronic stress. In response to acute restraint stress, PVN- targeted DSAP injection attenuated peak ACTH and corticosterone in controls, but only attenuated peak ACTH in CVS animals, suggesting that enhanced adrenal hypersensitivity compensated for reduced excitatory drive of the PVN. Collectively, our data suggest that PVNprojecting NE neurons for generation of systemic responses to chronic stress. 86

88 II. Introduction Chronic stress can cause physiologic and behavioral dyshomeostasis, potentially due to widespread changes in neuronal structure and function. Stress- induced functional deficits are associated with synaptic (Carvalho-Netto et al., 2011), dendritic (Cook and Wellman, 2004; Magarinos et al., 1999), and genetic (Andrus et al., 2010) alterations within the limbic system, suggesting that re- wiring of stress- sensitive circuitry underlies chronic stress- related pathology. This stress- induced re- organization of neural structures is likely mediated by tonic activation of recruited circuits, which have yet to be established. Previous studies suggest that dysregulation at the level of the paraventricular nucleus of the hypothalamus (PVN), in part, contributes to associated behavioral and physiological dysfunction (Herman et al., 2008). This hypothalamic nucleus contains corticotropin- releasing- hormone (CRH)- expressing medial parvocellular neurons that initiate a neuroendocrine cascade, triggering the release of ACTH from the anterior pituitary, and culminating in the synthesis and secretion of glucocorticoids from the adrenal gland (comprising the hypothalamo-pituitary-adrenocortical axis). Chronic stress alters PVN peptide (Herman et al., 1995b; Imaki et al., 1991; Kiss and Aguilera, 1993; Makino et al., 1995), receptor (Cullinan and Wolfe, 2000; Ziegler et al., 2005), and electrophysiological (Verkuyl et al., 2004) function in a manner consistent with enhanced central drive of the HPA axis by stress. In addition to these postsynaptic effects, chronic stress exposure increases the number of presynaptic excitatory neurotransmitter boutons in apposition to the CRH cell bodies and dendrites (Flak et al., 2009), indicating that chronic stress enhances afferent activation of the PVN. One neurotransmitter that plays a prominent role in PVN activation is norepinephrine (NE). NE terminals are known to directly contact PVN CRH neurons (Liposits et al., 1986), and the medial parvocellular PVN expresses alpha adrenergic receptors (Cummings and Seybold, 87

89 1988; Day et al., 1997), providing a means by which NE can initiate glucocorticoid release at the PVN. In support of these anatomical data, intraventricular or local infusion of NE induces PVN cfos and stimulates the HPA axis (Cole and Sawchenko, 2002; Szafarczyk et al., 1987), whereas local alpha adrenergic receptor antagonists both inhibit stress- induced PVN cfos and corticosterone release (Itoi et al., 1994; Leibowitz et al., 1989), indicating that stimulation of the alpha- adrenergic receptors can drive the HPA axis. Tract tracing studies have revealed that PVN NE is primarily supplied by neurons of the A2 catecholaminergic cell group, largely confined to the area of the nucleus of the solitary tract (NTS) (Cunningham and Sawchenko, 1988). Knife cuts of the ascending medullary PVN inputs reduces PVN DBH- straining, attenuates CRH immunoreactivity and blunts HPA axis responses to stress (Li et al., 1996; Sawchenko, 1988), supporting a role for the NTS in regulation of PVN activity. Previous studies indicate that injection of saporin (SAP) conjugated to a dopamine- betahydroxylase (DBH) antibody (DSAP) targeted for the PVN greatly reduces the NE/epinephrine (E) fiber density in this region accompanied with reductions in the number of DBHimmunoreactive neurons within NTS (Ritter et al., 2001), indicating that these neurons supply NE to the PVN. While E axons innervate the PVN, DBH immunoreactive fibers are much more dense than PNMT within the mppvn (Rinaman, 2001), indicating that NE is more prominent effector on the mppvn neuroendocrine neurons than E. These reductions in PVN NE/E innervation are associated with attenuated HPA axis responses to systemic stressors such as glucoprivation (Ritter et al., 2003), IL1-beta (Li et al., 1996), and LPS (Bienkowski and Rinaman, 2008), but not swim stress (Ritter et al., 2003), suggesting that PVN NE/E is necessary for appropriate HPA axis responses to systemic, but perhaps not psychogenic, stressors. Although these neurons are not necessary for acute psychogenic stress responding, there is 88

90 evidence to suggest that these neurons are recruited during chronic stress ((Zhang et al., 2010), and Chapter 3), and may be a critical mediator of stress pathlogy. Thus, enhancements in NTS NE output may materially contribute to physiological and behavioral dysfunction associated with chronic stress. Thus, we hypothesized that PVN- projecting NE/E neurons are necessary for the induction of the stress facilitatory effects of chronic stress. III. Materials and Methods Subjects. Male Sprague- Dawley rats from Harlan (Indianapolis, IN) weighing g upon arrival were housed one per cage for the duration of the experiment in clear polycarbonate cages containing granulated corncob bedding, with food and water available ad libitum. The colony room was temperature- and humidity- controlled with a 12 h light cycle (lights on 6:00 am; lights off 6:00 pm). Rats acclimated to the colony facility for one week prior to experimental manipulations. All experimental procedures were conducted in accordance with the National Institutes of Health Guidelines for the Care and Use of Animals and approved by the University of Cincinnati Institutional Animal Care and Use Committee. Hopper and body weights were also collected prior to surgery and following surgery in order to insure that the animals appropriately recovered from the procedure. The animals were weighed prior to chronic stress regimen and additionally every third day until the termination of the experiment. Following two weeks of experimental manipulation, the animals were exposed to novel restraint test and perfused two hours following the onset of stress. PVN- targeted DSAP injections. The animals, housed one per cage, were injected with ketamine and xylazine, and additionally butorphenol as a pre- emptive analgesic. A Hamilton syringe was 89

91 stereotaxically directed at the PVN. A preliminary study determined that the coordinates to be mm from Bregma, +/-.4 mm from the midline, and -8.2 mm from the skull. The animals were injected with either the phosphate buffered saline (PBS), the saporin toxin (SAP) (8.82ng/ 200nl, ph 7.4) (PR-01, Advanced Targeting Systems, San Diego, CA), or the saporin toxin conjugated to an antibody directed to DBH (DSAP) (42ng/ 200nl, ph 7.4) (MAB394, Millipore, Temecula, CA). The DSAP is endocytosed following binding at the synaptic cleft and transported back to the cell body (Wrenn et al., 1996). At the cell body, the toxin inactivates ribosomes (Ippoliti et al., 1992), leading to cell death within two weeks (Madden et al., 1999). DSAP has previously been reported to be specific for the ablation of NE/E neurons (Madden et al., 1999; Rinaman, 2003; Ritter et al., 2001). A 200 nl injected was gradually injected into the brain over 3 minutes. The parameters of the PVN DSAP injections were designed to be similar to previously published use (Bienkowski and Rinaman, 2008; Ritter et al., 2003). The animals were allowed to recovery for two weeks, since this amount of time is required to eliminate catecholamine neurons using DSAP. Chronic Stress Procedure. Subjects were randomly assigned to either CVS or control groups. Prior to chronic stress, the subjects were weighed and food intake recorded, and subsequently placed into groups such that groups did not differ in their pre- experiment body weights and food intake. The chronic stress protocol consisted of twice- daily (morning and afternoon) exposure to randomly assigned stressors for two weeks. Other than to record weight and food intake, control animals were not disturbed. Morning stressors were conducted between 8:00 am and 11:30 am and afternoon stressors were administered between 1:30 pm and 5:00 pm. Stressors consisted of rotation stress (1 h at 100 rpm on a platform orbital shaker), warm swim (20 min at 90

