Viewed Extensively (Reviewed in Aldskogius and Kozlova, 1998; Bedner Et Al., 2012)

MIAMI UNIVERSITY The Graduate School

Certificate for Approving the Dissertation

We hereby approve the Dissertation of Aminata Porhy Coulibaly

Candidate for the Degree: Doctor of Philosophy

______Director (Dr. Lori G. Isaacson)

______Reader (Dr. Kathleen Killian)

______Reader (Dr. Paul James)

______Reader (Dr. Yoshi Tomoyasu)

______Graduate school representative (Dr. Jennifer J. Quinn)

ABSTRACT

EFFECTS OF PERIPHERAL AXON TRANSECTION ON THE CENTRAL NERVOUS SYSTEM

by Aminata Porhy Coulibaly

This study focused on characterizing the effects of peripheral axon injury on the central nervous system (CNS). Following the transection of sympathetic preganglionic axons in the periphery, analyses of the injured neuron cell bodies and surrounding support cells were conducted in the spinal cord. Changes in the cell bodies of the injured neurons in the spinal cord observed at 1 week after injury included a decrease in neurotransmitter expression and reduced cell body volume. Plasticity also was observed in the oligodendrocyte (OL) population in the vicinity of the injured neurons. At 1 week after injury, both OL progenitor cells and OLs expressing the tyrosine kinase receptor TrkB were increased. The expression of the gap junction protein Cx32 in these cells also was increased. When uninjured neurons in the sympathetic chain were investigated, it was found that, at 1 week following the injury, the afferent inputs onto the injured neurons in the spinal cord were decreased. In addition, the neurons in the hypothalamus which provide afferent input to the injured neurons exhibited a loss of neurotransmitter expression as well as an increased number of synaptic inputs. The results of these studies demonstrate that peripheral axon injury can lead to changes in injured and uninjured neurons as well as glia cells in the central nervous system. Because no cell death was observed following the injury, these results suggest that acute cellular plasticity in CNS neurons plays an important role in neuronal survival after injury of axons in the periphery.

EFFECTS OF PERIPHERAL AXON TRANSECTION ON THE CENTRAL NERVOUS SYSTEM A dissertation

Submitted to the Faculty of Miami University in partial Fulfillment of the requirements For the degree of Doctor of Philosophy Department of Biology

Aminata Porhy Coulibaly Miami University Oxford, Ohio 2014

Dissertation Director: Dr. Lori G. Isaacson

TABLE OF CONTENTS

Table of contents……………………………………………………………………………...... ii List of Figures…………………………………………………………………………………vii Acknowledgements……………………………………………………………………………..x

Chapter 1: General Introduction………………………………………………………….....1 References……………………………………………………………………………………..9

Chapter 2: Transection of preganglionic axons leads to CNS neuronal plasticity followed by survival and target reinnervation……………………………………………………………...15 1. Introduction………………………………………………………………………….17 2. Materials and Methods……………………………………………………………...18 2.1. Animals…………………………………………………………………..18 2.2. Transection of cervical sympathetic trunk……………………………….19 2.3. Tissue collection and immunofluorescence studies……………………...20 2.4. TUNEL assay…………………………………………………………….20 2.5. Analysis of neurons in the IML…………………………………………..21 2.6. Semi-quantitative western analysis………………………………………22 2.7. Analysis of ChAT-ir profiles in the SCG………………………………...24 3. Results………………………………………………………………………………...24 3.1. Analysis of ptosis………………………………………………………....24 3.2. Changes in the IML following CST transection………………………….24 3.3. Changes in spinal cord BDNF and/or TrkB following CST transection....26 3.4. Change in the SCG following CST transection…………………………..27 4. Discussion…………………………………………………………………………….43 4.1. Cell survival following peripheral injury………………………………...43 4.2. Role of ATF3 following injury…………………………………………..44 4.3. Regulatory influences on spinal cord neurons: effects of injury………...45 4.4. Reinnervation of the SCG following CST transection…………………..46 5. Conclusions………………………………………………………………………….48

6. References……………………………………………………………………………50

Chapter 3: Distribution and phenotype of TrkB oligodendrocyte lineage cells in the adult rat spinal cord…………………………………………………………………………………….57 1. Introduction………………………………………………………………………….59 2. Materials and Methods……………………………………………………………...60 2.1. Tissue preparation and immunohistochemistry…………………………..60 2.2. Quantitative analysis……………………………………………………...61 3. Results………………………………………………………………………………...63 3.1. Distribution of TrkB cells in the adult rat spinal cord: comparison with OPCs and mature OLs……..……………………………………………...63 3.2. Expression of TrkB by OL lineage cells in the adult spinal cord………...64 3.3. Clustering of OL lineage cells in the adult spinal cord…………………..65 3.4. Perineuronal OL lineage cells in the adult spinal cord…………………...66 4. Discussion…………………………………………………………………………….79 4.1. Phenotype of TrkB nonneuronal cells in the intact spinal cord………….79 4.2. Distribution of OL lineage cells in the adult spinal cord…………………81 4.3. Density of OL lineage cells in the adult spinal cord……………………..82 5. Conclusions…………………………………………………………………………..82 6. References…………………………………………………………………………....85

Chapter 4: The effect of peripheral injury on oligodendrocyte subpopulations in the rat spinal cord……………………………….…………………………………………………....91 1. Introduction …………….…………………………………………………………...93 2. Materials and Methods.……………………………………………………………..94 2.1. Animal surgery and tissue collection…………………………………….94 2.2. Overall methods and data analysis……………………………………….95 2.3. Identification of OL lineage cells………………………………………...95 2.4. Analysis of Olig2 cells in the IML……………………………………….96 2.5. Analysis of changes in OL lineage cells………………………………….96

iii

2.6. Analysis of white matter-gray matter distribution of OL lineage cells in the spinal cord…………………………………………………………………...97 3. Results…………………………………………………………………………………..98 3.1. Olig2-ir cells in the IML were unchanged following CST transection….....98 3.2. Changes in OL lineage cells in the IML and LF following CST transection……………………………………………………………………....98 3.3. Analysis of NG2 and CC1 cells that expressed TrkB…………………….100 3.4. Comparison of cell density between GM and WM following injury……..100 4. Discussion ………………………………………………………………………….....116 4.1. Peripheral axon injury leads to NG2 cell plasticity……………………….116 4.2. Delayed effect of CST transection on the mature OL population………...118 4.3. Effects of peripheral injury are localized to the IML and LF……………..118 5. Conclusions…………………………………………………………………………....119 6. References………………………………………………………………………...…...120

Chapter 5: Peripheral axon injury leads to plasticity in connexin32 expression in the spinal cord…………………………………………………………………………………………….123 1. Introduction…………………………………………………………………………..125 2. Materials and Methods………………………………………………………………124 2.1. Surgery and tissue collection……………………………………………...126 2.2. General methods for Cx32 analyses………………………………………127 2.3. Analysis of Cx32 in the IML and LF following injury…………………...128 2.4. Identification of oligodendrocyte lineage cells that expressed Cx32……..129 3. Results…………………………………………………………………………………129 3.1. CST transection leads to increased expression of Cx32…………………..130 3.2. Differential expression of Cx32 in OL lineage cells following injury……131 4. Discussion……………………………………………………………………………..142 4.1. Effects of peripheral axon injury on Cx32 expression by OL lineage cells…………………………………………………………………………….142 4.2. Role for Cx32 expression in oligodendrocyte progenitor cells…………...143 4.3. Role for Cx32 expression in mature OLs……………………………...... 143

4.4. Peripheral injury increased the size of Cx32 plaques……………………..145 5. Conclusions…………………………………………………………………………....145 6. References……………………………………………………………………………..146

Chapter 6: Transneuronal effects of peripheral axon injury on the central nervous system.....151 1. Introduction…………………………………………………………………………...153 2. Materials and Methods……………………………………………………………….154 2.1. Animal surgery and tissue collection……………………………………...154 2.2. Procedures for locating the PVN of the hypothalamus…………………....156 2.3. Immunohistochemical procedures………………………………………...157 2.4. Analysis of syn and CRH in the spinal cord………………………………157 2.5. General analysis of brain tissues…………………………………………..159 2.6. Determination of overall CRH density and the number of CRH neurons in .the PVN…..………………………………………………………………...... 160 2.7. Determination of CRH neuronal volume in the PVN……………………..161 2.8. Analysis of syn-ir profiles in the PVN…………………………………….161 3. Results………………………………………………………………………………....162 3.1. Changes in syn and CRH immunoreactivity in the spinal cord…………...162 3.2. Changes in CRH in the PVN following peripheral injury………………...163 3.3. Changes in syn-ir profiles in the PVN following peripheral injury……….164 4. Discussion…………………………………………………………………………...... 179 4.1. Summary of findings……………………………………………………...179 4.2. Synaptic stripping in the IML of the spinal cord………………………….179 4.3. Changes in the number of CRH neurons in the PVN following CST transection……………………………………………………………………...181 4.4. Other models showing transneuronal effects following injury…………....183 4.5. Changes in synaptic inputs to CRH neurons in the hypothalamus………..183 4.6. Possible shortfalls of the present study……………………………………184 5. Conclusions…………………………………………………………………………....186 6. References……………………………………………………………………………..187

Chapter 7: General Discussion…………………….…………………………………..…….192 References……………………………………………………………………………………..199

LIST OF FIGURES CHAPTER1 Figure 1. Sympathetic injury model…………………………………………………………...7

CHAPTER 2. Figure 1. Transient changes in ChAT and ATF3 following CST transection………………..29 Figure 2. The total number of preganglionic neurons remained unchanged following CST transection……………………………………………………………………………………..31 Figure 3. Reduced soma volume observed at one week was reversed at the long term survival time points…………………………………………………………………………....33 Figure 4. No TUNEL positive neurons were observed in the IML following CST transection……………………………………………………………………………………..35 Figure 5. No caspase 3(casp3)-ir neurons were observed in the IML following CST transection……………………………………………………………………………………..37 Figure 6. Increased mature BDNF (mBDNF) and full length TrkB (TrkB.FL) in the spinal cord following CST transection…………………………………………………………….....39 Figure 7. ChAT and TH in the SCG following loss of afferent input………………………...41

CHAPTER 3. Figure 1. Localization and identification of OL lineage cells in the adult rat spinal cord…....71 Figure 2. Density of OL lineage cells within white matter (WM) and gray matter (GM) in the cervical (C4) and thoracic (T1) segments of the adult rat spinal cord…………………….73 Figure 3. The majority of mature OLs expressed TrkB at detectable levels……………….....75 Figure 4. Little overlap between OPC and TrkB populations in the intact spinal cord……....77 Figure 5. Clusters of TrkB and CC1 cells were observed primarily in the GM of the spinal cord…………………………………………………………………………………………….79 Figure 6. A population of OL lineage cells resided in close apposition to motor neurons in the ventral horn of adult rat spinal cord and was categorized as perineuronal………………...81

CHAPTER 4. Figure 1. Areas of quantification in the spinal cord of the rat……………………………….102

vii

Figure 2. Analysis of the number of Olig2 cells in the intermediolateral cell column (IML) following injury……………………………………………………………………………….104 Figure 3. Pattern of immunoreactivity for the three identified subpopulations of oligodendrocyte lineage cells in the intermediolateral cell column (IML) in the controls and following injury...... 106 Figure 4. Pattern of immunoreactivity for the three identified subpopulations of oligodendrocyte lineage cells in the lateral funiculus (LF) in the controls and following injury...... 108 Figure 5. Changes in OL lineage cells in the intermediolateral cell column (IML) and the lateral funiculus (LF) after peripheral axon injury……………………………………………110 Figure 6. Analysis of the expression of NG2 and CC1 with TrkB…………………………...112 Figure 7. Analysis of the density of OL lineage cells in white matter (WM) vs gray matter (GM) following injury………………………………………………………………………...114

CHAPTER 5. Figure 1. Cx32 expression in the IML………………………………………………………..132 Figure 2. Analysis of changes in Cx32 expression in the spinal cord following peripheral injury………………………………………………………………………………………….134 Figure 3. Analysis of the number of Cx32 plaques after peripheral injury………………….136 Figure 4. Cx32 plaques in association with OL lineage cells in the spinal cord…………….138 Figure 5. Analysis of the changes in Cx32 expression in OL lineage cells after injury……..140

CHAPTER 6. Figure 1. Immunohistochemical analysis of synaptophysin (syn) and corticotrophin releasing hormone (CRH) in the intermediolateral cell column (IML) at T1 level of the spinal cord….165 Figure 2. Evidence for synaptic stripping in the intermediolateral cell column (IML) of the spinal cord after peripheral axon transection…………………………………………………167 Figure 3. Experimental approach for quantification of changes in the paraventricular nucleus (PVN) after peripheral axon transection……………………………………………………...169 Figure 4. Analysis of CRH immunoreactive (-ir) neurons in PVN subdivisions following peripheral axon injury………………………………………………………………………...171

viii

Figure 5. Changes in CRH in the PVN after peripheral axon injury…………………………173 Figure 6. Syn immunoreactive (-ir) profiles in the PVN following peripheral axon injury….175 Figure 7. Synaptophysin (syn) immunoreactivity is increased in the PVN following peripheral axon injury……………………………………………………………………………………177

CHAPTER 7. Figure 1. Proposed model of the maturation of cells in the oligodendrocytes lineage……...197

AKNOWLEDGEMENTS

This study was supported by NIH NS051206 awarded to LGI and NSF DBI-0821211 awarded to the Center for Advanced Microscopic Imaging, Miami University, and in part by a Sigma Xi GIAR G20120315162215 awarded to APC. We thank Matt Duley and Richard Edelmann for their assistance with confocal microscopy and image analysis, Kiel Hawk, Zoe Hesp and Kara Francis for their help with the quantitative analysis of the SCG, Cassidy Lawrence and Matt Deer for their assistance with the cervical segments of the spinal cord, Sean Gannon for his help with the innervation of the thoracic level of the spinal cord, and James Oris, Miami University, for his help with the statistical analysis.

Chapter 1: General introduction

Peripheral nerve injury is a common source of nerve damage that can occur as a result of traumatic neuropathies (Ciaramitaro et al., 2010), upper limb injuries (McAllister et al., 1996), and compression injuries to the extremities (Burnett et al., 2004). However, very little is known about the effects that peripheral injury might have on the central nervous system (CNS). Recent studies have demonstrated that the parent neuronal cell bodies of transected peripheral axons, such as those of the facial (Rholmann et al., 1994; Jones et al., 2005; Kalla et al., 2007) and hypoglossal (Svensson et al., 1993) nuclei, and spinal cord ventral horn (Gilmore et al., 1990; Hajos et al., 1990; Coyle, 1998), undergo robust changes. Indeed these studies have led to the hypothesis that molecules originating from the peripheral injury site can influence the survival of the injured neuronal cell bodies in the CNS (Makwana and Raivich, 2005; Abe and Cavalli, 2008). This hypothesis provided the rationale for the experiments described in the chapters that follow. The peripheral injury model used in our laboratory involves the sympathetic preganglionic neurons that provide the sole innervation to postganglionic neurons of the superior cervical ganglion (SCG). The axons of these preganglionic neurons comprise a nerve bundle known as the cervical sympathetic trunk (CST). The CST is transected at its entry into the SCG (Fig. 1.). The postganglionic neurons of the SCG innervate peripheral targets such as the pineal gland, irides, submandibular gland, and extracerebral blood vessels (Arbab et al., 1986; Luebke and Wright, 1992). It has been suggested that these postganglionic axons are important for proper blood flow to the brain. Previous studies have demonstrated that the cell bodies of the preganglionic neurons innervating the SCG are located primarily in the first thoracic level of the spinal cord (T1), with the majority located in the intermediolateral cell column (IML), a nucleus found in the lateral horn of the spinal cord (Rando et al., 1981; Per Brodal, 1992; Pyner and Coote, 1994). Our lab has previously demonstrated the presence of robust and transient changes in the glial cell populations in the IML of the spinal cord following CST transection (Coulibaly and Isaacson, 2012). For example, we observed the activation of both astrocytes and microglia, as well as increased numbers of microglia and oligodendrocytes (OLs) expressing TrkB at 1 week after injury (Coulibaly and Isaacson, 2012). These changes subsided by 10 weeks after the injury (Coulibaly and Isaacson, 2012).

Other models have demonstrated changes in glial cells following injury to the facial (Kalla et al., 2001; Jones et al., 2005), hypoglossal (Svensson et al., 1993) and spinal nerves (Coyle, 1998). Both beneficial and detrimental roles have been attributed to these changes. In our injury model, it was unknown whether the changes in glial cells in the vicinity of the injured cell bodies were detrimental or beneficial. In Chapter 2 of this dissertation, the plasticity of the injured neuronal cell bodies in the IML as well as the reinnervation of their target, the SCG, was examined following CST transection. Growth factors have been shown to play an important role in both maintenance and survival of CNS neurons (Lu et al., 2005). Of these, brain derived neurotrophic factor (BDNF) in the SCG is thought to play a role in the regulation of the innervating IML neurons (Causing et al., 1997). The interaction of BDNF with the full length TrkB receptor (TrkB), elicits signal transduction in the expressing cells (Barbacid, 1995). The interaction of BDNF with the truncated TrkB receptor 1, which lacks the tyrosine kinase domain, is thought to play a role in BDNF sequestration (Biffo et al., 1995; Lu et al., 2005). Alterations in BDNF and TrkB have been observed in the spinal cord following injury (Dougherty et al., 2000; Gomez-Pinilla et al., 2012). Therefore, in Chapter 2 we also investigated whether protein levels of BDNF and/or TrkB receptors in the spinal cord were affected by the distal preganglionic axon transection. The finding that OLs expressing the full length TrkB receptor were increased in the spinal cord following CST transection (Coulibaly and Isaacson, 2012) led to the study in Chapter 3. The presence of TrkB OLs in the CNS, brain (Vondran et al., 2010) and spinal cord (Skup et al., 2002; McCartney et al., 2008) was previously demonstrated and, although these cells have shown plasticity under different conditions, exercise (Macias et al., 2005) and injury (Coulibaly and Isaacson, 2012), little was known regarding their roles in the spinal cord. Therefore, the goal of Chapter 3 of this dissertation was to determine which subpopulations of OLs expressed TrkB. Three subpopulations of OLs have been described in the CNS. The first population is comprised of oligodendrocyte progenitor cells (OPCs), a group of progenitor cells that expresses the chondroitin proteoglycan 4 commonly known as nerve/glia 2 or NG2 antigen on the plasma membrane (Kessaris et al., 2008; Armati and Mathey, 2010). NG2-expressing OPCs (referred to as NG2 cells) divide and differentiate into mature OLs upon stimulation, which may involve the presence of growth factors (McTigue et al., 1998). Once stimulated, the differentiating OPC down regulates NG2 expression and up regulates the expression of adenomatous polysis coli

(APC; commonly known as CC1), a marker for mature OLs (Miller et al., 2002; McTigue and Tripathi, 2008; Kessaris et al., 2008; Nishiyama et al., 2009; Armati and Mathey, 2010), which represent the second subpopulation of OL lineage cells. As the OPC transitions to the mature stage, an intermediate cell population has been described (Miller, 2002). This third population, which has only been reported in vitro exhibits neither NG2 nor CC1 expression (Miller, 2002; Nishiyama et al., 2009). The primary objective of the third chapter of this dissertation was to determine which subpopulations of OLs (OPC or mature) expressed TrkB in the normal adult spinal cord, and to characterize the location of TrkB OLs in the uninjured spinal cord. The goal of Chapter 4 was to understand how CST transection affected the distribution and phenotype of oligodendrocyte subpopulations in the spinal cord. It was known that TrkB OLs were increased in the IML (Coulibaly and Isaacson, 2012), yet it was unclear whether the progenitor and/or mature OL populations were affected by the injury. In addition, it was not known whether the effects of the peripheral injury were global or localized to a specific area of the spinal cord, and so the entire spinal cord was analyzed following CST transection. Intercellular communication is important for neuronal survival after peripheral injury. For example, intercellular communication between injured motor neurons and astrocytes, which is mediated through gap junctions (Chang et al., 2000; Aldskogius and Kozlova, 1998), play an important role in the provision of survival factors and removal of detrimental factors in the CNS after injury. Changes in intercellular communication between astrocytes following peripheral injury have been reviewed extensively (reviewed in Aldskogius and Kozlova, 1998; Bedner et al., 2012). For example, the plasticity in astrocyte gap junction protein expression was demonstrated in injury models of facial nucleus nerve (Rholmann et al., 1994), trigeminal nerve (Guo et al., 2007), spinal cord injury (Theriault et al., 1997; Lee et al., 2005), and brain injury (Ohsumi et al., 2006; Theodoric et al., 2012). However, little was known about the effects of injury, both peripheral and central, on the communication between OLs and other glial cells in the CNS. Therefore, the goal of Chapter 5 of this dissertation was to understand the effects of peripheral injury on the expression of intercellular communication channels called gap junctions expressed by OLs. Glial cells in the nervous system can communicate with each other through gap junctions (Bedner et al., 2012). Gap junctions permit the movement of small molecules, such as ions and signaling molecules, between adjacent cells (Bedner et al., 2012; Pereda et al., 2013). Gap

junctions are formed by a family of transmembrane proteins known as connexins (Cx; Bedner et al., 2012) which have the ability to form pores on the plasma membrane (Bedner et al., 2013; Pereda et al., 2013). Cx are identified by their molecular weight; for example, the weight of Cx30 is 30 kDa (Bedner et al., 2012). OLs have been shown to express Cx29, Cx32, and Cx47 (Bedner et al., 2012; Pereda et al., 2013). Cx29 is expressed only in the myelin sheath surrounding axons, while Cx32 and Cx47 are found around the cell membrane of neuronal cell bodies (Kleopa et al., 2004). Little is known regarding the roles of Cx29 and Cx47 in the CNS. In contrast, mutations in Cx32 have been implicated in the degenerative disorder X-linked Charcot Marie Tooth disease (Bedner et al., 2012; Pereda et al., 2012). In addition, changes in Cx32 have been shown to influence the severity of multiple sclerosis symptoms in the experimental autoimmune encephalomyelitis mouse (Markoullis et al., 2012). Therefore, the goal of Chapter 5 was to determine the effects of peripheral axon injury on Cx32 expression in OL subpopulations, including NG2, TrkB, and CC1 cells. The final chapter of this dissertation focused on the transneuronal effects of the CST transection in the brain, specifically the hypothalamus. Several recent studies have demonstrated plasticity in uninjured neurons with connections to injured neuronal cell bodies after peripheral injury (Scholz et al., 2005; Leong et al., 2011). Preganglionic neurons in the IML of the spinal cord receive afferent input from higher centers such as the paraventricular nucleus (PVN) of the hypothalamus (Fig. 1; Tucker and Saper, 1985; Hosoya et al., 1991). Neurons in the PVN can be divided into three different subnuclei: magnocellular neuroendocrine, parvocellular neuroendocrine and parvocellular preautonomic (Levy and Tasker, 2012). Many PVN neurons express corticotrophin-releasing hormone (CRH) (Olschowka et al., 1982; Merchenthaler et al., 1983). The preautonomic PVN neurons synapse directly onto preganglionic neurons in the IML (Smith and DeVito, 1984; Tucker and Saper, 1985; Hosoya et al., 1991), and have been shown to express CRH receptors (Korosi et al., 2007; Yamaguchi and Okada, 2009; Tanabe et al., 2012), providing evidence that PVN neurons expressing CRH can play a role in the sympathetic outflow. Therefore, it was feasible that the disruption of the sympathetic circuitry would affect these hypothalamic neurons. In Chapter 6, we analyzed the PVN for changes after peripheral injury to the IML preganglionic neurons. We hypothesized that the CRH preautonomic neurons in the PVN would show changes, indicating their response to the peripheral axon injury.

All of the studies in this dissertation investigated the possibility that both neurons and glial cells in the CNS exhibit plasticity following CST transection. The underlying hypothesis was that retrograde factors arising from the injury site can have far reaching effects on the spinal cord and the brain. In addition, these studies provided an in depth examination of OLs in the spinal cord and their response to CST transection.

Legend = CRH = NE = Ach Brain: = Satellite cells Paraventricular nucleus (PVN), = astrocytes Hypothalamus = Oligodendrocytes = microglial = transection BV = blood vessels

B Cervical sympathetic V trunk (CST)

Spinal cord: Superior cervical Extracerebral blood Intermediolateral ganglion (SCG) vessels cell column (IML)

Figure 1

Figure 1: Image depicting the hierarchy of neurons present in the sympathetic model used in this dissertation. In this model, a CRH neuron (orange) located in the hypothalamus innervates a preganglionic neuron (gray) located in the intermediolateral cell column (IML) of the spinal cord. The preganglionic neuron in turn innervates the postganglionic neuron of the superior cervical ganglion (SCG; purple) located in the periphery. The postganglionic neuron provides innervation to peripheral targets such as the extracerebral blood vessels. The injury in our model consists of the transection (red X) of the preganglionic axons located in the nerve bundle known as the cervical sympathetic trunk (CST) at their entry into the SCG.

References

Abe, N., Cavalli, V. (2008). Nerve injury signaling. Current Opinion in Neurobiology, 18(3), 276–83.

Aldskogius, H., Kozlova, E. N. (1998). Central neuron-glial and glial-glial interactions following axon injury. Progress in Neurobiology, 55(1), 1–26.

Arbab, M. a. R., Wiklund, L., Svendgaard, N. A. (1986). Origin and distribution of cerebral vascular innervation from superior cervical, trigeminal and spinal ganglia investigated with retrograde and anterograde WGA-HRP tracing in the rat. Neuroscience, 19(3), 695–708.

Armati PJ, Mathey EK. 2010. The biology of oligodendrocytes. Cambridge, UK. Cambridge University Press.

Barbacid, M. (1995). Neurotrophic factors and their receptors. Current Opinion in Cell Biology, 7(2), 148–155.

Bedner, P., Steinhäuser, C., Theis, M. (2012). Functional redundancy and compensation among members of gap junction protein families? Biochimica et Biophysica Acta, 1818(8), 1971–84.

Biffo, S., Offenhäuser, N., Carter, B. D., Barde, Y. A. (1995). Selective binding and internalisation by truncated receptors restrict the availability of BDNF during development. Development, 121(8), 2461–70.

Brodal P (1992). The central Nervous System: structure and function. New York, NY. Oxford University Press.

Burnett, M. G., Zager, E. L. (2004). Pathophysiology of peripheral nerve injury: a brief review. Neurosurgical Focus, 16(5), E1.

Causing, C. G., Gloster, A, Aloyz, R., Bamji, S. X., Chang, E., Fawcett, J., Miller, F. D. (1997). Synaptic innervation density is regulated by neuron-derived BDNF. Neuron, 18(2), 257–67.

Chang, Q., Pereda, A., Pinter, M. J., Balice-Gordon, R. J. (2000). Nerve injury induces gap junctional coupling among axotomized adult motor neurons. The Journal of Neuroscience, 20(2), 674–84.

Ciaramitaro, P., Mondelli, M., Logullo, F., Grimaldi, S., Battiston, B., Sard, A., Neuropathies, Italian Network for Traumatic. (2010). Traumatic peripheral nerve injuries: epidemiological

findings, neuropathic pain and quality of life in 158 patients. Journal of Peripheral Nervous System, 15(2), 120-127.

Coulibaly, A. P., Isaacson, L. G. (2012). Transient changes in spinal cord glial cells following transection of preganglionic sympathetic axons. Autonomic Neuroscience, 168(1-2), 32–42.

Coyle, D. E. (1998). Partial peripheral nerve injury leads to activation of astroglia and microglia which parallels the development of allodynic behavior. Glia, 23(1), 75–83.

Dougherty, K. D., Dreyfus, C. F., Black, I. B. (2000). Brain-derived neurotrophic factor in astrocytes, oligodendrocytes, and microglia/macrophages after spinal cord injury. Neurobiology of Disease, 7(6), 574–85.

Gilmore, S. A., Sims, T. J., & Leiting, J. E. (1990). Astrocytic reactions in spinal gray matter following sciatic axotomy. Glia, 3(5), 342-349.

Gomez-Pinilla, F., Ying, Z., & Zhuang, Y. (2012). Brain and spinal cord interaction: protective effects of exercise prior to spinal cord injury. PloS One, 7(2), 1-8.

Guo, W., Wang, H., Watanabe, M., Shimizu, K., Zou, S., LaGraize, S. C., Ren, K. (2007). Glial- cytokine-neuronal interactions underlying the mechanisms of persistent pain. Journal of Neuroscience, 27(22), 6006–18.

Hajós, F., Csillik, B., Knyihár-Csillik, E. (1990). Alterations in glial fibrillary acidic protein immunoreactivity in the upper dorsal horn of the rat spinal cord in the course of transganglionic degenerative atrophy and regenerative proliferation. Neuroscience Letters, 117(1-2), 8–13.

Hosoya, Y., Sugiura, Y., Okado, N., Loewy, A. D., Kohno, K. (1991). Descending input from the hypothalamic paraventricular nucleus to sympathetic preganglionic neurons in the rat. Experimental Brain Research, 85(1), 10–20.

Jones, K. J., Serpe, C. J., Byram, S. C., Deboy, C. A, Sanders, V. M. (2005). Role of the immune system in the maintenance of mouse facial motoneuron viability after nerve injury. Brain, Behavior, and Immunity, 19(1), 12–9.

Kalla, R., Liu, Z., Xu, S., Koppius, a, Imai, Y., Kloss, C. U., Raivich, G. (2001). Microglia and the early phase of immune surveillance in the axotomized facial motor nucleus: impaired microglial activation and lymphocyte recruitment but no effect on neuronal survival or axonal

regeneration in macrophage-colony stimulating factor-defici. The Journal of Comparative Neurology, 436(2), 182–201.

Kessaris, N., Pringle, N., Richardson, W. D. (2008). Specification of CNS glia from neural stem cells in the embryonic neuroepithelium. Philosophical Transactions of the Royal Society of London, 363(1489), 71–85.

Kleopa, K. a, Orthmann, J. L., Enriquez, A., Paul, D. L., Scherer, S. S. (2004). Unique distributions of the gap junction proteins connexin29, connexin32, and connexin47 in oligodendrocytes. Glia, 47(4), 346–57.

Korosi, A., Baram, T. Z. (2010). Plasticity of the stress response early in life: mechanisms and significance. Developmental Psychobiology, 52(7), 661–70.

Lee, I.-H., Lindqvist, E., Kiehn, O., Widenfalk, J., Olson, L. (2005). Glial and neuronal connexin expression patterns in the rat spinal cord during development and following injury. Journal of Comparative Neurology, 489(1), 1–10.

Leong, M. L., Gu, M., Speltz-Paiz, R., Stahura, E. I., Mottey, N., Steer, C. J., Wessendorf, M. (2011). Neuronal loss in the rostral ventromedial medulla in a rat model of neuropathic pain. Journal of Neuroscience, 31(47), 17028–39.

Levy, B. H., Tasker, J. G. (2012). Synaptic regulation of the hypothalamic-pituitary-adrenal axis and its modulation by glucocorticoids and stress. Frontiers in Cellular Neuroscience, 6, 24.

Lu, B., Pang, P. T., Woo, N. H. (2005). The yin and yang of neurotrophin action. Nature Reviews. Neuroscience, 6(8), 603–14.

Luebke, J. I., Wright, L. L. (1992). Characterization of superior cervical ganglion neurons that project to the submandibular glands, the eyes, and the pineal gland in rats. Brain Research, 589(1), 1–14.

Macias, M., Dwornik, A., Skup, M., Czarkowska-Bauch, J. (2005). Confocal visualization of the effect of short-term locomotor exercise on BDNF and TrkB distribution in the lumbar spinal cord of the rat: the enhancement of BDNF in dendrites? Acta Neurobiologiae Experimentalis, 65(2), 177–82.

Makwana, M., Werner, A., Acosta-Saltos, A., Gonitel, R., Pararajasingam, A., Pararajasingham, A., Raivich, G. (2010). Peripheral facial nerve axotomy in mice causes sprouting of motor axons

into perineuronal central white matter: time course and molecular characterization. Journal of Comparative Neurology, 518(5), 699–721.

McAllister, R. M., Gilbert, S. E., Calder, J. S., Smith, P. J. (1996). The epidemiology and management of upper limb peripheral nerve injuries in modern practice. Journal of Hand Surgery, 21(1), 4–13.

McTigue, D. M., Horner, P. J., Stokes, B. T., Gage, F. H. (1998). Neurotrophin-3 and brain- derived neurotrophic factor induce oligodendrocyte proliferation and myelination of regenerating axons in the contused adult rat spinal cord. Journal of Neuroscience, 18(14), 5354–65.

McTigue, D. M., Tripathi, R. B. (2008). The life, death, and replacement of oligodendrocytes in the adult CNS. Journal of Neurochemistry, 107(1), 1–19.

Merchenthaler, I. (1984). Corticotropin releasing factor (CRF)-like immunoreactivity in the rat central nervous system. Extrahypothalamic distribution. Peptides, 5, 53–69.

Miller, R. H. (2002). Regulation of oligodendrocyte development in the vertebrate CNS. Progress in Neurobiology, 67(6), 451–67.

Nishiyama, A., Komitova, M., Suzuki, R., Zhu, X. (2009). Polydendrocytes (NG2 cells): multifunctional cells with lineage plasticity. Nature Reviews. Neuroscience, 10(1), 9–22.

Novikov, L. N., Novikova, L. N., Holmberg, P., Kellerth, J. (2000). Exogenous brain-derived neurotrophic factor regulates the synaptic composition of axonally lesioned and normal adult rat motoneurons. Neuroscience, 100(1), 171–81.

Ohsumi, A., Nawashiro, H., Otani, N., Ooigawa, H., Toyooka, T., Yano, A., Shima, K. (2006). Alteration of gap junction proteins (connexins) following lateral fluid percussion injury in rats. Acta Neurochirurgica. Supplement, 96, 148–50.

Olschowka, J. A., O’Donohue, T. L., Mueller, G. P., Jacobowitz, D. M. (1982). The distribution of corticotropin releasing factor-like immunoreactive neurons in rat brain. Peptides, 3(6), 995– 1015.

Pereda, A. E., Curti, S., Hoge, G., Cachope, R., Flores, C. E., Rash, J. E. (2013). Gap junction- mediated electrical transmission: regulatory mechanisms and plasticity. Biochimica et Biophysica Acta, 1828(1), 134–46.

Pyner, S., Coote, J. H. (1994). A comparison between the adult rat and neonate rat of the architecture of sympathetic preganglionic neurones projecting to the superior cervical ganglion, stellate ganglion and adrenal medulla. Journal of the Autonomic Nervous System, 48(2), 153–66.

Rando, T. A., Bowers, C. W., Zigmond, R. E. (1981). Localization of neurons in the rat spinal cord which project to the superior cervical ganglion. Journal of Comparative Neurology, 196(1), 73–83.

Rohlmann, A., Laskawi, R., Hofer, A., Dermietzel, R., & Wolff, J. R. (1994). Astrocytes as rapid sensors of peripheral axotomy in the facial nucleus of rats. Neuroreport, 5(4), 409-412.

Scholz, J., Broom, D. C., Youn, D.-H., Mills, C. D., Kohno, T., Suter, M. R., Woolf, C. J. (2005). Blocking caspase activity prevents transsynaptic neuronal apoptosis and the loss of inhibition in lamina II of the dorsal horn after peripheral nerve injury. Journal of Neuroscience, 25(32), 7317–23.

Skup, M., Dwornik, A., Macias, M., Sulejczak, D., Wiater, M., Czarkowska-Bauch, J. (2002). Long-Term Locomotor Training Up-Regulates TrkBFL Receptor-like Proteins, Brain-Derived Neurotrophic Factor, and Neurotrophin 4 with Different Topographies of Expression in Oligodendroglia and Neurons in the Spinal Cord. Experimental Neurology, 176(2), 289–307.

Smith, O. A., DeVito, J. L. (1984). Central neural integration for the control of autonomic responses associated with emotion. Annual Review of Neuroscience, 7, 43–65.

Svensson, M., Eriksson, P., Persson, J. K., Molander, C., Arvidsson, J., Aldskogius, H. (1993). The response of central glia to peripheral nerve injury. Brain Research Bulletin, 30(3-4), 499– 506.

Tanabe, A., Onimaru, H., Suzuki, H., Takeyama, Y., Homma, I. (2012). Effects of corticotropin- releasing factor on intermediolateral cell column neurons of newborn rats. Autonomic Neuroscience : Basic & Clinical, 171(1-2), 36–40.

Theodoric, N., Bechberger, J. F., Naus, C. C., Sin, W.-C. (2012). Role of gap junction protein connexin43 in astrogliosis induced by brain injury. PloS One, 7(10), e47311.

Theriault, E., Frankenstein, U. N., Hertzberg, E. L., Nagy, J. I. (1997). Connexin43 and astrocytic gap junctions in the rat spinal cord after acute compression injury. Journal of Comparative Neurology, 382(2), 199–214.

Tucker, D. C., Saper, C. B. (1985). Specificity of spinal projections from hypothalamic and brainstem areas which innervate sympathetic preganglionic neurons. Brain Research, 360(1-2), 159–164.

VonDran, M. W., Singh, H., Honeywell, J. Z., Dreyfus, C. F. (2011). Levels of BDNF impact oligodendrocyte lineage cells following a cuprizone lesion. Journal of Neuroscience, 31(40), 14182–90.

Yamaguchi, N., Okada, S. (2009). Cyclooxygenase-1 and -2 in spinally projecting neurons are involved in CRF-induced sympathetic activation. Autonomic Neuroscience, 151(2), 82–9.

Chapter 2: Transection of preganglionic axons leads to CNS neuronal plasticity followed by survival and target reinnervation

Coulibaly et al., 2013. Autonomic Neuroscience. 179:49-59

ABSTRACT

The goals of the present study were to investigate the changes in sympathetic preganglionic neurons following transection of distal axons in the cervical sympathetic trunk (CST) that innervate the superior cervical ganglion (SCG) and to assess changes in the protein expression of brain derived neurotrophic factor (BDNF) and its receptor TrkB in the thoracic spinal cord. At 1 week, a significant decrease in soma volume and reduced soma expression of choline acetyltransferase (ChAT) in the intermediolateral cell column (IML) of T1 spinal cord were observed, with both ChAT-immunoreactive (-ir) and non-immunoreactive neurons expressing the injury marker activating transcription factor 3 (ATF3). These changes were transient, and at later time points, ChAT expression and soma volume returned to control values and the number of ATF3 neurons declined. No evidence for cell loss or neuronal apoptosis was detected at any time point. Protein levels of BDNF and/or full length TrkB in the T1 level of the spinal cord were increased throughout the survival period. In the SCG, both ChAT-ir axons and ChAT protein remained decreased at 16 weeks, but were increased compared to the 10 week time point. These results suggest that, though IML neurons show reduced ChAT expression and cell volume at 1 week following CST transection, at later time points, the neurons recovered and exhibited no significant signs of neurodegeneration. The alterations in BDNF and/or TrkB may have contributed to the survival of the IML neurons and the recovery of ChAT expression, as well as to the reinnervation of the SCG.