92 31 o C); cold swim (10 min at 18 o C), cold room stress (kept in 4 o C for one hour) and hypoxia (8% O 2 92% N 2 ). Stress testing. Following 14 days of CVS, blood was collected in EDTA at 0, 30, 60, and 120 minutes following the onset of 30 minutes of restraint stress. The animals were placed into a plastic restraint tube and blood collected. Following collection, blood was spun for 15 minutes at 6000 rpm. Plasma was collected and stored at -20 º C. Plasma ACTH and Corticosterone were quantified using radioimmunoassay. Corticosterone radioimmunoassasy used a kit from MP Biomedicals and ACTH utilized an antibody from Dr. Bill Engeland as previous described (Choi et al., 2007). Area under curve was calculated using equation for a trapezoid as previously described (Ulrich-Lai et al., 2011a). 2 ul of plasma were used to calculate plasma glucose concentration via Freestyle glucometer (Abbot Laboratories, Alameda Ca). In the morning following the last afternoon stressor, rats received an overdose of sodium pentobarbital and were perfused with phosphate buffered saline, followed by 4% paraformaldehyde. Brains were postfixed overnight in 4% paraformaldehyde and transferred to 30% sucrose at 4 o C until they were cut on a freezing microtome. Adipose tissue analysis. The carcasses of the animals were placed into a Plexiglas tube and inserted into an EchoMRI whole body composition analyzer system (Echo Medical Systems, Houston, TX). The EchoMRI provides estimations of fat and lean mass. Immunohistochemistry. Brains were cut at 35 µm on a sliding microtome throughout the brain and resulting sections were stored in cryoprotectant (0.1 M phosphate buffer, 30% sucrose, 1% 91

93 polyvinylpyrrolidone, and 30% ethylene glycol) at 20 C until used for immunohistochemistry. Sections were transferred from cryoprotectant to 50 mm potassium phosphate buffered saline (KPBS; 40 mm potassium phosphate dibasic, 10 mm potassium phosphate monobasic, and 0.9% sodium chloride) at room temperature (RT). Cryoprotectant was rinsed (5 x 5 min) in KPBS, the sections were transferred to KPBS + 1.0% H 2 O 2, and incubated for 10 minutes at room temperature (RT). Sections were then washed (5 x 5 min) in KPBS at RT, and placed in blocking solution (50 mm KPBS, 0.1% bovine serum albumin (BSA), and 0.2% Triton X-100) for 1 hour at RT. Sections were incubated overnight at 4 C in rabbit anti- CRH (rc70 courtesy of Wylie Vale) diluted 1:2500 and mouse anti- DBH (Millipore; Temecula, CA) at 1:2500 or rabbit anti-synaptophysin (Zymed Laboratories, San Francisco, CA) 1:300 and guinea pig antivesicular glutamate transporter 2 (Millipore; Temecula, CA) 1:1500 in blocking solution. The following morning sections were rinsed in KPBS (5 x 5 min) and incubated in biotinylated antirabbit secondary antibody (Vector Laboratories, Inc., Burlingame, CA) and Cy5 goat anti- mouse secondary antibody (Jackson Immuno Research; West Grove, PA) or Cy3 goat anti-rabbit (Jackson Immuno Research; West Grove, PA) and Alexa 488 donkey anti-guinea pig (Molecular Probes, Eugene, Oregon) diluted 1:500 in KPBS + 0.1% BSA for 1 hour at RT. Sections were rinsed in KPBS (5 x 5 min) and then treated with avidin- biotin complex (ABC, Vector Laboratories, Inc., Burlingame, CA) at 1:1,000 in KPBS + 0.1% BSA for 1 hour at RT. Following this incubation, sections were rinsed again in KPBS (5 x 5 min) and incubated with Cy3 Streptavidin. Sections were rinsed a final time in KPBS and coverslipped in Fluka Mounting Medium (Sigma Aldrich; St. Louis, MO). 92

94 Fiber Density. For each region, two images on each side were collected for image analysis. The images were collected on the smallest possible objective to both include distinguishable immunoreactivity from background and to cover the whole neural region within the coronally sliced section. Fiber density (the percent area occupied by immunoreactivity) was quantified with the Zeiss Axiovision Program. Appropriate parameters were determined for each region using a subset of randomly chosen images. For the PVN, CRH staining was used as a marker for where DBH- positive fiber densities were quantified. However, CRH staining was not present in synaptophysin and vglut2 stained tissue. Thus, z-stacks were collected within the medial parvocellular PVN, in a manner similar to what has been previously reported (Flak et al., 2009). Cell Counts. The number of DBH- immunoreactive neurons were counted within the rostral (- 13.7mm Bregma), medial (-14.0 mm Bregma), and caudal (-14.3 mm Bregma) NTS. Two images were collected on each side and immunoreactive neurons counted by hand by an observer blind to treatment. The locus coeruleus (LC) is an extremely cell dense region, and its high concentration of DBH makes it extremely difficult to separate neurons from each other. Thus, we quantified the area of the LC containing DBH immunoreactivity as an indirect method of cell loss. Numbers of cfos immunoreactive neurons within the PVN were determined using Zeiss Axiovision 4.8 software. Four images per animal were collected and subsequently analyzed. Both a threshold gray level and minimum pixel size were determined for using a random subset of images per region. The particle counting algorithm in Axiovision 4.8 was used to determine number of immunoreactive nuclei within the defined region of interest. 93

95 Statistical Analysis. Data are expressed as mean ± standard error. Outliers were determined if the value exceeded both 1.96 times the standard deviation and 1.5 times the interquartile range ( (McClave, 1994)). Factorial data were analyzed using two way ANOVA with Fisher s LSD post- hoc test, with PVN- targeted injection (DSAP and SAP) and stress (CVS and Control) as between subject factors. Hormone response data were analyzed by three way repeated measures ANOVA with Fisher s LSD post- hoc test. PVN- targeted injection (DSAP and SAP), stress (CVS and Control) and time (0, 30, 60, 120 minutes) were between subject factors. In order to insure homogeneity of variance, the data went under log transformation and then re- analyzed. Planned comparisons were performed to assess the effect of stress within PVN- injections (DSAP vs. SAP) and PVN- injections within stress (CVS vs. control). Since specific hypothesis tests were identified a priori, the planned comparisons were performed regardless of the outcome of the omnibus ANOVA (Maxwell and Delaney, 1989). One and two way ANOVAs were conducted using Sigma Stat (Systat Software, San Jose, California) and three way repeated measure ANOVAs, using GB stat (Dynamic Microsystems, Inc., Silver Springs, MD). IV. Results We microinjected DSAP and SAP into PVN using similar parameters to previously reported studies (Bienkowski and Rinaman, 2008; Ritter et al., 2003) and observed massive removal of PVN DBH immunoreactivity. This qualitative observation was supported by quantitative fiber density analysis that found a main effect of DSAP reducing PVN CRH DBHpositive fiber density {F(1,47)= ,p<.001} (Figure 4-1A). This removal of catecholaminergic fibers was not observed throughout the whole brain, as DBH- positive fiber density was not altered within the central nucleus of the amygdala (Figure 4-1B). However, we 94

96 did observe reductions in both supraoptic nucleus (SON) {F(1,41)=27.498,p<.001} (Figure 4-2A) and posterior cingulate gyrus {F(1,42)= , p<.001} (Figure 4-2B) DBH- positive fiber density, indicating that the effects of PVN- targeted DSAP injection were not solely confined to the PVN (Ritter et al., 2001). Specificity of DSAP was not expected, as NE/E neurons from the brainstem project throughout the hypothalamus and amygdala (Cunningham et al., 1990; Cunningham and Sawchenko, 1988; Delfs et al., 1998; Sawchenko and Swanson, 1981). PVNtargeted DSAP injection has previously been reported to induce widespread fiber loss (Ritter et al., 2001). PVN DBH- fiber density reductions were accompanied by reduced numbers of DBHimmunoreactive neurons in the NTS (Figure 4-3A-C) (rostral {F(1,42)=8.653,p<.001}, medial {F(1,42)=35.683, p<.001}, and caudal {F(1,42)=50.516,p<.001} divisions) and locus coeruleus (LC) {F(1,41)=35.611, p<.001} (Figure 4-3D), in accord with tract tracing studies showing that the NTS and LC project to portions of the parvocellular PVN (Cunningham and Sawchenko, 1988). In support of the inability for SAP to enter cells, fiber densities and neuron counts did not differ between SAP animals and animals injected with PBS, indicating there was not a cytotoxic effect of SAP. Following two weeks of recovery from surgery, half of the animals were put through chronic variable stress (CVS). CVS induced adrenal hypertrophy {F(1,44)=30.885, p<.001}, reductions in body weight gain {F(1,44)= ,p<.001} accompanied by losses in adipose {F(1,43)=19.681, p<.001}, and lean mass {F(1,45)=21.215,p<.001}. There was not an interactive effect within these measures, indicating that PVN- projecting NE/E neurons do not play a prominent role in chronic stress regulation of body and organ weight. Following two weeks of CVS, all animals were acutely restrained and blood collected in response to a novel stressor (assessment of facilitation). As expected, restraint elevated plasma ACTH and corticosterone in all groups, with a return to baseline within two hours after the onset 95