1. Introduction Following peripheral axon injury, it is thought that retrograde signals originating from the injury site can activate the parent cell bodies, possibly leading to survival and regenerative growth (Abe and Cavalli, 2008). In a previous study, a significant loss of IML neurons at spinal cord level T1 was reported at 10 weeks following CST transection (Tang and Brimijoin, 2002). This loss was based on a reduced number of choline acetyltransferase (ChAT) immunoreactive (- ir) neurons in the IML that also contained the retrograde tracer Fast Blue, which was applied to the SCG at the time of injury. Yet, in this same study, a terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay, which was carried out weekly throughout the survival period, failed to reveal any evidence of apoptosis in the IML (Tang and Brimijoin, 2002). Similarly, little evidence of neuronal cell death following transection of peripheral axons has been reported in other models. For example, the loss of ChAT expression, reduced soma size, and increased expression of activating transcription factor 3 (ATF3), a nuclear injury marker expressed by all injured neurons (Tsujino et al., 2000), was reported at the sacral level of the spinal cord following pelvic nerve transection (Peddie and Keast, 2011), but there was no evidence of cell death and these responses to injury subsided over time. In the nucleus ambiguus (Chang et al., 2004) and hypoglossal nucleus (Lams et al., 1988; Armstrong et al., 1991), loss of ChAT expression without significant cell death was observed after axonal transection, suggesting that, though ChAT immunoreactivity was reduced, the neurons were not lost. A slight neuronal loss in the hypoglossal nucleus following axonal transection was reported in a different study (Chang et al., 2004), yet there is general agreement that, following injury, an initial loss and then substantial recovery of ChAT expression occurs over time following axonal transection. In an attempt to understand whether similar changes occur in the IML following CST, we documented changes in ChAT expression as well as in the soma size of IML neurons at short term and long term survival points, and also determined whether the injured neurons in the IML expressed ATF3. Whether CST transection influenced neurotrophin expression in the spinal cord was a second objective of the present study. Because brain derived neurotrophic factor (BDNF) in the SCG is thought to play a role in the regulation of the innervating IML neurons (Causing et al., 1997), which in turn express the high affinity full length TrkB receptor (TrkB.FL; Skup et al., 2002; McCartney et al., 2008; Coulibaly and Isaacson, 2012), and alterations in BDNF and TrkB

17 have been observed in the spinal cord following injury (Dougherty et al., 2000; Morcuende et al., 2011; Gomes et al., 2012), we investigated whether BDNF and/or TrkB in the spinal cord were affected by the distal preganglionic axon transection. In the present study, the protein expression of BDNF, as well as TrkB.FL, which elicits signal transduction in the expressing cells (For review see Barbacid, 1995), and truncated TrkB receptor 1 (TrkB.T1), which lacks the tyrosine kinase domain and is thought to play a role in BDNF sequestration (Biffo et al., 1995; Lu et al., 2005), was compared across treatments. The third objective of this study was to examine the SCG in order to determine whether changes in IML neuronal cell bodies following CST transection were accompanied by alterations in ChAT protein and tyrosine hydroxylase (TH), the rate limiting enzyme in the production of norepinephrine. Because the regeneration of peripheral axons is estimated at ~1-3mm/day (Gordon et al., 2007), the reinnervation of the SCG following transection of the CST would be expected to occur quickly. Following CST transection, Nja and Purves (1977) noted full reinnervation of the guinea pig SCG at 3 months and the return of function in the cat SCG was observed as early as 11.4 days following CST transection (Butson, 1950). Yet, in other models, though ChAT protein expression in the cell bodies of injured neurons returned, the ChAT reinnervation of their peripheral targets remained altered (Wang et al., 1997). Indeed, ChAT activity in the SCG remained reduced at 10 weeks following CST transection (Tang and Brimijoin, 2002) and, at 6 months, though tyrosine hydroxylase (TH) activity appeared to return to control values Raisman et al. (1974) observed little ChAT activity and few synapses in the rat SCG, suggesting that the reinnervation of the rat SCG following CST transection proceeded slowly. Here we used western analysis and immunohistochemistry to monitor the changes in ChAT and TH protein expression in the SCG at short term and long term survival times following CST transection. Portions of this study were published in abstract form (Coulibaly et al., 2009; Coulibaly et al., 2011).

2. Materials and Methods 2.1 Animals Young adult (3 months of age) female Sprague Dawley rats (Harlan Labs, Indianapolis, IN) were housed in the Miami University Animal Facilities in a 12:12 light:dark environment at regulated temperature. The CST was bilaterally transected and animals survived for 1 day (n=8),

1 week (n=8), 3 weeks (n=13), 10 weeks (n=8), or 16 weeks (n=6). Data from these animals were compared with findings from respective age-matched controls that either received sham surgery or underwent no surgical procedures. Because no changes were observed in any of the sham cases when compared to unoperated control, data collected from shams and unoperated control groups were pooled to obtain a control groups that corresponded to 1 day-1 week (n=13), 3 week (n=13), 10-16 week (n=12) survival time points. All methods used in this study were approved by the Miami University Institutional Animal Care and Use Committee and efforts were taken to minimize discomfort and pain to the animals and to minimize the number of animals used in the study. Though previous retrograde labeling studies showed that preganglionic neurons projecting to the SCG were located primarily at levels C8- T2, with additional projections from levels T3 to T5 (Rando et al., 1981; Strack et al., 1988; Hosoya et al., 1991; Poulat et al., 1992), robust glial plasticity following CST transection in a previous study was reported primarily at T1 of the spinal cord (Coulibaly and Isaacson, 2012). Therefore, all analyses shown in this study were carried out at the T1 level of the spinal cord. T1 level of the spinal cord was identified by counting the nerve roots extending from the cord and the T1 level was verified by noting its location immediately caudal to the cervical enlargement.

2.2. Transection of the cervical sympathetic trunk Rats were anesthetized either with Ketamine (80 mg/kg):Rompun (14mg/kg) cocktail or via the inhalant 2.5% isofluorane. A 3 cm ventral incision was made on the neck region of the animal. The CST was exposed and gently separated from surrounding tissue and transected approximately 2 mm from the entry into the SCG (Sun and Zigmond, 1996). The proximal stump was placed carefully back into original position after the cut in close proximity to the distal stump. The procedure was repeated on the other side. The incision was closed using sutures and tissue glue (Nexaband, Phx, AZ). The CST was exposed, but not transected, in sham animals. Success of the surgery was assessed by the extent of ptosis following the surgery and post-surgical examination of the transection sites. The level of ptosis was documented bilaterally in each case and was ranked from 0 (no evidence) to 5 (most severe). Ptosis rankings were obtained immediately following the surgery, and on a daily basis for the first week. For longer

19 survival time points, ptosis was documented weekly until just prior to sacrifice, when a final ptosis ranking was obtained.

2.3. Tissue collection and immunofluorescence studies Animals were transcardially perfused with 0.9% saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB). Spinal cord and SCG tissues were removed and stored in 0.1M PB, cryoprotected by infiltrating with 30% sucrose in 0.1M PB, embedded in optimal cutting temperature medium (Ted Pella Inc), and cut using a MICROM HM 550 series cryostat. Coronal sections of spinal cord (18 m) and longitudinal sections of the SCG (12 µm) were mounted on Superfrost microscope slides, incubated overnight in 0.1M PBS-0.2% Triton-X solution, blocked with normal donkey serum, and then incubated for 48 hrs at 4oC in primary antibody. Spinal cord sections were incubated in a cocktail of goat anti-ChAT (1:200, Millipore) and rabbit anti-ATF3 (1:200; Santa Cruz Biotech), or in rabbit anti-caspase 3 (casp3; 1:200, Cell Signaling) followed by Neurotrace (1:200; Life Technologies), a Nissl stain conjugated to Alexa 500/525. SCG sections were incubated in goat anti-ChAT (1:200). Following a series of rinses, sections were incubated for 2 hours in AlexaFluor conjugated antibodies (1:200; Life Technologies) directed against the primary antibody host. Sections then were coverslipped using fluorescent mounting medium with 4',6-diamidino-2-phenylindole (DAPI; Vectashield). Images were captured with either an Olympus FV500 or a Zeiss 720 laser scanning confocal microscope. For casp3 controls, positive and negative control slides were obtained from Cell Signaling Technology (#8104). The prepared paraffin slides of Jurkat cells that were pre-treated with etoposide (positive controls) or received no treatment (negative controls) were de- paraffinized processed for casp3 immunohistochemistry as described above.

2.4. TUNEL assay Spinal cord sections from each survival time point were tested for apoptotic cell death using a TUNEL assay (Roche). Sections were processed according to protocol instructions, and then processed for ATF3, ChAT, or Iba1 (1:500; Abcam) immunolabeling as described above. To validate the protocol, both negative and positive controls were carried out using uninjured spinal cord tissue. The positive control was exposed to 6M HCl for 4 minutes to initiate DNA

20 fragmentation, while the negative controls were processed without terminal deoxynucleotidyl transferase followed by the normal protocol.

2.5. Analysis of neurons in the IML Cell counts were obtained from confocal images that were acquired at 200X magnification using an Olympus FV500 laser scanning confocal microscope and then enlarged to 500X for viewing on a computer monitor using Olympus Fluoview v4.3 software. For the total number of neurons (ChAT only + ATF3 only + ChAT/ATF3) and for the number of ChAT-ir neurons (ChAT only + ChAT/ATF3) in the IML, the number of cells in each category was recorded from the left and right IML from two different spinal cord sections per animal and then averaged to obtain a value for the spinal cord for each animal. In order to provide information regarding the extent of variability within the control sets, a mean was obtained for each set of controls and the values from each cord in the data set (both controls and treatments) were expressed as the proportion of control. This results in a mean that might be slightly more or less than 100% and allows for a calculated SEM for the control group. This is a more realistic demonstration of the controls rather than merely setting the control group at 100% and comparing treatments to this value. Therefore, a mean was obtained for each set of controls and then the values from each cord in the data set (both controls and treatments) were expressed as the proportion of control and then subjected to statistical analysis. In the first experiment, changes at 1 day and 1 week following transection were analyzed using a one way ANOVA followed by the Fisher post hoc comparison test to compare to one group of age-matched sham controls that survived from 1 day to 1 week following sham surgery. In the second experiment, data obtained from animals at 3 weeks following transection were analyzed using a Student’s test compared to age-matched sham controls, while the third experiment assessed long term changes using a one way ANOVA followed by the Fisher post hoc comparison test to compared data obtained at 10 and 16 weeks following the injury to age-matched sham controls that survived from 12-14 weeks following sham surgery. Because ATF3-ir cells were absent in the control cases, the number of ATF3 cells present at 1 week was compared with the number present at 3 weeks to determine whether any changes occurred between these 2 time points. Significance was reported at p<0.05. All values were reported as mean  SEM.

For the determination of cell volume, data were collected using the method previously described in Flak et al. (2009). In brief, two Z-stacks per spinal cord (one stack from the IML on each side) were obtained using images that were acquired at 0.5 µm intervals (30-40 images) with the 40X oil objective. The volume of at least 4 neurons from each stacked image (and thus at least 8 neurons per cord) was determined by obtaining a mean radius measurement for each neuron and then using the following equation for the volume of the sphere: V = 4/3 r3. The mean radius for each neuron was obtained by measuring the distance from the center of the nucleus to the edge of each cell at 4 different angles: 0, 90, 180 and 270 degrees. Each neuron was analyzed using this approach in 4 different images in the stack to yield 16 radius measurements, which then were averaged for use in the volume equation to obtain a final volume measurement for the neuron. Volumes from individual neurons from the left and right IML were averaged to obtain a value for each spinal cord. Values from each cord in the data set (both controls and treatments) were expressed as the percent control and then subjected to statistical analysis. Because few ChAT-ir neurons were present at 1 week following transection, soma volume was obtained from neuronal profiles in the IML that were stained with Neurotrace (1:200; Molecular Probes) and then compared with the volume of Neurotrace profiles from the age-matched controls. At later time points (3, 10, and 16 weeks) the neuronal volume was measured in ChAT-ir neurons. The values obtained from 1 week and 3 week treatments were compared to respective age-matched sham controls using Student’s t test. Data from 10 and 16 week treatments were compared to one group of age-matched sham controls that survived 12-14 weeks following sham surgery using one way ANOVA. Significance was reported at p<0.05. All values were reported as mean  SEM.

2.6. Semi-quantitative western analysis Protein from hemisections of T1 level of the spinal cord (Macias et al., 2007) and whole SCGs was extracted by sonicating tissue in 0.01M Tris-HCl buffer (pH 7.4) containing 1% sodium dodecyl phosphate (SDS) and 1% protease inhibitor cocktail. Protein concentration of entire SCG or T1 spinal cord hemisection was determined from the supernatant by BCA protein assay and sample preparation was performed in accordance with the Laemmli method (Laemmli, 1970). SDS-PAGE (5% stacking/10% resolving) was used to separate proteins, and a Precision Plus protein standard was used to confirm molecular mass. Following the transfer of 2400mA at

4oC in transfer buffer (25mM Tris, 192 mM glycine, 10% v/v methanol) to PVDF membrane, blots were submerged in methanol, allowed to dry, cut so that the standard was processed separately, and rehydrated in methanol. For the western analysis of TH (SCG), BDNF (cord), and TrkB (cord), the membrane containing protein samples was blocked for three hours in 4% nonfat dry milk in Tris-buffered saline with Tween20 (TBST) and incubated overnight at 4oC in primary antibody (SCG: mouse anti-TH, BD Biosciences, 1:200,000; cord: rabbit anti-BDNF, N- 20, Santa Cruz Biotech, 1:750; rabbit anti-TrkB.FL, 794, Santa Cruz Biotech, 1:500; rabbit anti- TrkB.T1, c13, Santa Cruz Biotech, 1:500) and the membrane containing the standard was incubated in TBST alone. Membranes were rinsed and incubated in appropriate secondary antibody (goat anti-mouse HRP IgG, 1:120,000; goat anti-rabbit HRP IgG, 1:10,000; rabbit anti- goat HRP IgG, 1:10,000; Chemicon) for 2 hours or in streptactin-HRP (for the standard; 1:500,000, Bio-Rad Labs), then rinsed and submerged in Supersignal West Pico Chemiluminescent Substrate for five minutes. Protein was visualized with X-ray film. For western analysis of ChAT (SCG), membranes were incubated in 4% BSA at overnight at 4oC followed by incubation (2.5 hours) with goat anti-ChAT (1:500, Millipore) and then processed as described above. Blots were then stripped at room temperature, blocked, and re-probed for GAPDH (1:100,000; Fitzgerald Industries International) followed by goat anti-mouse HRP IgG (1:100,000; Chemicon). Films were scanned and densitometry measurements were obtained using ImageQuant 5.2. A ratio of the protein of interest to GAPDH was determined for each case. The means for each treatment were expressed as the percent control. These proportional data were then square root transformed for statistical analysis and for presentation in the figures. For the western analysis of BDNF and TrkB in the spinal cord, changes at one week and 3 weeks following transection were compared to respective age-matched sham controls using Student’s t-test. For long term survivals, comparisons of 10 week and 16 week survivals were made to one age-matched control group (sacrificed approximately 12-14 weeks following sham surgery). For the western analysis of ChAT and TH in the SCG, proportional data obtained from comparisons to age-matched controls at each survival time point were compared across treatments using one way ANOVA followed by the Fisher post-hoc comparison test. As shown in Figure 7, significant differences of each injury time point with the control group as well as any

23 differences between the 10 week and the 16 week survival time point were reported. Significance was reported at p<0.05.

2.7. Analysis of ChAT-ir profiles in the SCG ChAT-ir profiles in the SCG were analyzed from images that were acquired at 200X magnification and then enlarged to 500X magnification for viewing on a computer monitor using Image Pro software. The number of fluorescent profiles in 3-4 different regions within each SCG was determined using Image Pro software and averaged to generate the overall axonal density value for each SCG. A mean density was obtained for the control group and then all values (controls and treatments) were expressed as the percent control. The proportional data obtained from the comparisons to age-matched controls at each survival time point were compared across treatments using one way ANOVA followed by the Fisher post-hoc comparison test. Significant differences with the control group as well as any difference between the 10 week and the 16 week survival time point were reported. Significance was reported at p<0.05. All values were reported as mean  SEM.

3. Results 3.1. Analysis of ptosis The severity of ptosis following CST transection was ranked over the survival period from 0 (no ptosis) to 5 (most severe ptosis). Ptosis typically ranged from 4.5-5.0 at 1 day following transection and remained relatively unchanged (rank of ~4.5) at one week following transection. By 3 weeks ptosis generally was moderately severe with a rank of ~3.5-4.0. Ptosis decreased to a rank of 2 by 10 weeks following transection. Little recovery of ptosis was observed between 10 weeks and 16 weeks following transection with the animals typically exhibiting mild ptosis with a rank of approximately 2 at the time of sacrifice.

3.2. Changes in the IML following CST transection ChAT-ir neurons were prevalent in the IML at cord level T1 taken from the control cases (Figs. 1, 2) with a mean of 7.8 +/- 0.8 cells per IML. At 1 day following injury, the number of ChAT-ir neurons was similar to controls and ATF3 neurons were rare, with only two ATF3-ir neurons present on one side of one of the four cases examined (not shown). At 1 week following

24 transection, a significant 67% decrease in the number of ChAT-ir neurons in the IML was observed (compared to controls) and ATF3 immunoreactivity was prevalent in both ChAT-ir and ChAT non-immunoreactive neurons. Of the total number of ATF3 neurons present at this time point, 72% did not co-localize ChAT, while the remaining 28% of the ATF3-ir neurons also exhibited ChAT immunoreactivity. Though ChAT-ir neurons were significantly reduced at the one week survival time point, no significant change was observed when the total number of IML neurons (ChAT only + ChAT/ATF3 + ATF only) was compared with that obtained from age- matched controls (Fig. 2). The changes observed at one week appeared to be restricted to spinal cord level T1, as no ATF3-ir neurons and no significant changes in ChAT-ir neurons were observed in cord levels T2 (control: 7.8 +/- 0.85 vs 1 week transection 6.8 +/- 0.97) or T3 (control 6.0 +/- 0.71 vs 6.8 +/-0.12). Values are provided as mean +/- SEM. No ATF3 immunoreactivity was observed in any glial cell type. At 3 weeks following transection, occasional ATF3-ir neurons were present in the IML at T1 level in 6 of the 9 injured animals, but there was no evidence of ATF3 in the remaining 3 injured animals (Figs.1, 2). When compared with the one week time point, a 79% decrease in ATF3-ir neurons was observed (Fig. 2), suggesting that the injury response was subsiding. Although 57% of the ATF3-ir neurons that were present did not co-localize ChAT, the number of ChAT-ir neurons at this survival time point was not significantly different from age-matched controls. Subsequently, the total number of neurons in the IML (ChAT only + ChAT/ATF3 + ATF only) also was not different from that observed in the controls (Figs.1, 2). Similar to that observed in the 1 week animals, the changes observed at 3 weeks appeared to be restricted to spinal cord level T1, as no ATF3-ir neurons and no significant changes in ChAT-ir neurons were observed in cord levels T2 (control: 7.8 +/- 0.85 vs 3 weeks transection 6.6 +/- 1.02 – ) or T3 (control 6.0 +/- 0.71 vs 7.3 +/-1.33). Values are provided as mean +/- SEM. At 10 and 16 weeks following the injury, no ATF3-ir cells were present in the IML at T1. Analysis of the number of ChAT-ir neurons revealed no significant differences compared to age- matched controls (Figs.1, 2). The neurons in the IML at one week following injury appeared to be smaller in size compared to the controls, similar to that reported in a similar injury model (Peddie and Keast, 2011). In order to assess alterations in cell volume in the IML due to injury and to determine whether any size alterations were sustained at longer survival time points, the cell volume of

IML neurons was quantified and compared to age-matched controls. The typical cell volume for preganglionic neurons in the each of the control groups averaged approximately 1,100 µm3. At 1 week following injury the volume of IML neurons was significantly decreased by 50% (Fig. 3). The volume of ChAT-ir neurons in the IML at later survival time points was similar to that of their age-matched controls (Fig. 3). Because previous reports have indicated the possibility of cell death in the IML following CST transection (Tang and Brimijoin, 2002), a TUNEL assay was carried out on tissues taken from animals surviving 1 week, 3 week, and 10 week and 16 weeks following injury. TUNEL positive neurons were observed in the positive control (HCl treated) tissues (Fig. 4). However, in the treatment cases, TUNEL positive neurons were essentially absent. At 1 week post injury, occasional small profiles showed positive staining for TUNEL and these typically co-localized Iba1 (Fig. 4C insets), a marker for microglia in the CNS (Imai et al., 1996). No TUNEL positive neuronal profiles were observed. At 3 weeks following transection, no TUNEL positive staining was observed in the IML (Fig. 4). At the 10 week survival time point, one neuron, which colocalized ChAT (not shown), was positive for TUNEL in the five cases studied and of the 25 neurons that were examined at this time point. No evidence for cell death was observed at the 16 week time point (Fig. 4) and no TUNEL staining was observed in any ATF3-ir neurons (not shown). The localization of casp3, an early and definite sign of apoptotic cell death (Deshmukh and Johnson, 1997; Radovic et al., 2008), was assessed in the IML to confirm the lack of cell death in the injured cases. The general neuronal cell body marker, Neurotrace, a fluorescent Nissl stain, was used to identify the IML neurons. Qualitative analysis showed that Neurotrace labeled most of the neurons in the IML as well as a few glial cells. Casp3-ir cells were observed in the positive controls (Fig. 5A), yet no casp3-ir neurons were found in IML neurons at any time points following injury (Fig. 5), supporting the idea that no neuronal death occurred in the IML following distal axonal transection. At the 10 week and 16 week time points, small casp3-ir cells, which did not co-localize Neurotrace, were present in both the controls and the treatment cases (Figs. 5C,G,H) and were likely glial in nature.

3.3. Changes in spinal cord BDNF and/or TrkB following CST transection

The effects of CST transection on BDNF in the spinal cord were evaluated using western analysis. Though unchanged at 1 week, the mature BDNF isoform (mBDNF; 15 kDa) was generally increased in the spinal cord over the survival period. Mature BDNF was significantly increased by 34% in the spinal cord at 3 weeks following injury, and though the increase in mBDNF at 10 weeks did not reach significance, a significant 80% increase in mBDNF was observed at the 16 week survival time point. The proBDNF isoform (32-37 kDa) was unchanged at 1 week following injury and, following an increase at the 3 week survival time point, returned to control values at 10 weeks. However, by 16 weeks, at a time point when mBDNF reached its highest levels, proBDNF protein was significantly decreased by 59% compared with age- matched controls (Fig. 6). The changes in BDNF observed in the spinal cord following CST transection were accompanied by changes in the TrkB receptor. TrkB.FL was significantly increased by 40% at 1 week following injury (Fig. 6), at a time when ATF3-ir expression in the IML was prevalent, the number of ChAT-ir neurons was significantly decreased, and the glial cells were shown in a previous study to be increased (Coulibaly and Isaacson, 2012). At the same survival time point, a significant 64% decrease was observed in protein levels of TrkB.T1. No changes in TrkB.FL or TrkB.T1 were observed at 3 weeks. While TrkB.FL protein was unchanged at 10 weeks and 16 weeks following injury, a significant decrease in TrkB.T1 was observed at 10 weeks, though by 16 weeks, the levels of TrkB.T1 protein expression rebounded and were similar to control values (Fig. 6).

3.4. Changes in the SCG following CST transection ChAT-ir profiles in the SCG from control cases were obvious with dense immunopositive fibers coursing through the ganglion and targeting neuronal cell bodies (Fig. 7). Following CST transection, ChAT-ir fibers in the SCG were significantly decreased at every time point examined. As expected, a depletion of ChAT-ir fibers was observed at 1 day and 1 week following CST transection. At later time points, a slight recovery in the number of ChAT-ir fibers was observed in the SCG (Fig. 7), and by 16 weeks, though still decreased compared with the controls, the density of ChAT-ir fibers was significantly increased compared to the 10 week time point (Fig. 7). Semi-quantitative western blot analysis of ChAT protein in the SCG showed similar trends, with little ChAT detected in the SCG at 1 day, 1 week, 3 weeks and 10 weeks

27 following injury. However, though significantly reduced compared to controls, ChAT protein in the SCG at 16 weeks was significantly increased compared to the 10 week time point, indicating that some recovery may be taking place. TH western analysis in the SCG revealed a significant decrease in TH in the SCG at early time points, but by 16 weeks, the levels of TH protein were similar to controls (Fig. 7).

Figure 1

Figure 1. Transient changes in ChAT and ATF3 following CST transection. All images depicted are coronal sections of T1 spinal cord. A.-C. ChAT-ir neuronal cell bodies (green, yellow arrows) were observed in the control IML, shown here at cord level T1. No ATF3 profiles were present (red). D.-F. At one week following injury, ChAT immunoreactivity was reduced and few ChAT-ir neurons (green, yellow arrows) were present. White arrows mark the location of IML neurons expressing the nuclear injury factor ATF3 (red), most of which lacked ChAT immunoreactivity. One neuron (asterisk) showed immunoreactivity for both ChAT and ATF3. G.-I. ChAT-ir neurons (green, yellow arrows) appeared similar to controls by 3 weeks following the injury. Few neurons expressed ATF3 immunoreactivity (red, white arrows). The ChAT-ir neurons in the IML at 10 weeks (J.-L.) and 16 weeks (M.-O.) appeared similar to controls. In the right column, merged images (ChAT + ATF3) were combined with DAPI to indicate the presence of cellular profiles. All images, scale bar = 50 m.

ChAT neurons A. 1.2

0.8

0.6 *

% control % 0.4

0.2

0 cont 1d 7d cont 3 wk cont 10 wk 16 wk

Number of ATF-3 neurons in the IML B. 7 * * 6 5 4

3 number of of cells number 2 1 0 cont 1d 7d cont 3 wk cont 10 wk 16 wk C. 1.2 Total number of neurons in the IML

0.8

0.6

control % 0.4

0.2

0 cont 1d 7d cont 3 wk cont 10 wk 16 wk

Figure 2

Figure 2. The total number of preganglionic neurons remained unchanged following CST transection. A. The number of ChAT-ir neurons was significantly decreased at 1 week following CST transection, but similar to age-matched sham controls at all other time points. B. Though ChAT-ir neurons were decreased, the preganglionic neurons were still present and expressed the nuclear injury marker ATF3. ATF3-ir neurons were present at 1 day, 1 week, and 3 weeks following the injury, and were most abundant at 1 week, showing a significant increase compared with the 1 day time point. ATF3 was diminished at 3 weeks (compared to 1 week). No ATF3-ir profiles were observed in controls or at 10 or 16 weeks following injury. C. Analysis of the total number neurons (ChAT only + ATF3 only + ChAT/ATF3) revealed no significant differences across the treatments following CST transection. Error bars represent the standard error of mean. n=3-7 cases per group. *, significantly different from controls at p < 0.05; np, not present.

1.6 Volume of ChAT-ir neurons

1.4

1.2

1 0.8

0.6 *

Proportion of control of Proportion 0.4

0.2

0 cont 7 d cont 3 wk cont 10 wk 16 wk

Figure 3

Figure 3. Reduced soma volume observed at one week was reversed at the long term survival time points. A significant decrease in the volume of ChAT-ir neurons was observed at 1 week following the injury. However volumes were similar to controls at later time points. Error bars represent the standard error of mean. n=3-6 cases per group. *, significantly different from controls at p < 0.05.

L + cont L 1 wk 3 wk 10 wk 16 wk

E E

U U

T T

A B C TUNEL D TUNEL E TUNEL F TUNEL

I P

P + cont 1 wk 3 wk 10 wk 16 wk

/ /

U U

T T TUNEL TUNEL TUNEL TUNEL A’ B’ C’ DAPI D’ DAPI E’ DAPI F’ DAPI

Figure 4

Figure 4. No TUNEL positive neurons were observed in the IML following CST transection. A. TUNEL staining in the positive control (HCL treated) showed numerous positive nuclear profiles. DAPI in A’ shows the cellular nature of the positive staining. B.-B’. No TUNEL staining was observed in the negative control tissues. C.-C’. At one week following injury, small TUNEL/DAPI profiles were frequently observed and were found to co-localize Iba1 (inset), indicating their microglial phenotype. D.-F. No TUNEL positive staining was observed at longer survival time points. n=3-5 cases per group. A.-F.’, scale bar = 50 m. Inset, scale bar = 10 m.

+ cont + cont - cont - cont 16 wk cont

casp3 casp3 A casp3 A’ DAPI B casp3 B’ DAPI C casp3 cont 1 wk 3 wk 10 wk 16 wk

D casp3 E casp3 F casp3 G casp3 H casp3 cont 1 wk 3 wk 10 wk 16 wk

casp3 casp3 casp3 casp3 casp3 Neuro Neuro Neuro Neuro Neuro D’ DAPI E’ DAPI F’ DAPI G’ DAPI H’ DAPI

Figure 5

Figure 5. No caspase 3 (casp3)-ir neurons were observed in the IML following CST transection. A. Casp3-ir staining in the positive controls showed numerous immunopositive profiles. DAPI in A’ shows the cellular nature of the positive profiles. B.-B’. No casp-ir profiles were observed in the negative control tissues. C. Small casp3 immunoreactive profiles (inset) were observed in the 16 week sham control cases. These profiles, which were present in both controls and the 16 week injury animals (see H.), did not co-localize Neurotrace (Neuro) (see H’. which shows lack of Neurotrace staining with casp3 profiles at 16 week survival time point in H.) and were likely glial in nature. D.-H. No evidence of casp3-ir neurons was observed at any time points examined. Neurons were counterstained using Neurotrace (Neuro). n=3-5 cases per control group and treatment. A.-H.’, scale bar = 50 m. Inset, scale bar = 10 m.

A. cont 3 wk cont 16 wk mBDNF c 14 kDa GAPDH 36 kDa cont 3 wk cont 16 wk proBDNF 32 kDa

2.5 GAPDH 36 kDa

2 * mBDNF proBDNF 1.5 * * 1

Proportion of of Proportioncontrol 0.5 *

0 cont 1 wk 3 wk 10 wk 16 wk

B. cont 1 wk cont 16 wk TrkB.FL 145 kDa GAPDH 36 kDa cont 1 wk cont 16 wk TrkB.T1 90 kDa GAPDH 36 kDa 2

TrkB.FL 1.5 TrkB.T1 *

1 *

Proportion of control 0.5 *

0 cont 1 wk 3 wk 10 wk 16 wk

Figure 6

Figure 6. Increased mature BDNF (mBDNF) and full length TrkB (TrkB.FL) in the spinal cord following CST transection. Representative bands from time points in which significance was observed are shown along with respective age-matched control taken from the same membrane. A. Western blots analysis of mBDNF revealed no changes at one week following injury, yet mBDNF levels generally were increased over longer survival time points. Protein expression of proBDNF was increased at 3 weeks, but the levels decreased over time and were significantly decreased at 16 weeks. B. At one week following injury, increased TrkB.FL was paralleled by a decrease in TrkB.T1. TrkB.T1 was decreased at 10 weeks following injury but was similar to controls by 16 weeks. n=3-5 cases per group were analyzed; Error bars represent the standard error of mean. *, significantly different from controls at p < 0.05.

Figure 7

Figure 7. ChAT and TH in the SCG following loss of afferent input. A. ChAT-ir fibers can be observed coursing through the SCG from controls (cont). At 1 day (1d) and 1 week (1 wk) following CST transection only a few ChAT-ir fibers were present. Frequent ChAT-ir fibers could be seen in the SCG at later time points, yet the numbers appeared much less than controls. Scale bar = 50 m. B. Quantitative analysis revealed that the number of ChAT-ir fibers was significantly decreased at each time point examined. However ChAT-ir fibers were increased at 16 weeks compared with the 10 week time point, indicating a potential recovery taking place. C.-D. Representative bands obtained from two independent samples following western blot analysis for ChAT (C.) or TH (D.). n=3-5 cases per group. Error bars represent the standard error of mean. *, significantly different from controls at p < 0.05; #, significantly different from 10 week survival time point.

4. Discussion

The effects of distal CST transection on preganglionic neurons in the IML were documented at short and long term survival time points. The most obvious changes in preganglionic neurons in the IML occurred at one week post injury, as demonstrated by decreased cell soma volume, decreased number of ChAT-ir neurons, and increased number of neurons that expressed the injury marker ATF3, the majority of which did not co-localize ChAT. When total number of ChAT-ir and ATF3-ir neurons was considered, no significant cell loss was observed. These changes were transient, and at later time points, ChAT expression and soma volume returned to control values and the number of ATF3 neurons declined. Because no evidence for neuronal loss or apoptosis was detected at any survival time point examined in this study, we conclude that no significant neuronal death occurred in the IML at short term or long term time points following CST transection. The morphological changes at early survival time points were accompanied by alterations in BDNF and TrkB protein in the spinal cord and, though no obvious morphological changes were observed at long term time survivals, BDNF and/or TrkB were altered at the longer survival times. The CST transection initially resulted in complete denervation of the SCG, and the SCG was not yet fully innervated by ChAT-ir fibers at the 16 week survival time point. Though significantly decreased at early time points, TH protein levels in the SCG were similar to controls at 16 weeks following the CST transection.

4.1. Cell survival following peripheral injury There was little evidence for cell death in the IML following CST transection at the time points examined in this study. These findings are similar to that observed in other models. For example, though the number of ChAT-ir neurons was reduced in the hypoglossal and vagal nuclei at 4 weeks following axotomy, there was no loss of neurons when evaluated using Nissl stain (Lams et al., 1988). Results in a subsequent study of the hypoglossal nucleus following peripheral axon transection paralleled these findings (Armstrong et al., 1991), although a slight reduction in ChAT-ir neurons in the hypoglossal was reported by Chang et al. (2004). Loss of ChAT immunoreactivity without cell death also was observed in the nucleus ambiguus after axonal transection (Chang et al., 2004). Similarly, the loss of ChAT expression without cell death was reported at the sacral level of the spinal cord following pelvic nerve transection (Peddie and Keast, 2011). In that study, a reduction in cell soma size, a reduction in ChAT

43 expression, and an increase in ATF3 were observed at one week. Yet by 4 weeks following the lesion, the neurons appeared similar to controls and no neuronal loss was observed. We also observed the most robust changes at the 1 week survival time point, with the return of ChAT expression and cell volume to control values occurring by 3 weeks following the injury. Yet, unlike our results in the IML and previous findings in other models, Tang and Brimijoin (2002) reported a significant loss of sympathetic preganglionic neurons in the IML at 10 weeks following transection of the CST. In their study, the fluorescent retrograde label True Blue was injected into the SCG at the time of injury and analysis of cell loss in the IML involved counting retrogradely labeled cellular profiles. In addition, when the number of ChAT-ir neurons in the IML was considered, a significant decrease at the T1 level of the cord was observed at 10 weeks with no significant loss was reported at T2 or T3. Though we observed a trend for a decrease in ChAT-ir neurons in the IML in T1 at 10 weeks following CST transection, the decrease did not reach significance, and by 16 weeks following the injury, the number of neurons in the IML was similar to controls. Thus, it appears that most neurons survived the injury, and by 16 weeks following the injury were able to begin restoring normal preganglionic innervation of the SCG. Similar to our results, in which a TUNEL assay yielded little evidence for neuronal cell death in the IML following CST transection, a TUNEL assay carried out by Tang and Brimijoin (2002) yielded no apoptotic neurons over the 10 week survival period. In addition, though we observed small casp 3-ir profiles that were not positive for Neurotrace at 10 weeks and 16 weeks in both controls and treatments, no neuronal profiles were detected. Overall, these findings lead to the conclusion that no significant loss of preganglionic neurons occurred in the IML over the 16 week survival following CST transection.

4.2. Role of ATF3 following injury Following injury, the neuronal cell body undergoes a variety of changes known as neuronal reaction and chromatolysis (Navarro et al., 2007). Some of the changes include synthesis of new proteins, such as the transcription factor ATF3, a protein that has been proposed as a marker for injured neurons (Tsujino et al., 2000; Kataoka et al., 2007). It is thought that ATF3, a member of the CREB/ATF family, and an immediate early gene (Tarn et al., 1999; Hai and Herman, 2001), is activated by retrograde signals from the injury site (Abe and Cavalli, 2008). As with many of the proteins involved in gene transcription following injury (Abe and

Cavalli, 2008), the role played by ATF3 in the survival of the injured neuron remains unclear. However there is evidence suggesting that ATF3 expression in neurons following injury has a neuroprotective effect, and that a relationship exists between ATF3 expression and axonal regeneration and neuroprotection (For review, see Hunt et al., 2012). Therefore, we hypothesize that ATF3 did not play a detrimental role in the IML following transection of the CST, but may have served a neuroprotective role to promote neuronal survival and regeneration. Similar to other models, the expression of ATF3 was transient in nature with a significant reduction in the number of ATF3-ir neurons in the IML at 3 weeks following the CST transection. Peddie and Keast (2011) observed that ATF3 protein peaked in the sacral spinal cord at 1 week following pelvic nerve injury, and then decreased by 4 weeks following the injury. Tsujino and colleagues (2002) found that ATF3 gene expression was relatively weak in the spinal cord at 1 day following sciatic nerve transection, peaked from 3-7 days, and gradually decreased so that it was diminished at 3 weeks, and then “faded away” by 6 weeks. It has been proposed that ATF-3 expression decreases once the regenerative process is underway (Seijffers et al., 2007), and that loss of ATF3 expression results from reconnection with the target. However, in our model, a decrease in ATF3 expression was observed at 3 weeks post-injury, yet ChAT reinnervation in the SCG was not complete even at 16 weeks post-injury, suggesting that the initiation of the regenerative process rather than the reinnervation of the target can lead to a decrease in ATF-3 expression. The slow reinnervation of the SCG was reflected by the ptosis, which was evident at 16 weeks following injury. The hypothesis that ATF-3 expression decreases following initiation of regeneration rather than target reinnervation is supported by other models (Matsuura et al., 1997; Wang et al., 1997).

4.3. Regulatory influences on spinal cord neurons: effects of injury In the present study, BDNF protein in the spinal cord was increased following CST transection and may have contributed to the neuronal response that took place following injury. Neurotrophin signaling appears to play an important role in the neuronal response to injury (Sofroniew et al., 1993; Yan et al., 1994; Yin et al., 2001; Spalding et al., 2005; Sahenk et al., 2008) by affecting cell survival (Hammond et al., 1999; Ichim et al., 2012), axonal regeneration (Oudega and Hagg, 1999; Yin et al., 2001) and target reinnervation (Michalski et al., 2008) and it is known that BDNF plays an important role in the maintenance of motoneurons in the adult

45 spinal cord (Koliatsos et al., 1993) and in the survival of motor (Koliatsos et al., 1993; Boyd and Gordon, 2001; Garraway et al., 2011) and sensory neurons (Song et al., 2008) following spinal cord injury. Therefore we postulate that BDNF and/or TrkB served a neuroprotective role to prevent neuronal death in the IML following the injury. Sympathetic preganglionic neurons in the IML express TrkB.FL (Skup et al., 2002; McCartney et al., 2008; Coulibaly and Isaacson, 2012) indicating their responsiveness to BDNF. BDNF mRNA is expressed in the SCG (Causing et al., 1997), and thus BDNF protein originating from the SCG may regulate sympathetic preganglionic neurons in the IML (Causing et al., 1997). The transection of the CST would interrupt any retrograde signals derived from the SCG and therefore affect BDNF protein levels in the spinal cord. Yet BDNF levels were increased in the spinal cord, even though reconnection with the SCG was incomplete. Other models have demonstrated that BDNF production is induced in glial cells, including microglia, astrocytes and CD4 cells in the central nervous system following peripheral injury (Dougherty et al., 2000; Jones et al., 2005). It is therefore possible that the activated glial cells which were documented previously in this injury model (Coulibaly and Isaacson, 2012) increased the extracellular availability of BDNF, thereby supplying trophic support to the injured IML neurons. Changes in Trk receptor expression can influence neurotrophin activities. In the course of this study, changes were observed in both full length and truncated BDNF receptors. An increase in TrkB.FL occurred at one week following injury, at a time many other responses in the IML were taking place. Others have shown that TrkB expression is altered following injury (Morcuende et al., 2011; Gomes et al., 2012), suggesting that it is important for the neuronal response following injury, possibly to facilitate regeneration (Hollis et al., 2009). A significant 64% decrease was observed in TrkB.T1 at the one week survival time point. Though TrkB.FL and TrkB.T1 isoforms, which were abundant in the brain and spinal cord, show 100% homology in their extracellular domains (Barbacid, 1995), TrkB.T1 possesses an intracellular domain that lacks tyrosine kinase activity. The increase in TrkB.FL may have contributed to neuronal survival (Koliatsos et al., 1993; McTigue et al., 1998), while decreased TrkB.T1 may reflect a reduced role in ligand sequestration (Biffo et al., 1995) to enhance BDNF availability for neuroprotective purposes.

The alterations in BDNF and TrkB may have played a role in restoring ChAT expression in the IML following CST transection. There is evidence that BDNF availability regulates the expression of ChAT (Sofroniew et al., 1993; Garcia et al., 2010).

4.4. Reinnervation of the SCG following CST transection The reinnervation of the SCG following CST transection can provide one measure of the regenerative potential of the IML neurons. Functional studies showed recovery (i.e. pupil size) in both the rabbit and cat by two weeks after transection of the CST (Langley, 1895; Butson, 1950). In addition, following CST transection, Nja and Purves (1977) noted full reinnervation of the guinea pig SCG at 3 months and return of function of the cat SCG was observed as early as 11.4 days (Butson, 1950). Therefore, we expected full reinnervation of the SCG at the 3 week post injury time point. Indeed, based on the proposed regenerative growth rate of 1-3 mm/day (Olson, 1969; Kirpekar et al., 1970; Gordon et al., 2007) and that the distance from the CST stump to the SCG is only a few millimeters, we were surprised to discover that, even at 4 months following transection, the number of ChAT-ir fibers and levels of ChAT protein in the SCG remained significantly reduced. It is curious that ChAT reinnervation remained significantly decreased at the 10 and 16 week time points, even when ChAT in the IML appeared similar to controls and no ATF3 neurons were present, yet this lack of reinnervation may explain the presence of moderate ptosis even at long term survival time points. Tang and Brimijoin (2002) also observed reduced ChAT activity in the SCG at 10 weeks following CST transection, and Raisman et al. (1974) observed little ChAT activity and few synapses in the rat SCG at long survival time points following CST transection. In fact, following compression injury of the CST revealed that connections within the SCG were still pathological after one year (Serebryakova, 2008). Others have shown in similar models that little target reinnervation took place even though cell soma ChAT re- expression tended to occur (Wang et al., 1997). Though ChAT protein levels remained depressed, TH protein appeared similar to controls at the 16 week time point. This is consistent with Raisman et al. (1974), who reported little ChAT activity and few synapses in the rat SCG at 6 months following CST injury while TH activity appeared to return to control values. We conclude that SCG reinnervation following CST transection was a slow process, and that, though overt changes in IML neurons had subsided, the IML neurons remained focused on recovery for

47 long periods following the injury. Yet the relatively slow reinnervation of the SCG, however, did not affect the recovery of TH protein expression by the SCG neurons.

5. Conclusions The neuronal cell bodies of transected axons in the IML showed transient changes in ChAT and ATF3 expression as well as soma volume that were paralleled by significant changes in BDNF and TrkB expression in the spinal cord. The most robust neuronal changes were observed at one week following the injury. TrkB.FL protein was increased at one week, suggesting that BNDF/TrkB signaling may play a role in both the neuronal and glial responses to distal axon injury. Though the morphological changes tended to subside at longer survival time points, BDNF remained increased throughout the survival period as the reinnervation of the SCG proceeded. We conclude that BDNF and/or TrkB may have played a role in the recovery of ChAT expression and neuroprotection in the IML at short term time points, and may have served to facilitate the reinnervation of the SCG, which proceeded slowly over the 16 week survival period.

6. References

Abe, N., Cavalli, V., 2008. Nerve injury signaling. Current Opinion in Neurobiology, 18(3), 276- 83.

Armstrong, D.M., Brady, R., Hersh, L.B., Hayes, R.C., Wiley, R.G., 1991. Expression of choline acetyltransferase and nerve growth factor receptor within hypoglossal motoneurons following nerve injury. Journal of Comparative Neurology, 304(4), 596-607.

Barbacid, M. 1995. Neurotrophic factors and their receptors. Current Opinion in Cell Biology, 7,148-155.

Biffo, S., Offenhäuser, N., Carter, B.D., Barde, Y., 1995. Selective binding and internalisation by truncated receptors restrict the availability of BDNF during development. Development, 121(8), 2461-70.

Blinzinger, K., Kreutzberg, G., 1968. Displacement of synaptic terminals from regenerating motoneurons by microglial cells. Zeitschrift fur Zellforschung und Mikroskopische Anatomie, 85, 145-157.

Boyd, J.G., Gordon, T., 2001. The neurotrophin receptors, trkB and p75, differentially regulate motor axonal regeneration. Journal of Neurobiology, 49, 314-325.

Butson, A.R.C., 1950. Regeneration of the cervical sympathetic. British Journal of Surgery, 38 (150), 223-39.

Causing, C.G., Gloster, A., Aloyz, R., Bamji, S.X., Chang, E., Fawcett, J., Kuchel, G., Miller, F.D., 1997. Synaptic innervation density is regulated by neuron-derived BDNF. Neuron, 18(2), 257-67.

Chang, H.-M., Wei, I.-H., Tseng, C.-Y., Lue, J.-H., Wen, C.-Y., Shieh, J.-Y., 2004. Differential expression of calcitonin gene-related peptide (CGRP) and choline acetyltransferase (ChAT) in the axotomized motoneurons of normoxic and hypoxic rats. Journal of Chemical Neuroanatomy, 28(4), 239-51.

Coulibaly, A.P., Isaacson, L.G., 2012. Transient changes in spinal cord glial cells following transection of preganglionic sympathetic axons. Autonomic Neuroscience, 168(1-2), 32-42.

Coulibaly, A.P., Morris, T.M., Walker, R.G., Isaacson., L.G., 2009. Changes in sympathetic preganglionic neurons and associated glial cells following injury to peripheral axons. Program

No. 442.6. 2009 Neuroscience Meeting Planner. Chicago, IL: Society for Neuroscience, 2009. Online.

Coulibaly, A.P., Walsh, B.F., Isaacson, L.G., 2011. Transient changes in BDNF parallel plasticity observed in the IML following distal axon injury. International Society for Autonomic Neuroscience, Buzios, Brazil. September, 2011.

Coyle, D.E., 1998. Partial peripheral nerve injury leads to activation of astroglia and microglia which parallels the development of allodynic behavior. Glia 23, 75-83.

Deshmukh, M., Johnson, E.M., 1997. Programmed cell death in neurons: focus on the pathway of nerve growth factor deprivation-induced death of sympathetic neurons. Molecular Pharmacology, 51(6), 897-906.

Dorsey, S.G., Bambrick, L.L., Balice-Gordon, R.J., Krueger, B.K., 2002. Failure of brain- derived neurotrophic factor-dependent neuron survival in mouse trisomy 16. Journal of Neuroscience, 22(7), 2571-8.

Dorsey, S.G., Lovering, R.M., Renn, C.L., Leitch, C.C., Liu, X., Tallon, L.J., Sadzewicz, L.D., Pratap, A., Ott, S., Sengamalay, N., Jones, K.M., Barrick, C., Fulgenzi, G., Becker, J., Voelker, K., Talmadge, R., Harvey, B.K., Wyatt, R.M., Vernon-Pitts, E., Zhang, C., Shokat, K., Fraser- Liggett, C., Balice-Gordon, R.J., Tessarollo, L., Ward, C.W., 2012. Genetic deletion of trkB.T1 increases neuromuscular function. American Journal of Physiology, 302, C141-C153.

Dougherty, K.D., Dreyfus, C.F., Black, I.B., 2000. Brain-derived neurotrophic factor in astrocytes, oligodendrocytes, and microglia/macrophages after spinal cord injury. Neurobiology of Disease, 7(6), 574-85.

Flak, J.N., Ostrander, M.M., Tasker, J. G., Herman, J.P., 2009. Chronic stress-induced neurotransmitter plasticity in the PVN. Journal of Comparative Neurology, 517(2), 156-65.

Garcia, N., Tomas, M., Santafe, M.M., Besalduch, N., Lanuza, M. A., Tomas, J., 2010. The interaction between tropomyosin-related kinase receptors B and presynaptic muscarinic receptors modulates transmitter release in adult rodent motor nerve terminals. Journal of Neuroscience, 30(49), 16514-16522.

Garraway, S. M., Turtle, J. D., Huie, J. R., Lee, K. H., Hook, M. a, Woller, S. a, & Grau, J. W. 2011). Intermittent noxious stimulation following spinal cord contusion injury impairs locomotor

50 recovery and reduces spinal brain-derived neurotrophic factor-tropomyosin-receptor kinase signaling in adult rats. Neuroscience, 199, 86-102.

Gomez-Pinilla, F., Ying, Z., Zhuang, Y., 2012. Brain and spinal cord interaction: protective effects of exercise prior to spinal cord injury. PloS One, 7(2), e32298.

Gordon, T., Brushart, T.M., Amirjani, N., Chan, K.M., 2007. The potential of electrical stimulation to promote functional recovery after peripheral nerve injury-comparisons between rats and humans. Acta Neurologica, Supplement 100, 3-11.

Hai, T., Hartman, M. G., 2001. The molecular biology and nomenclature of the activating transcription factor/cAMP responsive element binding family of transcription factors: activating transcription factor proteins and homeostasis. Gene, 273(1), 1-11.

Hammond, E.N., Tetzlaff, W., Mestres, P., Giehl, K.M., 1999. BDNF, but not NT-3, promotes long-term survival of axotomized adult rat corticospinal neurons in vivo. Neuroreport, 10(12), 2671-2675.

Hollis, E.R., Jamshidia, P., Lowa, K., Blesch, A., Tuszynski, M.H., 2009. Induction of corticospinal regeneration by lentiviral trkB-induced Erk activation. PNAS, 106, 7215-20.

Hosoya, Y., Sugiura, Y., Okado, N., Loewy, A.D., Kohno, K., 1991. Descending input from the hypothalamic paraventricular nucleus to sympathetic preganglionic neurons in the rat. Experimental Brain Research, 85, 10-20.

Hu, P., McLachlan, E.M. 2004. Inflammation in sympathetic ganglia proximal to sciatic nerve transection in rats. Neuroscience Letters, 365(1), 39-42.

Hunt, D., Raivich, G., Anderson, P.N., 2012. Activating transcription factor 3 and the nervous system. Frontiers in Molecular Neuroscience, 5, 1-17.

Ichim, G., Tauszig-Delamasure, S., Mehlen, P., 2012. Neurotrophins and cell death. Experimental Cell Research, 318 (11), 1221-1228.

Imai, Y., Ibata, I., Ito, D., Ohsawa, K., Kohsaka, S., 1996. A Novel gene iba1 in the major histocompatibility complex class III region encoding an EF hand protein expressed in a monocytic lineage. Biochemical Biophysical Research Communications, 224, 855-862.

Jones, K.J., Serpe, C. J., Byram, S.C., Deboy, C.,Sanders, V.M., 2005. Role of the immune system in the maintenance of mouse facial motoneuron viability after nerve injury. Brain Behavior and Immununology, 19(1), 12-19.

Kalla, R., Liu, Z., Xu, S., Koppius, A., Imai, Y., Kloss, C.U., Kohsaha, S., Gschwendtner, A., Moller, J.C., Werner, A., Raivich, G., 2001. Microglia and the early phase of immune surveillance in the axotomized facial motor nucleus: impaired microglial activation and lymphocyte recruitment but no effect on neuronal survival or axonal regeneration in macrophage-colony stimulating factor-deficient mice. Journal of Comparative Neurology, 436, 182-201.

Kataoka, K., Kanje, M., Dahlin, L. B., 2007. Induction of activating transcription factor 3 after different sciatic nerve injuries in adult rats. Scandinavian Journal of Plastic and Reconstructive Surgery and Hand, 41(4), 158-66.

Kirpekar, S.M., Wakade, A.R., Prat, J.C., 1970. Regeneration of sympathetic nerves to the vas deferens and spleen of the cat. Journal of Pharmacology and Experimental Therapies, 175, 197- 205.

Koliatsos, V.E., Clatterbuck, R.E., Winslow, J.W., Cayouette, M.H., and Price, D.L., 1993. Evidence that brain-derived neurotrophic factor is a trophic factor for motor neurons in vivo. Neuron, 10(3), 359-67.

Laemmli, U.K., 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, 227(5259), 680-5.

Langley, J.N., 1895. Note on regeneration of preganglionic fibres of the sympathetic. Journal of Physiology, 18(3), 280-4.

Lu, B., Pang, P.T., Woo, N.H., 2005. The yin and yang of neurotrophin action. Nature Reviews. Neuroscience, 6(8), 603-14.

Macias, M., Dwornik, A., Ziemlinska, E., Fehr, S., Schachner, M., Czarkowska-Bauch, J., Skup, M., 2007. Locomotor exercise alters expression of pro-brain-derived neurotrophic factor, brain- derived neurotrophic factor and its receptor TrkB in the spinal cord of adult rats. European Journal of Neuroscience, 25, 2425-2444.

Michalski, B., Bain, J.R., Fahnestock, M., 2008. Long-term changes in neurotrophic factor expression in distal nerve stump following denervation and reinnervation with motor or sensory nerve. Journal of Neurochemistry, 105(4), 1244-1252.

Morcuende, S., Materredona, E.R., Benitez-Temino, B., Munoz-Hernandez, R., Pastor, A.M., de la Cruz, R.R., 2011. Differential expression of the expression of neurotrophin receptors in rat extraocular motoneurons after lesion. Journal of Comparative Neurology, 519(12), 2335-2352.

McCartney, A.M., Abejuela, V.L., Isaacson, L.G., 2008. Characterization of trkB immunoreactive cells in the intermediolateral cell column of the rat spinal cord. Neuroscience Letters, 440(2), 103-8.

McTigue, D. M., Horner, P. J., Stokes, B. T., Gage, F. H., 1998. Neurotrophin-3 and brain- derived neurotrophic factor induce oligodendrocyte proliferation and myelination of regenerating axons in the contused adult rat spinal cord. Journal of Neuroscience, 18(14), 5354-65.

Navarro, X., Vivó, M., Valero-Cabré, A., 2007. Neural plasticity after peripheral nerve injury and regeneration. Progress in Neurobiology, 82(4), 163-201.

Nja, A., Purves, D., 1977. Re-innervation of guinea-pig superior cervical ganglion cells by preganglionic fibres arising from different levels of the spinal cord. Journal of Physiology, 272(3), 633-51.

Ohira, K., Funatsu, N., Homma, K. J., Sahara, Y., Hayashi, M., Kaneko, T., Nakamura S. 2007. Truncated TrkB-T1 regulates the morphology of neocortical layer I astrocytes in adult rat brain slices. European Journal of Neuroscience, 25(2), 406-416.

Olson, L., 1969. Intact and regenerating sympathetic noradrenaline axons in the rat sciatic nerve. Histochemistry, 17, 349-367.

Oudega, M., Hagg, T., 1999. Neurotrophins promote regeneration of sensory axons in the adult rat spinal cord. Brain Research, 818(2), 431-438.

Peddie, C. J., Keast, J. R., 2011. Pelvic nerve injury causes a rapid decrease in expression of choline acetyltransferase and upregulation of c-Jun and ATF-3 in a distinct population of sacral preganglionic neurons. Frontiers in Neuroscience, 5(6), 1-13.

Poulat, P., Marlier, L., Pellas, F., Rajaofetra, N., Privat, A., 1992. Effects of neonatal removal of superior cervical ganglion on serotonin and thyrotropin-releasing hormone immunoreactivity in the intermediolateral cell column of the rat spinal cord. Experimental Brain Research, 91, 21-28.

Radović, N., Cucić, S., Altarac, S., 2008. Molecular aspects of apoptosis. Acta medica Croatica : c̆ asopis Hravatske akademije medicinskih znanosti, 62(3), 249-56.

Raisman, G., Field, P.M., Ostberg, A.J.C., Iversen, L.L., Zigmond, R.E., 1974. A quantitative ultrastructural and biochemical analysis of the process of reinnervation of the superior cervical ganglion in the adult rat. Brain Research, 71, 1-16.

Rando, T.A., Bowers, C.W., Zigmond, R.E., 1981. Localization of neurons in the rat spinal cord which project to the superior cervical ganglion. Journal of Comparative Neurology, 196, 73-83.

Renn, C.L., Leitch, C.C., Dorsey, S.G., 2009. In vivo evidence that truncated trkB.T1 participates in nociception. Molecular Pain, 5, 61-69.

Seijffers, R., Mills, C.D., Woolf, C.J., 2007. ATF3 increases the intrinsic growth state of DRG neurons to enhance peripheral nerve regeneration. Journal of Neuroscience, 27(30), 7911-20.

Serebryakova, I.Y., 2008. Dynamics of the remodeling of neural connections in the superior cervical ganglion in rats after dosed compression of the preganglionic trunk. Neuroscience Behavior and Physiology, 38(8), 811-15.

Sofroniew, M. V., Cooper, J.D., Svendsen, C.N., Crossman, P., Ip, N.Y., Lindsay, R.M., Zafra, F., Lindholm, D., 1993. Atrophy but not death of adult septal cholinergic neurons after ablation of target capacity to produce mRNAs for NGF, BDNF, and NT3. Journal of Neuroscience, 13(12), 5263-5276.

Skup, M., Dwornik, A., Macias, M., Sulejczak, D., Wiater, M., Czarkowska-Bauch, J., 2002. Long-term locomotor training up-regulates TrkBFL receptor-like proteins, brain-derived neurotrophic factor, and neurotrophin 4 with different topographies of expression in oligodendroglia and neurons in the spinal cord. Experimental Neurology, 176, 289-307.

Song, D. Y., Yang, Y. C., Shin, D. H., Sugama, S., Kim, Y. S., Lee, B. H., Joh, T. H., Cho, B.P., 2008. Axotomy-induced dopaminergic neurodegeneration is accompanied with c-Jun phosphorylation and activation transcription factor 3 expression. Experimental Neurology, 209(1), 268-78.

Spalding, K.L., Cui, Q., Harvey, A.R., 2005. Retinal ganglion cell neurotrophin receptor levels and trophic requirements following target ablation in the neonatal rat. Neuroscience, 131(2), 387-395.

Strack, A.M., Sawyer, W.B., Marubio, L.M., Loewy, A.D., 1988. Spinal origin of sympathetic neurons in the rat. Brain Research, 455, 187-191.

Sun, Y., Zigmond, R.E., 1996. Involvement of leukemia inhibitory factor in the increases in galanin and vasoactive intestinal peptide mRNA and the decreases in neuropeptide Y and tyrosine hydroxylase mRNA in sympathetic neurons after axotomy. Journal of Neurochemistry, 67(4), 1751-60.

Svensson, M., Eriksson, P., Persson, J.K., Molander, C., Arvidsson, J., Aldskogius, H., 1993. The response of central glia to peripheral nerve injury. Brain Research Bulletin, 30(3-4), 499- 506.

Tang, H., Brimijoin, S., 2002. Death of preganglionic sympathetic neurons after surgical or immunologic lesion of the peripheral processes. Experimental Neurology, 177, 105-114.

Tarn, C., Bilodeau, M. L., Hullinger, R. L., Andrisani, O. M., 1999. Differential immediate early gene expression in conditional hepatitis B virus pX-transforming versus nontransforming hepatocyte cell lines. Journal of Biological Chemistry, 274(4), 2327-36.

Tsujino, H., Kondo, E., Fukuoka, T., Dai, Y., Tokunaga, A., Miki, K., Yonenobu, K., Ochi, T., Noguchi, K., 2000. Activating transcription factor 3 (ATF3) induction by axotomy in sensory and motoneurons: a novel neuronal marker of nerve injury. Molecular and Cellular Neuroscience, 15, 170-182.

Wang, W., Salvaterra, P.M., Loera, S., Chiu, A.Y., 1997. Brain-derived neurotrophic factor spares choline acetyltransferase mRNA following axotomy of motor neurons in vivo. Journal of Neuroscience Research, 47, 134-143.

Yamada, J., Nakanishi, H., Jinno, S., 2011. Differential involvement of perineuronal astrocytes and microglia in synaptic stripping after hypoglossal axotomy. Neuroscience, 182, 1-10.

Yan, Q., Matheson, C., Lopez., O., Miller, J., 1994. The biological responses of axotomized adult motoneurons to brain-derived neurotrophic factor. Journal of Neuroscience, 14(9), 5281- 5291.

Yin, Q., Kemp, G.J., Yu, L.G., Wagstaff, S.C., Frostick, S.P., 2001. Neurotrophin-4 delivered by fibrin glue promotes peripheral nerve regeneration. Muscle Nerve, 24(3), 345-351.

Zhou, X.-F., Song, X.-Y., Zhong, J.-H., Barati, S., Zhou, F. H.-H., Johnson, S. M., 2004. Distribution and localization of pro-brain-derived neurotrophic factor-like immunoreactivity in the peripheral and central nervous system of the adult rat. Journal of Neurochemistry, 91(3), 704-15.

Chapter 3: Distribution and phenotype of TrkB oligodendrocyte lineage cells in the adult rat spinal cord

Coulibaly et al., 2014. Brain Research 1582:21-33

ABSTRACT The distribution and phenotype of a previously undescribed population of nonneuronal cells in the intact spinal cord that expresses TrkB, the cognate receptor for brain derived neurotrophic factor (BDNF) and neurotrophin 4 (NT-4), were characterized by examining the extent of colocalization of TrkB with NG2, which identifies oligodendrocyte progenitors (OPCs) or CC1, a marker for mature oligodendrocytes (OLs). All TrkB nonneuronal cells expressed Olig2, confirming their role in the OL lineage. Similar to OPCs and OLs, TrkB cells resided in gray and white matter of the spinal cord in similar abundance. Less than 2% of TrkB cells expressed NG2, while over 80% of TrkB cells in the adult spinal cord co-expressed CC1. Most OPCs did not express detectable levels of TrkB, however a small OPC pool (~5%) showed TrkB immunoreactivity. The majority of mature OLs (~65%) expressed TrkB, but a population of mature OLs (~36%) did not express TrkB at detectable levels and 17% of TrkB nonneuronal cells did not express NG2 or CC1. Approximately 20% of the TrkB nonneuronal population in the ventral horn resided in close proximity to motor neurons and were categorized as perineuronal. We conclude that TrkB is expressed by several pools of OL lineage cells in the adult spinal cord. These findings are important in understanding the neurotrophin regulation of OL lineage cells in the adult spinal cord.

1. Introduction Oligodendrocytes (OLs) are supportive cells in the central nervous system (CNS) that are responsible for the myelination of axons and the establishment of proper communication between neurons in the brain and spinal cord. The dysregulation or loss of OLs has been implicated in disorders such as multiple sclerosis (Lee et al., 2012), schizophrenia (Karoutzou et al., 2008), amyotrophic lateral sclerosis (Lasiene and Yamanaka, 2011), X-linked Charcot-Marie Tooth (Sargiannidou et al., 2009), mood disorders (Roy et al., 2007) and the lack of recovery following brain injury or stroke (Dewar et al., 2003). Therefore, it is important to understand the biology of these cells in uninjured adult tissues. OLs are derived from a defined group of precursors known as oligodendrocyte progenitor cells (OPCs; Cohen et al., 1996; Scarisbrick et al., 2000; Nishiyama et al., 2002; Nishiyama et al., 2009). During development and into adulthood, factors such as neurotrophins and neurogulins regulate the proliferation and differentiation of OPCs, which express the proteoglycan marker nerve/glial antigen (NG2; Nishiyama et al., 2009), into mature OLs. In turn mature OLs are characterized by a down regulation of NG2 and up regulation of the adenomatous polyposis coli (APC; commonly known as CC1) antigen (Kessaris et al., 2008; McTigue and Tripathi, 2008; Nishiyama et al., 2009). OLs that express TrkB, the cognate receptor for brain derived neurotrophic factor (BDNF), have been reported in the adult spinal cord (Macias et al., 2007; McCartney et al., 2008; Coulibaly and Isaacson, 2012; Skup et al., 2002), suggesting that at least some cells in the OL lineage are responsive and/or regulated by BDNF. Indeed, numerous reports have indicated that BDNF can affect the cells in the OL lineage (Van’t Veer et al., 2009; VonDran et al., 2010). For instance, following spinal cord injury, the presence of BDNF grafts led to increased OPC proliferation and differentiation into OLs (McTigue et al., 1998). BDNF infusion into the rat spinal cord following injury led to decreased OL apoptosis (Koda et al., 2002). Though TrkB nonneuronal cells have been described in the intermediolateral cell column (IML; Coulibaly and Isaacson, 2012) and ventral horn (Macias et al., 2007; Skup et al., 2002) of the adult spinal cord, the distribution of TrkB cells and their overlap with cells in the OL cell lineage have not been investigated. The overall objective of this study was to characterize in detail the location and phenotype of TrkB cells in the adult spinal cord.

Perineuronal TrkB cells within the spinal cord gray matter (GM), which reside in close association with neuronal cell bodies, have been reported (Coulibaly and Isaacson, 2012; Skup et al., 2002), yet their frequency and phenotype in the spinal cord have not been studied. Therefore, perineuronal cells were included in our analysis of the spinal cord TrkB populations. Two levels of the spinal cord, the cervical and thoracic cord, were examined in the present study. The thoracic level was analyzed to assist in our understanding of TrkB cells at the level of the sympathetic intermediolateral cell column. For comparison, the cervical level was examined at the level of the phrenic nucleus. This level was selected due to its important role in respiration studies (Kuzuhara and Chou, 1980; Reviewed in Zimmer et al., 2007). The findings of this study revealed that, similar to OPCs and OLs, TrkB nonneuronal cells in the spinal cord were present in the GM and white matter (WM) in equal proportions. In addition, these cells typically exhibited characteristics of mature OLs, while TrkB cells that expressed markers for OPCs were relatively rare. Similarly, the majority of TrkB perineuronal cells in adult spinal cord expressed the OL phenotype and comprised approximately 15%-20% of all TrkB cells in the ventral horn. Portions of this study have been previously published in abstract form (Coulibaly and Isaacson, 2013).

2. Materials and Methods 2.1. Tissue preparation and immunohistochemistry Adult female Sprague Dawley rats (Harlan labs, Indianapolis, IN; n=7) approximately 3 months of age were housed in a 12:12 hour light:dark cycle at Miami University Animal Facility. All experiments were done in accordance with a protocol approved by the Miami University Institutional Animal Care and Use Committee. Animals were anesthetized with sodium pentobarbital (125 mg/kg), then transcardially perfused with 4% paraformaldehyde in 0.1 M phosphate buffer (PB). Spinal cords were removed and the fourth cervical (C4) and first thoracic (T1) segments were cut at 18µm in the coronal plane using a MICROM HM 550 series cryostat. Sections were mounted on Superfrost microscope slides (Fisher Scientific), allowed to dry, and then incubated overnight in 0.1M PBS- 0.2% Triton-X solution, followed by incubation with mouse anti-CC1 (1:500, Abcam #ab16794; for OLs, Bhat et al., 1996; De Nicola et al., 2009), mouse anti-NG2 (1:200, Millipore #MAB5384; for OPCs, Nishiyama et al., 2002), goat anti-olig 2 (1:100, R&D Systems

#AF2418; Barnabe-Heider et al., 2010), and/or rabbit anti-TrkB (1:200, sc-12; Santa Cruz Biotechnologies, directed against the full length receptor) antibodies diluted in 0.1M PBS for 24 hours at 4oC. After rinsing, sections were incubated in Alexa Fluor conjugated antibodies (Molecular Probes; 1:200) directed toward the primary host antibody for 2 hours. Sections were then rinsed and coverslipped using fluorescent mounting medium with DAPI (Vector Labs).

2.2. Quantitative analysis Images were captured with a Zeiss 710 confocal laser scanning microscope and Zen 2009 software. Using the microscope ‘Tile’ function, a single mosaic of the entire spinal cord (n=7) was automatically ‘stitched’ from partial images or ‘tiles’ that were each acquired at 200X magnification in a 4x7 configuration. The number and distribution of TrkB-ir, NG2-ir, CC1-ir profiles (referred to in this study as TrkB, NG2, CC1 cells) were analyzed within the white matter (WM) and gray matter (GM) areas of the spinal cord from tiled images that were magnified to approximately 500X on a computer monitor. A standardized measuring box (approx. 100 m2) was used to characterize the number and location of OPCs and OLs that expressed TrkB in the spinal cord. The size of the box was determined using the approximate width of the dorsal horn as a guide. The box was placed over 14 gray and white matter regions of the spinal cord. In the WM, the box was placed over the lateral funiculus (LH) and the ventral funiculus (VF) just ventral to the ventral horn on both sides, and a second central region of the VF (Fig. 1.A.) as well as a central region of the dorsal funiculus (DF; Fig. 1.A.). In the GM, the box was placed bilaterally over the dorsal horn (DH) just ventral to Lamina II, the lateral horn (LH) and ventral horn (VH) and a region adjacent to central canal (CC; Fig. 1.A.). The number of cells within each region expressing each marker was expressed as a proportion of number of cells in the entire cord expressing that marker. Counts were combined from left and right sides for bilateral regions while the VF was totaled from the two sides and a third, more central region. The values were converted to mm2 for comparison across regions and demonstration in the figures and Venn diagrams (basic and stacked). Size exclusion criteria were used to exclude TrkB neurons from the analysis. In the ventral horn neurons that expressed TrkB were easily identified due to their large size. TrkB neurons in the lateral horn averaged ~1,100 m3 in volume (Coulibaly and Isaacson 2013). In addition to their larger size, the identity of TrkB neurons was verified by labeling with the NeuN

61 marker (McCartney et al., 2008) which shows specificity for neurons (Mullen et al., 1992). In contrast TrkB nonneuronal cells were smaller and their volume averaged ~53 m3 (unpublished information). It should also be noted that the DAPI labeled nuclei exhibited by neurons typically were larger and paler when compared to dense, bright blue nuclei of the glial cells (See Figs. 1, 5, 6). The minor proportion of small neurons in Lamina II of the dorsal horn that express TrkB (Salio et al., 2005) likely were not included in our analysis as the measuring box for the dorsal horn encompassed only the lower portion of Lamina II and focused primarily on Laminae III, IV,V. For the proportion of cells expressing CC1 (CC1+/TrkB- and CC1+/TrkB+) within the double labeled CC1/TrkB population, as shown in the basic Venn (Fig. 3.C.), the total number of CC1+ cells was divided by the total number of immunolabelled cells (CC1+/TrkB- and CC1- /TrkB+ and CC1+/TrkB+). To determine the proportion of CC1 only cells, the number of CC1 only cells (CC1+/TrkB-) was divided by the total number of all CC1 immunolabeled cells (CC1+/TrkB- and CC1+/TrkB+ ; Fig. 3.C., basic Venn). For the proportion of overlap between the two populations (CC1+/TrkB+), the number of CC1+/TrkB+ cells was divided by the total number of immunolabeled cells (CC1+/TrkB- and CC1-/TrkB+ and CC1+/TrkB+ ; Fig. 3.C., basic Venn). To determine the proportion of cells expressing TrkB within the double labeled population as shown in the basic Venn (TrkB+/CC1- and TrkB+/CC1-), the total number of TrkB+ cells was divided by the total number of immunolabelled cells (TrkB+/CC1+ and TrkB+/CC1- and TrkB-/CC1+). For the proportion of TrkB only cells, the number of TrkB+/CC1- cells was divided by the total number of TrkB cells (TrkB+/CC1- and TrkB+/CC1+; Fig. 3.C., basic Venn). To determine the proportion of TrkB cells within the TrkB population that co-expressed CC1, the number of TrkB+/CC1+ cells was divided by the total number of TrkB+ cells (TrkB+/CC1- and TrkB+/CC1+; Fig. 3.C., stacked Venn). Within the TrkB population, the proportion of TrkB cells within the TrkB population that did not express CC1 was determined by dividing the number of TrkB only cells (TrkB+/CC1- ) by the total number of TrkB immunolabeled cells (TrkB+/CC1- and TrkB+/CC1+; Fig. 3.C., stacked Venn). Similar analyses were carried out for the NG2 and TrkB data shown in Figure 4. To assess any differences in cell density between WM and GM, the mean number of profiles per mm2 in the WM and GM was compared across animals (n=7) using Student’s t-test with significance reported at p< 0.05. NG2-ir vascular cells such as pericytes exhibiting nuclear

62 staining in the shape of a quarter circle (McTigue et al., 2001) and associating with blood vessels (McTigue et al., 2001; Ligon et al., 2006) were excluded from the analysis. All NG2 non- vascular cells, regardless of their unipolar, bipolar, or multipolar morphologies (Horner et al., 2000), were included in the analysis. The proportion of cells that were observed in direct apposition in a cluster of two or more cells in the GM (DH, LH, and VH) and WM (DF, LF, and VF) matter was calculated (n=5) at both cervical and thoracic segments. The proportions from GM and WM were compared across animals using a Student’s t-test with significance reported at p< 0.05. For the analysis of perineuronal cells, sections of the ventral horn were triple-labeled with TrkB/CC1/ChAT or TrkB/NG2/ChAT as shown in Figure 6. The images were acquired at 400X magnification and then magnified to 800X on the computer monitor for quantitative analysis. The proportion of ChAT-ir ventral horn neurons in association with at least one perineuronal OL lineage profile was calculated (n=7). For the number of cells in apposition to ventral horn motor neurons, the number of TrkB, CC1, or NG2 cell bodies that were directly apposed to a ventral motor neuron was expressed as a proportion of the total cells expressing that marker. The frequency of CC1+/TrkB+ or NG2+/TrkB+ profiles in the perineuronal cell population also was determined.

3. Results 3.1. Distribution of TrkB cells in the adult rat spinal cord: comparison with OPCs and mature OLs In order to determine whether TrkB nonneuronal cells were of the OL lineage, TrkB cells were co-labeled with Olig2 (Fig. 1.C.), a transcription factor that is expressed by cells in the OL cell lineage (Nishiyama et al., 2009). TrkB cells exhibited a thin rim of TrkB immunoreactivity surrounding a rounded nucleus (Fig. 1.B.; Skup et al., 2002) and all of the TrkB nonneuronal cells also expressed Olig2 (Fig. 1.C.). Frequent TrkB cells that did not express Olig2 exhibited characteristics of neurons and are denoted by asterisks in Figure 1.C. Some Olig2 cells did not express TrkB (Fig. 1.C.) and likely represented the NG2 or CC1 populations that did not express detectable levels of TrkB (see below). The density of TrkB nonneuronal cells in the spinal cord and their location throughout the GM and WM of the spinal cord were examined. The density of TrkB cells in the cervical

63 segment averaged 143 cells/mm2 in the WM and 204 cells/mm2 in GM (Fig. 2.A.). In the thoracic segment, TrkB density averaged 240 cells/mm2 in the WM and 232 cells/mm2 in GM (Fig. 2.A.). No significant differences were observed at either segment when TrkB cell density was compared between WM and GM (Fig. 2.A.). Furthermore, when TrkB cell density was compared between the cervical and thoracic segments, no differences were observed. The density and distribution of TrkB cells were compared with that of mature OLs and OPCs. Similar to TrkB cells, mature OLs showed CC1 immunoreactivity along the perimeter of a rounded cell body (Fig. 1.B.). In the cervical segment, CC1 cells averaged 240 cells/mm2 in the WM and 286 cells/mm2 in GM (Fig. 2.B.). In the thoracic segment, 258 cells/mm2 and 276 cells/mm2 were observed in the WM and GM, respectively (Fig. 2.B.). No significant differences were observed at either segment when the density of CC1 cells was compared between WM and GM (Fig. 2.B.). Further, no differences were observed when CC1 cell density was compared between the two segments (Fig. 2.B.). NG2 cells were characterized by a ring of immunoreactivity around the cell nucleus and fine immunoreactive processes that emanated from the cell body (Ligon et al., 2006; McTigue et al., 2001; Fig. 1.B.). The values for NG2 cell density in the spinal cord generally were lower than TrkB and CC1 cell density. NG2 density averaged 35 cells/mm2 in the WM and 41 cells/mm2 in the GM in the cervical segment (Fig. 2.C.). In the thoracic segment, NG2 density averaged 58 cells/mm2 in WM and 73 cells/mm2 in GM (Fig. 2.C.). No significant differences were observed in either segment when NG2 cell density was compared between the GM and WM (Fig. 2.C.). However, NG2 cell density in the cervical GM was significantly less when compared to the thoracic segment GM (Fig. 2.C.).