97 of stress. There was a significant stress X injection X time effect in corticosterone {F(3,44)=2.95,p=.03}, but only an injection X time effect in ACTH {F(3,43)=9.02,p<.001}. DSAP (p<.001), but not SAP injected animals displayed chronic stress-induced elevations in basal glucocorticoids, suggesting of PVN- projecting NE/E neurons normally limit HPA axis hypersecretion following chronic stress. PVN- targeted DSAP injection blunted peak ACTH (p<.01) and corticosterone (p<.001) levels in control animals, but only attenuated peak ACTH (p<.01) levels in CVS animals, suggestive for chronic stress enhancement of adrenal sensitivity. In support of these data, CVS exposure elevated plasma corticosterone/log plasma ACTH levels, an indirect measure of adrenal sensitivity, in animals with PVN- targeted DSAP injection (p<.01), but not SAP. PVN- targeted DSAP injection also reduced 60 minute plasma corticosterone levels in both CVS (p<.01) and control (p<.01) animals, suggesting that PVN NE/E attenuates glucocorticoid negative feedback. In support of dissociated ACTH and corticosterone levels between DSAP CVS and DSAP controls, ACTH area under the curve was reduced in CVS animals (p=.04), while corticosterone area under the curve was reduced in controls (p<.01) relative to their SAP counterparts. Since PVN- targeted DSAP injection has not been shown to regulate basal PVN CRH mrna (Ritter et al., 2003), we expected that the ablation of PVN- projecting NE/E neurons would either 1) reduce CRH stores within the median eminence or 2) attenuate the drive of the PVN in response to restraint. However, neither CRH median eminence fiber density nor PVN cfos induction differed between groups, suggesting that PVN- targeted DSAP injection may alter central drive of the HPA axis in a subtle manner. Since previous studies indicate that PVN- targeted DSAP injection attenuates HPA axis responses to glucoprivation (Ritter et al., 2003), we predicted that PVN- projecting NE/E neurons would block chronic stress attenuation of stress- induced hyperglycemia. There was a 96

98 significant main effect of stress on plasma glucose levels {F(1,43)=7.49,p<.01}. As previously described, CVS attenuated peak restraint- induced hyperglycemia in SAP animals (p<.01) on control (SAP- injected) animals. The effects of CVS on acute stress-induced glucose production were blocked by PVN- targeted DSAP- injection. However, PVN- targeted DSAP injection reduced glucose levels in CVS animals during the recovery from stress-induced hyperglycemia, indicating that PVN- projecting NE/E neurons regulate the recovery of the glucose response to stress. While PVN- projecting NE/E neurons do not appear to play a prominent role in chronic stress- induced dysregulation of central HPA axis drive, they may be important in maintenance of glucose homeostasis following CVS. Since chronic stress increases both PVN density of synaptophysin staining and the number of direct excitatory contacts in apposition to PVN CRH neurons, we hypothesized that PVN-projecting noradrenergic neurons would be necessary for the induction of chronic stress PVN neurotransmitter plasticity. Following fiber density analyses, there was a significant stress X injection effect in both PVN synaptophysin immunoreactivity {F(1,45)=10.691,p<.001} and vglut2 immunoreactivity {F(1,42)=7.045,p<.01}. As previously shown (Carvalho-Netto et al., 2011; Flak et al., 2009), chronic stress increased the density of synaptophysin staining in the PVN, but this effect was abolished by PVN- targeted DSAP injection, indicating that these cells are necessary for the induction of this effect following chronic stress. In a similar trend to what has previously been reported, CVS increased the density of PVN vglut2 immunoreactivity. Similar to synaptophysin, CVS- induced elevation in the density of vglut2 was also abolished following PVN-targeted DSAP injection, suggesting that the chronic stress enhancement of central PVN drive may have been eliminated following PVN-targeted DSAP injection. Importantly, these results also indicate that there was not a compensatory increase in 97

99 glutamatergic innervation of the PVN or a significant removal of total presynaptic PVN innervations following DSAP lesions. V. Discussion Collectively, our data suggest that PVN- projecting NE/E neurons are not necessary for chronic stress- induced HPA axis facilitation. PVN- targeted DSAP injection attenuated peak ACTH and not corticosterone in CVS animals, suggestive of CVS enhancement of adrenal sensitivity in CVS DSAP animals. Previous studies indicate that CVS enhances adrenal sensitivity (Ulrich-Lai et al., 2006b), which likely contributes to post- stress hypercortisolemia and HPA facilitation (Akana et al., 1992). In addition to being unnecessary for chronic stress modification of HPA axis activation, the PVN- projecting NE/E neurons are not required for chronic stress- induced alterations to organ or body weight, indicating that the free glucocorticoid exposure in DSAP CVS and SAP CVS was similar. Surprisingly, the ablation of PVN- projecting NE/E neurons prevented chronic stress attenuations in peak glucose levels following acute psychogenic stress, suggesting a role of PVN- projecting NE/E on stress-induced metabolic regulation. NE/E NTS neurons are recruited by both acute and chronic stress (Cullinan et al., 1995; Teppema et al., 1997; Zhang et al., 2010) and are involved in HPA axis responding to glucoprivation (Ritter et al., 2003). A more thorough investigation is needed to determine how recruitment of NTS neurons is involved in metabolic responses to psychogenic stimulation. The destruction of PVN- projecting NE/E neurons did not appear to ameliorate responses to chronic stress, since cumulative glucocorticoid exposure between SAP and DSAP animals was likely similar. Attenuated ACTH responses may not be a driving force to produce the 98

100 physiological indices of chronic stress. As noted by the acute stress responses in CVS DSAP animals, peak ACTH levels were attenuated, but corticosterone unchanged, suggesting that a loss of PVN NE enhanced CVS- induced adrenal hypersensitivity. In CVS DSAP animals, tonic ACTH stimulation may have produced adrenal hypersensitivity, as acute fluxes in ACTH are not a contributing factor toward tuning adrenal sensitivity. Body weight and food intake are also likely dependent on free glucocorticoids, as chronic glucocorticoid treatment reduces body weight and food intake (Bush et al., 2003; Lerman et al., 1997). Despite the inability of PVNtargeted DSAP injection to alter our typical physiological indices of chronic stress, these results suggest that the mechanisms of chronic stress regulation are distinct from factors that mediate acute stress responses, and may work through alternative or overlapping brain regions/neurotransmitters to impair physiological and behavioral function. Previous studies from our lab have reported glucocorticoid- independent increases in NTS tyrosine- hydoxylase (TH) mrna following CVS (Zhang et al., 2010), suggesting an enhancement of NTS NE/E output to the PVN by unpredictable stress. By taking into account these previous and present results, it would suggest that despite clear changes in biosynthetic activity and PVN innervation, NTS NE/E neurons are not necessary for chronic stress- induced HPA facilitation. Thus, other transmitter systems (e.g., glutamate), working independently or in concert with NE/E, likely control facilitation of the HPA axis mediated by chronic stress. However, 50%, at most, of the catecholaminergic NTS neurons were removed by PVN- targeted DSAP injection, suggesting that hyper- responsiveness of remaining NTS NE/E neurons may culminate in HPA facilitation via indirect effects on PVN activation. Tracing studies indicate that NTS neurons that project to the hypothalamus are separate from those that project to the spinal cord, triggering sympathetic activation (Ritter et al., 2001). 99