3.2 Expression of TrkB by OL lineage cells in the adult spinal cord In order to determine the overlap between the TrkB and mature OL populations, double label immunohistochemistry was carried out using CC1 and TrkB antibodies. A substantial overlap between CC1 and TrkB pools was observed in both the cervical and thoracic segments of the spinal cord (Fig. 3.A., B., C.), yet CC1 only (Fig. 3.B., C.) and TrkB only (Fig. 3.C.) cells also were observed. In the cervical segment, 88% of the cells expressed CC1 (CC1+/TrkB- or CC1+/TrkB+) and 64% expressed TrkB (TrkB+/CC1- or TrkB+/CC1+; Fig. 3.C.). As shown in the basic Venn diagram in Figure 3.C., the overlap between the two populations was substantial,

64 with 52% of the population labeled with both CC1 and TrkB (Fig. 3.C.). However, 36% of the cells were CC1 only (CC1+/TrkB-) and 12% expressed only TrkB (TrkB+/CC1-). As shown in the stacked Venn diagram on the right panel of Figure 3.C., within the TrkB population, 81% of the TrkB cells co-expressed CC1 (TrkB+/CC1+). In the thoracic segment (Fig. 3.A., B.), 86% of the cells expressed CC1 (CC1+/TrkB- or CC1 +/TrkB+) while 70% expressed TrkB (TrkB+/CC1- or TrkB+/CC1+; Fig. 3.C., basic Venn). Also as shown in the basic Venn, 57% of the population expressed both CC1 and TrkB (CC1+/TrkB+), 30% of the cells were CC1 only (CC1+/TrkB-) while 14% expressed only TrkB (TrkB+/CC1-; Fig. 3.C.). Similar to the cervical segment, within the TrkB population, 81% of the TrkB cells co-expressed CC1 (TrkB+/CC1+; Fig. 3.C., stacked Venn), while 19% were TrkB+/CC1-. In order to determine the overlap between TrkB and OPC populations, double label immunohistochemistry was carried out using NG2 and TrkB antibodies. As shown in the confocal images in Figure 4.A. and Figure 4.B. and the Venn diagrams in Figure 4.C., there was little overlap between the NG2 and TrkB pools. In the cervical segment, 19% expressed NG2 (NG2+/TrkB- or NG2+/TrkB+), while 82% expressed TrkB (NG2-/TrkB+ or NG2+/TrkB+; Fig. 4.A., B.). As shown in the basic Venn diagram in Figure 4.B., overlap between the TrkB and NG2 pools (NG2+/TrkB+) was only 1.3%. 18% of the population expressed NG2 only (NG2+/TrkB-) while 81% expressed only TrkB (TrkB+/CC1-; Fig. 4.B.). The stacked Venn diagram on the right panel of Figure 4.B. revealed that, within the TrkB population, only 1.6% of the TrkB cells co-expressed NG2 (TrkB+/NG2+; Fig. 4.B.). In the thoracic segment, 26% of the NG2/TrkB labeled population expressed NG2 (NG2+/TrkB- or NG2+/TrkB+), while 75% expressed TrkB (NG2-/TrkB+ or NG2+/TrkB+; Fig. 4.B., basic Venn). As shown in the basic Venn, only 1.6% of the cells were NG2+/TrkB+ (Fig. 4.B.) and 25% of the cells expressed only NG2 (NG2+/TrkB-) while 74% expressed only TrkB (TrkB+/CC1-; Fig. 4.B.). The stacked Venn diagram depicts that, within the TrkB population, only 2.1% of the TrkB cells co-expressed NG2 (TrkB+/CC1+; Fig. 4.B.).

3.3. Clustering of OL lineage cells in the adult spinal cord While analyzing the distribution of OL lineage cells, it was observed that mature OLs tended to cluster or aggregate in groups of 2 or more cells, particularly in the gray matter. In order to better characterize this phenomenon, the clustering of OL lineage cells in the spinal cord

65 was quantified. Clustering was exhibited primarily by CC1 and TrkB cells (Fig. 5.A., B.; See also figure 3B and 6B for examples of clustering), particularly in the GM, and this phenomenon was observed at both the cervical and thoracic levels. The proportion of cells in clusters ranged from 11-22% in the cervical segment (WM=11%; GM=22%; Fig. 5.B.) and 7-23% in the thoracic segment (WM=7%; GM=23%; Fig. 5.B.). In the thoracic segment the number of GM clusters (both TrkB and CC1) was significantly higher compared with the number in the WM (Fig. 5.B.). No difference was observed when comparing the number of clusters between the two spinal segments.

3.4. Perineuronal OL lineage cells in the adult spinal cord As described in previous studies (Skup et al., 2002; Macias et al., 2007; McCartney et al., 2008; Coulibaly et al., 2012), TrkB cells were frequently observed in close proximity to neuronal cell bodies in the IML and the ventral horn and thus were categorized as perineuronal. In order to determine the overlap of these cells with OL lineage cells markers, the expression of NG2, CC1 and/or TrkB by perineuronal cells and their association with ChAT immunoreactive (- ir) ventral horn neurons were analyzed. Approximately 57% of the ChAT-ir neurons in the ventral horn of the cervical and thoracic cord was associated with at least one perineuronal cell (Fig. 6). CC1 and TrkB perineuronal cells both resided in close proximity to the neuronal cell body membrane (Fig. 6. A. B.) but perineuronal NG2 cells rarely were observed (Fig. 6.C.). Approximately 12% of the CC1 population was perineuronal while the number of TrkB perineuronal cells ranged from 15%-20% in the cervical and thoracic segments. Though some overlap existed between the CC1 and TrkB perineuronal populations, it was not complete and approximately 65% and 75% of the TrkB perineuronal cells co-expressed CC1 in the respective cervical and thoracic segments. In the cervical segment, 5% of the NG2 population was perineuronal, while no perineuronal NG2 cells were observed in the thoracic segment. Both segments showed fine NG2-ir processes in close proximity to the perimeter of the ventral horn motor neurons (Fig. 6.C.).

TrkB TrkB A. B. DAPI

DH DH DF Cc 1 Cc 1 DAPI LF LH CC CC LH LF

VH VF VH Ng 2 Ng 2 DAPI VF VF

C. Olig2 Olig2 Olig2 * * TrkB * TrkB DAPI

* * *

Figure 1

Figure 1: Localization and identification of OL lineage cells in the adult rat spinal cord. A. Diagrammatic representation of the adult rat spinal cord showing the location of the measuring boxes that were used in this study. DH, dorsal horn; LH, lateral horn; VH, ventral horn; CC, central canal; DF, dorsal funiculus; LF, lateral funiculus; VF, ventral funiculus. B. Confocal micrographs depicting the characteristic immunolabeling used to identify TrkB, CC1, and NG2 cells in the spinal cord. C. Double immunolabeling with TrkB (red) and Olig2 (green) revealed that all nonneuronal TrkB-immunoreactive (-ir) cells also expressed Olig2 (white arrows). Olig2 cells that were not TrkB-ir (yellow arrow) may represent mature OLs (CC1+/TrkB-) and OPCs (NG2+/TrkB-). TrkB-ir neurons (yellow asterisks), identified by larger cell volume, were observed in the gray matter and did not express Olig2. In merged images (far right, B., C.), DAPI was used to demonstrate the cellular nature of the staining. Scale bar in B. = 5µm; Scale bar in C. = 20 µm.

A. B. TrkB distribution in spinal cord CC1 distribution in spinal cord 300 WM 350 WM

GM GM 2

250 300

mm mm 2 250 200 200 150 150

Average per cells of Average # 100

100 Average # of cells per mm per cells mm of Average # 50 50

0 0 C4 T1 C4 T1

C. NG2 distribution in the spinal cord 90 WM #

80 GM 2 70 60 50 40 30

Average # of cells per mm per cells mm of Average # 20 10 0 C4 T1

Figure 2

Figure 2: Density of OL lineage cells within white matter (WM) and gray matter (GM) in cervical (C4) and thoracic (T1) segments of the adult rat spinal cord. Comparison of cell densities between the WM and GM within segments revealed no differences in the number of TrkB (A.), CC1 (B.), or NG2 (C.) cells. In addition, no differences were observed in the number of TrkB and CC1 cells when the cervical and thoracic segments were compared. However, the density of NG2 cells in the GM in the thoracic segment was significantly higher when compared with GM density in the cervical segment. Data shown as mean +/- SEM. #, significance reported at p<0.05. WM, white matter, GM, gray matter.

Figure 3

Figure 3: The majority of mature OLs expressed TrkB at detectable levels. A.-A.’’ Confocal micrographs of coronal sections through the T1 segment of the spinal cord reveal the colocalization of CC1 and TrkB in nonneuronal cells (yellow arrows). Note one cell expressing CC1 only (green arrows), but only faintly immunoreactive for TrkB. Scale bar = 50 µm. B.-B.’’ Most cells expressed both CC1 and TrkB (yellow arrows), yet cells expressing CC1 only (green arrows) also were observed. DAPI was used to demonstrate the cellular nature of the staining. Scale bar = 20 µm. C. Basic Venn diagrams (left panel) represent the percentage of cells that expressed CC1 only (green), TrkB only (red), or CC1 and TrkB (yellow) in cervical (C4) and thoracic (T1) segments. Stacked Venn diagrams (right panel) reveal that ~81% of the TrkB cells also expressed CC1 in both of the segments.

Figure 4

Figure 4: Little overlap between OPC and TrkB populations in the intact spinal cord. A.-A.” Confocal micrographs of coronal sections at T1 segment reveal the lack of co-localization between NG2 and TrkB cells. NG2 cells (green arrows) and TrkB cells (red arrows) showed no overlap. DAPI was used to demonstrate the cellular nature of the staining. Scale bar = 50 µm. B. Basic Venn diagrams (left panel) represent the percentage of cells expressing NG2 only (green), TrkB only (red), or NG2 and TrkB (yellow) in the spinal cord in cervical (C4) and thoracic (T1) segments. Stacked Venn diagrams (right panel) reveal that ~2% of the TrkB cells also expressed NG2 in both of segments.

A. A.’ * *

* *

A.’’ B. 30 TrkB in clusters 25 WM * GM ) * % 20

(

l e 15

r 10

T 5 0 * C4 T1

30 CC1 in clusters WM * 25 GM

)

( 20

C 10 5 0 C4 T1

Figure 5

Figure 5: Clusters of TrkB and CC1 cells were observed primarily in the GM of the spinal cord. A.-A.’’ Both TrkB (red) and CC1 (green) cells in the ventral horn (delineated by yellow lines) frequently were observed in aggregates or clusters of two or more cells (white arrows). Motor neurons (yellow asterisks) abutted by CC1+/TrkB+ cells were present. DAPI used to demonstrate the cellular nature of each cell type. Scale bar = 50 m. B. Graphs showing the number of TrkB (upper) and CC1 (lower) cells that were observed in clusters at cervical (C4) and thoracic (T1) segments of the spinal cord. In the thoracic segment, the number of CC1 and TrkB clusters was significantly higher in the GM compared with WM. Data shown as mean +/- SEM. *, significance at p<0.05 when comparing WM and GM at T1.

A. A.’ A.’’ * * * * * * * * * TrkB TrkB CC1 CC1 * ChAT * ChAT * ChAT DAPI DAPI DAPI B. B.’ B.’’

TrkB * TrkB * TrkB * ChAT DAPI ChAT DAPI C. * C.’ * C.’’ *

TrkB TrkB NG2 NG2 * ChAT * ChAT * ChAT DAPI DAPI DAPI

Figure 6

Figure 6: A population of OL lineage cells resided in close apposition to motor neurons in the ventral horn of adult rat spinal cord and was categorized as perineuronal. A.-A.’’ The majority of perineuronal cells expressed both CC1 and TrkB (yellow arrows). CC1 cells that were not in close proximity to motor neuron cell bodies (*) also were observed (white arrows). B.-B.’’ TrkB perineuronal cells (white arrows) are shown in close proximity to ChAT-ir ventral horn motor neurons (*). Yellow arrows indicate the TrkB nonneuronal cells that were not in direct apposition to neuronal cell bodies. C.-C.’’ NG2 cell cell bodies (yellow arrows) rarely were observed in direct apposition to ventral horn neuronal cell bodies (i.e. perineuronal), yet NG2-ir processes (white arrows) frequently were observed in close proximity to the ventral horn neuronal cell bodies (*). Scale bar = 50 µm.

4. Discussion 4.1. Phenotype of TrkB nonneuronal cells in the intact spinal cord TrkB cells in the rat spinal cord were first described by Skup et al. (2002), who noted their presence throughout the cord, and reported their co-localization with OL markers such as GalC, RIP, and/or NG2. Subsequent studies from our laboratory confirmed that the TrkB cells were nonneuronal (McCartney et al., 2008), that many of the cells expressed CC1 (Coulibaly and Isaacson, 2012), and that they did not express markers for astrocytes or microglia (Skup et al. 2002; Coulibaly, unpublished). Others report little colocalization of full length TrkB and GFAP in uninjured tissues though reactive astrocytes can express full length TrkB following CNS injury (McKeon et al., 1997) and in models of multiple sclerosis (Colombo et al., 2012). Furthermore, all TrkB nonneuronal cells co-expressed Olig2, a transcription factor that has been shown by others to be expressed by OL lineage cells (Miller, 2002), indicating that they indeed are part of the OL cell lineage. Plasticity of TrkB cells in the spinal cord has been reported. For example the TrkB population in the ventral horn has been shown to exhibited plasticity following exercise (Skup et al., 2002), and TrkB cells in the IML, which houses sympathetic preganglionic neuronal cell bodies, were increased following distal transection of preganglionic axons (Coulibaly and Isaacson, 2012), possibly to contribute to neuronal survival and regeneration (Coulibaly et al., 2013). In our assessment of the TrkB population in the spinal cord of intact uninjured animals, we expected that most cells in the OL cell lineage would express TrkB. Indeed, VonDran and colleagues (2010) demonstrated that both OPCs and mature OLs in the mouse basal forebrain expressed TrkB and that the cells in the OL lineage were dependent upon BDNF for survival. In addition, OPCs, which give rise to OLs during development (Miller, 2002) and in the adult CNS (Nishiyama et al., 2009; Barnabe-Heider et al., 2010), can be induced to differentiate in the presence of BDNF both in vitro (VonDran et al., 2010; VonDran et al., 2011) and in vivo (McTigue et al., 1998). Yet our analysis revealed that only a relatively small proportion of OPCs in the adult spinal cord expressed TrkB at detectable levels. Horner and colleagues (2002) reported that ~3% of the NG2 population in the spinal cord was in the cell cycle over a 12 day period, while 97% of the cells were quiescent or carrying out other activities. These numbers compare favorably with

79 the small proportion of NG2+/TrkB+ cells observed in our study. Because BDNF appears to regulate OPC proliferation and differentiation, we propose that the OPCs showing detectable levels of TrkB in the present study may represent the pool that has committed to either self- renew or to differentiate into OLs (Barnabe-Heider et al., 2008). Rather than exist in a quiescent state, cells in this stage would be actively dividing and/or maturing into OLs similar to the model proposed by Baumann and Pham-Dinh (2001). It should be noted that at least some of the OPCs that were not expressing detectable levels of TrkB may have been involved in functions other than renewal or OL differentiation and/or might be regulated by other stimulatory molecules, such as glutamate, FGF, PDGF, NGF and/or other neurotrophins (Miller, 2002; Nishiyama et al., 2009). Regardless of their exact function, our data support the existence of a heterogeneous NG2 cell population in the adult spinal cord GM and WM. Our results are supported by previous findings that the population of NG2 cells is heterogeneous in the adult spinal cord (Horner et al., 2002). As expected, a majority of the CC1 cells co-expressed TrkB, suggesting that a large proportion of mature OLs are regulated by BDNF and/or NT-4. Yet a significant subset (~36%) of mature OLs either expressed TrkB at very low levels or did not express TrkB. It is possible that a subset of the mature OL subpopulation within the spinal cord loses responsivity to, or possibly is not regulated by, BDNF or NT4. When considering the phenotype of the TrkB population in the spinal cord, 81% of the TrkB cells expressed the mature OL marker, CC1, while less than 2% of TrkB cells expressed the OPC marker NG2. Therefore, approximately 17% of the TrkB cells did not express detectable levels of NG2 or CC1. Our studies as well as others show no localization of TrkB in other nonneuronal cells such as astrocytes or microglia (Skup et al., 2002; Garraway et al., 2011) and these TrkB only are not in the size range of neurons. While it is possible that these cells expressed NG2 or CC1 below the level of detection of our antibodies, they also may represent a subpopulation of TrkB cells in transition from the precursor (NG2+/TrkB+) to the mature stage (CC1+/TrkB+). Indeed, the presence of an “immature OL” stage, one that occurs between the precursor and mature stages, in which NG2 is down-regulated, but detectable levels of CC1 are not evident, has been suggested (Baumann and Pham-Dinh, 2001; Miller, 2002; Nishiyama et al., 2009).

4.2. Distribution of OL lineage cells in the adult spinal cord OLs typically are known for their role in myelination and thus would be expected to be most prevalent in the WM. However, the results of the present study suggest that TrkB cells as well as OPCs and OLs are found in similar proportions in WM and GM throughout the spinal cord. The equal distribution of OPCs has been reported previously in the spinal cord (Horner et al., 2002) as well as in the brain (Staugaitis and Trapp, 2009). To our knowledge, we provide the first report of a similar distribution of TrkB cells throughout the GM and WM of the intact adult spinal cord. Little is known regarding the function of OLs that reside in the GM. They may simply participate in the myelination of descending axonal tracts that pass through spinal cord GM (Peters et al., 1970; Shinoda et al., 1986). However, functions for OLs other than myelination have been reported. For example, OLs may play a role in neuroprotection (Taniike et al., 2002) and/or provide important metabolic support to CNS neurons in the form of lactate (Lee et al., 2012). Approximately 20% of the TrkB population in the GM ventral horn was located within the perineuronal space of motor neurons and categorized as perineuronal. Furthermore, ~75% of these TrkB perineuronal cells expressed the mature OL marker. In the cerebral cortex, perineuronal OLs in association with cortical neuronal cell bodies comprised ~44% of the total number of OLs (Takasaki et al., 2010). Though the role of perineuronal OLs in the spinal cord is not yet known, it is possible that they provide metabolic support to the neurons, as observed in the mouse cortex (Takasaki et al., 2010), and/or possibly protect the neurons from apoptosis (Taniike et al., 2002). Interestingly, the number of cortical perineuronal OLs was found to be significantly reduced in individuals with schizophrenia and mood disorders (Vostrikov et al., 2007). Though rare, perineuronal OPCs were observed in the cervical segment and numerous NG2-ir processes were observed in close proximity to the perimeter of ventral horn motor neurons in both segments. We postulate that the processes emanate from nearby OPCs rather than from vascular pericytes. Though NG2 is also expressed by vascular pericytes, which in turn might be the source of these processes, most reports indicate that vascular NG2 cells interact only with blood vessels and blood brain barrier elements in the CNS (For review, see Sa-Pereira et al., 2012; Lange et al., 2013).

Specific interactions between ventral horn neurons and OPCs have not been described, yet others have reported NG2 cells in the GM as well as WM CNS regions (Dimou et al., 2008; Horner et al., 2000), and have postulated that they may serve roles other than differentiation into OLs. In the WM, NG2 cells can make contact with nodes of Ranvier (Butt et al., 1999), and they express functional GABA and glutamate receptors (Bergles et al., 2000; Gallo et al., 2008), indicating a possible role in the propagation of action potentials. NG2 cells can receive synaptic input from neurons (Bergles et al., 2000; Karadottir et al., 2008; Kukley et al., 2008; for review see Maldonado et al., 2011), and it has been proposed that NG2 processes located in close proximity to the synaptic cleft can produce action potentials as a result of neurotransmitter spillover (Maldonado et al., 2011). NG2 cells adjacent to neurons in the brain GM may play a role in signal integration (Mangin et al., 2008). It is unclear why perineuronal OPCs would be observed in the cervical segment, but not in the thoracic segment. Differences in brain regions also have been reported, in which OPCs associated readily with interneurons of the dentate gyrus (Mangin et al., 2008), while no perineuronal OPCs were observed in the cerebral cortex (Takasaki et al., 2010). Mangin and colleagues (2008) suggested that perineuronal OPCs in the dentate gyrus played a role in fine tuning the local neuronal network. Whether these cells play a similar role in the cervical spinal cord is a topic for further investigation. NG2 cells never were observed in clusters, while TrkB and CC1 cells frequently were observed in small aggregates or clusters of 2 or more, particularly in the GM. The singular distribution of NG2 cells, which has been described previously in the spinal cord (Dawson et al., 2003) as well as the cerebral cortex (Hill et al., 2011; Dawson et al., 2000; Dawson et al., 2003; Staugaitis and Trapp, 2009), has been attributed to contact inhibition properties between the NG2 cells (Staugaitis and Trapp, 2009), and in turn provides a network of uniformly distributed NG2 cells throughout both the GM and WM. This singular distribution, together with the Notch pathway and interactions with other CNS elements, results in the creation of individual niches or domains in the CNS and may facilitate the proliferative abilities of the OPCs (McTigue et al., 1998). Such contact inhibition may be specific to the progenitor, undifferentiated status of the NG2 cells. It is unknown whether the clustering serves a role in mature OL function.

4.3.Density of OL lineage cells in the adult spinal cord In the upper thoracic segment the density of OPCs averaged 58 and 73 NG2 cells/mm2 in the WM and GM, respectively. This compares favorably with Schonberg et al. (2012) who reported 34 and 80 NG2 cells/mm2 respectively in the WM and GM of the lower thoracic segment. In addition, we observed that the NG2 cell density in the GM of the cervical segment was significantly less than that of the thoracic segment. Indeed, NG2 cell density is reportedly quite variable throughout the CNS. For example, Wilson and colleagues (2006) reported an overall WM density of 171 NG2 cells/mm2 in the CNS. On the other hand, Levine and colleagues (1993) reported 14 NG2 cells per every 40,000 mm2 in the normal cerebellum. Such variability might reflect different region-specific functions for NG2 cells throughout the brain and spinal cord. Furthermore, no differences were observed in the WM versus GM distribution of any of the OL lineage cells examined in this study, suggesting that these cells may serve functional roles in both the WM and GM. It is interesting to note that the density of mature OLs and TrkB cells in the spinal cord was approximately 10 fold higher than the NG2 cell density. Barnabe-Heider et al. (2010) reported that the number of OLs in the spinal cord increased over time, while the number of OPCs did not. This activity could explain the higher density of mature OLs compared with OPCs observed in the present study.

5. Conclusions Though TrkB nonneuronal cells in the spinal cord were previously classified as OLs, little was known regarding their distribution and/or phenotype. The results of the present study revealed that the nonneuronal TrkB cells expressed Olig2, verifying that they are indeed OL lineage cells. These TrkB cells were distributed uniformly throughout the WM and GM of the adult spinal cord. The widespread distribution of TrkB cells suggests that BDNF and/or NT-4 utilization by these cells may occur throughout all regions of the adult spinal cord. Local sources of BDNF might originate from neurons (Lu et al., 2005), nonneuronal cells such as astrocytes (Friedman et al., 1998; Dougherty et al., 2000) and/or microglia (Dougherty et al., 2000). Consistent with previous findings (Horner et al., 2002; Staugaitis and Trapp, 2009), OPCs and mature OLs also showed equal distribution in the WM and GM.

Interestingly, the TrkB population did not overlap completely with either OPC or OL populations. TrkB cells exhibited morphological characteristics more similar to mature OLs, and TrkB was expressed by a substantial proportion of mature OLs, while few OPCs expressed detectable levels of TrkB. Only a small proportion of OPCs expressed detectable levels of TrkB, and this TrkB expressing pool of OPCs may represent a small population of OPCs preparing for renewal or the differentiation and maturation into OLs. While the majority of OLs expressed TrkB, a subset of mature OLs did not express TrkB, suggesting that some mature OLs reduce TrkB expression. Furthermore, approximately 17% of TrkB cells expressed neither NG2 nor CC1. This ‘TrkB only’ population may represent an immature stage of OL maturation that has down-regulated NG2, but not yet up-regulated mature markers such as CC1. Mature OLs were observed to form clusters with like cells in the spinal cord, and these clusters were more abundant in the GM. This cellular association suggests a communication between OL lineage cells and a supportive glial network to neurons. Indeed, a recent study demonstrated that mature OLs can provide lactate as an energy source to nearby neurons (Lee et al., 2012). Approximately 56-58% of ventral horn motor neurons associated with perineuronal cells of the OL lineage. The majority of nestling cells expressed both CC1 and TrkB. The role of these nestling OLs is unknown, but they may play a neuroprotective role (Taniike et al., 2002). Our data suggest that BDNF/NT-4 (known ligands of TrkB) may play a role in the oligodendrocyte-neuronal interactions. While OPC cell bodies were rarely observed nestling with neurons, their processes frequently were in close proximity to neuron cell bodies. The presence of NG2-immunoreactive processes found in close proximity to the neuronal cell bodies indicates a possible interaction between the OPCs and neuronal cell bodies in the spinal cord GM.

6. References

Barnabé-Heider F, Göritz C, Sabelström H, Takebayashi H, Pfrieger FW, Meletis K, Frisén J. 2010. Origin of new glial cells in intact and injured adult spinal cord. Cell Stem Cell, 7, 470-82.

Baumann N, Pham-Dinh D. 2001. Biology of oligodendrocyte and myelin in the mammalian central nervous system. Physiological Reviews, 81, 871-927.

Bergles DE, Roberts JD, Somogyi P, Jahr CE. 2000. Glutamatergic synapses on oligodendrocyte precursor cells in the hippocampus. Nature, 405,187-91.

Bhat RV, Axt KJ, Fosnaugh JS, Smith KJ, Johnson KA, Hill DE, Kinzler KW, Baraban JM. 1996. Expression of the APC tumor suppressor protein in oligodendroglia. Glia, 17, 169-74.

Butt AM, Duncan A, Hornby MF, Kirvell SL, Hunter A, Levine JM, Berry M. 1999. Cells expressing the NG2 antigen contact nodes of Ranvier in adult CNS white matter. Glia, 26,84- 91.

Cohen RI, Marmur R, Norton WT, Mehler MF, Kessler JA. 1996. Nerve growth factor and neurotrophin-3 differentially regulate the proliferation and survival of developing rat brain oligodendrocytes. Journal of Neuroscience, 16, 6433-42.

Colombo E, Cordiglieri C, Melli G, Newcombe J, Krumbholz M, Parada LF, Medico E, Hohlfeld R, Meinl E, Farina C. 2012. Stimulation of the neurotrophin receptor TrkB on astrocytes drives nitric oxide production and neurodegeneration. Journal of Experimental Medecine, 209, 521-535.

Coulibaly AP, Gannon SM, Hawk K, Walsh BF, Isaacson LG. 2013. Transection of preganglionic axons leads to CNS neuronal plasticity followed by survival and target reinnervation. Autonomic Neuroscience, 179, 49-59.

Coulibaly AP, Isaacson LG. 2012. Transient changes in spinal cord glial cells following transection of preganglionic sympathetic axons. Autonomic Neuroscience, 168, 32-42.

Coulibaly AP, Isaacson LG. 2013. Characterization of oligodendrocyte lineage cells in the normal adult rat spinal cord: Distribution, association with neurons, and TrkB expression Program 328.05. 2013 Neuroscience Meeting Planner. San Diego, CA. Society for Neuroscience, 2013. Online.

Dawson MR, Levine JM, Reynolds R. 2000. NG2-expressing cells in the central nervous system: are they oligodendroglial progenitors? Journal of Neuroscience Research, 61, 471-9.

Dawson MR, Polito A, Levine JM, Reynolds R. 2003. NG2-expressing glial progenitor cells: an abundant and widespread population of cycling cells in the adult rat CNS. Molecular and Cellular Neuroscience, 24, 476-88.

Dewar D, Underhill SM, Goldberg MP. 2003. Oligodendrocytes and ischemic brain injury. Journal of Cerebral Blood Flow and Metabolism, 23, 263-74.

Dimou L, Simon C, Kirchhoff F, Takebayashi H, Götz M. 2008. Progeny of Olig2-expressing progenitors in the gray and white matter of the adult mouse cerebral cortex. Journal of Neuroscience, 28, 10434-42.

Dougherty KD, Dreyfus CF, Black IB. 2000. Brain-derived neurotrophic factor in astrocytes, oligodendrocytes, and microglia/macrophages after spinal cord injury. Neurobiology of Disease, 7, 574-85.

Friedman WJ, Black IB, Kaplan DR. 1998. Distribution of the neurotrophins brain-derived neurotrophic factor, neurotrophin-3, and neurotrophin-4/5 in the postnatal rat brain: an immunocytochemical study. Neuroscience, 84,101-14.

Garraway SM, Turtle JD, Huie JR, Lee KH, Hook MA, Woller SA, Grau JW. 2011. Intermittent noxious stimulation following spinal cord contusion injury impairs locomotor recovery and reduces spinal brain-derived neurotrophic factor-tropomyosin-receptor kinase signaling in adult rats. Neuroscience, 199, 86-102

Gallo V, Mangin JM, Kukley M, Dietrich D. 2008. Synapses on NG2-expressing progenitors in the brain: multiple functions? Journal of Physiology, 586, 3767-81.

Hill RA, Natsume R, Sakimura K, Nishiyama A. 2011. NG2 cells are uniformly distributed and NG2 is not required for barrel formation in the somatosensory cortex. Molecular and Cellular Neuroscience, 46, 689-98.

Horner PJ, Power AE, Kempermann G, Kuhn HG, Palmer TD, Winkler J, Thal LJ, Gage FH. 2000. Proliferation and differentiation of progenitor cells throughout the intact adult rat spinal cord. Journal of Neuroscience, 20, 2218-28.

Horner PJ, Thallmair M, Gage FH. 2002. Defining the NG2-expressing cell of the adult CNS. Journal of Neurocytology, 31, 469-80.

Karadottir, R, Hamilton, NB, Bakiri, Y, Attwell, D. 2008. Spiking and nonspiking classes of oligodendrocyte precursor glia in CNS white matter. Nature Neuroscience, 11, 450-456.

Karoutzou G, Emrich HM, Dietrich DE. 2008. The myelin-pathogenesis puzzle in schizophrenia: a literature review. Molecular Psychiatry, 13, 245-60.

Kessaris N, Pringle N, Richardson WD. 2008. Specification of CNS glia from neural stem cells in the embryonic neuroepithelium. Philosophical Transactions of the Royal Society, 363, 71-85.

Koda M, Murakami M, Ino H, Yoshinaga K, Ikeda O, Hashimoto M, Yamazaki M, Nakayama C, Moriya H. 2002. Brain-derived neurotrophic factor suppresses delayed apoptosis of oligodendrocytes after spinal cord injury in rats. Journal of Neurotrauma, 19, 777-85.

Koenning M, Jackson S, Hay CM, Faux C, Kilpatrick TJ, Willingham M, Emery B. 2012. Myelin gene regulatory factor is required for maintenance of myelin and mature oligodendrocyte identity in the adult CNS. Journal of Neuroscience, 32, 12528-42.

Kukley M, Kiladze M, Tognatta R, Hans M, Swandulla D, Schramm J, Dietrich D. 2008. Glial cells are born with synapses. FASEB Journal, 22, 2957-69.

Kuzuhara S, Chou SM. 1980. Localization of the phrenic nucleus in the rat: a HRP study. Neuroscience Letters, 16, 119-124.

Lange S, Trost A, Tempfer H, Bauer H-C, Bauer H, Rohle E, Reitsamer HA, Franklin RJM, Aigner L, Rivera FJ. 2013. Brain pericytes plasticity as a potential drug target in CNS repair. Drug Discovery Today, 18, 456-63.

Lasiene J, Yamanaka K. 2011. Glial cells in amyotrophic lateral sclerosis. Neurology Research International, 2011, 718987.

Lee Y, Morrison BM, Li Y, Lengacher S, Farah MH, Hoffman PN, Liu Y, Tsingalia A, Jin L, Zhang PW and others. 2012. Oligodendroglia metabolically support axons and contribute to neurodegeneration. Nature, 487, 443-8.

Levine JM. 1994. Increased expression of the NG2 chondroitin-sulfate proteoglycan after brain injury. Journal of Neuroscience, 14, 4716-30.

Ligon KL, Kesari S, Kitada M, Sun T, Arnett HA, Alberta JA, Anderson DJ, Stiles CD, Rowitch DH. 2006. Development of NG2 neural progenitor cells requires Olig gene function. Proceeds of the National Academy of Sciences, 103, 7853-58.

Lu B, Pang PT, Woo NH. 2005. The yin and yang of neurotrophin action. Nature Reviews Neuroscience, 6, 603-14.

Macias M, Dwornik A, Ziemlinska E, Fehr S, Schachner M, Czarkowska-Bauch J, Skup M. 2007. Locomotor exercise alters expression of pro-brain-derived neurotrophic factor, brain- derived neurotrophic factor and its receptor TrkB in the spinal cord of adult rats. European Journal of Neuroscience, 25, 2425-44.

Maldonado PP, Velez-Fort M, Angulo MC. 2011. Is Neuronal communication with NG2 cells synaptic or extrasynaptic? Journal of Anatomy, 219, 8-17.

Mangin JM, Kunze A, Chittajallu R, Gallo V. 2008. Satellite NG2 progenitor cells share common glutamatergic inputs with associated interneurons in the mouse dentate gyrus. Journal of Neuroscience, 28,7610-23.

McCartney AM, Abejuela VL, Isaacson LG. 2008. Characterization of trkB immunoreactive cells in the intermediolateral cell column of the rat spinal cord. Neuroscience Letters, 440, 103- 108.

McKeon RJ, Silver J, Large TH. 1997. Expression of full-length trkB receptor by reactive astrocytes after chronic CNS injury. Experimental Neurology, 148, 558-567.

McTigue DM, Horner PJ, Stokes BT, Gage FH. 1998. Neurotrophin-3 and brain-derived neurotrophic factor induce oligodendrocyte proliferation and myelination of regenerating axons in the contused adult rat spinal cord. Journal of Neuroscience, 18, 5354-65.

McTigue DM, Tripathi RB. 2008. The life, death, and replacement of oligodendrocytes in the adult CNS. Journal of Neurochemistry, 107, 1-19.

McTigue DM, Wei P, Stokes BT. 2001. Proliferation of NG2-positive cells and altered oligodendrocyte numbers in the contused rat spinal cord. Journal of Neuroscience, 21, 3392- 400.

Miller RH. 2002. Regulation of oligodendrocyte development in the vertebrate CNS. Progress in Neurobiology, 67, 451-67.

Mullen RJ, Buck CR, Smith AM. 1992. NeuN, a neuronal specific nuclear protein in vertebrates. Development, 116, 201-211.

Nishiyama A, Komitova M, Suzuki R, Zhu X. 2009. Polydendrocytes (NG2 cells): multifunctional cells with lineage plasticity. Nature Reviews Neuroscience, 10, 9-22.

Nishiyama A, Watanabe M, Yang Z, Bu J. 2002. Identity, distribution, and development of polydendrocytes: NG2-expressing glial cells. Journal of Neurocytology, 31, 437-55.

Peters A, Palay SL, Webster H. 1970. The fine structure of the nervous system: neurons and their supporting cells. New York, NY: Oxford University Press.

Roy K, Murtie JC, El-Khodor BF, Edgar N, Sardi SP, Hooks BM, Benoit-Marand M, Chen C, Moore H, O'Donnell P, Brunner D, Corfas G. 2007. Loss of erbB signaling in oligodendrocytes alters myelin and dopaminergic function, a potential mechanism for neuropsychiatric disorders. Proceeds of the National Academy of Science, 104, 8131-36.

Salio C, Lossi L, Ferrini F, Merighi A. 2005. Ultrastructural evidence for a pre- and postsynaptic localization of full-length trkB receptors in substantia gelatinosa (Lamina II) of rat and mouse spinal cord. European Journal of Neuroscience, 22, 1951-66.

Sa-Pereira I, Brites D, Brito MA. 2012. Neurovascular unit: a focus on pericytes. Molecular Neurobiology, 45, 327-47.

Sargiannidou I, Vavlitou N, Aristodemou S, Hadjisavvas A, Kyriacou K, Scherer SS, Kleopa KA. 2009. Connexin32 mutations cause loss of function in Schwann cells and oligodendrocytes leading to PNS and CNS myelination defects. Journal of Neuroscience, 29, 4736-49.

Scarisbrick IA, Asakura K, Rodriguez M. 2000. Neurotrophin-4/5 promotes proliferation of oligodendrocyte-type-2 astrocytes (O-2A). Developmental Brain Research, 123, 87-90.

Shinoda Y, Yamaguchi T, Futami T. 1986. Multiple axon collaterals of single corticospinal axons in the cat spinal cord. Journal of Neurophysiology, 55, 425-48.

Shonberg DL, Goldstein EZ, Sahinkaya FR, Wei P, Popovich PG, McTigue DM. 2012. Ferritin stimulates oligodendrocytes genesis in the adult spinal cord and can be transferred from macrophages to NG2 cells in vivo. Journal of Neuroscience, 32, 5374-84.

Skup M, Dwornik A, Macias M, Sulejczak D, Wiater M, Czarkowska-Bauch J. 2002. Long- term locomotor training up-regulates TrkB (FL) receptor-like proteins, brain-derived

89 neurotrophic factor, and neurotrophin 4 with different topographies of expression in oligodendroglia and neurons in the spinal cord. Experimental Neurology, 176, 289-307.

Staugaitis SM, Trapp BD. 2009. NG2-positive glia in the human central nervous system. Neuron Glia Bioliogy, 5, 35-44.

Takasaki C, Yamasaki M, Uchigashima M, Konno K, Yanagawa Y, Watanabe M. 2010. Cytochemical and cytological properties of perineuronal oligodendrocytes in the mouse cortex. European Journal of Neuroscience, 32, 1326-36.

Taniike M, Mohri I, Eguchi N, Beuckmann CT, Suzuki K, Urade Y. 2002. Perineuronal oligodendrocytes protect against neuronal apoptosis through the production of lipocalin-type prostaglandin D synthase in a genetic demyelinating model. Journal of Neuroscience, 22, 4885- 96.

Van't Veer A, Du Y, Fischer TZ, Boetig DR, Wood MR, Dreyfus CF. 2009. Brain-derived neurotrophic factor effects on oligodendrocyte progenitors of the basal forebrain are mediated through trkB and the MAP kinase pathway. Journal of Neuroscience Research, 87, 69-78.

Vostrikov VM, Uranova NA, Orlovskaya DD. 2007. Deficit of perineuronal oligodendrocytes in the prefrontal cortex in schizophrenia and mood disorders. Schizophrenia Research, 94(1-3), 273-80.

VonDran MW, Clinton-Luke P, Honeywell JZ, Dreyfus CF. 2010. BDNF+/- mice exhibit deficits in oligodendrocyte lineage cells of the basal forebrain. Glia, 58, 848-56.

VonDran MW, Singh H, Honeywell JZ, Dreyfus CF. 2011. Levels of BDNF impact oligodendrocyte lineage cells following a cuprizone lesion. Journal of Neuroscience, 31, 14182- 90.

Wilson HC, Scolding NJ, Raine CS. 2006. Co-expression of PDGF α receptor and NG2 by oligodendrocytes precursors in human CNS and multiple sclerosis lesions. Journal of Neuroimmunology, 176, 162-73.