101 Differential projections likely account for the sparing of some 50% of DBH- positive neurons in the NTS. If preserved, NTS projections to the intermediolateral spinal cord or ventrolateral medulla may be sufficient to trigger chronic stress- induced changes in autonomic function (Grippo et al., 2002), which may contribute to HPA axis hyperactivity via autonomic- HPA axis cross- talk (Ulrich-Lai et al., 2006a; Ulrich-Lai and Engeland, 2002). To answer this question, future studies with require complete ablation of NTS NE/E prior to CVS exposure. The altered glucose responses may be indicative of regulation of the sympathetic responses to stress. Some of the neurons of the PVN project to the spinal cord and brainstem, regulating sympathetic (as well as parasympathetic) activity (Kreier et al., 2006; Swanson and Kuypers, 1980). However, glucose responses are attenuated following chronic stress, whereas cardiovascular responses are not (Deboer et al., 1990; Flak et al., 2011), suggesting that these responses are not representative of a general sympathetic response but specific end organ effects of chronic stress. If the glucose levels are not indicative of the sympathetic response, they may be due to effects on glucose clearance or storage. A more thorough analysis will be needed to decipher the potential autonomic alterations due to PVN- targeted DSAP injection. Our results indicated that PVN- targeted DSAP injection reduced the number of DBHimmunoreactive neurons in and the LC as well as the NTS. While tract tracing studies have revealed that the NTS supplies the majority of the medial parvocellular PVN with NE (Cunningham and Sawchenko, 1988; Sawchenko and Swanson, 1982), this is not the sole location that these neurons project. For example, NTS NE neurons also projects to the SON (Cunningham and Sawchenko, 1988; Sawchenko and Swanson, 1982), another areas that we observed a reduction in DBH- positive fiber density. Therefore, we cannot assume that our effects can be entirely due to removal PVN NE/E afferents alone. 100

102 The chronic stress data fall in line with previous observations that NTS is responsive to systemic but not psychogenic stress (Li et al., 1996; Ritter et al., 2003), since unpredictable stress is, at least in part, a chronic psychogenic stressor. With CVS, the stressors are placed in a random sequence in order to minimize habituation. Therefore, the unpredictability of the regimen is stressful for the animals independent of the stressors themselves and primarily recruit limbic sites, rather than brainstem pathways, such as the one we disrupted with DSAP injection (Emmert and Herman, 1999; Herman and Cullinan, 1997). Loss of DBH- positive cells was also noted within the LC. Indeed, the LC sends projections to the extreme medial component of the parvocellular PVN and the periventricular zone, largely targeting dopamine-, somatostatin-, and thyrotropin-releasing- hormoneexpressing neurons (Sawchenko and Swanson, 1982). In addition to the PVN CRH neurons, these additional parvocellular PVN projecting axons likely took up DSAP, leading to LC NE removal. Axons of LC NE neurons collateralize extensively throughout the brain, which may account for loss of immunoreactive terminals in regions outside the PVN (e.g., the posterior cingulate gyrus (Fallon and Loughlin, 1982; Loughlin et al., 1982) or any area shown to be partially denervated in our studies). However, previous studies have suggested that the LC is not responsive to unpredictable stress. Unlike chronic social stress (Watanabe et al., 1995) and repeated restraint (Mamalaki et al., 1992), LC TH content does not change (Ziegler et al., 1999), suggesting that if anything, the LC way be involved in stress habituation rather than facilitation. However, LC NE loss may regulate acute stress responding. Ablation of the LC NE attenuates HPA axis responses to acute stress (Ziegler et al., 1999), suggesting that PVN- targeted DSAP injection HPA axis blunting may be via the LC, as well as or perhaps instead of the NTS. 101

103 In conclusion, the PVN- targeted DSAP injection greatly reduced PVN DBH- fiber density, accompanied by reductions in the number of DBH- immunoreactive neurons within the NTS and LC. Collectively, the data suggest that PVN- projecting NE/E neurons are not necessary for chronic stress- induced HPA axis dysregulation. However, the interpretation of our data is difficult due to the inherent limitations of DSAP. Ideally, we would have selectively removed CRH- innervating NE neurons, but we ablated all NE/E neurons projecting into the PVN. Injecting DSAP into the NTS would selectively removed NTS NE/E neurons, but at the expense of additionally destroying spinally projecting NE/E neurons. As viral technology advances, we will be able selectively ablate projections from NTS NE to the PVN CRH neurons. 102

104 Table 4-1: Organ and Body Weight Measures SAP Control SAP CVS DSAP Control DSAP CVS Adrenal Weight (normalized) 15.4 ± ± 0.5* 15.5 ± ± 0.5* Thymus Weight (normalized) 87.5 ± ± ± ± 3.8 Body Weight Gain (grams) 23.4 ± ± 1.4* 22.5 ± ± 2.1* Lean Mass (grams) ± ± 2.3* ± ± 2.3* Adipose Mass (grams) 39.6 ± ± 1.4* 38.3 ± ± 0.7* CRH fiber density (% area immunoreactive) 21.3 ± ± ± ± 1.4 CVS exposure induced adrenal hypertrophy and attenuated body weight gain through reductions in both lean and adipose mass. However, there was not an interactive effect of injection X stress. * denotes significant main effect of stress. Table 4-2: PVN cfos and CRH fiber density SAP Control SAP CVS DSAP Control DSAP CVS CRH fiber density (% area immunoreactive) 21.3 ± ± ± ± 1.4 PVN cfos (# immunoreactive nuclei) ± ± ± ± 11.3 Despite there being an effect of PVN- targeted DSAP injection attenuating peak levels of ACTH in response to restraint, there was no effect of DSAP on CRH fiber density and PVN cfos induction 103

105 Figure 4-1 Figure 4-1: PVN and CeA Fiber Densities. PVN- targeted DSAP injection reduced DBHimmunoreactivity within the PVN, but not the central nucleus of the amygdala, indicating that there was not complete removal of NE/E throughout the forebrain. * denotes group different from corresponding control group. The scale bar refers to 100 µm. 104

106 Figure 4-2 Figure 4-2: SON and Posterior Cingulate Gyrus Fiber Densities. PVN- targeted DSAP injection did not solely reduce PVN DBH-positive fiber density. PVN- targeted DSAP injection reduced DBH-fiber density in the supraoptic nucleus (SON) and the posterior cingulate gyrus. * denotes group different from corresponding control group. The scale bar refers to 100 µm. 105

107 Figure 4-3 Figure 4-3: NTS and LC DBH- immunoreactive cell loss. Reductions in forebrain DBH-positive fiber density were accompanied with loss of DBH- immunoreactive neurons in the NTS (rostral, medial, and caudal) and LC. * denotes group different from corresponding control group. 106

108 Figure 4-4 Figure 4-4: HPA axis responses to restraint. PVN- targeted DSAP injection blunted 60 minute corticosterone levels, indicating that norepinephrine hastens the recovery to basal glucocorticoid levels following stress. In addition, PVN- targeted DSAP injection elevated basal corticosterone levels in CVS animals, which indicate that norepinephrine attenuates chronic stress- induced hypercortisolemia. PVN- targeted DSAP injection also attenuated peak levels of ACTH and corticosterone in control animals, but only ACTH in CVS animals. This dissociation between CVS and control DSAP animals is associated with a difference in corticosterone/log ACTH, an indirect method for determining adrenal sensitivity. The data suggest that PVN norepinephrine activates central drive of the HPA axis, but is overcome by adrenal compensation to yield no difference in peak glucocorticoid levels. * denotes group different from corresponding control group. 107

109 Figure 4-5 Figure 4-5: Restraint- induced hyperglycemia. PVN- targeted DSAP injection ameliorated the effect of chronic stress to attenuate peak stress- induced hyperglycemia. However, CVSexposed DSAP injected animals exhibited reduced glucose levels during the recovery from restraint stress, indicating that PVN norepinephrine/epinephrine controls chronic stress modulation of homeostatic recovery of basal glucose levels. * denotes group different from corresponding control group. # denotes group different from all. 108

110 Figure 4-6 Figure 4-6: PVN Bouton Density PVN- targeted DSAP injection eliminated the chronic stress enhancement in PVN bouton density in the PVN, indicating that PVN-projecting noradrenergic neurons are necessary for chronic stress alteration of PVN innervations. Additionally, these results demonstrate that there is not a compensatory elevation in glutamatergic contacts following noradrenergic removal. * denotes group different from corresponding control group. 109

111 Chapter 5 Opposing Effects of Chronic Stress and Weight Restriction on Cardiovascular, Neuroendocrine and Metabolic Function 110

112 I. Abstract Chronic stress is associated with dysregulation of energy homeostasis, but the link between the two is largely unknown. For most rodents, periods of chronic stress reduce weight gain. We hypothesized that these reductions in weight are an additional homeostatic challenge, contributing to the chronic stress syndrome. Experiment #1 examined cardiovascular responsivity following exposure to prolonged intermittent stress. We used radio- telemetry to monitor mean arterial pressure and heart rate in freely moving, conscious rats. Three groups of animals were tested: chronic variable stress (CVS), weight- matched (WM), and controls. Using this design, we can distinguish between effects due to stress and effects due to the changing body weight. WM, but not CVS, markedly reduced basal heart rate. Although an acute stress challenge elicited similar peak heart rate, WM expedited the recovery to baseline heart rate. The data suggest that CVS prevents the weight- induced attenuation of cardiovascular stress reactivity. Experiment #2 investigated hypothalamic- pituitary- adrenocortical axis and metabolic hormone reactivity to novel psychogenic stress. WM increased corticosterone area under the curve. CVS blunted plasma glucose, leptin, and insulin levels in response to restraint. Experiment #3 tested the effects of WM and CVS on PVN oxytocin and corticotropin- releasing hormone mrna expression. CVS increased, while WM reduced PVN CRH mrna expression, whereas both CVS and WM reduced dorsal parvocellular PVN oxytocin mrna. Overall, the data suggest that weight loss is unlikely to account for the deleterious effects of chronic stress on the organism, but in fact produces beneficial effects that are effectively absent or indeed, reversed in the face of chronic stress exposure. 111

113 II. Introduction Humans, as well as animals, experience random, unpredictable bouts of stress throughout daily life induced by a myriad of psychogenic and systemic threats to their livelihood. In reaction to these challenges, their bodies have evolved two main physiological responses to stress: the sympathetic nervous system (SNS) and the hypothalamic- pituitary- adrenocortical (HPA) axis responses. Within seconds of stress onset, catecholamines are released into the bloodstream from sympathetic ganglia and adrenal medulla, triggering the SNS ( fight or flight ) response. Concurrently, corticotropin- releasing- hormone (CRH)- containing neurons within the paraventricular nucleus of the hypothalamus (PVN) are activated, which evokes a neuroendocrine cascade culminating in the synthesis and release of glucocorticoids from the adrenal cortex. The temporal differences between these two responses produce very different means by which the SNS and HPA axis can influence present and future stress reactivity. Peak SNS responses can be seen within seconds of stress onset, but it takes minutes to yield an increase in plasma glucocorticoid levels. However, SNS responses terminate more expeditiously and do not typically produce lasting changes in whole- organism physiology. Glucocorticoids can produce lasting effects because they act on steroid receptors throughout the body, which can have delayed and persistent changes in gene expression (De Nicola et al., 1998) and cellular plasticity (McEwen et al., 1991). Repeated activation of physiological stress responses modulates future stress reactivity. For example, following chronic stress, corticosterone responses to novel stressors are significantly exaggerated, likely due to enhanced adrenal sensitivity (Ulrich-Lai et al., 2006b) and/or reduced glucocorticoid negative feedback (Mizoguchi et al., 2003). In addition, following chronic stress, PVN CRH neurons exhibit electrophysiological (Verkuyl et al., 2004), genetic 112

114 (Cullinan and Wolfe, 2000; Ziegler et al., 2005), and morphological changes (Flak et al., 2009) that lead to enhanced central drive of the HPA axis (Herman et al., 2008). While adaptations of physiological responses can be seen as adaptive, dysregulated stress responses are a facet of mood and anxiety disorders, which underscores the importance of proper stress control. Stress responses can affect numerous central and peripheral systems that impact physiological status. Both the SNS and HPA axis release glucose from energy stores, which could affect energy balance. In rodents, chronic stress regimens, such as social subordination (Tamashiro et al., 2007b), variable stress (Solomon et al., 2010), or homotypic stress (Rybkin et al., 1997), reduce food intake, body weight gain, and adiposity. As a result, plasma leptin and insulin are also reduced (Solomon, 2010; Tamashiro et al., 2007a). In turn, metabolic state can also influence stress responding. Metabolic challenges activate both the HPA axis and SNS (Jahng et al., 2007; Stamp et al., 2008). Furthermore, leptin blunts HPA axis and behavioral responses to stress (Clark et al., 2008; Heiman et al., 1997; Lu et al., 2006), but can shift heart rate variability toward sympathetic outflow (do Carmo et al., 2008). Following the cessation of the stress regimen, body weight eventually recovers, but is delayed relative to food- restricted animals (Harris et al., 2002), indicating that stress attenuates weight gain independent of effects on metabolic parameters. Our group has demonstrated that chronic stress (using a chronic variable stress (CVS) regimen) reduces food intake, body weight gain, and adiposity (Solomon et al., 2010). Given that there are clear connections between metabolism and stress responding, modulations in metabolic state due to chronic stress- induced weight loss may represent a substantial component of the physiological response to chronic stress exposure. Therefore, the present study tested the 113

115 hypothesis that weight loss alone would contribute to chronic stress- related pathology and chronic stress facilitation of HPA axis and sympathetic responses to a novel stressor. III. Materials and Methods Subjects. Male Sprague- Dawley rats from Harlan (Indianapolis, IN) weighing g upon arrival were group- housed two per cage for the duration of the experiment in clear polycarbonate cages containing granulated corncob bedding, with food and water available ad libitum. The colony room was temperature- and humidity- controlled with a 12 h light cycle (lights on 6:00 am; lights off 6:00 pm). Rats acclimated to the colony facility for one week prior to experimental manipulations. All experimental procedures were conducted in accordance with the National Institutes of Health Guidelines for the Care and Use of Animals and approved by the University of Cincinnati Institutional Animal Care and Use Committee. Hopper and body weights were collected once per day prior to stress exposure in all animals. In experiment 1, rats (CVS (n=7), WM (n=7), and Control (n=5)) were implanted with radio- telemetry devices and allowed to recover from surgery. Following recovery, one group of animals was placed through two weeks of CVS, and their cardiovascular response (for details see Cardiovascular recording after novel (footshock) stress.) to a novel stressor (acute footshock) was analyzed. In experiments 2 and 3, rats were exposed to CVS immediately following the acclimation period to the colony facility. At the end of each experiment, the rats were perfused. In experiment 2, the rats (CVS (n=8), WM (n=9), and Control (n=8)) were exposed to novel restraint test and perfused two hours following the onset of stress. In experiment 3, the rats (CVS (n=8), WM (n=9), and Control (n=8)) were sacrificed without acute stimulus to assess basal hypothalamic mrna analyses. 114

116 Chronic Stress Procedure. Subjects were randomly assigned to either CVS, weight- matched (WM), or control groups. The chronic stress protocol consisted of twice- daily (morning and afternoon) exposure to randomly assigned stressors for two weeks. Morning stressors were conducted between 8:00 am and 11:30 am and afternoon stressors were administered between 1:30 pm and 5:00 pm. Stressors consisted of rotation stress (1 h at 100 rpm on a platform orbital shaker), warm swim (20 min at 31 o C); cold swim (10 min at 18 o C), cold room stress (kept in 4 o C for one hour) and hypoxia (8% O 2 92% N 2 ). In the morning following the last afternoon stressor of experiment 2 and 3, rats received an overdose of sodium pentobarbital and were perfused with phosphate buffered saline, followed by 4% paraformaldehyde. Brains were post- fixed overnight in 4% paraformaldehyde and transferred to 30% sucrose at 4 o C until they were cut on a freezing microtome. Weight- matching. Animals were fed a sufficient amount of chow to produce a similar reduction in weight gain compared to the chronically stressed animals. In order to provide a less stressful means of food restriction, animals were fed 5 grams of chow in the morning at a random time between 7am and 11am and the rest between 5:30 and 6pm. Feeding the animals just before lights off does not shift their circadian rhythm. The morning feeding was added in to limit the total amount of time without food (since rats typically eat small amounts during the light period of the circadian cycle). Feeding them in this manner did not shift their circadian pattern of heart rate and blood pressure (data not shown). WM and CVS animals did not differ in body weight on any day of either experiment (Figure 5-1A). 115