Zimmer MB, Nantwi K, Goshgarian HG. 2007. Effect of spinal cord injury on the respiratory system: basic research and current clinical treatment options. Journal of Spinal Cord Medicine, 30, 319-330

Chapter 4 The effect of peripheral injury on oligodendrocyte subpopulations in the rat spinal cord

Abstract: Glial cells in the intermediolateral cell column (IML) of the spinal cord exhibited robust plasticity following the transection of preganglionic axons in the periphery. One observed change was an increase in the number of TrkB immunoreactive (-ir) oligodendrocytes (OL). Because TrkB can be expressed by both NG2 (marker for OL progenitors or OPCs) cells and CC1 (marker for mature OLs) cells, it was not known which of the OL subpopulations was responsible for the increased number of TrkB cells. The objectives of the present study were to determine whether spinal cord OLs that express NG2 and CC1 were affected by preganglionic transection, and then to relate these findings to our previous results that showed an increased number of TrkB cells. At 1 week after injury, though the number of CC1 cells was unchanged, the number of NG2 cells was increased in both the IML and adjacent lateral funiculus (LF). At 3 weeks, the number of NG2 cells, TrkB cells, and CC1 cells in the spinal cord were similar to controls. While at 16 weeks no changes were observed in NG2 or TrkB cells when compared to controls, the number of CC1 cells in the LF was decreased. NG2+/TrkB+ cells were increased at 1 week and 3 weeks while no changes were observed in CC1+/TrkB+ cells at any time point. The results of this study reveal that peripheral injury led to transient plasticity in OPCs, including those that expressed TrkB, in the spinal cord, and that these changes were localized to the IML and LF, locations specific to the injured neurons. Though OPCs were increased, the decrease in the mature OL population at 16 weeks suggested that any newly formed OLs may not have survived over a long time period after injury.

1. Introduction Our lab has previously demonstrated that transection of preganglionic axons innervating the superior cervical ganglion leads to changes in the glial cells in the intermediolateral cell column (IML) of the spinal cord (Coulibaly and Isaacson, 2012), the location of the parent cell bodies. Such changes are thought to be the result of retrograde influences of the peripheral injury on the local environment of the injured cell bodies. Indeed, in the vicinity of the injured cell bodies in the IML, astrocytes and microglia showed reactive morphology, along with close associations with the injured cell bodies, and microglia showed significant increases in number (Coulibaly and Isaacson, 2012). In addition, the number of oligodendrocytes (OL) that expressed TrkB, the cognate receptor for the neurotrophin brain derived neurotrophic factor (BDNF) also was significantly increased after injury (Coulibaly and Isaacson, 2012). Though the TrkB OL subpopulation responded to the injury, it was not clear whether other cells in the OL cell lineage were affected by the retrograde influences of peripheral injury. Two main OL subpopulations, progenitor and mature, have been described in the adult CNS (Nishiyama et al., 2009). OL progenitor cells (OPCs) are cells in the CNS that retain the ability to produce additional OLs upon stimulation (McTigue et al., 1998). These cells are identified by the expression of the proteoglycan nerve/ glia (NG2; Nishiyama et al., 2009), a protein proposed to play a role in cell migration and proliferation (Trotter et al., 2010). in vivo, OPCs can give rise to mature OLs (McTigue et al., 1998; Trotter et al.,2010), yet these cells have also been shown to play an important role in the modulation of neural outputs (Nishiyama et al., 2009), from integration of excitatory inputs (Bergles et al., 2000) to the formation of functional synapses (Kukley et al, 2008). The mature OL subpopulation expresses the adenomatous polyposis coli (APC; commonly known as CC1; Lang et al., 2013) antigen and myelin proteins (Kessaris et al, 2008; Nishiyama et al., 2009). These cells myelinate central axons (Lang et al., 2013) and provide lactate as an energy source to neurons (Lee et al., 2012). In the present study, we used Olig2, a transcription factor expressed in all cells in the OL cell lineage (Barnabe-Heider et al., 2010), as well as NG2 and CC1 as markers for OL subpopulations. We set out to determine whether the transection of the preganglionic axons of the cervical sympathetic trunk (CST) affects either of these OL populations in the spinal cord. Because we have already shown that TrkB OLs are increased following the injury (Coulibaly and Isaacson, 2012), and that TrkB is expressed in both NG2 cells as well as mature OLs (Chapter

3), we investigated whether the expression of TrkB in either of these populations was influenced by the injury.

2. Materials and Methods 2.1. Animal surgery and tissue collection Young adult (3 months of age) female Sprague Dawley rats (Harlan Labs, Indianapolis, IN) were housed in the Miami University Animal Facilities in a 12:12 light:dark environment at regulated temperature. Animals were anesthetized using the inhalant isoflurane (2.5%). A ventral incision approximately 3 cm in length was made on the neck region of the animal. The cervical sympathetic trunk (CST) was exposed and gently separated from surrounding tissue and transected approximately 2 mm from the entry into the superior cervical ganglion (SCG; Sun and Zigmond, 1996). After the transection, the proximal stump was placed carefully back into original position in close proximity to the distal stump. The procedure was repeated on the other side. The incision was closed using sutures and tissue glue (Nexaband, Phoenix, AZ). The CST was exposed, but not transected, in sham animals. The CST was bilaterally transected and animals survived for 1 week (n=18), 3 weeks (n=20) or 16 weeks (n=12) and compared with shams in which the CST was dissected and exposed but was not cut (1 week, n=16; 3 week, n=14; 16 week, n=12). All methods used in this study were approved by the Miami University Institutional Animal Care and Use Committee Protocol 825 and efforts were taken to minimize discomfort and pain to the animals and to minimize the number of animals used in the study. After the survival period, the animals were transcardially perfused with 0.9% saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB). Spinal cord and SCG tissues were removed, and stored in 0.1M PB, cryoprotected by infiltrating with 30% sucrose in 0.1M PB, embedded in optimal cutting temperature medium (Ted Pella Inc.), and cut using a MICROM HM 550 series cryostat. Because the robust glial plasticity following CST transection observed in a previous study was confined primarily to T1 of the spinal cord (Coulibaly and Isaacson, 2012), all analyses in this study were carried out at the T1 level, which was identified by counting the nerve roots extending from the cord and was verified by noting its location immediately caudal to the cervical enlargement. Coronal sections of spinal cord (18 m) were mounted on Fisherbrand Superfrost microscope slides, incubated overnight in 0.1M PBS-0.2% Triton-X solution, blocked with

94 normal donkey serum, and then incubated for 48 hrs at 4oC with a cocktail of goat anti-Olig2 (1:100; R&D Systems, AF2418), mouse anti-NG2 (1:200; Millipore, MAB5384; to identify oligodendrocyte precursors), mouse anti-CC1 (1:500; Abcam, ab16794; to identify mature oligodendrocytes), and rabbit anti-TrkB (Santa Cruz Biotechnologies, SC-12; to identify a subpopulation of oligodendrocytes). Following a series of rinses, sections were incubated for 2 hours in AlexaFluor conjugated antibodies (1:200; Molecular Probes) directed against the primary antibody host. Sections then were coverslipped using fluorescent mounting medium with 4',6-diamidino-2-phenylindole (DAPI; Vectashield). The following antibody combinations were used for the analyses described below: Olig2 with Alexa 488 only; NG2 Alexa 488/TrkB Alexa 594; CC1 Alexa 488/ TrkB Alexa 594; NG2 with Alexa 488 only; CC1 with Alexa 488 only; TrkB with Alexa 594 only.

2.2. Overall methods and data analysis All images were captured with a Zeiss 720 laser scanning confocal microscope. The left and right IML from each animal were considered separate cases since the peripheral axons on each side of the animals were transected and the response to peripheral transection in the spinal cord has been determined to be unilateral. Because the controls from each survival time point were found to be significantly different from each other, each injury group was compared to its age matched control group. For each statistical analysis described below, a mean was obtained for the control group and all values (control and injury) were expressed as proportion of the control mean at each survival time point, square root transformed, and comparisons between treatments and age-matched controls were made using the Mann-Whitney test. Significance was ascribed at p < 0.05. The graphs shown in Figures 2, 5, 6 depict square root transformed data.

2.3. Identification of OL lineage cells As described in Chapter 3, Olig2 immunoreactivity was localized to the nucleus of OL cells in the spinal cord (Fig. 2.A.). NG2 immunoreactivity was localized to the processes of oligodendrocyte precursor cells as well as around the DAPI positive nucleus (Figs. 3, 4). Because NG2 is also a marker for pericytes, any cells with a quarter shaped nucleus and/or associating with blood vessels (McTigue et al. 2001; Ligon et al., 2006) were excluded from our counts. Mature oligodendrocytes were characterized by a ring of CC1 imunoreactivity around

95 the nucleus of the cells (Figs. 3, 4; Ch.2, this thesis). Similar to CC1 oligodendrocytes, TrkB oligodendrocytes, a mixed pool of mature and progenitor OLs (Chapter 3), also were characterized by ring of TrkB immunoreactivity around the nucleus (McCartney and Isaacson, 2008; Coulibaly and Isaacson, 2012; Chapter 3).

2.4. Analysis of Olig2 cells in the IML Olig2 is a transcription factor that is expressed by cells in the OL lineage (Nishiyama et al., 2009), therefore, its expression was used to determine whether peripheral axon transection affected the overall oligodendrocyte population. The number of Olig2 positive cells was quantified at each survival time point from two confocal images of the IML per case which were acquired at 200X magnification. Using Image Pro 6.3 Software, the images were enlarged to 200% on the computer monitor. Every cell colocalizing Olig2 and DAPI within the field of view (which encompassed an area of approximately 0.22 mm2) was tallied and a mean was obtained from the two images to generate a mean number of cells for that case. These data were used to generate the graph shown in Figure 2.B.

2.5. Analysis of changes in OL lineage cells The number of cells expressing NG2 and CC1 was analyzed in the entire section of the spinal cord (denoted as “overall”) as well as in specific spinal cord subregions. Images of the entire spinal cord sections were acquired at 200X using the “tile” feature of the Zeiss 710 laser scanning microscope and Zen 2009 software. Tiling of the spinal cord allowed for the acquisition of a patchwork of images of the spinal cord that were ‘stitched’ together. The tiled images were acquired in a 4 X 7 configuration. Using the software Image Pro 6.3, these images were used to quantify the number of NG2-, and CC1- immunoreactive (-ir) cells present in the spinal cord as shown in Figure 5A.-D, as well as cells that coexpressed NG2 and TrkB or CC1 and TrkB as shown in Figure 6. The number of NG2-ir and CC1-ir profiles (referred to in this study as NG2 and CC1 cells) as well as the number of NG2 cells expressing TrkB (NG2+/TrkB+) and CC1 cells expressing TrkB (CC1+/TrkB+) were determined within the white matter (WM) and gray matter (GM) areas of the spinal cord from tiled images that were magnified to approximately 500X on a computer monitor. A standardized measuring box (approx. 0.1 mm2; Fig. 1, white dashed boxes)

96 was used to characterize the number and location of the cell populations. The size of the box was determined using the approximate width of the dorsal horn as a guide. The box was placed over 14 gray and white matter regions of the spinal cord. In the WM, the box was placed over the lateral funiculus (LH) and the ventral funiculus (VF) just ventral to the ventral horn on both sides, and a second central region of the VF, as well as a central region of the dorsal funiculus (DF). In the GM, the box was placed bilaterally over the dorsal horn (DH) just ventral to Lamina II, the lateral horn (LH) and ventral horn (VH) and a region adjacent to central canal (CC; Fig. 1). Counts were combined from left and right sides for bilateral regions while the VF was totaled from the two sides and the third, more central region. To obtain the number of cells in GM and WM, data from GM and WM subregions were averaged to obtain a value for the overall GM and WM, respectively. The values were converted to mm2 for comparison across regions. These data were used to generate the graphs shown in Figure 5. To calculate the number of NG2+/TrkB+ and CC1+/TrkB+ cells, the percent overlap of the populations was determined and these data are shown in the scatter plots and graphs in Figure 6. Though the location of the preganglionic neuronal cell bodies in the IML of the spinal cord has been clearly defined (Rando et al., 1981; Strack et al., 1988; Hosoya et al., 1991), the location of the injured axons is not as clearly defined in the LF of the spinal cord. Therefore, to increase the likelihood that the analysis of the LF included the injured axons, images that encompassed a larger field of view of the LF were acquired using the 400X oil objective, using both the “Z-stack” and “tile” features of the microscope (resulting in a 0.2 mm2 area; Fig. 1, orange dashed box). Z-stacks (6 optical sections at 3 µm intervals) of the LF were acquired in a 2 X 2 tiled configuration. Using a composite image of collapsed Z-stacks (Zen 2009 software), the number of NG2, CC1 and TrkB cells in the entire field of view (which encompassed the LF), was tallied using Image Pro 6.3 software. These data were used to generate the graph shown in Figure 5.E.

2.6. Analysis of white matter-gray matter distribution of OL lineage cells in the spinal cord In Chapter 2 we reported that NG2, CC1, and TrkB cells were uniformly distributed in the normal adult spinal cord. This analysis compared the distribution of cells between GM and WM subdivisions, and the results revealed that the density of these OL subpopulations was similar between the GM and WM. Therefore, we examined whether the uniformity in

97 distribution was affected by the peripheral injury. Using ‘tiled’ images of the entire spinal cord obtained at 200X, the number of NG2, CC1, and TrkB cells were counted in each subregion of the spinal cord. Counts were combined from left and right sides for bilateral regions while the VF was totaled from the two sides and the third, more central region. To obtain the number of cells in GM and WM, data from GM and WM subregions were averaged to obtain a value for the overall GM and WM, respectively. Within each treatment, GM values were compared to the WM values using the Mann Whitney test. These data were used to generate the graphs shown in Figure 7.

3. Results 3.1. Olig2-ir cells in the IML were unchanged following CST transection Our lab has reported an increase in the number of TrkB-ir OLs in the IML at 1 week after injury (Coulibaly and Isaacson, 2012). These results led to the inquiry as to whether these changes were observed in the general population of OLs in the IML or was specific to one OL subpopulation. Olig2, a transcription factor expressed by all OL lineage cells (Nishiyama et al., 2009), was used as a marker to label all OLs in the IML. Cells exhibiting Olig2-ir nuclei were observed throughout the spinal cord (Fig. 2.A). These cells were easily identifiable in both GM and WM. Analysis of the number of Olig2-ir cells in the spinal cord revealed no significant changes in the IML at any of the survival time points (Fig. 2.B.).

3.2. Changes in OL lineage cells in the IML and LF following CST transection As described in Chapter 3 of this thesis, NG2 immunoreactivity was observed in the cell body and extended into the processes of the OPCs (Fig. 3, 4). These cells were found in both theGM and WM of the spinal cord. Using the rectangular measuring boxes that encompassed an area of 0.1mm2, the typical number of NG2 cells in the overall spinal cord ranged from 36-54 cells/mm2 in the controls (Fig. 7.A). No significant changes were observed in the overall number of NG2 cells in the spinal cord when comparing control and injury groups at each of the survival time points (Fig. 5.A.). When considering the number of NG2 cells in the overall GM and overall WM, the typical number of cells ranged from 40-79 cells/mm2 in the GM and 34-58 cells/mm2 in the WM of the control cases and no changes were observed following injury (Fig. 5.A.).

Most GM subregions showed no significant differences in NG2 cells following injury (Fig. 5.C.). However a significant 34% increase in the number of NG2 cells was observed in the IML (Fig. 5.C.; cont=44 cells/mm2; 1 wk=79 cells/mm2) at 1 week following injury. Interestingly, the IML is the specific GM subregion that houses the cell bodies of the injured neurons (Rando et al., 1981; Coulibaly and Isaacson, 2012), suggesting that the response was specific to the location of the injured cell bodies. A similar approach (use of 0.1mm2 rectangular boxes) was used to determine whether specific WM spinal cord subregions were affected by CST transection. No changes in the number of NG2 cells were observed in any of the WM subregions, including the LF, the location of the injured axons just as they emerge from their cell bodies in the IML, at any survival time point (Fig. 5.C.). However, when a larger sample area of the LF was analyzed (field of view=0.2 mm2), a significant 14% increase in the number of NG2 cells was observed (cont=48 cells/mm2; 1 wk=64 cells cells/mm2; Fig. 5.E.), suggesting that the rectangular measuring box that included the LF may not have encompassed all of the injured axons and, therefore, may have not been large enough to detect changes in the number of NG2 cells. A similar series of analyses was carried out for the CC1 cells. Using the 0.1mm2 rectangular measuring boxes, the typical number of CC1 cells in the overall spinal cord ranged from 226-238 cells/mm2 in the controls (Fig. 7.B). No significant changes were observed in the number of CC1 cells in the overall spinal cord when comparing control and injury at each of the survival time points examined (Fig. 5.B.). In addition, the typical number of cells in GM ranged from 250-310 cells/mm2 and in the WM ranged from 197-296 cells/mm2 in the control cases. No changes in the number of CC1 cells in GM or WM were observed following injury (Fig. 5.D.). Analysis of the number of CC1 cells in the larger sample area revealed no significant changes in the number of CC1 cells at either 1 or 3 weeks after injury. However, a significant 22% decrease in the number of CC1 cells was observed in the LF at 16 weeks after injury (from 663 cells to 448 cells per mm2; Fig. 5.E.), when the larger sample area was analyzed. A significant increase in the number of TrkB OLs in the IML of the spinal cord at 1 week after injury was reported previously, with numbers similar to control values at 3 and 16 weeks after injury (Coulibaly and Isaacson, 2012). Using the larger sample area for direct comparison to the number of NG2 and CC1 cells, the number of TrkB cells in the LF was analyzed. A

99 significant 11% increase in the number of TrkB cells was observed (cont=468 cells/mm2; 1 wk= 581 cells/mm2; Fig. 5.E.). No changes were observed at 3 or 16 weeks after injury.

3.3. Analysis of NG2 and CC1 cells that expressed TrkB Previous colocalization studies from our lab revealed that ~1.5% of NG2 cells and ~56% of CC1 cells express the full length TrkB receptor in the control spinal cord (Ch2; this thesis). Because TrkB OLs were increased in the IML (Coulibaly et al., 2012) and the LF at 1 week after injury (Fig. 5.E.), we investigated whether CST transection affected the number of OPCs and mature OLs that expressed TrkB and found that NG2+/TrkB+ cells were increased by 2-fold in the overall section of the spinal cord at 1 week after injury (Fig. 6.A.), from ~1% NG2 cells expressing TrkB in the control to ~3% in the injured cases. However, analysis of the overall GM and WM as well as the specific spinal cord subregions yielded no significant changes in NG2+/TrkB+ cells following injury. Though NG2+/TrkB+ cells were increased in the overall section of the spinal cord, these cells were absent in many of the spinal cord subregions, particularly in the control cases, making it impossible to carry out comparisons using standard statistical methods. Therefore, scatter plots were generated showing the number of cases within each treatment in which NG2+/TrkB+ cells were present. The data depicted in scatter plots by WM (Fig. 6.C.) and GM (Fig. 6.D.) subregions revealed trends for an increase in NG2+/TrkB+ cells in the WM subregions. At 1 week after injury, all WM subregions showed a trend toward increased numbers of NG2+/TrkB+ cells (Fig. 6.C.). This trend was also observed in all WM subregions at the 3 week time point (Fig. 6.C.). At 16 weeks, the dorsal funiculus (DF) still showed a trend for an increase while cells in the lateral funiculus (LF) were similar to control values, and those in the ventral funiculus (VF) trended toward a possible decrease. No specific trends were observed in the GM subregions (Fig. 6.D.). Unlike the NG2+/TrkB+ cell population, all cases showed an abundant number of cells coexpressing CC1 and TrkB (CC1+/TrkB+; Fig. 6.B.). No changes were observed in the number of cells coexpressing both markers in the spinal cord or its subregions after injury at any of the time points studied (Fig. 6.B.).

3.4. Comparison of cell density between GM and WM following injury

100

In a previous study we demonstrated that NG2, CC1, and TrkB cells in the uninjured rat spinal cord were distributed uniformly when cell density comparisons between GM and WM were made (Chapter 3). Here, we investigated whether this uniform distribution was affected by CST transection. Comparisons of GM and WM cell densities at each survival time point revealed that CST transection had no effect on the density of NG2 (Fig. 7.A.), CC1 (Fig. 7.B.), or TrkB (Fig. 7.C.) cells when the overall GM and WM were compared. Similarly, the cells coexpressing NG2 and TrkB (NG2+/TrkB+) or CC1 and TrkB (CC1+/TrkB+) were uniformly distributed in the control cases when overall GM and WM cell densities were compared. Following CST transection, the only change observed was in the density of NG2+/TrkB+ cells in the WM when compared with GM at the 3 week survival time point (WM=5%; GM=1%; Fig. 7.D.). The density of CC1+/TrkB+ cells was similar in the GM and WM in the control cases and no changes were detected following injury (Fig. 7.E.).

101

NG2/ DAPI

DF DH

IML LF

VFa

VFb

Figure 1

102

Figure 1: Areas of quantification in the spinal cord of the rat. Confocal micrograph of a hemisection of the rat spinal cord at level T1 denoting the areas that were used to analyze oligodendrocyte lineage cells. The white boxes represent the 0.1 mm2 areas used to analyze spinal cord subregions. The orange box represents the larger sample area (0.2 mm2) of the lateral funiculus (LF) that was included in the analysis. Scale bar = 200µm. DF= dorsal funiculus, DH= dorsal horn, IML= intermediolateral cell column, CC= central canal, VH= ventral horn, VF= ventral funiculus. DAPI (blue) shows cellular nature of labeled profiles.

103

A. Olig2/ DAPI cont Olig2/ DAPI 1 wk

IML

B. Olig2 cells in the IML of the spinal cord

1.4 )

rt 1.2 sq 1.0 0.8 0.6 0.4

0.2 Proportion of ( control 0.0 1 wk 3 wk 16 wk

Figure 2

104

Figure 2: Analysis of the number of Olig2 cells in the intermediolateral cell column (IML) following injury. A. Confocal micrographs of Olig2-ir cells (white arrows) in the IML of the spinal cord in control and at 1 week (1wk) after injury. Scale bar for all images = 100µm. B. Graph representing the number of cells expressing Olig2 in the IML of the spinal cord at each survival time point. No significant changes were observed in the number of cells expressing Olig2 in the IML of the spinal cord at 1 week, 3 weeks, or 16 weeks after transection of the cervical sympathetic trunk when compared to age-matched controls. DAPI (blue) shows cellular nature of labeled profiles.

105

NG2 / DAPI CC1 / DAPI TrkB / DAPI

A. E. I.

Sham

B. F. J.

1 1 wk

C. G. K.

3 3 wk

D. H. L. 16 16 wk

Figure 3

106

Figure 3: Pattern of immunoreactivity for the three identified subpopulations of oligodendrocyte lineage cells in the intermediolateral cell column (IML) in the controls and following injury. Confocal micrographs from the IML show cells expressing the oligodendrocyte progenitor marker NG2 (white arrows in A.-D.), the mature marker CC1 (white arrows in E.-H.) and the tyrosine kinase B receptor (TrkB) (white arrows in I. - L.) across the survival time points that were analyzed in this study. Few changes in the pattern of immunoreactivity were observed. Scale bar for all images = 25µm. DAPI (blue) shows cellular nature of labeled profiles.

107

NG2/ DAPI CC1/ DAPI TrkB/ DAPI

A. E. I. Control

B. F. J.

1 1 wk

C. G. K.

3 3 wk

D. H. L. 16 16 wk

Figure 4

108

Figure 4: Pattern of immunoreactivity for the three identified subpopulations of oligodendrocyte lineage cells in the lateral funiculus (LF) in the controls and following injury. Confocal micrographs show cells expressing the OL progenitor marker NG2 (white arrows, A. - D.), the mature marker CC1 (white arrows, E.-H.) and tyrosine kinase B receptor (TrkB) (white arrows, I.-L.) across the survival time points that were analyzed in this study. Few changes in the pattern of immunoreactivity were observed. Scale bar for all images = 25µm.

109

A. B. Number of NG2 cells in spinal cord Number of CC1 cells in spinal cord

1.5 1.5

)

sq (

1.0 1.0 control control

of of 0.5 0.5

roportion Proportion of ( control

P 0.0 0.0 overall GM WM overall GM WM 1 wk 3 wk 16 wk 1 wk 3 wk 16 wk C. Number of NG2 cells in spinal cord D. 2.0 Number of CC1 in spinal cord ) subregions

rt 1.5 subregions

) rt sq 1.5 * sq 1.0 1.0

0.5 0.5

Proportion of ( control 0.0 0.0 DH IML CC VH DF LF VF Proportion of ( control DH IML CC VH DF LF VF GM WM GM WM 1 wk 3 wk 16 wk 1 wk 3 wk 16 wk

E. OL lineage cells in larger sample size of LF

1.5 )

rt * *

sq ( 1.0

* control 0.5

0.0

Proportion of NG2 CC1 TrkB

1 wk 3 wk 16 wk Figure 5

110

Figure 5: Changes in OL lineage cells in the intermediolateral cell column (IML) and the lateral funiculus (LF) after peripheral axon injury. No changes were observed in the number of NG2 cells (A.) or CC1 cells (B.) in the overall spinal cord, or in the overall gray matter (GM) or white matter (WM) at any time point after injury. C. Analysis of spinal cord subregions revealed a significant increase in the number of NG2-ir cells in the IML at 1 week (1 wk) following injury while no significant changes were observed in the number of CC1-ir cells (D.). E. Analysis of the lateral funiculus (LF) using the larger sample area revealed a significant increase in the number of NG2 and TrkB cells at 1wk. Although no changes were observed at 1 and 3 weeks after injury, the number of CC1-ir cells was decreased at the16 week time point. DF=dorsal funiculus, DH=dorsal horn, CC=central canal, VH=ventral horn, VF=ventral funiculus. *, p < 0.05.

111

A. B.

Colocalization of NG2 and TrkB Colocalization of CC1 and TrkB * 1.5 2.5

2.0 1.0 1.5

1.0 0.5

0.5 Proportion of (sqcontrol rt) Proportion of (sqcontrol rt) 0.0 0.0 overall GM WM overall GM WM 1 wk 3 wk 16 wk 1 wk 3 wk 16 wk

C. D. NG2+/TrkB+ colocalization in WM NG2+/TrkB+ colocalization in GM 100 subregions 100 subregions DH cont DH injury DF cont DF injury IML cont IML inury

LF cont LF injury cells cells CC cont CC injury VF cont VF injury

75 75

colocalized colocalized 50 50

25 25

Percent cases with Percent cases with Percent cases with Percentwith cases 0 0 1 wk 3 wk 16 wk 1 wk 3 wk 16 wk

Figure 6

112

Figure 6: Analysis of the expression of NG2 or CC1 with TrkB. A. The number of NG2+/TrkB+ cells in the overall spinal cord was increased at 1 week following injury. However no changes were observed in this population in the overall gray matter (GM) or white matter (WM). B. The number of CC1+/TrkB+ cells in the spinal cord was unchanged following injury. C. Scatter plot representation of the proportion of cases having NG2+/TrkB+ cells in WM subregions. Though not significant, all three WM subregions (DF, LF, VF) trended toward an increase at 1 week (1 wk; squares) and 3 weeks (3 wk; squares) after the injury compared to controls (circles), while at 16 weeks (16 wk), injury cases (squares) trended toward a decrease when compared to controls (circles). D. No discernable trends were observed in any of the GM subregions in the number of cases with NG2+/TrkB+. DF=dorsal funiculus, DH=dorsal horn, IML=intermediolateral cell column, LF=lateral funiculus, CC=central canal, VH=ventral horn, VF=ventral funiculus. *, p < 0.05.

113

A. NG2 cell density in GM and WM

90 2

80 mm 70 60 50 40 30 20 D. NG2+/TrkB+ cell density in GM and WM 10

Number Number NG2of cells per 8 0 7 * cont 1 wk 3 wk 16 wk 6 GM WM 5 B. cells CC1 cell density in GM and WM 4

400 3

2 % total % 350 2 300 1 250 0 cont inj cont inj cont inj 200 1 wk 3 wk 16 wk 150 100 GM WM 50 E. Number CC1 in Number of cells mm CC1+/TrkB+ cell density in GM and WM 0 70 cont 1 wk 3 wk 16 wk 60 GM WM 50

C. TrkB cell density in GM and WM cells 40 300

2 % total %

mm 250 20 10 200 0 150 cont inj cont inj cont inj

100 1 wk 3 wk 16 wk GM WM 50

Number TrkB Number perof cells 0 cont 1 wk 3 wk 16 wk GM WM

Figure 7

114

Figure 7: Analysis of the density of OL lineage cells in white matter (WM) vs gray matter (GM) following injury. Comparisons of WM and GM density of NG2 cells (A.), CC1 cells (B.), and TrkB cells (C.) in the spinal cord revealed no significant differences at any time points examined, while NG2+/TrkB+ cells were increased in the WM at 3 weeks (3 wk) (D.). E. No changes were observed in the distribution of CC1+/TrkB+ cells in the spinal cord GM compared with WM. *, p < 0.05.

115

4. Discussion The results of the present study demonstrate changes in OLs subpopulations of the spinal cord as a result of peripheral axon transection. Specifically, we provide evidence that, although there were no changes in the overall OL population (Olig2-ir cells), the number of NG2 cells was increased specifically in the IML and LF at 1 week after injury. Furthermore, the number of NG2 cells expressing TrkB in the overall spinal cord at 1 week after injury was increased and although the change could not be attributed to a specific subregion, there were trends for increased NG2+/TrkB+ cells in each of the WM subregions that were analyzed. Interestingly, the density of NG2 cells expressing TrkB in the WM of the spinal cord was significantly increased at 3 weeks after injury. Our lab has previously demonstrated that TrkB OLs are increased in the IML at 1 week after peripheral injury (Coulibaly and Isaacson, 2012). In the present study, we provide evidence that TrkB OLs are also increased in the LF of the spinal cord at the same time point. These changes were shown to have subsided by 3 weeks and no changes were observed at 16 weeks after injury. Finally, while no changes were observed in the number of CC1 cells in the IML at any of the time points analyzed, a significant decrease in these cells was observed in the LF at 16 weeks after injury. No changes were observed in CC1 cell distribution or density of CC1+/TrkB+ cells at any of the time points after injury.

4.1. Peripheral axon injury leads to NG2 cell plasticity NG2 cells were increased in the IML and LF of the spinal cord at 1 week after injury. These regions contain the cell bodies (IML) and axons (LF) of the injured neurons. The fact that no other subregions of the spinal cord showed changes in NG2 cells provides evidence for the specificity of the response and supports the idea that retrograde signals from the injured axons had a profound effect on the NG2 population in the local vicinity of the injured neuronal cell bodies. The plasticity of NG2 cells has also been demonstrated in several different experimental paradigms. Stress (Siefi et al., 2014) and injury (McTigue et al., 1998; Barnabe-Heider et al., 2010) can increase the proliferation rate of NG2 cells. In our model, the increase in the number of NG2 cells observed at 1 week after injury could be the result of cell migration and/or increased proliferation. Because no changes were

116 observed in the distribution of NG2 cells in spinal cord subregions or in the NG2 cell density throughout the spinal cord at this time point, there is little evidence to support the idea that NG2 cell migration took place at T1 level of the spinal cord. However, because the other levels of the spinal cord were not examined, migration from other segments is a possibility. Since we have no evidence for cell migration, it is more likely that NG2 cells were increased due to proliferation. NG2 cell proliferation has been reported in the spinal cord following spinal cord injury (Barnabe-Heider et al., 2010), suggesting that injury does indeed stimulate the proliferation of NG2 cells. In addition, Barnabe-Heider and colleagues (2010) demonstrated that the increase proliferation of NG2 cells led to an increase in number of mature OLs present in the spinal cord. Therefore, it is possible that in our model the increase in NG2 cells led to an increase in the mature OL population. The NG2 cell population has been shown to be diverse. It has been proposed that only a small proportion (~3%) of the cells are actively engaged in proliferation and/or differentiation, while the remaining cells play an alternate role in spinal cord maintenance (Horner et al., 2000; Horner et al., 2002). Furthermore, these cells have been shown to be responsive to growth factors. McTigue and colleagues (1998) demonstrated that NG2 cells proliferate in the injured spinal cord in the presence of BDNF and NT3. We previously proposed that NG2 cells expressing TrkB may be the subpopulation of NG2 cells that actively proliferate and differentiate in the spinal cord (Chapter 3). Indeed, the 2 fold increase in the number of NG2 cells expressing TrkB suggests that the NG2+/TrkB+ subpopulation may comprise the small proportion of NG2 cells actively proliferating and/or differentiating. Upon stimulation by either changes in BDNF expression in the spinal cord and/or retrograde injury factors from the periphery, these cells reenter the cell cycle to produce more mature OLs. As these cells differentiate and take on the morphology of mature OLs, they down regulate their expression of NG2 (Nishiyama et al., 2009) but retain their expression of TrkB, accounting for the increase in TrkB cells observed in the spinal cord after injury. The density of NG2/TrkB cells was increased at 1 week following injury. Although we observed no changes in mature BDNF expression at 1 week, the disconnection of the preganglionic axons with the superior cervical ganglion, a BDNF source (Causing et al., 1997) for the preganglionic neurons, would limit BDNF availability following injury. The increase in TrkB expression in the NG2 cell population at 1 week might serve as a compensatory

117 mechanism to maximize their use of the limited supply of BDNF. Indeed, we observed an increase in TrkB protein at 1 week following injury (Chapter 2) which parallels the increase in TrkB cells as well as NG2/TrkB cells observed at this time point. At 3 weeks after injury, the density of NG2+/TrkB+ cells were increased in the WM following injury. Interestingly, this parallels the significant increase in BDNF protein expression in the spinal cord (Chapter 2), suggesting that the increased TrkB expression in NG2 cells at this time point may result from enhanced BDNF availability and changes in the activity of the NG2 cells. From our results, it is clear that BDNF signaling plays a role in the CNS response to the peripheral injury and that NG2 cells are responsive to these changes in BDNF.

4.2. Delayed effect of CST transection on the mature OL population The mature OL population, those cells that expressed CC1 (Bhat et al., 1996; Lang et al., 2013) showed little change in the spinal cord at 1 week following CST transection, when robust changes in other glial populations were observed. CC1 has been reported to be a pan marker for all mature OLs, both myelinating and non-myelinating (Miller et al., 2002; Nishiyama et al., 2009). Yet, we demonstrated that a subset of mature OLs express TrkB (Chapter 3) and it is the TrkB population that shows plasticity in our model. Therefore, it is possible that the use of CC1 may mask small changes in subpopulations of mature OLs. Interestingly, at 16 weeks after injury, the number of mature CC1 cells was decreased in the LF with no changes observed in the TrkB population at this time point. Decreases in mature CC1 OLs have been demonstrated in the CNS only in cases of degenerative diseases such as multiple sclerosis and Huntington’s disease, and following direct injury to the CNS (McTigue et al., 2001; McTigue and Tripathi, 2008; Markoullis et al., 2012). Because cell death assays showed no apoptotic cells in the spinal cord following CST transection (Chapter 2), the decrease in CC1 cells at 16 weeks is likely not due to cell death. It is possible that the loss of cells could be through other mechanisms, such as necrosis or autophagy, which were not detected in our cell death assays. Alternatively, CC1 cells might have migrated away from the injured neurons although the reason for this is unclear. It would be interesting to use other OL markers, such as proteolipid protein, myelin basic protein and/or galactocerebroside to determine which subpopulation of OLs decreases at this time point. This decrease, however, does suggest that the increase in the OPC population does not lead to an increased number of mature OLs.

118

4.3 Effects of peripheral injury are localized to the IML and LF The results of this study demonstrate that after peripheral transection of preganglionic axons, plasticity is observed in the IML and the LF of the spinal cord. The IML has been previously demonstrated to be the location of the preganglionic neuronal cell bodies in the spinal cord (Rando et al., 1981; Strack et al., 1988; Hosoya et al., 1991). Although little is known regarding the path of exit of the preganglionic axons in the spinal cord, the present study suggests that a portion of the preganglionic axons can be found in the LF. The localized effect of the injury also suggests that the retrograde injury factors released by the injured neuronal cell bodies do not have high diffusibility in the spinal cord. Indeed, no changes were observed in any other GM subregions in the spinal cord. In addition, the changes observed in the LF may suggest that the axons of these neurons may also release injury factors as a result of the injury. In support of this idea, it has been shown that, following axon transection, many molecules, such as calpain and calcium ions, are increased in the proximal process of the axon (Conforti et al., 2014). Injured axons might be able to release factors that facilitate repair as it has been shown that activated cytoskeletal factors are important in the axon repairing process (Brosius and Barres, 2014).

5. Conclusions In the present study, we showed that CST transection resulted in an early increase in OPCs and TrkB cells in the IML and LF of the spinal cord, which are the spinal cord areas in which the injured neurons are located. Our lab has previously demonstrated changes in both astrocytes and microglia in the same regions of the spinal cord after injury (Coulibaly and Isaacson, 2012). It appears that these glial changes play a neuroprotective role, since no cell death was observed in at any of the time points analyzed (Chapter 2). Indeed, the reinnervation of the SCG, the sole target of the IML neurons, was observed (Chapter 2), suggesting that these neurons are regenerating and reinnervating their original target tissues. We also provide evidence that BDNF played an important role in the events that occurred in the spinal cord following CST transection. Here, we confirmed the increase in TrkB cells previously reported (Coulibaly and Isaacson, 2012) and, as shown in Chapter 2, TrkB protein levels were increased following the injury, suggesting that these OLs might be the source of the

119

TrkB protein increase. In addition, NG2 cells upregulated TrkB expression, which suggested that the observed changes in OPCs were mediated by BDNF. Because BDNF has been shown to play a role in both the proliferation and differentiation of NG2 cells (McTigue et al., 1998; Vondran et al., 2010), we conclude that CST transection resulted in increased numbers of NG2 cells that was likely the result of proliferation, and that BDNF signaling mediated this event. However, the increase in OPCs did not lead to an increase in the mature OL population, suggesting that the events did not affect the overall number of OLs in the spinal cord. The use of injury models such as in this study may provide an unprecedented tool to better understand the roles these cells play in the normal adult nervous system.

6. References

Barnabé-Heider, F., Göritz, C., Sabelström, H., Takebayashi, H., Pfrieger, F. W., Meletis, K., Frisén, J. (2010). Origin of new glial cells in intact and injured adult spinal cord. Cell Stem Cell, 7(4), 470–82.

Bergles, D. E., Roberts, J. D., Somogyi, P., Jahr, C. E. (2000). Glutamatergic synapses on oligodendrocyte precursor cells in the hippocampus. Nature, 405(6783), 187–91.

Bhat, R. V, Axt, K. J., Fosnaugh, J. S., Smith, K. J., Johnson, K. A., Hill, D. E., Baraban, J. M. (1996). Expression of the APC tumor suppressor protein in oligodendroglia. Glia, 17(2), 169–74

Brosius Lutz, A., Barres, B. A. (2014). Contrasting the glial response to axon injury in the central and peripheral nervous systems. Developmental Cell, 28(1), 7–17.

Causing, C. G., Gloster, a, Aloyz, R., Bamji, S. X., Chang, E., Fawcett, J., Miller, F. D. (1997). Synaptic innervation density is regulated by neuron-derived BDNF. Neuron, 18(2), 257–67.