117 Blood Collection. In experiment 2, blood was collected in EDTA at 0, 30, 60, and 120 minutes following the onset of 30 minutes of restraint stress. The animals were placed into a plastic restraint tube. Following collection, blood was spun for 15 minutes at 6000 rpm. Plasma was collected and stored at -20 º C. Corticosterone was quantified on these samples by radioimmunoassay using kit from MP Biomedicals. Area under curve was calculated using equation for a trapezoid as previously described (Choi et al., 2008a; Choi et al., 2008b). Additional aliquots were analyzed using a luminex assay for insulin and leptin analysis, with 2 ul used to calculate plasma glucose concentration via Freestyle glucometer (Abbot Laboratories, Alameda Ca). Telemetry surgery. The use of radiotelemetry to measure cardiovascular parameters and activity allows the continuous recording of mean arterial pressure (MAP), heart rate (HR), and activity in conscious freely moving animals. Rats were anesthetized with isoflurane and implanted with a radiotelemetry transmitter (TA11PA-C40, Data Sciences International (DSI), St Paul, MN, USA). The descending aorta was exposed via an abdominal incision, and a catheter extending from the transmitter capsule was placed into the descending aorta and secured with tissue adhesive (Vetbond; St Paul, MN, USA) and a cellulose patch. The capsule was sutured to the abdominal musculature, the abdominal musculature was sutured, and wound clips were applied to the skin. Following surgery, animals were monitored for two weeks to insure that the animals recovered properly. Animals were removed from the study if they did not recover their presurgery body weight. Baseline cardiovascular parameters were recorded for two prior days the beginning of experimentation. Animals were divided into groups such that there was no difference in starting body weight, MAP, and HR. The radio- telemetry system produced data 116

118 points at 10 second intervals throughout the day. For the daily time points, data was averaged throughout the 24 hour period, excluding an hour following the termination of each stressor. MAP, HR, and activity were determined using A.R.T. Platinum software (DSI, St. Paul, MN) Cardiovascular recording after novel (footshock) stress. After 14 days of CVS in experiment #1, the cardiovascular response to footshock stress was monitored. Baseline values were defined as the average measure taken over the hour immediately prior to placement in a Gemini Shock Apparatus (San Diego Instruments). The animals were allowed to explore for five minutes, after which they were given a series of five 1.5 ma shocks over a period of two and a half minutes. The shocks were administered at five minutes, six minutes, six minutes and 10 seconds, seven minutes, and seven minutes and 30 seconds following exposure to the shock chamber. The animals were then returned to their homecage, where cardiovascular recordings were collected. Average MAP and HR were calculated across five minute time bins over the 120 minute poststress period. Adipose tissue analysis. The carcasses of the animals were placed into a plexiglass tube and inserted into an EchoMRI whole body composition analyzer system (Echo Medical Systems, Houston, TX). The EchoMRI provides estimations of fat and lean mass. After completion, the animals were pelted in order to separate subcutaneous from visceral fat and subsequently reanalyzed in the same EchoMRI. The pelt contains inguinal white adipose tissue fat pads and any other subcutaneous fat. The remaining carcass contained all internal fat pads and intramyocellular adipose tissue (Table 5-1). 117

119 In Situ Hybridization. Animals were overdosed with Pentobarbital and perfused 0.9% saline followed by 4% phosphate buffered formaldehyde. The brains were sectioned on a microtome at 35 micrometers in a one in twelve series. Two wells of one in twelve series of tissue were used for analysis of oxytocin (OT) and CRH mrna expression. Sections were immersed in 0.25% acetic anhydride [suspended in 0.1 M triethanolamine (ph 8)] for 10 min, rinsed twice in 2x saline sodium citrate buffer (SSC) for 5 min, then dehydrated through graded alcohols. Antisense crna probes complementary to CRH (765 bp) and OT (477 bp) were generated by in vitro transcription using 35 S-UTP. The CRH probe was synthesized as previously described (Choi et al., 2008a; Choi et al., 2008b; Solomon et al., 2010). The OT fragment was cloned into a BSSK vector, linearized with HindIII, and transcribed with T7 RNA polymerase (Fisher Scientific Co., Pittsburgh, PA). Each transcription reaction (15 µl) consisted of 1x transcription buffer, 62.5 µci 35 S-UTP, 330 µm ATP, 330 µm GTP, 330 µm CTP, 10 µm cold UTP, 66.6 mm dithiothreitol, 40 U ribonuclease inhibitor, 20 U T7 RNA polymerase, and 2.5 µg linearized DNA. The transcription reaction was incubated at 37ºC for 60 min, and labeled probe separated from free nucleotide by ammonium acetate precipitation. 35 S-probes were diluted in hybridization buffer [50% formamide, 20 mm Tris-HCl (ph 7.5), 1 mm EDTA, 335 mm NaCl, 1x Denhardt s solution, 200 µg/ml herring sperm DNA, 100 µg/ml yeast trna, 20 mm dithiothreitol, and 10% dextran sulfate] at an activity count of 1,000,000 cpm per 50 µl buffer. 50 µl of buffer was added to each slide. Slides were coverslipped then incubated overnight at 55ºC in humidified chambers containing 50% formamide. The coverslips were removed in 2x SSC and the slides incubated in 100 µg/ml ribonuclease A for 30 min at 37ºC. Slides were rinsed in 2x SSC, incubated in 0.2x SSC (65ºC) for 1 h then dehydrated through graded alcohols. Slides were exposed to Kodak Biomax MR-2 film (Eastman Kodak, Rochester, NY) for 7 days for CRH 118

120 and 8 hours for OT. Scion Image software (Scion, Frederick, MD) was used for semiquantitative analyses of autoradiographs. The anatomical regions of interest (CRH: PVN; OT: dorsal parvocellular PVN (dppvn), lateral parvocellular PVN (lppvn), medial parvocellular PVN (mppvn), magnocellular PVN (mgpvn), and supra optic nucleus (SON)) were determined based on Paxinos and Watson s rat brain atlas (Paxinos and Watson, 2005b). Gray level units were collected for specified brain regions and the background signal over a non- hybridized area within the same section subtracted to obtain corrected gray level units (CGL). With each film, 14C radioactive standards (ARC) were included to assure that all signal intensities were within the linear range of detection. Analysis of autoradiographs was completed by a researcher blind to experimental groups. Statistical Analyses. Statistics were analyzed using Sigma Stat (Systat Software, San Jose, California). Data are expressed as mean ± standard error. Outliers were determined if the value exceeded both 1.96 times the standard deviation and 1.5 times the interquartile range (McClave, 1994). The data in figures 5-1A, 5-2, 5-3A, and 5-4 were analyzed by two way repeated measure ANOVA and Fisher s LSD post- hoc test with group (CVS, WM, and control) as between subject factors and time as repeating within subjects factor. The data in figure 5-3B, 5-5, table 5-1, and table 5-2 were analyzed by one way ANOVA with a Fisher s LSD post- hoc test with group (CVS, WM, and control) as a between subject factor. Since the food intake values of the WM animals did not differ from animal to animal, Figure 5-1B was analyzed by one way ANOVA on rank. When necessary, the data went under log transformation and then reanalyzed. 119

121 IV. Results Experiment #1 To test the hypothesis that the reductions in weight gain produced by CVS are sufficient to drive the physiological effects of chronic stress, we used three different groups: CVS, WM, and controls. WM animals were fed a sufficient amount of food to produce similar reductions in weight gain to CVS animals. Notably, we fed the WM animals significantly less food to produce similar reductions in body weight gain to the CVS group (Figure 5-1B) {H(1,25)= , p<.01}, suggesting that CVS elevates energy expenditure to a greater degree than would be predicted by weight loss alone. The animals mean arterial pressure, heart rate, activity, body weight, and 24 hour food intake were monitored during recovery from telemetry surgery. Animals were assigned to CVS, WM and control groups on the basis of these parameters to normalize baseline values. On the first day of CVS, 24 hour HR (Figure 5-2A), MAP (Figure 2B), and activity did not differ between groups. Starting at day six, 24 hour average heart rate was reduced in WM animals relative to CVS and control animals {F(27,284)= 4.303, p<.01} (Figure 2A). Reduced heart rat persisted through the end of the experiment (Figure 5-2A). However, neither MAP (Figure 5-2B) nor activity differed between the groups over the course of the study. On day 14, we tested cardiovascular parameters following exposure to a brief novel stressor (footshock). Footshock elicited an increase in heart rate that did not differ between the groups (Figure 5-2C). However, the WM animals recovered to their baseline heart rate more quickly than unstressed controls {F(19,208)=1.65, p=.047} (Figure 5-2C), indicating that weight loss affects the recovery of cardiovascular responses to psychogenic stress. Despite a similar reduction in body weight gain, CVS did not reduce the duration of the HR response to novel 120

122 stress to similar degree than WM animals (Figure 5-2C), indicating that chronic stress negates the effects of weight loss alone on return to baseline HR. Experiment #2 In experiment #2, we tested the impact of CVS or WM on post- stress facilitation of HPA axis responses to psychogenic stress. Groups of CVS, WM, and control animals were prepared using the same stress and weight- matching protocol outlined in experiment #1. As expected, novel restraint elicited a significant increase in plasma corticosterone in all groups (Figure 5-3B). Food restriction increased the area under the curve of the corticosterone response, whereas there was no main effect of chronic stress {F(1,24)=6.335, p<.01} (Figure 5-3B). These results suggest that weight restriction may be sufficient to facilitate HPA axis responses to psychogenic stress. Despite elevated plasma corticosterone, it is important to note that food restriction reduced adrenal weight, whereas chronic stress induced adrenal hypertrophy {F(2,32)=7.614, p<.01} and thymic involution {F(2,32)=4.297, p=.013} (table 5-1) (as previously noted (Choi et al., 2008a; Choi et al., 2008b; Flak et al., 2009)). These data suggest that chronic stress has longterm HPA axis effects above and beyond those of weight restriction alone. Since weight restriction modulated both the cardiovascular and glucocorticoid responses to psychogenic stress, we also assessed secretion of leptin and insulin, two major factors that are known to be modulated by energy stores. Importantly, chronic stress and food restriction reduced adiposity in both the subcutaneous {F(2,25)=5.387, p=.01} and visceral {F(2,25)=3.728, p=.036} depots of the rats (table 5-1), suggesting a possible connection between changes in stress responsiveness and altered adipose signaling. We examined plasma leptin and insulin in the same samples used for corticosterone measures. Despite a similar amount of fat to that of WM animals, chronic stress reduced plasma leptin {F(5,98)=5.325, p<.01} and insulin 121

123 {F(5,96)=3.342, p<.01}, suggesting that chronic stress reduces release/storage of leptin/insulin independent of the effects of weight loss (Figure 5-4). Thus, we predicted that chronic stress would facilitate stress- induced hyperglycemia, but CVS attenuated stress- induced hyperglycemia {F(2,96)=62.873, p<.01} (Figure 5-4), suggesting either an enhancement in leptin/insulin sensitivity or reduction in hepatic glucose output. Experiment #3 Additional groups of CVS (n=8), WM (n=9) and control (n=8) animals were prepared to test the impact of CVS and WM on key neuropeptidergic systems involved in stress excitation (CRH) (Ulrich-Lai and Herman, 2009) and body weight regulation (OT) (Blevins et al., 2004) Indeed, CVS (n=8) increased and WM (n=9) reduced PVN CRH mrna (Figure 5-5A) {F(2,19)=3.482, p<.032}. PVN OT was not altered within the SON, mgpvn, mppvn, or lppvn (Table 5-2). However, both CVS and WM reduced dppvn OT mrna {F(1,22)=6.143, p<.01}, suggesting a weight- specific effect on OT regulation in this pre- autonomic cell group (Figure 5-5B). V. Discussion Our data indicate that body weight alone is not sufficient to account for the deleterious physiological effects of chronic stress. In fact, reductions in weight gain generally reduced resting heart rate, suggestive of a beneficial effect on the organism. Basal heart rate was reduced in WM animals beginning at day six, and was accompanied by a more expeditious recovery of baseline heart rate following a psychogenic challenge. The therapeutic properties of dieting for hypertensive patients have long been accepted, but the mechanism is not known (Aronne and Isoldi, 2007; Laederach-Hofmann et al., 2000; Van Gaal et al., 2006). Currently published 122

124 reviews would presume that the connection is due to a change in circulating hormones, that alters sympathetic/parasympathetic reactivity. Since previously published work utilized months of intermittent food deprivation (Wan et al., 2003), it is somewhat surprising that our food restriction model attenuated heart rate. In addition, our ~20% food restriction is mild compared to that of other groups (usually employing 40% restriction) (Mattson and Wan, 2005). Interestingly, CVS did not attenuate cardiovascular responses like WM despite a similar reduction in body weight gain. This suggests that chronic stress exposure blocks the cardiovascular effect of reduced body weight, or alternatively, that weight loss protects against a more severe cardiovascular effect of chronic drive. Since chronic unpredictable stress can produce hypertension in rats without altering body weight (Grippo et al., 2002), we assume that the latter possibility is more likely. Future studies will specifically test these two possibilities. Chronic stress- induced regulation of metabolic parameters appears to go above and beyond the effect of weight alone. For instance, chronic stress reduced insulin and leptin levels during the restraint challenge to a greater degree than weight restriction alone, indicating that chronic stress- specific regulation of insulin and leptin transcends that driven by negative energy balance alone. Other studies have shown reductions in leptin and insulin following chronic stress regimens (Lin et al., 2005; Lu et al., 2006; Solomon, 2010), but those studies did not include a weight- matched control. These reductions in leptin and insulin could have subsequent effects on stress responsiveness, since they both can modulate HPA axis reactivity (Heiman et al., 1997; Holden, 1999), but we do not know the current state of leptin and insulin sensitivity following CVS. Dysregulation in leptin and insulin action (Aronne and Isoldi, 2007; Laederach- Hofmann et al., 2000; Van Gaal et al., 2006) is proposed to mediate obesity- associated cardiovascular disease, but our data suggest that these hormones are not critical components in 123

125 differentiating the cardiovascular impact of weight restriction and stress. We should note that for the purposes of studying the HPA axis endpoints, animals were not fasted prior to leptin and insulin measures, which may have some bearing on the observed basal hormone levels. For example, some of these changes could possibly be due to different feeding schedules of the animals, which limits the overall interpretation of the basal values. True basal levels require fasting of the animals for a substantial time prior to testing, which could not be performed within the context of a stress paradigm (food depriving animals clearly influences stress responses (Akana et al., 1994; Kiss et al., 1994)). The current studies queries whether stress- related changes in leptin and insulin may influence the differential regulation of stress responses, which does not appear to the case in our paradigm. Both CVS and WM reduced PVN OT mrna in the dppvn, suggesting a weight- related regulation of OT mrna. This effect was not observed in other PVN subregions or the SON. The dppvn neurons send pre- autonomic projections to either the brainstem (e.g., nucleus of the solitary tract (NTS) or the intermediolateral division of the spinal column (Petersson, 2002). OT is known to have anorectic actions at the NTS (Blevins et al., 2004), and thus reductions in OT mrna may occur as a reaction to negative energy balance in both WM and CVS animals. Food restriction reduced PVN CRH mrna. Current data suggests that PVN CRH is both anorectic (Britton et al., 1982; Gosnell et al., 1983) and stimulatory for sympathetic activity (Arase et al., 1988; LeFeuvre et al., 1987). The weight- related reduction in CRH may, like OT, be involved in reducing satiation and curtailing sympathetic activity, perhaps in concert with decrements in OT. This stands in contrast to the effects of CVS, which enhance PVN CRH mrna expression (see also (Choi et al., 2008a; Choi et al., 2008b; Solomon et al., 2010)) and are associated with anorexia and, relative to WM animals, increased HR responsiveness to stress. 124