Conforti, L., Gilley, J., Coleman, M. P. (2014). Wallerian degeneration: an emerging axon death pathway linking injury and disease. Nature Reviews. Neuroscience, 15(6), 394–409.

Coulibaly, A. P., Isaacson, L. G. (2012). Transient changes in spinal cord glial cells following transection of preganglionic sympathetic axons. Autonomic Neuroscience , 168(1-2), 32–42.

Horner, P. J., Power, A. E., Kempermann, G., Kuhn, H. G., Palmer, T. D., Winkler, J., Gage, F. H. (2000). Proliferation and differentiation of progenitor cells throughout the intact adult rat spinal cord. Journal of Neuroscience, 20(6), 2218–28.

120

Horner, P. J., Thallmair, M., Gage, F. H. (2003). Defining the NG2-expressing cell of the adult CNS. Journal of Neurocytology, 31(6-7), 469–80.

Lang, J., Maeda, Y., Bannerman, P., Xu, J., Horiuchi, M., Pleasure, D., Guo, F. (2013). Adenomatous polyposis coli regulates oligodendroglial development. Journal of Neuroscience, 33(7), 3113–30.

Lee, Y., Morrison, B. M., Li, Y., Lengacher, S., Farah, M. H., Hoffman, P. N., Rothstein, J. D. (2012). Oligodendroglia metabolically support axons and contribute to neurodegeneration. Nature, 487(7408), 443–8.

Ligon, K. L., Kesari, S., Kitada, M., Sun, T., Arnett, H. A, Alberta, J. A, Rowitch, D. H. (2006). Development of NG2 neural progenitor cells requires Olig gene function. Proceedings of the National Academy of Sciences of the United States of America, 103(20), 7853–8.

McTigue, D. M., Tripathi, R. B. (2008). The life, death, and replacement of oligodendrocytes in the adult CNS. Journal of Neurochemistry, 107(1), 1–19.

McTigue, D. M., Wei, P., Stokes, B. T. (2001). Proliferation of NG2-positive cells and altered oligodendrocyte numbers in the contused rat spinal cord. Journal of Neuroscience, 21(10), 3392–400.

Miller, R. H. (2002). Regulation of oligodendrocyte development in the vertebrate CNS. Progress in Neurobiology, 67(6), 451–67

Nishiyama, A., Komitova, M., Suzuki, R., Zhu, X. (2009). Polydendrocytes (NG2 cells): multifunctional cells with lineage plasticity. Nature Reviews. Neuroscience, 10(1), 9–22.

121

Rando, T. A., Bowers, C. W., Zigmond, R. E. (1981). Localization of neurons in the rat spinal cord which project to the superior cervical ganglion. Journal of Comparative Neurology, 196(1), 73–83.

Seifi, M., Corteen, N. L., van der Want, J. J., Metzger, F., & Swinny, J. D. (2014). Localization of NG2 immunoreactive neuroglia cells in the rat locus coeruleus and their plasticity in response to stress. Frontiers in Neuroanatomy, 8(May), 31.

Strack, A. M., Sawyer, W. B., Hughes, J. H., Platt, K. B., Loewy, A. D. (1989). A general pattern of CNS innervation of the sympathetic outflow demonstrated by transneuronal pseudorabies viral infections. Brain Research, 491(1), 156–162.

Sun, Y., Zigmond, R. E. (1996). Involvement of leukemia inhibitory factor in the increases in galanin and vasoactive intestinal peptide mRNA and the decreases in neuropeptide Y and tyrosine hydroxylase mRNA in sympathetic neurons after axotomy. Journal of Neurochemistry, 67(4), 1751–60

Trotter, J., Karram, K., Nishiyama, A. (2010). NG2 cells: Properties, progeny and origin. Brain Research Reviews, 63(1-2), 72–82.

122

Chapter 5: Peripheral axon injury leads to plasticity in connexin32 expression in the spinal cord

123

Abstract Glial cells such as oligodendrocytes (OLs) in the intermediolateral cell column (IML) of the spinal cord exhibit robust plasticity following the transection of preganglionic axons of the cervical sympathetic trunk (CST). This plasticity suggested that glial cell-cell communication may be influenced by the injury. Glial cells are coupled by gap junctions such as connexin (Cx) 32 which is expressed by exclusively by OL lineage cells. The objective of this study was to determine if the plasticity in OLs observed following peripheral injury was accompanied by changes in Cx32 protein expression. At 1 and 3 weeks following injury, Cx32 expression in the IML, the location of the injured cell bodies, was significantly increased. The increased Cx32 expression was attributed to an increase in the number of Cx32 plaques per cell since no changes were observed in the number of Cx32 positive cells at these time points. At 3 weeks, Cx32 was increased in the lateral funiculus, the location of the injured axons. Colocalization studies using Cx32 and OL markers demonstrated that Cx32 was decreased in oligodendrocyte progenitor cells (OPCs) in the IML at 1 week, while Cx32 expression in OLs expressing TrkB was increased. Though no changes were observed at earlier time points, Cx32 expression was increased in the mature OL population at 16 weeks after injury. These results demonstrate that retrograde factors expressed following peripheral axon injury can affect the expression of gap junction proteins in OL lineage cells in the spinal cord and that the plasticity in OLs observed following CST transection may be mediated by Cx32.

124

1. Introduction Understanding how neurons and glial cells communicate, particularly following injury, is the fundamental basis for understanding neuronal survival. Following injury to motor axons in the periphery, retrograde influences from the injury site lead to plasticity in the centrally located cell bodies (Lams et al., 1988; Armstrong et al., 1991; Chang et al., 2004; Chapter 2). In addition to exhibiting robust neurotransmitter and morphological plasticity (Abe and Cavalli, 2008), it has been proposed that the injured cell bodies release factors into the local environment which in turn serve to activate nearby glial cells (Jones et al., 2005; reviewed in Vallejo et al. 2010; Pekny and Pekna, 2014). While these changes appear to contribute to cell survival and regeneration (Barron et al. 1990; Coulibaly and Isaacson, 2012; Pekny and Pekna, 2014), the specific roles served by the activation of astrocytes, microglia and oligodendrocytes (OLs) following peripheral injury are poorly understood. In particular, the plasticity of OL lineage cells following injury is not well studied, yet the dysregulation of OLs contributes to demyelinating disorders (Lasiene and Yamanaka, 2011), mood disorders (Roy et al., 2007; Karoutzou et al., 2008), and lack of recovery following both traumatic brain injury and spinal cord injury (McTigue et al., 1998; Dewar et al., 2009). Therefore, a better understanding of the factors that influence these cells has important clinical implications. We recently reported that a population of OLs expressing full length TrkB, the cognate receptor for brain derived neurotrophic factor (BDNF), was increased in the vicinity of injured sympathetic preganglionic neuronal cell bodies in the intermediolateral cell column following the transection of the axons in the cervical sympathetic trunk (CST; Coulibaly and Isaacson, 2012). This suggests that some OL cells are responsive and/or regulated by BDNF. Indeed, BDNF has been shown to play an important role in OL regulation after injury. For example, after spinal cord injury, the presence of BDNF grafts led to increased oligodendrocyte progenitor cell (OPC) proliferation and differentiation into mature OLs (McTigue et al., 1998). BDNF infusion into the rat spinal cord following injury led to decreased OL apoptosis (Koda et al., 2002). Such robust oligodendrocyte plasticity in the spinal cord following CST transection suggested that glial cell-cell communication may be influenced by the peripheral injury. Indeed, following direct injury to the central nervous system, traumatic brain injury and/or spinal cord injury, a multitude of studies have demonstrated that glial cell communication is critical to neuronal survival (Reviewed in Proshnow, 2014).

125

Glial cells are coupled by gap junction channels that allow for intercellular transfer of ions and small signaling molecules (Reviewed in Bedner et al., 2012). Gap junctions are comprised of a family of connexin (Cx) membrane proteins which form hemichannels that dock with compatible hemichannels on adjacent glial cells to form gap junctions (Bedner et al., 2012). Cx32 and Cx47 (based on respective MW of 32kDa and 47 kDa) appear to be exclusive to OLs (Bedner et al., 2012), and OL Cx32 associates mainly with astrocyte Cx26 or Cx30 (Theis and Giaume, 2012) to form heteorotypic channels for communication with astrocytes. The primary objective of this study was to determine whether Cx32 expression in the spinal cord was influenced by injury to peripheral axons of the CST. Here, we show that Cx32 expression in the spinal cord is increased following peripheral axon injury and that the increased expression was localized specifically to the subpopulation of OLs that expresses TrkB rather than other cell types in the OL cell lineage.

2. Materials and methods 2.1. Surgery and tissue collection Young adult (3 months of age) female Sprague Dawley rats (Harlan Labs, Indianapolis, IN) were housed in the Miami University Animal Facilities in a 12:12 light:dark environment at regulated temperature. The cervical sympathetic trunk (CST) was exposed by making a 3 cm ventral incision in the neck region of the animal. After being separated from surrounding tissue, the CST was transected about 2 mm from its entry into the superior cervical ganglion (SCG; Sun and Zigmond, 1996). After the transection, both the proximal and distal stumps were placed carefully back into original position close to each other. The procedure was repeated on the other side. The incision was closed using sutures and tissue glue (Nexaband, Phx, AZ). In sham control animal, the CST was exposed but not transected. The CST was bilaterally transected and animals survived for 1 week (n=4), 3 weeks (n=5), 16 weeks (n=4) and compared with shams in which the CST was dissected and exposed but was not cut (1 week, n=4; 3 weeks, n=3; 16 weeks, n=3). All methods used in this study were approved by the Miami University Institutional Animal Care and Use Committee and efforts were taken to minimize discomfort and pain to the animals and to minimize the number of animals used in the study.

126

Because the robust glial plasticity following CST transection observed in a previous study was confined primarily to T1 of the spinal cord (Coulibaly and Isaacson, 2012), all analyses in this study were carried out at the T1 level, which was identified by counting the nerve roots extending from the cord and was verified by noting its location immediately caudal to the cervical enlargement. After each survival period has passed, animals were transcardially perfused with 0.9% saline followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB). Spinal cords were removed, rinsed in 0.1 M PB, and stored in 0.1M PB at 4oC. Prior to sectioning, T1 segments were blocked off from the spinal cord and cryoprotected by using a 30% sucrose solution in 0.1M PB. The T1 segment was then embedded in optimal cutting temperature medium (Ted Pella Inc.), and cut using a MICROM HM 550 series cryostat. Coronal sections of spinal cord (18 m) were serially collected and mounted onto Fisherbrand Superfrost microscope slides. For immunohistochemical analysis, sections were incubated overnight in 0.1M PBS-0.2% Triton-X solution, blocked with normal donkey serum, and incubated for 48 hrs at 4oC with a cocktail of rabbit-anti TrkB (1:200; Santa Cruz Biotechnologies, sc-12), mouse-anti Cx32 (1:200; Invitrogen, 358900), mouse anti-CC1 (1:500; Abcam, ab16794), mouse anti-NG2 (1:200; Millipore, MAB5384), and rabbit anti-Cx32 (1:200; Invitrogen, 710600). Sections were rinsed in 0.1M PBS and incubated for 2 hours in AlexaFluor conjugated antibodies (1:200; Molecular Probes) directed against the primary antibody host. Sections then were coverslipped using fluorescent mounting medium with 4', 6-diamidino-2-phenylindole (DAPI; Vectashield). Images were captured with a Zeiss 720 laser scanning confocal microscope and Zen 2000 software (Zeiss).

2.2. General methods for Cx32 analyses 400X oil confocal images were obtained of the intermediolateral cell column (IML; location of the IML neuronal cell bodies; Rando et al., 1981) or the lateral funiculus (LF; white matter area immediately adjacent to the IML which contains axons of the IML neurons). Images were obtained from each side of the spinal cord at the T1 level. The left and right IML from each animal were considered separate cases since the peripheral axons on each side of the animals were transected and the response to peripheral transection in the spinal cord has been determined to be unilateral.

127

All image settings were kept the same within each time point. Images were acquired using the “Tile” and “Z stack” features of the microscope. Tile images were taken in a 2 x 2 configuration and optical sections of the Z stacks were acquired at 3 µm intervals. The 6 optical sections were collapsed to form a composite image using Zen 2000 software. Separate composite images were obtained for the IML and LF from each side of an animal. Each composite image was enlarged to 200% on a computer monitor and analyzed using Image Pro 6.3 software. Because the shams from each survival time point were found to be significantly different from each other, each injury group was compared to its age-matched control group. For each statistical analysis described below, a mean was obtained for the sham group and all values (sham and injury) were expressed as proportion of the sham mean at each survival time point. The data were square root transformed and statistically analyzed using the Mann Whitney test with significance ascribed at p<0.05. Square root transformed data are shown in the graphs in Figures 2 and 5. In each analysis described below, data from 1 week (n=6), 3 week (n=6), and 16 week (n=8) survival times were directly compared to age matched controls.

2.3. Analysis of Cx32 in the IML and LF following injury The analysis of Cx32 was based on methodology described previously by Markoullis and colleagues (2012). In the present study, a Cx32 plaque was defined as a focal accumulation with a diameter larger than 0.0002 µm. To determine the effects of peripheral injury on Cx32 plaques, the overall density of Cx32 plaques, the number of cells that expressed Cx32, the number of Cx32 plaques per cell, and the size of the Cx32 plaques in the IML and LF were assessed by counting all immunolabeled profiles in the field of view, which encompassed an area of approximately 0.2 mm2. The density of Cx32 plaques and the number of Cx32 plaques per cell in the IML and LF were determined using Image Pro 6.3 software. Any Cx32 immunopositive profiles in the field of view that exhibited Cx32 protein expression above the background expression were included in the analysis of plaque density. In order to obtain the number of Cx32 plaques per cell for each case, the plaque density value was divided by the total number of DAPI nuclei. For determination of Cx32 plaque size, the diameter of each Cx32 plaque was measured using Image Pro 6.3 software. The plaque sizes were grouped into bins at 0.2 µm intervals,

128

ranging from 0.5 µm to 2.6 µm and greater than 2.6 µm. The number of plaques in each bin was divided by the total number of plaques and this value was multiplied by 100 to determine the percent of plaques within each bin. The percentage of plaques within each bin in each injury group was compared to that of the age-matched controls. These data are represented in Figure 3.

2.4. Identification of oligodendrocyte lineage cells that expressed Cx32 Several oligodendrocyte (OL) subpopulations have been described in the literature: 1) an OPC population that expresses the proteoglycan nerve/glia (NG) 2 and has been shown to produce OLs once stimulated (Nishiyama et al., 2009); 2) a mature OL population, which expresses the adenomatous polyposis coli (APC, commonly known as CC1) and has been shown to myelinate central axons (Baumann and Pham-Dinh, 2001; Miller et al., 2002; Lang et al., 2013); and 3) A subset of OLs that expresses the TrkB receptor (Coulibaly and Isaacson 2012; Chapter 3). In order to determine whether any changes in Cx32 expression was specific to a particular subpopulation of OL lineage cells, Cx32 was colocalized with NG2, CC1, or TrkB. Using the confocal images which were obtained from Z-stacks taken at 400X magnification in a 2 X 2 configuration, the number of cells expressing Cx32 and NG2 (NG2+/Cx32+), Cx32 and CC1 (CC1+/Cx32+), and Cx32 and TrkB (Cx32+/TrkB+) was manually counted in images of the IML and the LF, and Image Pro 6.3 was used to keep track of the counts. All cells were counted within a field of view, which accounted for an area of ~ 0.2 µm2. In all cases, only cells with clearly defined DAPI nuclei were included in the analysis. To determine the proportion of cells coexpressing NG2 and Cx32, the number of cells expressing NG2 only, Cx32 only, and both NG2 and Cx32 was obtained in the field of view. The total number of cells was then determined by adding the number of cells in each subtype (NG2+/Cx32- + NG2-/Cx32+ + NG2+/Cx32+). The number of NG2+/Cx32+ cells was then divided by the total number of cells. This analysis was performed for each IML and LF. A similar analysis was performed to determine the proportion of cells expressing Cx32 and CC1 (CC1+/Cx32+), and Cx32 and TrkB (Cx32+/TrkB+).

3. Results 3.1. CST transection leads to increased expression of Cx32

129

Cx32 plaques were observed throughout both the IML and the LF of the spinal cord and resembled the plaques reported in Sargiannidou et al. 2009 (Fig. 1.A.). The density of Cx32 plaques was compared between controls and injured cases at each time point. At 1 week following injury, a significant 228% increase in the density of Cx32 plaques in the IML was observed, from ~4,000 plaques/mm2 in the control IML to ~9,000 plaques/mm2 in the injured IML (Fig. 1; 2.A.). At 1 week no changes were observed in the density of Cx32 plaques in the LF (Fig. 2.A.). At 3 weeks following injury, the density of Cx32 plaques in the IML was significantly increased by ~400% from ~80 plaques/mm2 to ~8,073 plaques/mm2 (Fig. 2.A.). At this time point, the density of Cx32 in the LF also was increased by almost 800%, from ~50 plaques to over 13,000 plaques/mm2 (Fig. 2.A.). At 16 weeks after injury, the density of Cx32 plaques was similar to control values in both the IML and LF (Fig. 2.A.). The increased density of Cx32 plaques in the IML and LF could result from an upregulation of Cx32 protein expression per cell and/or an increase in the number of OL lineage cells in the spinal cord that expressed Cx32. Analysis of the number of Cx32 plaques per cell revealed a significant 30% increase in the IML at 1 week after injury (Fig. 2.B.). The control cases averaged approximately 2.0 plaques/cell while the injury cases ~3.5 plaques/cell (Fig. 2.B.). No changes were observed in the number of plaques per cell in the LF at this time point At 3 weeks, a significant increase (Fig. 2.B.) was observed in the number of Cx32 plaques per cell in the IML (cont=1.0 plaque/cell; injury=~3.0 plaques/cell) and in the LF (cont=1.0 plaque/cell; injury=6.5 plaques/cell). At 16 weeks, the number of plaques per cell was similar to control values in both the IML and LF (Fig. 2.B.). When the number of cells that expressed Cx32 was assessed, no significant changes were observed in the number of Cx32 cells in either the IML or the LF at any of the time points analyzed (Fig. 2.C.) Analysis of the diameter of Cx32 plaques revealed no changes in the size distribution of Cx32 plaques in either the IML or LF at 1 week following injury. At 3 weeks, a significant increase was observed in the IML in the number of plaques falling into the category ranging from 1.4 -1.7µm as well as plaques larger than 2.6µm in diameter (Fig. 3.C.). No changes were observed in the plaque sizes in the LF at this time point (Fig. 3.D). At 16 weeks after injury Cx32 plaque sizes were similar to controls (Fig. 3.E., F.).

3.2. Differential expression of Cx32 in OL lineage cells following injury

130

To determine whether the changes in Cx32 expression were specific to a particular OL subpopulation, double label studies were carried out combining Cx32 with NG2, TrkB, or CC1. Cx32 immunofluorescent plaques that associated with NG2 cells were found on the cellular processes (Fig. 4C.-C’’.) while the Cx32 immunoreactive plaques associating with CC1 (Fig. 4.A.-A.’’) as well as TrkB (Fig. 4B.-B’’.) cells were observed predominantly around the cell body. The proportion of Cx32+/NG2+ cells was similar in both control and injury in the IML at 1 week after injury (Fig. 5.C.). However in the LF, a significant 16% decrease in the number of Cx32+/NG2+ cells was observed (Fig. 5.C). No changes were observed in Cx32+/NG2+ cells at either 3 or 16 weeks post injury. Analysis of the number of Cx32+/TrkB+ cells revealed a significant 6% increase in the IML at 1 week after injury (Fig. 5.A.), while no changes in these cells were observed in the LF (Fig. 5.A.). Although the number of Cx32+/TrkB+ cells was similar to control values in the IML at 3 weeks, a significant 34% increase was observed in the LF at this time point (Fig. 5.A.). At 16 weeks after injury, no changes were observed in the number of Cx32+/TrkB+ in either the IML or LF. The number of Cx32+/CC1+ cells was unchanged at 1 and 3 weeks post injury compared to controls (Fig. 5.B.). However, at 16 weeks, a significant 40% increase was observed in Cx32+/CC1+ cells in the LF (Fig. 5.B.).

131

Sham

cont Cx32 Cx32/ DAPI

1 wk Cx32 Cx32/ DAPI

3 wk Cx32 Cx32/ DAPI

16 wk Cx32 Cx32/ DAPI

Figure 1

132

Figure 1: Cx32 expression in the IML. Confocal micrographs of Cx32 (green) in the IML in the controls and following injury. Cx32 (green) was observed along the perimeter of small cells (arrows) throughout the field of view. The number of Cx32 immunoreactive profiles appeared to be increased in the IML at 1 week and 3 weeks after injury in the IML. DAPI (blue) shows cellular nature of labeled profiles. Scale bar for all images = 100 µm.

133

B. A. Density of Cx32-ir Cx32 plaques per cell 11 * 11 10 10 * 9 9 8 8 7 7 * * 6 6 5 5 4 4

3 3 Proportion of sham of Proportion(sqrt) sham Proportion of sham Proportion of (sham sq rt) 2 2 * * 1 1 0 0 IML LF IML LF 1 wk 3 wk 16 wk 1 wk 3 wk 16 wk

C. Cx32-ir cells 80 70 60 50 40 30

number of number cells 20 10 0 cont inj cont inj IML LF

1 wk 3 wk 16 wk

Figure 2

134

Figure 2: Analysis of changes in Cx32 expression in the spinal cord following peripheral injury. A. The number of Cx32 plaques was increased in the intermediolateral cell column (IML) at 1 week (1wk) and 3 weeks (3wk) after injury. The number of plaques was increased in the lateral funiculus (LF) at 3 wk. B. The number of Cx32 plaques per cell was increased in the IML at 1 wk and 3 wk and also was increased in the LF at 3 wk after injury. C. The number of cells that express Cx32 was similar to controls. Dashed lines represent the mean of the controls. *, p < 0.05.

135

A. Cx32 plaque size in the IML B. 50 50 Cx32 plaque size in the LF cont 1 wk 40 40 cont 1 wk

30 30

20 20

Frequency Frequency Frequency Frequency

10 10

0 0 0.5 0.7 1 1.2 1.4 1.7 1.9 2.2 2.4 2.6 0.5 0.7 1 1.2 1.4 1.7 1.9 2.2 2.4 2.6 Plaque size (µm) Plaque size (µm)

C. Cx32 plaque size in IML D. Cx32 plaque size in LF 60 50 cont 3 wk 50 cont 3 wk 40 40 30 30

20 Frequency Frequency Frequency Frequency 20 * * 10 10

0 0 0.5 0.7 1 1.2 1.4 1.7 1.9 2.2 2.4 2.6 0.5 0.7 1 1.2 1.4 1.7 1.9 2.2 2.4 2.6 Plaque size (µm) Plaque size (µm)

Cx32 plaque size in LF E. Cx32 plaque size in IML F. 60 60

50 cont 16 wk 50 cont 16 wk

40 40

30 30 Frequency Frequency

Frequency Frequency 20 20

10 10

0 0 0.5 0.7 1 1.2 1.4 1.7 1.9 2.2 2.4 2.6 0.5 0.7 1 1.2 1.4 1.7 1.9 2.2 2.4 2.6 Plaque size (µm) Plaque size (µm) Figure 3

136

Figure 3: Analysis of the number of Cx32 plaques after peripheral injury. At 1 week (1wk), no changes were observed in the size of Cx32 plaques in either the intermediolateral cell column (IML; A.) or lateral funiculus (LF; B.). C. At 3 weeks (3wk), the number of plaques with a diameter of 1.7 µm and those with diameters larger than 2.6 µm in the IML were increased. D. No changes were observed in plaque size in the LF at 3 wk following injury. E.-F. Plaque size was unchanged at 16 weeks (16wk). *, p < 0.05.

137

TrkB/DAPI Cx32/ DAPI TrkB/ Cx32/ DAPI

A. A’. A’’.

CC1/DAPI Cx32/ DAPI CC1/ Cx32/ DAPI

B. B’. B’’.

NG2/DAPI Cx32/ DAPI NG2/ Cx32/ DAPI

C. C’. C’’.

Figure 4

138

Figure 4: Cx32 plaques in association with OL lineage cells in the spinal cord. A.-A”. Cx32 expression (green; arrows) was found on the perimeter of the cell body of TrkB cells (red). B.- B’’. Similarly Cx32 expression (green; arrows) was found on the perimeter of CC1 cells (red). C.-C”. Cx32 (green) was found primarily associated with the processes of NG2 cells (yellow). DAPI (blue) shows cellular nature of stained profiles. Scale bar for all images = 12.5 µm.

139

A. Colocalization of Cx32 and TrkB in OL B. Colocalization of Cx32 and CC1 in OL lineage cells lineage cells 1.6 1.6 * * 1.4 1.4 * 1.2 )

) 1.2

rt sq sq 1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4

Proportion of control( Proportion of ( control 0.2 0.2

0.0 0.0 IML LF IML LF

1 wk 3 wk 16 wk 1 wk 3 wk 16 wk

C. Colocalization of Cx32 and NG2 in OL 1.2 lineage cells

1.0

) * rt

0.8 sq

0.6

0.4

Proportion of ( control 0.2

0.0 IML LF

1 wk 3 wk 16 wk Figure 5

140

Figure 5: Analysis of the changes in Cx32 expression in OL lineage cells after injury. A. A significant increase was observed in the number of TrkB cells expressing Cx32 in the intermediolateral cell column (IML) at 1 week (1wk) and lateral funiculus (LF) at 3 weeks (3wk) after injury. B. An increase was observed in the number of CC1 cells expressing Cx32 in the LF at 16 weeks (16wk) after injury. C. At 1 wk, Cx32 expression in NG2 cells was decreased in the LF. Dashed lines represent the mean of the controls. *, p<0.05.

141

4. Discussion 4.1. Effects of peripheral axon injury on Cx32 expression by OL lineage cells The results of the present study suggest that retrograde factors released by neurons following injury (Jones et al., 2005; Vallejo et al. 2010; Pekny and Pekna, 2014) can modulate the expression of Cx32 in the vicinity of the injured neurons, and that the changes in Cx32 were specific to different OL subpopulations. To our knowledge, we are the first to show that peripheral axon injury can lead to an increase in Cx32 protein expression in the central nervous system and that in the spinal cord, OL progenitor cells express Cx32. Similar to previous reports (Kleopa et al., 2004), we observed Cx32 expression in both GM and WM subregions of the spinal cord. In addition, Cx32 expression was found in the processes and cell bodies of OPCs and on the perimeter of mature OLs expressing CC1 or TrkB, similar to that described by Melanson-Drapeau and colleagues (2003), and Kleopa and colleagues (2004), respectively. The role of Cx protein on OPCs is yet unknown. Until recently, it was believed that OPCs did not couple to other cells in the CNS. However, it has been shown both in vitro (Imbeault et al., 2009) and in vivo (Melanson-Drapeau et al., 2003; Maglione et al., 2010) that these cells do indeed express connexins that are exclusive to OLs, specifically Cx32. Dye coupling experiments in brain slices have demonstrated that OPCs are coupled to mature OL cells in the CNS (Maglione et al., 2010). Therefore, it is possible that in the spinal cord, the expression of Cx32 by NG2 cells is evidence of electrical coupling between OL progenitor cells and mature OLs. Little is known regarding the role of Cx32 expression in the central nervous system. However, many studies have been conducted on the roles of astrocytic Cxs in homeostatic maintenance of the CNS (reviewed in Dbouk et al., 2009; reviewed in Rouach et al., 2002). For example, the disruption of astrocytic gap junctions in a mixed culture proved to be detrimental to neuronal survival (Blanc et al., 1998). Blanc and colleagues (1998) showed that the suppression of astrocytic Cx43 led to decreased removal of extracellular calcium ions, and increased mitochondria disruption, in the neurons. Also, gap junctions have been shown to play a role in the dissipation of excess potassium ions by astrocytes in the extracellular domain, which is critical for normal neuronal activity (Newman, 1985; Ballanyi et al., 1987; Rouach et al., 2002). In addition, it has been shown that the presence of gap junctions between astrocytic end feet plays a critical role in the formation of microdomains at the capillaries, which may assist the

142

astrocytes as they perform their role in the blood brain barrier (Giaume and McCarthy, 1996; Wolff et al., 1998; Rouach et al., 2002). Though all these functions are attributed to astrocytic Cxs, it is possible that OL Cx32 plays similar roles.

4.2. Role for Cx32 expression in oligodendrocyte progenitor cells In the present model, a decrease was observed in the number of Cx32+/NG2+ cells at 1 week following injury. This decrease parallels the increased number of NG2 cells in the IML and LF that was reported in Chapter 4 of this dissertation. It is possible that the decrease in Cx32+/NG2+ cells indicates a shift in the NG2 cells to a proliferative state. Indeed, a previous study revealed that, in the absence of Cx32, NG2 cells readily re-entered the cell cycle, leading to an increased proliferation rate (Melanson-Drapeau et al., 2003). Therefore, it is possible that the loss of Cx32 by NG2 cells in our model is indicative of proliferation. Indeed, the increase in NG2 cells observed at 1 week in our model supports this idea. Whether CST transection activates NG2 cell proliferation is a topic for future investigation.

4.3. Role for Cx32 expression in mature OLs Our results revealed that injury to preganglionic axons in the periphery led to increase Cx32 expression on mature OLs. The expression of Cx32 by mature OLs have been shown to mediate the electrical coupling between mature OLs and OPCs, mature OLs and mature OLs, and mature OLs and astrocytes (Pastor et al., 1998; Maglione et al., 2010). In the present model, an increase in Cx32 expression could be due to an increase in any of the above mentioned interactions. However, from our results we can surmise that the increased interaction may not be due to an increased in electrical coupling between mature OLs and OPCs at 1 week, since there was a decrease in the expression of Cx32 in OPCs at this time point. It has been previously demonstrated that increase electrical coupling can lead to the removal of neurotoxic factors in the vicinity of neuronal cell bodies (Giardina et al., 2007; Rash, 2010; Nualart-Marti et al., 2013). Therefore, it is possible that this may be the case in our model. Additional experiments are needed in order to confirm or refute this assumption. Our results showed a significant increase in Cx32 plaques per cells in the spinal cord at both 1 week and 3 weeks after injury. Our analysis of OL lineage cells revealed that only TrkB OLs increased Cx32 expression at those time points. Therefore, we postulate that the increase in

143

Cx32 is due primarily to TrkB OLs. Our lab has previously demonstrated that TrkB OLs make up approximately 56% of all mature OLs (Chapter 3). An increase in Cx expression in this cell subset suggests that these cells are forming more connections with neighboring cells after injury. Because the role of OL gap junctions is still unknown, the possible roles of altered interaction between TrkB OLs and other cells is a mystery. Due to the changes observed in our system and results from other injury models, we propose the following scenario. Increased Cx32 protein expression in TrkB cells during a time when a source of BDNF is lost might help to propagate and/or maximize the effects of BDNF on these cells. This would be an important function due to the low availability of BDNF. When coupling is increased following the injury, calcium and IP3 in one cell can be propagated throughout the interacting OLs and astrocytes, thereby maximizing the effects of BDNF and possibly inducing the release of survival factors by the glial cells in the vicinity of the injured neurons, thereby affecting their survival. At 3 weeks after injury, we documented an increase in Cx32 expression in TrkB OLs in the LF. It has been previously demonstrated that Cx32 expression can be found in the external layers of the myelin sheath around axons (Nagy et al., 2003; Kleopa et al., 2004). In addition, it has been demonstrated that although white matter OLs express Cx32, they do not form functional synapses with other cells in the white matter (Pastor et al., 1998), suggesting that these are hemichannels. Hemichannels have been shown to play a role in neurite outgrowth (Belliveau et al., 2006; Dbouk et al., 2009) and the movement of the coenzyme nicotinamide adenine dinuceleotide (NAD) within cells (Bruzzone et al., 2001; Dbouk et al., 2009). It is possible that the increased Cx32 expression in TrkB cells in the LF at 3 weeks may facilitate the regeneration of the injured axons. Indeed, at 3 weeks, we documented an increase in ChAT-ir fibers present in the SCG (Chapter 2). This would suggest that the increase in Cx32 expression observed in mature OLs at 16 week may also play a role in axonal regeneration in our model. In addition, the presence of Cx32 in the myelin sheath is thought to play an important role in the removal of potassium ions and the maintenance of osmotic balance in the axons (Rash, 2010). Therefore, an increase in Cx32 expression in the LF may a role in stabilizing and/or maintaining optimal conditions in the axons in order to promote neuronal survival and axon regeneration.

144

4.4. Peripheral injury increased the size of Cx32 plaques Our results demonstrated that at 3 weeks after injury, there was a significant increase in large Cx32 plaques in the IML. Under normal conditions, Cxs are folded and undergo posttranslational modifications in the ER (Dbouk et al., 2009). Following which they are transported through vesicles to the plasma membrane where they are inserted (Segratain et al., 2004; Dbouk et al., 2009). Interestingly, connexin proteins are usually inserted in relatively close proximity in the plasma membrane, which are then termed plaques (Dbouk et al., 2009). It has been demonstrated that novel Cx insertion happens in the outer edge of preexisting plaques (Segratain et al., 2004; Dbouk et al., 2009). Degradation of Cxs plaques often occurs from the center, where older Cx proteins are located, outward. In addition, Cx protein has a considerably short half-life in the plasma membrane (Segratain et al., 2004). It has been shown that Cxs are only present in the plasma membrane for a few of hours before they are internalized and degraded (Laird et al., 1991; Segratain et al, 2004; Dbouk et al., 2009). Therefore, it is possible that in our model, an increase in large Cx plaques may be an indication of either an increase in the translation and insertion of these molecules in the plasma membrane or a decrease in their degradation. Although more works need to be done to verify this hypothesis, the increase in plaque size suggests that retrograde factors from the injury may affect regulatory proteins and/or proteins that are important for protein synthesis in OLs.

5. Conclusions To our knowledge, the present study is the first to show Cx32 expression in OPCs in the spinal cord, and to show an increase in Cx32 expression in OL lineage cells after peripheral injury. A decrease in Cx32 expression in the OPCs suggests a decrease in the interaction of these cells with their neighboring cells. This in turn suggests that these cells may have become more proliferative as a result of the injury. On the other hand, an increase in Cx32 expression in mature OLs after injury points to an increased interactions between mature OLs and other OLs or astrocytes. Coupling between glial cells has been shown to play a critical role in neuronal survival in both normal and pathological states (Dbouk et al., 2009; Bedner et al., 2012). In our model it is possible that an increase in electrical coupling between OLs and other cells helped to facilitate both the survival and regeneration of the injured preganglionic neurons. In summary,

145

the present study provides evidence that retrograde influences from the periphery can alter gap junction expression in OL lineage cells.

6. References

Abe, N., Cavalli, V. (2008). Nerve injury signaling. Current Opinion in Neurobiology, 18, 276– 83.

Armstrong, D. M., Brady, R., Hersh, L. B., Hayes, R. C., Wiley, R. G. (1991). Expression of choline acetyltransferase and nerve growth factor receptor within hypoglossal motoneurons following nerve injury. Journal of Comparative Neurology, 304, 596–607.

Ballanyi, K., Grafe, P., Bruggencate, G. (1987). Ion activities and potassium uptake mechanisms of glial cells in guinea-pig olfactory cortex slices. Journal of Physiology, 382, 159–74.

Baumann, N., Pham-dinh, D. (2001). Biology of Oligodendrocyte and Myelin in the Mammalian Central Nervous System. Physiological reviews, 81, 871–927.

Bedner, P., Steinhäuser, C., Theis, M. (2012). Functional redundancy and compensation among members of gap junction protein families? Biochimica et Biophysica Acta, 1818, 1971–84.

Belliveau, D. J., Bani-Yaghoub, M., McGirr, B., Naus, C. C. G., Rushlow, W. J. (2006). Enhanced neurite outgrowth in PC12 cells mediated by connexin hemichannels and ATP. Journal of Biological Chemistry, 281, 20920–31.

Blanc, E. M., Bruce-Keller, A. J., Mattson, M. P. (2002). Astrocytic Gap Junctional Communication Decreases Neuronal Vulnerability to Oxidative Stress-Induced Disruption of Ca2+ Homeostasis and Cell Death. Journal of Neurochemistry, 70, 958–970.

Bruzzone, S., Franco, L., Guida, L., Zocchi, E., Contini, P., Bisso, A, De Flora, A. (2001). A self-restricted CD38-connexin 43 cross-talk affects NAD+ and cyclic ADP-ribose metabolism and regulates intracellular calcium in 3T3 fibroblasts. Journal of Biological Chemistry, 276, 48300–8.

Chang, H.-M., Wei, I.-H., Tseng, C.-Y., Lue, J.-H., Wen, C.-Y., Shieh, J.-Y. (2004). Differential expression of calcitonin gene-related peptide (CGRP) and choline acetyltransferase (ChAT) in the axotomized motoneurons of normoxic and hypoxic rats. Journal of Chemical Neuroanatomy, 28, 239–51.

146

Chang, Q., Gonzalez, M., Pinter, M. J., Balice-gordon, R. J. (1999). Gap Junctional Coupling and Patterns of Connexin Expression among Neonatal Rat Lumbar Spinal Motor Neurons. Journal of Neuroscience, 19, 10813–10828.

Coulibaly, A. P., Isaacson, L. G. (2012). Transient changes in spinal cord glial cells following transection of preganglionic sympathetic axons. Autonomic Neuroscience, 168, 32–42.

Dansereau, M.-A., Gosselin, R.-D., Pohl, M., Pommier, B., Mechighel, P., Mauborgne, A., Melik-Parsadaniantz, S. (2008). Spinal CCL2 pronociceptive action is no longer effective in CCR2 receptor antagonist-treated rats. Journal of Neurochemistry, 106, 757–69.

Dbouk, H.Aa, Mroue, R. M., El-Sabban, M. E., Talhouk, R. S. (2009). Connexins: a myriad of functions extending beyond assembly of gap junction channels. Cell Communication and Signaling, 7, 4.

Dewar, D., Underhill, S. M., Goldberg, M. P. (2003). Oligodendrocytes and Ischemic Brain Injury. Journal of Cerebral Blood Flow and Metabolism, 23, 263–274.

Giardina, S. F., Mikami, M., Goubaeva, F., Yang, J. (2007). Connexin 43 confers resistance to hydrogen peroxide-mediated apoptosis. Biochemical and Biophysical Research Communications, 362, 747–52.

Giaume, C., McCarthy, K. D. (1996). Control of gap-junctional communication in astrocytic networks. Trends in Neurosciences, 19, 319–25.

Imbeault, S., Gauvin, L. G., Toeg, H. D., Pettit, A., Sorbara, C. D., Migahed, L., Bennett, S. Al. (2009). The extracellular matrix controls gap junction protein expression and function in postnatal hippocampal neural progenitor cells. BMC Neuroscience, 10, 13.