126 Thus, stress- induced changes in CRH may at least in part, drive the dissociation in cardiovascular responsiveness that are observed between CVS animals and their weightmatched (but otherwise unstressed) counterparts. Food restriction elevated resting corticosterone levels (as previously observed (Jahng et al., 2007; Wan et al., 2003)) and increased the overall HPA axis response to restraint stress (area under the curve). These data are not consistent with reduced adrenal weight and decreased PVN CRH mrna expression observed in the WM group relative to controls, suggesting that the increase in corticosterone is due to enhanced drive of central or pituitary limbs of the HPA axis, and may be adrenal in origin. In line with this interpretation, previous studies indicate that ACTH responses are blunted, but glucocorticoids are increased following 24 hour food deprivation (Akana et al., 1994), which is sufficient to reduce body weight. These observed changes in HPA axis responsiveness to a novel challenge are correlated with reduced clearance of glucocorticoids (Kiss et al., 1994), consistent with peripheral enhancement of adrenal hormone availability. With the addition of the data collected within this manuscript, the evidence suggests that these post- pituitary changes in HPA responsivity are not unique to food deprivation, but sensitive to changes in metabolic state. Overall, the data suggest peripheral enhancement of both basal and post- stress corticosterone under mild weight restriction that could be a compensation for reduced energy availability. This compensatory mechanism does not appear to be in place in CVS animals, perhaps as a result of enhanced central HPA axis and cardiovascular activation. It is important to note that the CVS animals did not exhibit an exaggerated glucocorticoid response to acute novel restraint. However, Dallman defines facilitation as a maintained or enhanced response despite a history of negative feedback signals 125

127 caused by intermittently elevated plasma glucocorticoids (Dallman, 1993). By this definition, the maintained response to a novel stressor after CVS is consistent with HPA axis facilitation. The corticosterone increases seen in the weight- matched animal suggests that animals may be under some degree of chronic stress. However, whereas glucocorticoid hypersecretion can be seen during chronic stress, it does not define the condition, as elevated glucocorticoid levels can be driven by negative energy balance (Akana et al., 1994; Kiss et al., 1994; Wan et al., 2003), which is the case with the WM group. Indeed, CRH mrna levels and adrenal weights are both reduced in the WM group, whereas these endpoints are consistently increased in a variety of chronic stress models (Cullinan and Wolfe, 2000; Herman et al., 1995a; Ulrich-Lai et al., 2006b; Ziegler et al., 2005). These data suggest a reduced central HPA axis drive in the WM group, inconsistent with a state of chronic stress. Moreover, the spectrum of data in the WM group are not consistent with a habituated response to stress (Bhatnagar and Meaney, 1995), given that habituating models typically show normal adrenal weights and CRH expression, within the context of HPA axis facilitation (not observed in our WM group). However, it is important to note that the weight loss of CVS animals is due to voluntary means, while it is involuntary in WM animals. This may have ramifications independent of altered body weight, as reward- aversion related pathways may modulate both HPA axis and sympathetic responses to novel stressors (Ulrich-Lai et al., 2011a). In conclusion, our data demonstrate that reductions in weight gain produce changes in cardiovascular tone and reactivity. The observed changes could be beneficial to the organism, and possibly reversed during chronic stress. In addition, reduced body weight decreased adrenal weight and attenuated PVN CRH expression, indicative of blunted central HPA drive, whereas chronic stress has the opposite effects. Overall, the data suggest that weight loss is unlikely to 126

128 account for the deleterious effects of chronic stress on the organism, but in fact produces beneficial effects that are effectively absent or indeed, reversed in the face of chronic stress. Additional experiments will be required to fully understand the interplay between metabolism and stress in regulation of physiological responses to chronic adversity. Table 5-1. Experiment 2 body, organ, and adipose weight Body Weight (Pre) (grams) Body Weight (Post) (grams) Control CVS WM ± 3.03 g 336 ± 3.09 g ± 2.95 g ± 6.38 g ± 3.72 g* 348 ± 1.57 g* Adrenal Weight (mg) 70 ± 3.35 mg ± 2.37 mg* 60.9 ± 1.79 mg* Thymus Weight (mg) ± mg ± mg* ± mg Total Lean (grams) ± 2.11 g ± 1.66 g* ± 3.94 g* Total Fat (grams) ± 1.17 g 22.7 ±.9 g* ±.84 g* Subcutaneous Fat (grams) 13.56±.89 g ±.65 g* 9.86 ±.7 g* Visceral Fat (grams) ± 1.03 g ± 1.34 g* ±.65 g* Data are expressed as mean + SEM. * group is significantly different from control at P <

129 Table 5-2. PVN and SON oxytocin mrna expression Control CVS WM lppvn (CGL % Control) 100 ± 10.22% ± 12.46% ± 13.73% mppvn (CGL % Control) 100 ± 8.51% ± 9.42% ± 9.77% mgpvn (CGL % Control) 100 ± 3.6% ± 5.11% ± 12.69% SON (CGL % Control) 100 ± 3.28% ± 4.75% ± 11.07% Data are expressed as mean + SEM. Figure 5-1 Figure 5-1: Experiment #1: Body weight and Food Intake. Figure 5-1A demonstrates the body weight gain throughout experiment 1 and Figure 5-1B displays total food intake over the two week experiment. CVS exposure reduced body weight (A), but not food intake (B). WM animals did not differ from CVS animals on any day of the experiments. We fed the WM animals significantly less food than the control and CVS animals despite a similar trajectory in weight gain. * denotes p.05 compared to control animals. 128

130 Figure 5-2 Figure 5-2: Experiment #1: Basal cardiovascular parameters. Figures 5-2A and 5-2B summarize the basal cardiovascular activity data. The 0 time point refers to the day prior to chronic stress exposure/ weight restriction. WM reduced heart rate beginning at day 6 and continuing through the rest of the experiment (B). Despite a similar body weight, CVS did not exhibit recapitulate this effect. However, mean arterial pressure did not differ between groups (A). * denotes p.05 compared to control animals. Figure 5-3 Figure 5-3: Experiment #1: Cardiovascular responses to acute psychogenic stress.5-3a and 5-3B illustrate cardiovascular responses to acute psychogenic stress. 0 time point refers to the period before anyone entered the animal housing room. The arrow indicates the time where the animals experienced footshock. WM expedited the heart rate recovery to baseline following acute shock exposure (B) without altering mean arterial pressure (A). In this regard, the CVS animals displayed an intermediate recovery to baseline heart rate. * denotes p.05 compared to control animals. 129

131 Figure 5-4 Figure 5-4: Experiment #2: Corticosterone responses to acute psychogenic stress. Figure 5-3A illustrates glucocorticoid responses to acute psychogenic stress. The animals were restrained for 30 minutes and then released from the restraint tube after this time period. Blood samples were collected prior to restraint, after 30 minutes of restraint, after 30 minutes of recovery, and after 90 minutes of recovery. The 30 minute time point refers to peak glucocorticoid levels. Figure5-3B summarizes area under the curve analysis. There was a main effect of WM to increase plasma corticosterone (A), as well as the integrated stress response (area under the curve) (B). * denotes p.05 compared to control animals. 130

132 Figure 5-5 Figure 5-5: Experiment #2: Metabolic responses to acute psychogenic stress. Figure 5-4 summarizes plasma leptin (A), insulin (B), and glucose (C) responses to acute psychogenic stress. CVS reduced leptin 0 and 60 minute time point following 30 minutes of restraint stress, but WM only reduced leptin at the 0 minute time point (A). CVS reduced plasma insulin at the 0, 30, and 60 minute time points (B). WM only increased plasma insulin at the 0 minute time point (B). There was a main effect of CVS to attenuate stress- induced hyperglycemia (C). * denotes p.05 compared to control animals. 131

133 Figure 5-6 Figure 5-6: Experiment #3: Basal PVN peptide mrna. Figure 5-6 illustrates PVN CRH (A) and OT (B) expression. CVS increased, while WM reduced PVN CRH mrna (A). Both CVS and WM reduced dppvn OT mrna (B). * denotes p.05 compared to control animals. 132