Karoutzou, G., Emrich, H. M., Dietrich, D. E. (2008). The myelin-pathogenesis puzzle in schizophrenia: a literature review. Molecular Psychiatry, 13, 245–60.

Kleopa, K. A, Orthmann, J. L., Enriquez, A., Paul, D. L., Scherer, S. S. (2004). Unique distributions of the gap junction proteins connexin29, connexin32, and connexin47 in oligodendrocytes. Glia, 47, 346–57.

147

Koda, M., Murakami, M., Ino, H., Yoshinaga, K., Ikeda, O., Hashimoto, M., Moriya, H. (2002). Brain-derived neurotrophic factor suppresses delayed apoptosis of oligodendrocytes after spinal cord injury in rats. Journal of Neurotrauma, 19, 777–85.

Laird, D. W., Puranam, K. L., Revel, J. (1991). Turnover and phosphorylation dynamics of. Biocehmical Journal273, 67–72.

Lang, J., Maeda, Y., Bannerman, P., Xu, J., Horiuchi, M., Pleasure, D., Guo, F. (2013). Adenomatous polyposis coli regulates oligodendroglial development. Journal of Neuroscience, 33, 3113–30.

Lasiene, J., Yamanaka, K. (2011). Glial cells in amyotrophic lateral sclerosis. Neurology Research International, 2011, 718987.

Maglione, M., Tress, O., Haas, B., Karram, K., Trotter, J., Willecke, K., Kettenmann, H. (2010). Oligodendrocytes in mouse corpus callosum are coupled via gap junction channels formed by connexin47 and connexin32. Glia, 58, 1104–17.

Markoullis, K., Sargiannidou, I., Gardner, C., Hadjisavvas, A., Reynolds, R., Kleopa, K. A. (2012). Disruption of oligodendrocyte gap junctions in experimental autoimmune encephalomyelitis. Glia, 60, 1053–66.

McTigue, D. M., Horner, P. J., Stokes, B. T., & Gage, F. H. (1998). Neurotrophin-3 and Brain- Derived Neurotrophic Factor induce oligodendrocyte proliferation and myelination of regenerating axons in the contused adult rat spinal cord. Journal of Neuroscience, 18, 5354– 5365.

Melanson-drapeau, L., Beyko, S., Dave, S., Hebb, A. L. O., Franks, D. J., Sellitto, C., Bennett, S. A. L. (2003). Oligodendrocyte Progenitor Enrichment in the Connexin32 Null-Mutant Mouse. Journal of Neuroscience, 23, 1759–1768.

Miller, R. H. (2002). Regulation of oligodendrocyte development in the vertebrate CNS. Progress in Neurobiology, 67, 451–67

148

Nagy, J. I., Ionescu, A. V, Lynn, B. D., Rash, J. E. (2003). Connexin29 and connexin32 at oligodendrocyte and astrocyte gap junctions and in myelin of the mouse central nervous system. Journal of Comparative Neurology, 464, 356–70.

Newman, E. A. (1985). Membrane physiology of retinal glial (Müller) cells. Journal of Neuroscience, 5, 2225–39.

Nishiyama, A., Komitova, M., Suzuki, R., Zhu, X. (2009). Polydendrocytes (NG2 cells): multifunctional cells with lineage plasticity. Nature Reviews: Neuroscience, 10, 9–22.

Nualart-Marti, A., Solsona, C., Fields, R. D. (2013). Gap junction communication in myelinating glia. Biochimica et Biophysica Acta, 1828, 69–78.

Ochalski, P. A., Frankenstein, U. N., Hertzberg, E. L., Nagy, J. I. (1997). Connexin-43 in rat spinal cord: localization in astrocytes and identification of heterotypic astro-oligodendrocytic gap junctions. Neuroscience, 76, 931–45.

Pekny, M., Pekna, M. (2014). Astrocyte Reactivity and Reactive Astrogliosis: Costs and Benefits. Physiological Reviews, 94, 1077–1098.

Prochnow, N. (2014). Relevance of gap junctions and large pore channels in traumatic brain injury. Frontiers in Physiology, 5, 1-8. Rando, T. A., Bowers, C. W., Zigmond, R. E. (1981). Localization of neurons in the rat spinal cord which project to the superior cervical ganglion. Journal of Comparative Neurology, 196, 73–83.

Rash, J. E. (2010). Molecular disruptions of the panglial syncytium block potassium siphoning and axonal saltatory conduction: pertinence to neuromyelitis optica and other demyelinating diseases of the central nervous system. Neuroscience, 168, 982–1008.

Rash, J. E., Kamasawa, N., Davidson, K. G. V, Yasumura, T., Pereda, A. E., Nagy, J. I. (2012). Connexin composition in apposed gap junction hemiplaques revealed by matched double-replica freeze-fracture replica immunogold labeling. Journal of Membrane Biology, 245, 333–44.

Rash, J. E., Staines, W. A., Yasumura, T., Patel, D., Furman, C. S., Stelmack, G. L., & Nagy, J. I. (2000). Immunogold evidence that neuronal gap junctions in adult rat brain and spinal cord contain connexin-36 but not connexin-32 or connexin-43. Proceedings of the National Academy of Sciences, 97, 7573–7578.

149

Rouach, N., Avignone, E., Même, W., Koulakoff, A, Venance, L., Blomstrand, F., Giaume, C. (2002). Gap junctions and connexin expression in the normal and pathological central nervous system. Biology of the Cell, 94, 457–475.

Roy, K., Murtie, J. C., El-Khodor, B. F., Edgar, N., Sardi, S. P., Hooks, B. M., Corfas, G. (2007). Loss of erbB signaling in oligodendrocytes alters myelin and dopaminergic function, a potential mechanism for neuropsychiatric disorders. Proceedings of the National Academy of Sciences of the United States of America, 104, 8131–6.

Sargiannidou, I., Vavlitou, N., Aristodemou, S., Hadjisavvas, A., Kyriacou, K., Scherer, S. S., Kleopa, K. A. (2009). Connexin32 mutations cause loss of function in Schwann cells and oligodendrocytes leading to PNS and CNS myelination defects. Journal of Neuroscience, 29, 4736–49.

Segretain, D., Falk, M. M. (2004). Regulation of connexin biosynthesis, assembly, gap junction formation, and removal. Biochimica et Biophysica Acta, 1662, 3–21.

Theis, M., Giaume, C. (2012). Connexin-based intercellular communication and astrocyte heterogeneity. Brain Research, 1487, 88–98.

Vallejo, R., Tilley, D. M., Vogel, L., Benyamin, R. (2010). The role of glia and the immune system in the development and maintenance of neuropathic pain. Pain Practice, 10, 167–84.

Wolff, J. R., Stuke, K., Missler, M., Tytko, H. (1998). Autocellular coupling by gap junctions in cultured astrocytes : a new view on cellular autoregulation during process formation. Glia, 140, 121–140.

150

Chapter 6: Transneuronal effects of peripheral axon injury on the central nervous system

151

Abstract Following the transection of preganglionic axons in the periphery, we have shown that the parent cell bodies of the injured neurons housed in the intermediolateral cell column (IML) undergo robust plasticity, particularly at 1 week following survival. The preautonomic neurons of the paraventricular nucleus (PVN) of the hypothalamus have been shown to provide input to the IML neurons. In the present study we investigated whether any transneuronal effects of peripheral axon injury could be detected in uninjured neurons that provide input to the IML. At 1 week and 16 weeks following injury, a decrease in the number of synaptic inputs onto the injured preganglionic neuronal cell bodies in the IML was observed. In addition, while the number of neurons in the PVN that expressed corticotrophin releasing hormone (CRH) was decreased at 1 week after injury, the number of synaptic boutons onto CRH neurons in the PVN was increased. The results of this study reveal that retrograde signaling following injury to preganglionic axons can lead to robust changes in the chain of preautonomic neurons and suggest that uninjured neurons in the central nervous system can respond to injury to peripheral axons.

152

1. Introduction Our lab has previously reported glial cell activation in the intermediolateral cell column (IML) of the first thoracic segment of the spinal cord after transection of the axons in the cervical sympathetic trunk (CST) (Coulibaly and Isaacson, 2012). At 1 week following CST transection, we demonstrated the activation of astrocytes, and increased numbers of microglia and oligodendrocytes in the vicinity of the injured parent neuronal cell bodies located in the IML (Coulibaly and Isaacson, 2012). Our findings also revealed that astrocytic and microglia processes enveloped the injured neuronal cell bodies in the IML (Coulibaly and Isaacson, 2012). Such altered interactions between activated glia cells and injured neurons have been reported in other models of peripheral nerve injury, such as sciatic nerve transection (Coyle, 1998), as well as axotomy of the hypoglossal nerve (Svensson et al., 1993) and facial nerve (Kalla et al., 2001). The effects of this altered association of astrocytes and microglia with injured neurons are unknown, yet it has been suggested that altered glial-neuronal interactions contribute to the phenomenon of synaptic stripping, or loss of synaptic input onto the cell bodies and dendrites of injured neurons (Moreno- Lopez et al., 2011). The presence of synaptic stripping associated with the parent cell bodies of injured axons has been documented in other models of peripheral injury (Svensson et al., 1993; Kalla et al., 2001; Berg et al., 2013), yet it was not known whether this phenomenon occurred in the IML following CST transection. The first objective of this study was to determine whether transection of the CST resulted in synaptic stripping of the injured neuronal cell bodies located in the IML. Most studies on synaptic stripping have focused on the cell types involved in the displacement of synaptic input, the time lines involved in the plasticity, and the possible mechanisms surrounding these changes (Trapp et al., 2007; Gonzalez-Ferero and Moreno-Lopez, 2014). Though these studies enhance our understanding of the events that lead to injury induced synaptic stripping, few studies have addressed the effects of such synaptic remodeling on the parent neurons, those neurons on which the synapses have been displaced. Therefore, the second goal of this study was to determine whether CST transection, which occurs in the periphery, led to changes in brain regions that provide afferent input to the injured IML neurons.

153

Previous studies have demonstrated that the preganglionic sympathetic neurons located in the IML of the spinal cord receive afferent inputs from numerous regions of the brain, including the raphe nucleus, the nucleus solitarius and the paraventricular nucleus (PVN) of the hypothalamus (Loewy et al., 1978; reviewed in Smith and DeVito, 1984; Strack et al. 1989; Tucker and Saper, 1985; Hosoya et al., 1991; Affleck et al., 2012). These afferent inputs, specifically those located in the PVN, have been shown to play a role in regulating sympathetic output (Yamaguchi et al., 2009; Yamaguchi and Okada, 2009). The preautonomic neurons in the PVN synapse directly onto the preganglionic neurons in the IML of the thoracic spinal cord (Yamaguchi et al., 2009) and express corticotrophin releasing hormone (CRH; Yamaguchi and Okada, 2009), a neuropeptide responsible for the PVN regulation of sympathetic output (Yamaguchi and Okada, 2009). Indeed preganglionic sympathetic neurons in the thoracic segment of the spinal cord have been shown to express the CRH receptor (Korosi et al., 2007). Because recent studies have demonstrated transneuronal effects of peripheral axon transection (Scholz et al., 2005; Cowey et al., 2011; Leong et al., 2011), we hypothesized that the transection of the CST would lead to the plasticity of afferent input to preganglionic neurons in the IML, and that these effects would be discernable in PVN regions having projections to the spinal cord. Therefore, in the present study we examined both the IML and the PVN at several survival time points for evidence that peripheral injury induced changes in uninjured hypothalamic neurons. Because recent studies have demonstrated changes in central neurons that were two or more synapses removed from the injured neurons (Fuccio et al., 2009; Leong et al., 2011), the synaptic inputs onto CRH immunoreactive (-ir) neurons in the PVN also were analyzed.

2. Materials and Methods 2.1. Animal surgery and tissue collection Young adult (3 months of age) female Sprague Dawley rats (Harlan Labs, Indianapolis, IN) were kept in the Miami University Animal Facilities in a 12:12 light:dark cycle at regulated temperature. Animals were anesthetized using the inhalant isofluorane (2.5%). To access the CST for transection, a 3 cm ventral incision was made on the neck region of the animal. The CST was exposed, gently separated from surrounding tissue and transected 2 mm from its entry into the superior cervical ganglion (Sun and Zigmond, 1996). After the cut, both the proximal

154 and distal stumps were placed carefully next to each other, as close to the original position as possible. The procedure was repeated on the other side. The incision was closed using sutures and tissue glue (Nexaband, Phx, AZ). The transection surgeries were considered successful when the rats exhibited ptosis or eyelid droopiness. In the sham animals, the CST was exposed, but not transected. Following the surgery, animals were left to survive for 1 week (sham, n = 24; injury, n=24), 3 weeks (sham, n = 6; injury, n = 14) or 16 weeks (sham, n = 16; injury, n= 12). The number of animals in each group represents the total number of animals used in this study. Because not all tissues were suitable for every analysis, a detailed listing is provided below describing the number of animals that were used for each analysis. All methods used in this study were approved by the Miami University Institutional Animal Care and Use Committee (Protocol #825) and efforts were taken to minimize discomfort and pain to the animals. At each survival time point, animals were anesthetized with sodium pentobarbital (125 mg/kg), then transcardially perfused with 0.9% saline followed by 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB). Brain and spinal cord were removed from the animals. Brains were post-fixed in 4% PFA in 0.1 M PB solution for an additional 1 hour, rinsed and stored in 0.1M PB. Spinal cords were placed directly into 0.1M PB. Both spinal cord and brain were stored in PB at 4oC until further processing. The T1 segment was trimmed from the spinal cord since previous work in our lab demonstrated a robust glia activation only at the first thoracic (T1) segment of the spinal cord (Coulibaly and Isaacson, 2012). Prior to sectioning with the cryostat, brain blocks and the T1 segment of the spinal cord were cryoprotected by infiltrating with 30% sucrose in 0.1M PB at 4o C for 3-7 days, embedded in optimal cutting temperature medium (Ted Pella Inc.), frozen and sectioned using a MICROM HM 550 series cryostat. Spinal cord sections were cut in the coronal plane at a thickness of ~18 µm and mounted in series directly on Fisherbrand Superfrost Plus slides, with the first section on slide 1, the second on slide 2, etc. The T1 segment of the spinal cord generated approximately 15 slides with each slide containing a rostral-caudal series of 8-12 sections, with approximately 270 µm between each section. Prior to sectioning, the brain was trimmed to remove all regions anterior to the optic chiasm and all regions caudal to mammillary bodies. This resulted in a brain block containing the hypothalamus. The brain block was cut in the coronal plane at a thickness of ~25 um in a

155 rostral to caudal fashion and collected as floating sections in groups of five per well and stored in wells containing 0.1M PB at 4o C until further processing. 2.2. Procedures for locating the PVN of the hypothalamus Previous studies have demonstrated that the axons innervating the superior cervical ganglion are located predominantly in the IML of the thoracic level of the spinal cord (Rando et al., 1981; Strack et al., 1989; Poulat et al., 1992). In turn, these IML neurons receive synaptic input from the paraventricular nucleus (PVN) of the hypothalamus (Tucker and Saper, 1985; Strack et al. 1989; Hosoya et al., 1991; Affleck et al., 2012). Therefore, the PVN of the hypothalamus was the region of focus for determining whether transneuronal changes took place following the peripheral axon injury. Prior to immunohistochemical analysis, a rostral to caudal mapping of the brain was conducted to identify the location of the PVN and to ensure that the analysis was carried out at the same level in each brain. As mentioned above, the brain was blocked so that only tissue caudal to the optic chiasm and rostral to the mammillary bodies (~3.5 mm rostral-caudal extent) was sectioned. We found that this block contained the entire hypothalamus. One of the five sections per well collected during sectioning was processed for Nissl staining by first mounting the section onto gelatin coated glass slides and drying for 48 hours at room temperature. The slides then were dipped in cresyl violet acetate (0.1%), rinsed with double distilled water and dehydrated in an ethanol series, rinsed in xylene and coverslipped using Permount mounting medium (Fisher Scientific). The coverslipped sections were left to dry at room temperature overnight and viewed with an Olympus BX50 light microscope. Using the Rat Brain Atlas in Stereotaxic Coordinates (Paxinos and Watson, 1986), each Nissl stained section, which was representative of the remaining sections in that well, was mapped to a specific brain level. The presence of three anatomical landmarks was used for the identification of the sections that contained the PVN (Fig. 2.A.): the third ventricle (3V), the optic tract (OT), and the rostral aspects of the hippocampal formation. These landmarks were observed at 1.80 mm caudal to Bregma (-1.80 mm) and closely matched Figure 25 of the brain atlas (Paxinos and Watson, 1986). This brain level has been used in previous studies of the PVN (Flak et al., 2009) and thus was selected for analysis in this study. The remaining four brain sections from the well in which the Nissl section identified as -1.80 Bregma were then used for immunohistochemical analyses.

156

For the analysis of the volume of CRH neurons, brain sections from wells immediately rostral and caudal to -1.80 Bregma also were included in the analysis (three different sections) in order to include the rostral caudal extent of the PVN subnuclei. All other analyses described below were carried out on the section that was mapped to -1.80 Bregma (Fig. 3.A.; Fig. 25, Paxinos and Watson, 1986).

2.3. Immunohistochemical procedures Both mounted spinal cord sections and free floating brain sections were incubated overnight in 0.1M PBS-0.2% Triton-X solution, blocked with normal donkey serum. Spinal cord sections were incubated for 48 hrs at 4oC with a cocktail of mouse anti-synaptophysin (syn; 1:500; Millipore), rabbit anti-CRH (1:10,000; Gift from Dr. Wylie Vale) and goat anti-choline acetyltransferase (ChAT; 1:200; Millipore; to identify IML neurons in cord sections). Brain sections were incubated in a cocktail of mouse anti-syn (1:500) and rabbit anti-CRH (1:10,000). Following a series of rinses, sections were incubated for 2 hours in AlexaFluor conjugated antibodies (1:200; Molecular Probes) directed against the primary antibody host. The spinal cord sections were triple labeled with CRH, syn, and ChAT using donkey anti-rabbit AlexaFluor 594, donkey anti-mouse AlexaFluor 488, and donkey anti-goat AlexaFluor 647, respectively. Brain sections were double labeled with CRH and syn using donkey anti-rabbit AlexaFluor 594 and donkey anti-mouse AlexaFluor 488, respectively. Following three rinses in PB, mounted cord sections were immediately coverslipped using fluorescent mounting medium with 4', 6- diamidino-2-phenylindole (DAPI; Vectashield; Vector Labs). Free floating brain sections were mounted onto gelatin coated slides and coverslipped using the same mounting medium. Slides were stored at 4oC and in the dark until analyzed using confocal microscopy. Because the CRF antibody was being used for the first time, a series of sections was processed with all steps of the immunohistochemical labeling except for the primary antibody. Comparisons were made of the no primary antibody sections with sections that received primary antibody in the same experiment.

2.4. Analysis of syn and CRH in the spinal cord The spinal cords were examined for overall density of syn-ir and CRH-ir profiles as well as syn-ir and CRH-ir boutons contacting ChAT neurons in the IML. Z-stacks (~20 optical

157 sections taken at 1 µm intervals) of left and right IML from T1 spinal cord were obtained using the Zeiss 710 laser confocal microscope at 400X magnification with the 40X oil objective. The left and right IMLs from each animal were considered separate cases since the peripheral axons on each side of the animals were transected and the response to peripheral transection in the spinal cord has been determined to be unilateral. While acquiring the images, all microscope settings were kept the same for both CRH and syn across the sham and injury tissues at each time point. The settings were determined by quickly examining all stained tissues at a particular time point and then using the tissue with the lowest staining intensity (from either sham or injured group) to set the parameters. Using a composite image of collapsed Z-stacks taken at 400X, the density of CRH and syn immunoreactivity in the spinal cord was determined using Image Pro 6.3 software as previously described in Chapter 2 of this Thesis. Any syn-ir or CRH-ir profiles in the field of view that exhibited expression above the background expression were included in the analysis. Data from 1 week (n=4), and 16 week (n=3) survival times were compared to a combined control group (n=13). These findings are shown in Figure 2. The number of syn-ir or CRH-ir synaptic inputs onto ChAT-ir IML neurons was determined using the methods described by Flak and colleagues (2009). Four ChAT neurons per case used for analysis were chosen using the following three criteria: 1) cell showed obvious ChAT immunoreactivity; 2) cell was totally contained within the Z-stack; and 3) cell exhibited a DAPI positive nucleus. When scrolling through the Z-stack, the first four neurons that met these criteria were chosen for the analysis. Once chosen, the number of syn-ir or CRH-ir boutons directly apposed to the ChAT neuron was counted from three optical sections in the Z-stack: a scan in the middle of the stack, a scan above the middle scan and a scan below the middle scan. Boutons were included in the counts only when no discernable space was found between the bouton and the edge of the ChAT neuron (For reference, see syn-ir profiles in the brain in Fig. 2.C). The number of boutons contacting the neurons throughout the three optical sections was averaged to obtain an average number of boutons contacting that neuron. The number of boutons for the four different neurons was averaged to obtain a mean for each IML. These findings are shown in Figure 2. In each analysis of the spinal cord, the mean was obtained for the sham group at each time point and then each value (sham and injury values for that time point) was expressed as a

158 proportion of the sham mean. Because the shams from each time point were not significantly different from each other, the square root transformed values from the shams (1 week, n=4; 16 week, n=3) were combined to generate one control group. Animals from the survival time points of 1 week (n=4) and 16 week (n=3) were analyzed and compared to the combined control group (n=13) using a one way ANOVA followed by a Fisher’s post hoc comparison. Significance was ascribed at p < 0.05.

2.5. General analysis of brain tissues Images of brain sections were acquired at both 200X (20x objective) and 400X (40x oil objective) magnification using the Zeiss 710 laser confocal microscope. The images taken at 200X magnification contained both left and right sides of the PVN and were used to determine the number of CRH-ir neuronal cell bodies (which will be referred to in the rest of this document as ‘CRH neurons’) present in the PVN (Figs. 3; 4). Images that were taken at 400X contained only one side of the PVN and were used to characterize the volume of the CRH neurons (Fig. 4), the number of synaptic boutons onto CRH neurons (Figs. 5, 6) and the overall density of CRH-ir (Fig. 4) and syn-ir profiles (Fig. 6) in the PVN. All images (200X and 400X) were captured using both the “Z-stack” and “Tile” features of the microscope. Z-stacks of each PVN were obtained by collecting 9 -10 optical sections at 3 µm intervals. Tiling of the brain involved the acquisition of a patchwork of images of the brain that were ‘stitched’ together. Depending on the appearance and orientation of the brain section on the slide, 400X tiled images were acquired in either a 4 X 3 or 3 X 3 configuration while 200X images were acquired in a 3 X 3 or 3 X 2 configuration. The ‘stitched’ composite image resulted in a field of view that encompassed ~0.22 mm2 (12 tiles) or 0.17 mm2 (9 tiles) at 400X and 1.56 mm2 (9 tiles) or 1.06 mm2 (6 tiles) at 200X. The 200X images were used to analyze the density of CRH (Fig. 5.A.) and syn (Fig. 7.A.) immunolabeling. The density values obtained using Image Pro were divided by the area contained within each field of view that was analyzed. All other analyses were carried out using 400X images. Based on previous reports describing the location of CRH neurons having projections to the spinal cord (Yamaguchi et al., 2009; Yamaguchi and Okada, 2009), three specific subregions of the PVN were analyzed: 1) ventromedial parvocellular PVN (VM), a ventral portion of the PVN typically found in close proximity to the midregion of the third ventricle (Fig. 2.B.); 2)

159 dorsomedial parvocellular PVN (DM), a region within 200 µm from the dorsal tip of the third ventricle; 3) dorsolateral PVN (DL), a region located greater than 200 µm lateral to the dorsal tip of the third ventricle and lateral to DM (Fig. 2.B.). These regions were superimposed on the PVN as diagrammed in Paxinos and Watson (1986). At each survival time point, the tissue with the lowest staining intensity of syn or CRH (from either sham or injured group) was used to set the microscope parameters prior to acquiring images. These parameters were kept constant as images were acquired across all sham and injury cases. In all analyses, the left and right PVN from each animal were considered separate cases since the peripheral axons on each side of the animals were transected and the response has been determined to be unilateral. Similar to the analysis of the spinal cord, a mean was obtained for the sham group at each survival time point and all values (sham and injury) were expressed as proportion of the sham mean. Because the shams from each time point were not significantly different from each other, the square root transformed values of the shams (1 week, 3 week, 16 week) were combined to generate one control group. Analyses were carried out using a one way ANOVA followed by a Fisher’s post hoc comparison. Significance was ascribed at p<0.05. The number of cases analyzed for each specific data set is described below.

2.6. Determination of overall CRH density and the number of CRH neurons in the PVN Using a composite image of collapsed Z-stacks taken at 400X, the overall density of CRH immunoreactivity in the hypothalamus was determined using Image Pro software as previously described in Chapter 2. Any CRH-ir profiles in the field of view that exhibited CRH protein expression above the background expression and thus included CRH- neuronal cell bodies as well as CRH-ir dendritic and axonal profiles. Data from 1 week (n=6), 3 week (n=6), and 16 week (n=8) survival times were compared to a combined control group (n=22). Using a composite image of collapsed Z-stacks taken at 200X, the total number of CRH- ir neurons exhibiting a distinct DAPI nucleus in each PVN was determined using Image Pro 6.3. The effects of injury were determined in the overall PVN as well as in individual PVN subregions. The number of neurons in the overall PVN was obtained by adding the number of neurons counted in all three subregions (VM, DM, and DL). Data from 1 week (n=6), 3 week

160

(n=6) and 16 week (n=8) survival times were analyzed and compared to a combined control group (n=22). These findings are shown in Figure 5.B. and 5.C.

2.7. Determination of CRH neuronal volume in the PVN Z-stack images taken at 400X magnification were used to determine whether the injury affected the volume of CRH neuronal soma in the hypothalamus. The methods were based on previously used methods (Chapter 2) which were originally modified from Flak et al. (2009). The CRH neurons used for analysis were chosen using three criteria: 1) the cell showed obvious CRH immunoreactivity; 2) the cell was totally contained within the Z-stack; 3) the cell exhibited a DAPI positive nucleus. The first three cells per subdivision in each section that met these criteria were chosen for analysis. Sections from wells immediately rostral and caudal to -1.80 Bregma were included in the analysis. Therefore, for the three sections per case, a total of 9 cells for each subdivision was analyzed for cell volume. Because the number of CRH neurons was significantly reduced at 1 week, only ~6 cells per subdivision were analyzed at this time point. Using this approach the proportion of the total number of neurons that were present was ~75% of the total number of CRH neurons in the control cases, ~100% of the cells at 1 week survival time point, ~60% of the CRH neurons at the 3 week time point, and ~50% of the cells present at 16 weeks after injury. For each cell that was chosen for analysis, the radius was obtained by taking 16 total radii measurements within the Z stack. Specifically, four different measurements were taken from the nucleolus to the edge of the cell in four consecutive optical levels through the cell. These 16 measurements were then averaged to determine the mean radius of each cell, which was then entered into the formula for the volume of a sphere, (4/3) πr3 to obtain a volume for the neuron. The volumes taken from the neurons in each subdivision were averaged to obtain a value for the subdivision of that case. Animals from the survival time points of 1 week (n=6), 3 week (n=6), and 16 week (n=8) were analyzed and compared to the combined control group (n=22).

2.8. Analysis of syn-ir profiles in the PVN Using a composite image of collapsed Z-stacks taken at 200X, the density of syn immunoreactivity in the hypothalamus was determined using Image Pro 6.3 software as previously described in Chapter 2 of this thesis. Any profiles in the field of view that exhibited

161 syn protein expression above the background expression were included in the analysis of overall density of syn in the PVN as well as syn-ir profiles contacting CRH neurons. Data from 1 week (n=6), 3 week (n=6), and 16 week (n=8) survival times were compared to a combined control group (n=22). These findings are shown in Figure 7.A. The number of synaptic inputs onto CRH neurons and the number of syn-ir boutons in association with CRH neurons were characterized using Z-stack images taken at 400X magnification. The CRH neurons chosen for analysis met the following criteria: 1) the CRH neuron was contained within the Z-stack; 2) the CRH neuron had detectable cellular boundaries with discernable syn-ir puncta in apposition; 3) no space existed between the syn-ir puncta and the CRH-ir cell body; 4) the syn-ir profile was at least 2 pixels in diameter (See Fig. 3.C.). After scrolling through the Z-stack within each of the three PVN subdivisions, the analysis was carried out on the first two CRH neurons in each subdivision that met these criteria, leading to a total of 6 neurons analyzed per PVN. This approach led to the analysis of ~50% of the CRH neurons that were fully contained within the controls, ~100% of the neurons at 1 week, ~40% in the 3 week survival time points, and ~33% at the 16 week survival time point. The number of syn-ir boutons in apposition to each identified neuron was determined at three consecutive optical sections of the Z-stack, starting with the first optical section in which the cell exhibited a visible DAPI nucleus. The number of syn-ir boutons from the three optical sections through the cell were added to obtain the total number of inputs onto the neuron. Data obtained from the survival time points of 1 week (n=6), 3 week (n=6), and 16 week (n=6) were analyzed and compared to the combined control group (n=22). These findings are shown in Figure 7.

3. Results 3.1. Changes in syn and CRH immunoreactivity in the spinal cord Both syn and CRH immunopositive profiles in the IML were observed as punctate structures. Syn-ir puncta were abundant and could be found in close association with ChAT neurons as well as in the neuropil (Fig. 1.). CRH-ir profiles were rare and when present typically were in association with ChAT-ir neurons and, less frequently, within the neuropil (Fig. 1.). The overall density of syn immunoreactivity in the IML was unchanged at 1 week and 16 weeks following injury when compared to the control group (Fig. 2.A.). However, the number of syn-ir boutons making contact onto ChAT-ir neurons was significantly decreased at both 1 and 16

162 weeks after the injury (Fig. 2.B.). No changes were observed in the overall density of CRH immunoreactivity in the IML at either survival time point (Fig. 2.C.). However, the number of CRH-ir boutons contacting ChAT-ir neurons was decreased by 33% at 1 week after injury (Fig. 2.D.). At 16 weeks this value was similar to controls (Fig. 2.D.). When the number of boutons that co-localized CRH and syn was considered, no significant changes were observed at 1 week or 16 weeks (Fig. 2.E.).

3.2. Changes in CRH in the PVN following peripheral injury In the hypothalamus, CRH immunoreactivity was localized to cell bodies as well as to numerous profiles throughout the PVN that were likely dendrites and axons (Figs. 3, 4). The Image Pro software settings and the ‘density’ analysis thus included the analysis of all CRH immunopositive profiles in the PVN. As shown in Figure 5.A., no changes in the density of CRH-ir in the PVN were observed at 1 week after injury. However, at the 3 week time point, a significant 59% decrease was observed in the density of CRH profiles in the PVN (Fig. 5.A.). At 16 weeks after the injury, the density of CRH in the PVN returned to control values (Fig. 5.A.). CRH neurons, which exhibited a ring of CRH immunoreactivity that surrounded the nucleus, were numerous in the PVN (Fig. 4.). In the controls, CRH positive neurons averaged ~67 CRH-ir neurons per section in the PVN (Fig. 5.). At 1 week following injury, the number of CRH neurons in the entire PVN was significantly decreased by 45% (Figs. 4., 5.B.). At this survival time point, the number of CRH neurons was decreased in each of the PVN subregions, with respective decreases of 72%, 33%, and 49% in the VM, DM, DL (Fig. 5.C.). At 3 weeks and 16 weeks following injury, the number of CRH neurons was similar to control values (Fig. 5.C.). The volume of CRH neurons in the PVN averaged ~573 µm3 in the control cases and no changes in the cell volume were observed at any survival time point when the entire PVN was taken into account (Fig. 5.D.). Similarly the analysis of CRH cell volume in each of the PVN subdivisions revealed little change in volume with the exception of a small 8% decrease in cell volume in the dorsomedial PVN at 16 weeks after injury (Fig. 5.D.).

3.3. Changes in syn-ir profiles in the PVN following peripheral injury

163

To determine the effects of injury on synaptic inputs in the hypothalamus, the overall density of syn immunoreactivity and the number of synaptic boutons contacting CRH neurons were analyzed. Syn-ir puncta were observed throughout the hypothalamus and only a subset of these puncta made direct contact onto CRH-ir neurons as shown in Figure 3 and the quantification of these inputs is shown in Figure 7. At 1 week following injury, the density of syn-ir profiles throughout the hypothalamus was increased by 400% (Figs. 6; 7.A.). At 3 and 16 weeks after injury the density of syn-ir puncta in the hypothalamus was similar to control cases (Fig. 7.A.). The number of syn-ir boutons contacting CRH neurons was analyzed in the overall PVN as well as in each of the subregions. At 1 week following injury, when the entire PVN was taken into account, the number of syn-ir boutons contacting CRH-ir neurons was increased by 38% (Fig. 7.B.). Similar increases were observed in each subregion, with respective increases of 35%, 32%, and 44% in the VM, DM, and DL (Fig. 7.C.). Interestingly, at 3 weeks, the number of syn boutons contacting CRH neurons was significantly decreased by 42% in the entire PVN when compared to controls (Fig. 7.B.). Yet no changes were detected in any of the individual subregions. By 16 weeks after injury, no change in the number of syn-ir inputs was observed (Fig. 7.C).

164

syn syn ChAT DAPI

A. A’. A’’.

* * Cont

B. B’. B’’.

* *

1 1 wk

C. C’. C’’.

* * 16 16 wk

Figure 1

165

Figure 1: Immunohistochemical analysis of synaptophysin (syn) and corticotrophin releasing hormone (CRH) in the intermediolateral cell column (IML) at T1 level of the spinal cord following injury. CRH (white; white arrows) and syn (green; yellow arrows) immunoreactive (- ir) puncta were observed in the neuropil as well as in association with ChAT-ir preganglionic neurons (asterisks) in the controls (sham; A.-A’’.), as well as the 1 week (1 wk; B.-B”.), and 16 week (16 wk; C.-C”.) survival time points. DAPI (blue) was used to show the cellular nature of labeled profiles. Scale bar for all images = 100 µm.

166

A. B. Density of syn-ir in IML Syn-ir profiles onto ChAT-ir

1.2 1.2 )

) neurons in IML rt

rt 1.0 1.0

sq ( 0.8 0.8 * *

0.6 0.6 control

of 0.4 0.4 0.2 0.2

roportion 0.0 0.0 Proportion of control ( controlof Proportion P shamcont 1 wk 16 wk cont 1 wk 16 wk

C. Density of CRH-ir in IML D. CRH-ir profiles onto ChAT-ir 1.2

) 1.2 neurons in IML rt

sq 1.0 1.0 0.8 0.8 *

control ( control 0.6 0.6 0.4 0.4

0.2 0.2 Proportion of control (sq rt)(sq controlof Proportion

Proportion of Proportion 0.0 0.0 contsham 1 wk 16 wk cont 1 wk 16 wk

E. Profiles colocalizing CRH and

1.2 syn in IML

) rt

sq 1.0 0.8 0.6 0.4 0.2

Proportion of control ( controlof Proportion 0.0 cont 1 wk 16 wk

Figure 2

167

Figure 2: Evidence for synaptic stripping in the intermediolateral cell column (IML) of the spinal cord after peripheral axon transection. A. The overall density of synaptophysin (syn) immunoreactivity in the IML of the spinal cord was unchanged following injury. B. While overall density was not changed, the number of syn immunoreactive (-ir) inputs onto cholinergic (ChAT-ir) preganglionic neuronal cell bodies in the IML was decreased at 1 week (1 wk) and 16 weeks (16 wk) following injury. C. The overall density of corticotrophin releasing hormone (CRH) immunoreactivity in the IML of the spinal cord was unchanged at 1 and 16 weeks after injury. D. While the overall density was unchanged, the number of CRH terminals contacting ChAT neurons was decreased at 1 week after injury. E. No changes in the number of synapses colocalizing CRH and syn were observed following injury. *, p< 0.05.

168

A. B.

DL DM 3 PVN VM PVN 3V

OT OT

200µm C. D.

C. . DL DM

VM 3V 3V

* * * *

* CRH * * * CRH SYN SYN DAPI DAPI DAPI

Figure 3

169

Figure 3: Experimental approach for quantification of changes in the paraventricular nucleus (PVN) after peripheral axon transection. A. Plate from Paxinos and Watson (1986) rat brain atlas showing the level of the brain used in the analysis. The plate corresponds to Bregma –1.80 mm of the rat brain (Paxinos and Watson, 1986). B. Higher magnification view of the red boxed area on the left. The represented plate shows the location of the paraventricular nucleus (PVN) as well as the optic tract (OT), a landmark used in the identification of the PVN. The majority of the analyses were performed at this level. The three subdivisions: ventromedial PVN (VM; red), dorsomedial PVN (DM; green), and dorsolateral PVN (DL; blue) analyzed in this study can be seen inside the yellow boxed region. C. Coronal sections through the level in A. stained for Nissl. Landmarks such as optic tract (OT), third ventricle (3V) were used to identify the appropriate brain level. Scale bar = 1 mm. D. Higher magnification of the area within the yellow box shows additional details of the 3 subdivisions of the PVN used for quantification. Scale bar = 500 µm. E. Confocal micrographs demonstrating CRH-ir neurons (white asterisks) and syn immunoreactive (-ir) profile (green) in the PVN. Quantification of syn-ir inputs onto CRH neurons was conducted by counting the number of syn-ir profiles observed in close proximity to the perimeter of the CRH neurons (arrows). Dashed arrows represent boutons that were not counted. DAPI (blue) shows cellular nature of labeled profiles. Scale bar = 25 µm.

170

VM DM DL A. cont E. I.

CRH DAPI

B. 1 wk F. J.

CRH DAPI 3 wk C. G. K.

CRH DAPI

D. 16 wk H. L.

CRH DAPI

Figure 4

171

Figure 4: Analysis of CRH immunoreactive (-ir) neurons (arrows) in PVN subdivisions following peripheral axon injury. Representative confocal micrographs from the ventromedial (VM; A.-D.), dorsomedial (DM; E.-H.), and dorsolateral (DL; I.-L.) subdivisions of the PVN. At 1 week post injury, the number of CRH-ir neurons appeared to be decreased in each of the three PVN subdivisions. (B.-B’’.). DAPI (blue) shows the cellular nature of labeled profiles. Scale bar for all images = 25 µm.

172

A. Density of CRH in hypothalamus

1.4 )

rt 1.2

sq 1 0.8

of of ( contorl 0.6 * 0.4

Proportion 0.2 0 cont 1 wk 3 wk 16 wk

Number of CRH-ir neurons in B. C. Number of CRH-ir Neurons in PVN subdivision overall PVN 1.2

) 1.4 rt

sq 1.0 1.2 0.8 1.0 * 0.8 0.6 * * 0.6 0.4 * 0.4 0.2

Proportion controlof ( 0.2 Proportion of (sqcontrol rt) 0.0 0.0 contsham 1 wk 3 wk 16 wk VM DM DL cont 1 wk 3 wk 16 wk

D. Volume of CRH-ir neurons in overall PVN E. Number of CRH-ir neurons in PVn subdivisions

1.0 1.4 ) ) rt 1.2 sq 0.8 * 1.0 0.6 0.8

of of ( control 0.4 0.6 0.4 0.2

roportion 0.2

P Proportion of (sqcontrol rt) 0.0 0.0 shamcont 1 wk 3 wks 16 wk VM DM DL cont 1 wk 3 wk 16 wk Figure 5

173

Figure 5: Changes in CRH in the PVN after peripheral axon injury. A. The density of CRH immunoreactivity in the hypothalamus was decreased in CRH density at 3 weeks (3wk) following injury with no changes observed at 1 week (1wk) or 16 weeks (16 wk). B. At 1wk the number of CRH neurons in the overall PVN was decreased while numbers were similar to controls at 3 wk and 16 wk following the injury. C. At 1 wk, the number of CRH neurons was decreased in each of the PVN subdivisions, with no changes observed at 3 wk or 16 wk following the injury. D. No changes were observed in the volume of CRH neurons in the overall PVN. E. A slight but significant decrease was observed in the volume of CRH neurons in the dorsomedial (DM) at 16 week after injury. *, p< 0.05. DL, dorsolateral; VM, ventromedial.

174

VM DM DL A. Ccont E’. I.

SYN DAPI

B. 1 wk F. J.

C. 3 wk G. K.

D. 16 wk H. L.

Figure 6

175

Figure 6: Syn immunoreactive (-ir) profiles in the PVN following peripheral axon injury. Representative confocal micrographs from the ventromedial (VM; A. -D.), dorsomedial (DM; E.-H.), and dorsolateral (DL; I.-L.) subdivisions of the PVN. At 1 week (1wk), the number of syn-ir puncta (arrows) appeared to be increased in each of the three subdivisions (B., F., J.). DAPI (blue) shows cellular nature of labeled profiles. Scale bar for all images = 25 µm.

176

A. Density of syn-ir profiles in hypothalamus

) 6.0

rt * sq ( 5.0 4.0

control control 3.0 of 2.0

1.0

roportion P 0.0 cont 1 wk 3 wk 16 wk

Syn-ir profiles onto B. CRH neurons in C. Syn-ir profiles onto CRH neurons in the PVN overall PVN subdivision 1.6 * 1.6 ) ) * * * rt 1.4

1.4 sq ( 1.2 1.2 1.0 1.0 0.8

control control 0.8 * of 0.6 0.6 0.4 0.4

0.2 0.2

Proportion of control (sq rt)(sq controlof Proportion roportion

P 0.0 0.0 cont 1 wk 3 wk 16 wk VM DM DL cont 1 wk 3 wk 16 wk

Figure 7

177

Figure 7: Synaptophysin (syn) immunoreactivity is increased in the PVN following peripheral axon injury. A. At 1 week (1 wk) after injury, the overall density of syn-ir profiles in the hypothalamus was increased. No changes were observed at 3 weeks (3wk) or 16 weeks (16 wk). B. The number of syn immunoreactive (-ir) profiles contacting CRH neurons was increased in the overall PVN, followed by a decrease at 3 weeks (3 wk). Numbers were similar to controls at 16 weeks (16 wk). C. At 1 wk the number of syn-ir profiles contacting CRH neurons was increased in each PVN subdivision with no changes at 3wk or 16wk. *, p< 0.05. DM, dorsomedial; DL, dorsolateral; VM, ventromedial.

178

4. Discussion 4.1. Summary of findings The results of this study revealed that the number of synaptic boutons contacting the injured IML neurons in the spinal cord was dramatically reduced at 1 week following transection of the CST. This synaptic remodeling was long lasting, as the number of boutons contacting IML neurons remained decreased 16 weeks after injury, indicating chronic changes in the afferent input to the injured neurons in the IML, and thus providing evidence for synaptic stripping, a phenomenon observed in other injury models. At the 1 week survival time point, the number of CRH-ir boutons in the IML also was significantly decreased. In turn, CRH neurons in the PVN, which provide afferent input to the IML neurons (reviewed in Smith and DeVito, 1984; Strack et al. 1989; Tucker and Saper, 1985; Hosoya et al., 1991; Affleck et al., 2012), were dramatically impacted by the peripheral injury. Indeed, the number of CRH neurons in the PVN was significantly decreased at 1 week following the injury, at a time when syn-ir and CRH-ir boutons in the IML also were decreased. The afferent input to PVN neurons also was affected, as the number of synaptic boutons in the PVN were increased. To our knowledge, we are the first to demonstrate transneuronal changes in the PVN following transection of preganglionic axons in the periphery. Our findings indicate that injury to peripheral axons can have retrograde effects on uninjured neurons in the central nervous system.

4.2. Synaptic stripping in the IML of the spinal cord. In the present study, the density of syn immunoreactivity in the IML and the number of syn-ir boutons contacting IML neurons in the spinal cord were used to determine whether any evidence of synaptic stripping was present. Indeed, others have used the loss or displacement of syn-ir boutons as a measure of synapse removal from injured neuronal cell bodies or dendrites (reviewed by Moreno-Lopez et al., 2011). Following CST transection, the overall density of syn-ir profiles in the IML was unchanged, yet the number of syn-ir boutons making contact onto IML neurons was significantly reduced. The removal of afferent synapses from an injured neuronal cell body is a phenomenon that has been documented in several injury models (Moreno-Lopez et al., 2011), such as the hypoglossal nucleus (Svensson et al., 1993) and facial nucleus (Kalla et al., 2001), following transection of the respective nerves, as well as in the ventral horn following sciatic nerve transection (Berg et al., 2013).

179

Following CST transection, we previously demonstrated that both astrocyte and microglia undergo morphological changes from quiescent to reactive and also exhibit cytoplasmic processes that surround the cell body of the injured neurons (Coulibaly and Isaacson, 2012). It is possible that these enhanced glial-neuronal interactions played an important role in displacing afferent inputs from the cell body of the injured neuron (Svensson et al., 1993; Moreno-Lopez et al., 2011). Indeed, the removal of synaptic inputs onto injured neurons has been attributed to glial cell activation in the central nervous system. For example, activated or reactive astrocytes (Kapadia and LaMotte, 1987; Svensson et al., 1993) and microglia (Trapp et al., 2007; Kettenmann et al., 2012; Berg et al., 2013) appear to play a role in synaptic remodeling in the ventral horn after axotomy of the motorneurons. Though we observed changes in the synaptic input to IML cell bodies, we were not able to determine whether input to dendrites was affected with our experimental approach. Brannstrom and Kellerth (1998) demonstrated that synaptic stripping was more prominent on neuronal cell bodies compared on dendrites (1998), yet it is possible that synaptic input onto dendrites also was affected by the CST transection. However, because there were no changes in the overall density of syn-ir or CRH-ir profiles in the IML, we conclude that the synaptic input onto the injured cell bodies, rather than dendrites, was primarily affected by the CST injury. The decrease in synaptic inputs onto ChAT neurons in the IML was still present at 16 weeks following injury even though no reactive gliosis was observed past the 3 week survival time point (Coulibaly and Isaacson, 2012). These findings suggest that subtle effects of injury remain well past the appearance of gliosis. Brannstrom and Kellerth (1999) demonstrated that synaptic remodeling subsided only after motorneurons had reinnervated their targets, which in their cat axotomy model occurred after 12 weeks of survival. In our model, the reinnervation of the SCG after CST transection appears to be continuing at 16 weeks following the injury (Chapter 2), supporting the idea that changes in the spinal cord, though subtle, would be expected even at the 16 week survival time point. Although no changes were observed in the number of boutons coexpressing CRH and syn in the IML, there were trends for a decrease at both 1 week and 16 weeks after the injury. Although CRH inputs onto IML neurons have been described (Yamagucchi and Okada, 2009), CHR inputs onto the IML neurons were difficult to visualize in our model. Though beaded CRH axonal profiles were present, few were colocalized with syn near the neuronal cell bodies. The

180 apparent reduced number of CRH/syn boutons might have reached significance with a larger sample size. In addition, the use of antigen retrieval protocols may have provided better CRH localization.

4.3. Changes in the number of CRH neurons in the PVN following CST transection Preganglionic sympathetic neurons located in the IML receive afferent inputs from numerous regions of the brain, including the nucleus solitarius and the paraventricular nucleus (PVN) of the hypothalamus (Loewy et al., 1978; reviewed in Smith and DeVito, 1984; Tucker and Saper, 1985; Strack et al. 1989; Hosoya et al., 1991; Affleck et al., 2012). A significant decrease in the number of CRH positive neurons in the PVN was observed at 1 week following injury and this decrease was observed in all of the subdivisions that were examined. However, by 3 weeks the number of CRH neurons was similar to control values, suggesting that these cells did not die but only down regulated their expression of CRH. The loss of neurotransmitter expression is a phenomenon previously described in the axotomized neurons of the hypoglossal nucleus (Lams et al., 1988; Armstrong et al., 1991), nucleus ambiguous (Chang et al., 2004), and the motorneurons of the spinal cord (Peddie and Keast, 2011) and in the IML neurons following CST transection (Chapter 2). Such injury-related changes in neurotransmitter expression in the uninjured neurons in the PVN support the idea that a retrograde signal from the injured neurons in the IML influences the uninjured PVN neurons that provide afferent input to the IML. Though the number of CRH positive neurons was decreased in the PVN following CST transection, the cell volume of PVN neurons was relatively unchanged. Changes in cell size have been reported in neurons directly affected by injury (Peddie and Keast, 2011; Chapter 2). Because the PVN neurons were not directly affected by the injury, our data support the idea that neurotransmitter plasticity, but not alterations in cell volume, may be more indicative of the transneuronal effects of injury. However, it is possible that undetected volume changes in preautonomic PVN neurons occurred at the 1 week time point. Because the number of CRH positive neurons was decreased and the volume measurements were taken only of the CRH positive neurons, we likely did not analyze the volume of the neurons that responded to the injury. A similar problem was seen in a previous study in which the volume of IML neurons was measured following CST transection (Chapter 2). Since ChAT expression was decreased in IML

181 neurons at 1 week following injury, many of the neurons affected by the injury were not labeled with ChAT. In order to visualize the injured IML neurons, they were labeled with NeuN for obtaining the volume measurements. This approach worked well because the IML contained only neurons that were injured. However, the hypothalamus contains numerous populations of neurons in addition to the preautonomic CRH neurons (Levy and Tasker, 2012) and all of these neurons would label with NeuN. Therefore, this would not be a useful approach for identifying the PVN neurons affected by the injury. The discrepancy observed between the number of CRH neurons in the PVN and the overall density of CRH immunoreactivity in the hypothalamus was interesting. At 1 week, no changes were observed in the overall density of CRH immunoreactivity in the hypothalamus, but significantly fewer CRH positive neurons were present. On the other hand, at 3 weeks no difference in the number of CRH positive neurons in the PVN was observed but the overall density of CRH in the hypothalamus was decreased. The CRH density measurements included CRH positive neurons as well as dendritic and axonal processes in the hypothalamus. Because there were no changes in CRH positive neurons at 3 weeks, the decrease in overall CRH density likely was the result of reduced CRH in the immunoreactive processes. Possible origins for these processes include the CRH neurons in the PVN, or possibly the amygdala or hippocampus, both of which contain CRH neurons and project to the hypothalamus. Though speculative, it may be that a molecule from the injury site was retrogradely transported to the cell body of the preautonomic CRH neurons in the PVN, where it affected neurotransmitter levels. Unlike small neurotransmitters, neuropeptides such as CRH are translated in the neuronal cell bodies and then transported through vesicles to the axon terminal (Delcomyn, 1998). Because CRH is actively transported to the axon terminal, at 1 week after injury, when the protein levels are down in the cell body, the CRH protein already in transit along the axons might not yet be affected and remained abundant. However, at later time points, such as 3 weeks, CRH present in the axons might be either utilized or degraded, resulting in a decrease in the axons. Alternatively, it is possible that the change in CRH immunoreactivity observed at 3 week is a consequence of a downregulation in the amygdala and the hippocampus and/or a decrease in CRH transport from these regions to the hypothalamus. Indeed, it has been shown that PVN

182 neurons receive afferent inputs from these regions, which in turn regulate sympathetic (Levy and Tasker, 2012) and HPA axis outputs (Jankord and Herman, 2008).

4.4. Other models showing transneuronal effects following injury The transneuronal effects of injury have been demonstrated in other injury models. For example, decreased neuronal size was observed in the lateral geniculate nucleus and the superior colliculus following degeneration of the optic nerve (Shibuya et al., 1993). These findings demonstrated changes in target cells following loss of afferent input. In another model, Leong and colleagues (2011) observed that ligation of the L5 spinal nerve led to neuronal cell death and glial activation in the dorsal horn, the termination of sensory afferents of the spinal nerve, as well as death in the RVM, which would likely represent the second neuron in the relay of neurons entering the CNS, showing an anterograde transneuronal effect of the injury. They proposed that these changes contributed to the onset and maintenance of neuropathic pain in their model (Leong et al., 2011). In our model, we observed changes in CRH expression and CRH neurons in the PVN of the hypothalamus after peripheral transection of the preganglionic axons. Unlike the above models, our results show that injury can retrogradely affect uninjured neurons in the CNS.

4.5. Changes in synaptic input to CRH neurons in hypothalamus The overall density of syn-ir profiles in the PVN as well as number of syn-ir inputs specifically onto CRH neurons showed plasticity after injury, suggesting that the peripheral injury may have affected the neurons that innervate the CRH neurons of the PVN. Previous studies have demonstrated multiple levels of plasticity in the central nervous system as a result of peripheral axon injury. For example, Fuccio and colleagues (2005) demonstrated that injury to the sciatic nerve can lead to changes in expression of caspases in the orbito-frontal cortex of mice, which is more than 3 neurons removed from the injured neurons. Previous studies have demonstrated that CRH neurons in the PVN receive inputs from many regions of the brain, including the hippocampus and amygdala (Reviewed in Jankord and Herman, 2008). Some of these inputs have been shown to either increase or decrease the production of CRH in the hypothalamus. For example, the stimulation of the ventral hippocampus inhibits CRH expression (Casady and Taylor, 1976), while stimulation of the

183 amygdala increases CRH in the hypothalamus (Redgate and Fahringer, 1973). Indeed, the activation of the amygdala has been shown to interact with the stress nuclei of the hypothalamus (Janhord and Herman, 2008). Furthermore, the activity of the basolateral amygdala was increased in response to a variety of stress paradigms (Sawchenko et al., 2000). It is possible that the input from the amygdala might be affected by the peripheral axon injury, resulting in increased activation of the PVN. Additional experiments will need to be done using the syn antibody in conjunction with inhibitory and excitatory neurotransmitter markers to determine which specific inputs are affected by the injury. Changes in the density or number of synaptic boutons appear to coincide with alterations in the number of synapses (Moreno-Lopez et al., 2006). Therefore, the increase in syn immunoreactivity observed in the hypothalamus may indicate an increase in the number of synapses after injury, which in turn could be attributed to an increased number of inputs onto the CRH positive neuronal cell body. It has been recently shown that chronic stress can increase the number of excitatory inputs onto CRH neurons in the hypothalamus (Flak et al., 2009). In the present model, the peripheral axon injury would be considered a stressor for the injured animals and, therefore, the increase in synaptic input in the PVN in our model could result from an increase in excitatory inputs. The analysis of synaptic inputs onto CRH neurons at 1 week likely did not include many of the neurons of interest in the PVN, since the number of CRH immunopositive neurons was dramatically decreased. Therefore, we do not know whether the synaptic input to neurons with undetectable CRH also was altered. While the neurons with decreased CRH may also show increased synaptic input similar to the remaining CRH immunopositive neurons in the PVN, it also is possible that the neurons with decreased CRH experienced a decrease in synaptic input, and the synaptic boutons remodeled to hyperinnervate the remaining CRH neurons. The use of retrograde tracers administered to the spinal cord at the time of injury may help us to identify the neurons synapsing onto IML neurons, and also to determine whether the injury affected the synaptic inputs onto those preautonomic neurons.

4.6. Possible shortfalls of the present study One of the primary problems with the analyses carried out the present study was the use of CRH to identify the preautonomic neurons in the PVN. The neurons in the PVN can be

184 divided into three specific groups, the magnocellular neurodendocrine neurons, the parvocellular neuroendocrine neurons, and the parvocellular autonomic neurons (Levy and Tasker, 2012). The parvocellular neurons, which are either neuroendocrine or autonomic, express CRH (Jankord and Herman, 2008). Studies have shown that the two different CRH neuronal subnuclei are found in distinct locales within the PVN. For example, the majority of CRH neuroendocrine neurons are located primarily in the region labelled dorsomedial in the present study (Rho and Swanson, 1999), while the location of the CRH preautonomic neurons corresponds to the dorsolateral region in our study (Strack et al., 1989; Rho and Swanson, 1999). In addition, the ventromedial PVN seems to house both neuroendocrine and preautonomic CRH neurons (Rho and Swanson, 1999). Therefore, it is very possible that in the present study the effects of peripheral injury on the preautonomic neurons were diluted by the inclusion of the neuroendocrine CRH neurons within our counts. It would be interesting to perform the same analysis conducted in the present study while focusing on the preautonomic population alone. That could be done by using another marker in addition to CRH to label the preautonomic neurons. Indeed, it has been shown that oxytocin is expressed by preautonomic CRH neurons in the PVN (Swanson et al., 1980; Jansen et al., 1995). Finally, our use of CRH as a preautonomic marker most likely affected our analyses of the number of CRH neurons, their cellular volume, and the number of synapses found on their cell bodies since the number of CRH neurons was decreased in response to the injury and prevented us from examining their anatomy and synaptic inputs. Though the use of CRH may not have been optimal, it did not impede our ability to detect changes in the population of CRH neurons after injury. Indeed, a 40% decrease in the total number of cells expressing CRH in the PVN was documented at 1 week. If we assume that only the CRH preautonomic neurons will respond to the injury, then we postulate that the CRH neurons with reduced CRH expression, ~40% of the total number of CRH neurons, are those directly innervating the preganglionic neurons of the IML. The loss of CRH expression in these neurons may have led to their exclusion in many of our analyses. . Because the expression of CRH was affected in PVN neurons, and we are certain that our analyses included both neuroendocrine and preautonomic CRH neuronal populations, we cannot say with 100% confidence that the peripheral injury had no effect on the volume of preautonomic CRH neurons. Indeed at 1 week, if we assume that the neurons that lost their CRH expression are those projecting to the IML, then at this time point it is very likely that our

185 morphometric analysis of CRH neurons excluded the specific neurons of interest. In addition, using the same logic, the analysis of synaptic inputs onto CRH neurons at 1 week, did not include any of the neurons of interest. Therefore, the increase observed at this time point is specific to the inputs on the neuroendocrine CRH population, which raises the question as to why peripheral injury affected these neurons. It could be due to their proximity to the preautonomic neurons, or maybe the stress of the injury led to changes in the stress pathway, which includes the neuroendocrine CRH neurons (Jankord and Herman, 2008).

5. Conclusions This study provides evidence that the transection of preganglionic axons in the periphery led to decreased synaptic inputs onto the injured cell bodies in the IML. Interestingly, the time course for the observed changes in synaptic input seems to parallel the time course of glial activation in the spinal cord as well as the reinnervation of the SCG (Chapter 2). Much of the robust plasticity in the hypothalamus also occurred at 1 week after injury. This parallels the changes observed in spinal glial cells (Coulibaly and Isaacson, 2012; Chapter 4, Chapter 5) and injured IML neurons (Chapter 2). These changes in both the hypothalamus and spinal cord suggest an early acute and transient phase in the injury induce plasticity. Indeed, all the changes in CRH observed in the hypothalamus at 1 week subsided by 3 weeks after injury. The number of synaptic inputs onto the CRH neurons was increased soon after injury, suggesting that peripheral axon injury can affect synaptic inputs to neurons in the hypothalamus. Similar to the spinal cord, changes were observed in the hypothalamus at 16 weeks, suggesting possible long term effects of the injury. All of these results suggest the presence of retrograde injury molecules that influence the injured neurons as well as their connections in the CNS. Though yet unknown, these injury molecules may be key to understanding why some neurons survive and regenerate after injury and others do not.

186

6. References

Affleck, V. S., Coote, J. H., Pyner, S. (2012). The projection and synaptic organization of NTS afferent connections with presympathetic neurons, GABA and nNOS neurons in the paraventricular nucleus of the hypothalamus. Neuroscience, 219, 48–61.

Berg, A., Zelano, J., Pekna, M., Wilhelmsson, U., Pekny, M., Cullheim, S. (2013). Axonal regeneration after sciatic nerve lesion is delayed but complete in GFAP- and vimentin-deficient mice. PloS One, 8(11), e79395.

Brännström, T., Kellerth, J. O. (1998). Changes in synaptology of adult cat spinal alpha- motoneurons after axotomy. Experimental Brain Research, 118(1), 1–13.

Casady, R. L., Taylor, A. N. (1976). Effect of electrical stimulation of the hippocampus upon corticosteroid levels in the freely-behaving, non-stressed rat. Neuroendocrinology, 20(1), 68–78.

Coulibaly, A. P., Isaacson, L. G. (2012). Transient changes in spinal cord glial cells following transection of preganglionic sympathetic axons. Autonomic Neuroscience, 168(1-2), 32–42.

Cowey, A., Alexander, I., Stoerig, P. (2011). Transneuronal retrograde degeneration of retinal ganglion cells and optic tract in hemianopic monkeys and humans. Brain, 134(Pt 7), 2149–57.

Coyle, D. E. (1998). Partial peripheral nerve injury leads to activation of astroglia and microglia which parallels the development of allodynic behavior. Glia, 83, 75–83.

Delcomyn, F. (1998). Foundations of neurobiology. New York, NY. WH Freeman and Company.

Flak, J. N., Ostrander, M. M., Tasker, J. G., & Herman, J. P. (2009). Chronic stress-induced neurotransmitter plasticity in the PVN. Journal of Comparative Neurology, 517(2), 156–65.

187

Fuccio, C., Luongo, C., Capodanno, P., Giordano, C., Scafuro, M. A., Siniscalco, D., Berrino, L. (2009). A single subcutaneous injection of ozone prevents allodynia and decreases the overexpression of pro-inflammatory caspases in the orbito-frontal cortex of neuropathic mice. European Journal of Pharmacology, 603(1-3), 42–9.

González-Forero, D., & Moreno-López, B. (2014). Retrograde response in axotomized motoneurons: Nitric oxide as a key player in triggering reversion toward a dedifferentiated phenotype. Neuroscience, 283, 138-165.

Hosoya, Y., Sugiura, Y., Okado, N., Loewy, A. D., & Kohno, K. (1991). Descending input from the hypothalamic paraventricular nucleus to sympathetic preganglionic neurons in the rat. Experimental Brain Research, 85, 10–20.

Jankord, R., & Herman, J. P. (2008). Limbic regulation of hypothalamo-pituitary-adrenocortical function during acute and chronic stress. Annals of the New York Academy of Sciences, 1148, 64–73.

Jansen, A. S., Nguyen, X. Van, Karpitskiy, V., Mettenleiter, T. C., & Loewy, A. D. (1995). Central command neurons of the sympathetic nervous system: basis of the fight-or-flight response. Science, 270, 644–6.

Kapadia, S. E., LaMotte, C. C. (1987). Deafferentation-induced alterations in the rat dorsal horn: I. Comparison of peripheral nerve injury vs. rhizotomy effects on presynaptic, postsynaptic, and glial processes. Journal of Comparative Neurology, 266, 183–97.

Kettenmann, H., Kirchhoff, F., Verkhratsky, A. (2013). Microglia: new roles for the synaptic stripper. Neuron, 77, 10–8.

Korosi, A., Baram, T. Z. (2010). Plasticity of the stress response early in life: mechanisms and significance. Developmental Psychobiology, 52, 661–70.

188

Lams, B. E., Isacson, O., Sofroniew, M. V. (1988). Loss of transmitter-associated enzyme staining following axotomy does not indicate death of brainstem cholinergic neurons. Brain Research, 475, 401–6.

Levy, B. H., Tasker, J. G. (2012). Synaptic regulation of the hypothalamic-pituitary-adrenal axis and its modulation by glucocorticoids and stress. Frontiers in Cellular Neuroscience, 6, 24.

Loewy, A. D. (1991). Forebrain nuclei involved in autonomic control. Progress in Brain Research, 87, 253–68.

Loewy, A. D., Burton, H. (1978). Nuclei of the solitary tract: efferent projections to the lower brain stem and spinal cord of the cat. Journal of Comparative Neurology, 181(2), 421–49.

Moreno-López, B., Sunico, C. R., González-Forero, D. (2011). NO orchestrates the loss of synaptic boutons from adult “sick” motoneurons: modeling a molecular mechanism. Molecular Neurobiology, 43, 41–66.

Paxinos, G., Watson, C. (1986). Rat Brain: in Sterotaxic coordinates. Orlando, FL. Academic Press, Inc.

Peddie, C. J., Keast, J. R. (2011). Pelvic Nerve Injury Causes a Rapid Decrease in Expression of Choline Acetyltransferase and Upregulation of c-Jun and ATF-3 in a distinct population of sacral preganglionic neurons. Frontiers in Neuroscience, 5, 1-6.

Poulat, P., Marlier, L., Pellas, F., Rajaofetra, N., Privat, A. (1992). Effects of neonatal removal of superior cervical ganglion on serotonin and thyrotropin-releasing hormone immunoreactivity in the intermediolateral cell column of the rat spinal cord. Experimental Brain Research, 91, 21–8.

Rando, T. A., Bowers, C. W., Zigmond, R. E. (1981). Localization of neurons in the rat spinal cord which project to the superior cervical ganglion. Journal of Comparative Neurology, 196(1), 73–83.

Redgate, E. S., Fahringer, E. E. (1973). A comparison of the pituitary adrenal activity elicited by electrical stimulation of preoptic, amygdaloid and hypothalamic sites in the rat brain. Neuroendocrinology, 12, 334–43

189

Rho, J. H., Swanson, L. W. (1989). A morphometric analysis of functionally defined subpopulations of neurons in the paraventricular nucleus of the rat with observations on the effects of colchicine. Journal of Neuroscience, 9, 1375–88.

Sawchenko, P. E., Li, H. Y., Ericsson, A. (2000). Circuits and mechanisms governing hypothalamic responses to stress: a tale of two paradigms. Progress in Brain Research, 122, 61– 78.

Smith, O. A., DeVito, J. L. (1984). Central neural integration for the control of autonomic responses associated with emotion. Annual Review of Neuroscience, 7, 43–65.

Svensson, M., Eriksson, P., Persson, J. K., Molander, C., Arvidsson, J., Aldskogius, H. (1993). The response of central glia to peripheral nerve injury. Brain Research Bulletin, 30, 499–506.

Swanson, L. W., Hartman, B. K. (1980). Biochemical specificity in central pathways related to peripheral and intracerebral homeostatic functions. Neuroscience Letters, 16, 55–60.

Trapp, B. D., Wujek, J. R., Criste, G. A., Jalabi, W., Yin, X., Kidd, G. J.,Ransohoff, R. (2007). Evidence for synaptic stripping by cortical microglia. Glia, 55, 360–8.

Tucker, D. C., Saper, C. B. (1985). Specificity of spinal projections from hypothalamic and brainstem areas which innervate sympathetic preganglionic neurons. Brain Research, 360, 159– 164.

190

Yamaguchi, N., Okada, S. (2009). Cyclooxygenase-1 and -2 in spinally projecting neurons are involved in CRF-induced sympathetic activation. Autonomic Neuroscience, 151, 82–9.

Yamaguchi, N., Okada, S., Usui, D., Yokotani, K. (2009). Nitric oxide synthase isozymes in spinally projecting PVN neurons are involved in CRF-induced sympathetic activation. Autonomic Neuroscience, 148, 83–9.

191

Chapter 7: General discussion

192

The possible role of BDNF in retrograde effects of peripheral injury in the spinal cord Using our injury model, we documented changes in BDNF expression in the spinal cord after injury (Chapter 2). Specifically, we observed an increase in the level of BDNF protein at 3 weeks following the injury, with increased TrkB levels at one week paralleling an increase in the number of OLs that express TrkB (Chapters 2, 4). The neurotrophin BDNF has been shown to be important for the survival and maintenance of neurons in the central nervous system (reviewed in Lu et al., 2005). Previous studies have demonstrated that administration of exogenous BDNF to injured CNS tissues led to increase neuronal survival (Lu et al., 2005). In our model, endogenous levels of BDNF were increased together with an increase in the BDNF receptor, TrkB. The source of the endogenous BDNF in the spinal cord following injury is unknown. Similar to many other neurotrophins, BDNF is thought to be retrogradely transported in neurons (Lu et al., 2005), suggesting that neuronal and non-neuronal target tissues are sources of these molecules in the CNS. Neurons and possibly the satellite cells in the superior cervical ganglion (SCG; i.e. the target cells of the preganglionic axons) can produce BDNF (Causing et al., 1997), indicating that the SCG typically provides BDNF to the preganglionic neurons in the IML. Following CST transection, which results in the disconnection of the preganglionic neurons with their target-derived source of BDNF, the preganglionic neurons lose their target-derived source of BDNF. Yet an increase in BDNF and TrkB was observed. Since activated microglia and astrocytes are thought to provide BDNF to neurons in other models (Jones et al., 2005), it is possible that the activated glia observed in our model (Coulibaly et al., 2012) are the source of the increase BDNF signaling present in the spinal cord. Because BDNF has been shown to be a survival molecule (Lu et al., 2005), we presume that the increased protein expression in our model played a neuroprotective role. In the next few paragraphs, the possible mechanisms through which BDNF may be influencing the survival of the injured preganglionic neurons are described. We, as well as others, have demonstrated that TrkB is expressed by oligodendrocyte (OL) cells in the spinal cord (Skup et al., 2002; McCartney et al., 2008; Coulibaly and Isaacson, 2012; Chapters 3, 4). Therefore the changes in BDNF expression in the spinal cord may affect not only the expression of TrkB in these OLs, but likely also affects the physiology of these cells. Our results demonstrated an increase in the number of TrkB cells in the spinal cord following injury

193

(Coulibaly and Isaacson, 2012; Chapter 3) and an increase in TrkB expression was in OL progenitor cells (OPC). Because BDNF can promote the proliferation of OPCs in the spinal cord (McTigue et al., 1998), it is possible that the changes in BDNF observed in our model may also have led to an increase in OPC proliferation. Indeed, we did observe an increase in the number of OPCs in the spinal cord after injury (Chapter 4) as well as an increase in the number of OPCs that express TrkB. Furthermore, the increase in the number of TrkB OLs (Coulibaly et al., 2012; Chapter 3), which may have been OPCs, mature OLs or even cells in transition between these two stages, may have indicated that changes in BDNF levels in the spinal cord can enhanced OPC proliferation and differentiation.. However because an increase in mature OLs did not occur at long term survival time points, the physiological role of the OPC proliferation is not clear. While the injury-induced changes in OL lineage cells were the primary focus of this study, the findings that we obtained have increased the understanding of the role of BDNF in the regulation of cells in the OL lineage. Others have shown that the activity of OPCs is regulated by BDNF (McTigue et al., 1998; Du et al., 2003; McTigue and Tripathi, 2008). We have incorporated this information into our studies to propose a model, as shown in Figure 1, that suggests that a population of quiescent OPCs in the spinal cord (NG2+ only) becomes activated following the peripheral injury (or some other stimulus) to proliferate and/or differentiate into mature OLs. This small population of activated OPCs then upregulates TrkB expression to become NG2+/TrkB+ cells. As NG2+/TrkB+ cells begin the differentiation process, they lose NG2 expression, but continue to express TrkB, resulting in the population of TrkB+ only cells. The OLs would continue to mature and thus acquire CC1 expression (TrkB+/CC1+). Later stages of the mature OLs would involve the downregulation of TrkB, resulting in the population of CC1+ only cells. BDNF appears to also play a role in the plasticity of the TrkB OLs observed following peripheral axon injury. An increase in the expression of the gap junction protein Cx32 was observed in the IML following injury, with the majority of the changes occurring in TrkB OL population (Chapter 5). This suggests that BDNF may play a role in the regulation of this molecule. Survival roles have been attributed to increased gap junctions in the CNS after injury. Indeed, gap junctions have been shown to play a role in regulating the levels of extracellular ions in the vicinity of injured neurons as well as provide growth factors and energy molecules to the

194 injured neurons (Bruzzone et al., 2001; Belliveau et al., 2006; Dbouk et al., 2009; Rash, 2010). Therefore, the increased expression of the gap junction protein Cx32 in TrkB OLs may be neuroprotective in our model. Indeed, no neuronal loss was observed in the spinal cord after injury (Chapter 2). Therefore, in the present model the survival role BDNF plays may be an indirect one. Through its actions on the glial population, microglia, astrocytes, and OLs, BDNF may have led to the provision of the necessary survival molecules to the injured neurons. It would be interesting to perform knockout and overexpression studies in this model to fully determine which glial response resulted from BDNF and whether the removal of BDNF would affect the survival of the neurons and possibly abolish the glial response.

Retrograde molecules play a role in CNS effects of peripheral injury We obtained convincing evidence of transneuronal changes in our model, suggesting the activity of retrogradely transported injury factors affecting uninjured neurons in the brain. Specifically, we documented changes in the preautonomic neurons in the PVN of the hypothalamus (Chapter 6). Such alterations suggest the presence of factors derived from the injured neurons (and/or injury site) that have physiological effects on the afferents to the injured neurons. Both positive and negative signals retrogradely transported from the injury site can lead to changes in the injured neurons (reviewed in Abe and Cavalli, 2008). The cytokine interleukin, a molecule that has been shown to play an important role in regeneration, has been proposed as a positive retrograde signal that is derived from the injury site and activates the JAK/STAT pathway in the injured neurons (reviewed in Abe and Cavalli, 2008). In addition, the loss of activating transcription factor 2, which normally represses neuronal growth, is another possible mechanism through which retrograde influences of the injury can affect neurons in the CNS (Reviewed in Abe and Cavalli, 2008). It is yet unclear whether the peripheral injury factors are the same molecules being released by the injured neurons, or if the injured neurons produce specific signaling molecules that in turn affect the surrounding cells. With regard to how the injured neurons might affect their surroundings, it has been proposed that nitric oxide (NO) may be a candidate molecule in this process (Moreno-Lopez et al., 2011). The NO model suggests that NO is produced and release by injured motorneurons in the CNS (Moreno-Lopez et al., 2011). Once released, NO acts in an autocrine fashion to down

195 regulate the expression and production of BDNF in the injured neurons. Loss of this growth factor production then leads to the retraction of afferent inputs onto the injured neurons (Moreno- Lopez et al., 2011). Indeed, it has been shown that local injection of exogenous BDNF in areas of synaptic stripping help reestablish synaptic contacts (Gonzalez-Forero and Moreno-Lopez, 2014). However, in our model, at 16 weeks when there are high levels of BDNF in the spinal cord (Chapter 2), there is still evidence of synaptic stripping (Chapter 6), which raises the question as to where the increase in BDNF is occurring in our model. On the other hand, the restorative aspect of BDNF seem to be specific to inhibitory afferents. For example, Novikov and colleagues (2000) demonstrated that administration of BDNF to the lumbar spinal cord after transection of L5 ventral root only rescued the number of inhibitory synapses on the injured neurons. Therefore, since CRH has been demonstrated to have excitatory effects on IML neurons in vitro (Tanabe et al., 2012), it is possible that in our model the increase in BDNF observed in the spinal cord at 16 weeks (Chapter 2) would have no effect on the number of CRH inputs onto the preganglionic neurons. Also, it has been demonstrated that growth factors can influence neuronal phenotype (Patel et al., 2000). Therefore, a downregulation of BDNF expression in the injured neurons may have led to the loss of CRH phenotype observed in the preautonomic neurons in the PVN. It would be very interesting to characterize the level of NO in our model and determine whether there is a correlation in the level of this molecule and BDNF in our system.

196

Oligodendrocyte progenitor pools Oligodendrocytes

Self renew Differentiation Immature Mature TrkB/CC1 NG2 only NG2/TrkB TrkB only CC1 only

Figure 1

197

Figure 1: Proposed model of the maturation of cells in the oligodendrocytes lineage. In accordance with other studies, we propose that there two populations of NG2-ir OPCs reside in the spinal cord. The first is self-renewing and the second population undergoes differentiation as demonstrated by Barnabe-Heider and Colleagues (2010). From the findings in the present study, we propose that the readily differentiating population of NG2-ir OPCs is the population that coexpressed TrkB. Although not detected in this study, it is probable that the self-renewing population expressed TrkB at very low levels. As the cells mature and down regulate their expression of NG2 (Kessaris et al. 2008; Nishiyama et al. 2009), we postulate that they retain their expression of TrkB. As maturation progresses, the cells upregulate/express the mature oligodendrocyte marker CC1.

198

References

Abe, N., Cavalli, V. (2008). Nerve injury signaling. Current Opinion in Neurobiology, 18(3), 276–83.

Causing, C. G., Gloster, A, Aloyz, R., Bamji, S. X., Chang, E., Fawcett, J., Miller, F. D. (1997). Synaptic innervation density is regulated by neuron-derived BDNF. Neuron, 18(2), 257–67.

Coulibaly, A. P., Isaacson, L. G. (2012). Transient changes in spinal cord glial cells following transection of preganglionic sympathetic axons. Autonomic Neuroscience, 168(1-2), 32–42.

Dbouk, H. a, Mroue, R. M., El-Sabban, M. E., Talhouk, R. S. (2009). Connexins: a myriad of functions extending beyond assembly of gap junction channels. Cell Communication and Signaling , 7, 4.

Du, Y., Fischer, T. Z., Lee, L. N., Lercher, L. D., Dreyfus, C. F. (2003). Regionally specific effects of BDNF on oligodendrocytes. Developmental Neuroscience, 25(2-4), 116–26.

González-Forero, D., Moreno-López, B. (2014). Retrograde response in axotomized motoneurons: Nitric oxide as a key player in triggering reversion toward a dedifferentiated phenotype. Neuroscience. 283. 138-165.

Lu, B., Pang, P. T., Woo, N. H. (2005). The yin and yang of neurotrophin action. Nature Reviews. Neuroscience, 6(8), 603–14.

199

McCartney, A. M., Abejuela, V. L., Isaacson, L. G. (2008). Characterization of TrkB immunoreactive cells in the intermediolateral cell column of the rat spinal cord. Neuroscience Letters, 440(2), 103–8.

McTigue, D. M., Tripathi, R. B. (2008). The life, death, and replacement of oligodendrocytes in the adult CNS. Journal of Neurochemistry, 107(1), 1–19.

Patel, T. D., Jackman, A., Rice, F. L., Kucera, J., Snider, W. D. (2000). Development of sensory neurons in the absence of NGF/TrkA signaling in vivo. Neuron, 25(2), 345–57.

200