Regeneration of sensory but not motor axons following visceral nerve injury

Experimental Neurology 266 (2015) 127–142

Contents lists available at ScienceDirect

Experimental Neurology journal homepage: www.elsevier.com/locate/yexnr

Regular Article

Regeneration of sensory but not motor axons following visceral nerve injury Sophie C. Payne ⁎, Philip J. Belleville, Janet R. Keast Department of Anatomy and Neuroscience, The University of Melbourne, Victoria 3010, Australia

a r t i c l e

i n f o

Article history: Received 16 October 2014 Revised 13 February 2015 Accepted 17 February 2015 Available online 26 February 2015 Keywords: Axotomy Bladder innervation Inferior hypogastric plexus Nerve regeneration Pelvic ganglia Pelvic surgery Collateral sprouting

a b s t r a c t Following peripheral nerve injury, restoration of function may occur via the regeneration of injured axons or compensatory sprouting of spared axons. Injury to visceral nerves that control urogenital organs is a common consequence of pelvic surgery, however their capacity to reinnervate organs is poorly understood. To determine if and how sensory and motor connections to the bladder are re-established, a novel surgical model of visceral nerve injury was performed unilaterally in adult male Wistar rats. Bladder-projecting motor and sensory neurons in pelvic ganglia and lumbosacral dorsal root ganglia, respectively, were identified and characterised by retrograde tracing and immunofluorescence. Application of tracers ipsi- and contralateral to injury distinguished the projection pathways of new connections in the bladder. In naive animals, the majority of sensory and motor neurons project ipsilaterally to the bladder, while ~20 % project contralaterally and ~5 % bilaterally. Injured axons of motor neurons were unable to regenerate by 4 weeks after transection. In contrast, by this time many injured sensory neurons regrew axons to reform a substantial plexus within the detrusor and suburothelial tissues. These regeneration responses were also indicated by upregulation of activating transcription factor-3 (ATF-3), which was sustained in motor neurons but transient in sensory bladder-projecting neurons. Axotomy had little or no effect on the survival of bladder-projecting sensory and motor neurons. We also found evidence that uninjured motor and sensory neurons develop additional projections to the denervated bladder tissue and return connectivity, likely by undergoing compensatory growth. In conclusion, our results show that visceral sensory and motor neurons have a different capacity to regenerate axons following axotomy, however in both components of the circuit uninjured bladder neurons spontaneously grow new axon collaterals to replace the lost terminal field within the organ. For a full functional recovery, understanding the environmental and cellular mechanisms that reduce the ability of pelvic ganglion cells to undergo axonal regeneration is needed. © 2015 Elsevier Inc. All rights reserved.

Introduction Peripheral nerve injury studies have extensively investigated the effects of axotomy on somatic nerves that innervate targets such as skin and skeletal muscle (Bloechlinger et al., 2004; Groves et al., 1997; Shin et al., 2014; Tandrup et al., 2000). However, comparatively little is known of the responses and molecular events that occur following injury to visceral nerves that innervate urogenital organs. Damage to visceral nerves is difficult to avoid during pelvic surgery, such as prostatectomy and resection of bowel tumours (Lange et al., 2008; Nishizawa et al., 2011; Wallner et al., 2008). Such damage invariably leads to bladder, bowel and/or sexual dysfunction because autonomic neurons that innervate urogenital organs are located in a network of ganglia Abbreviations: ATF-3, activating transcription factor-3; CGRP, calcitonin gene-related peptide; DRG, dorsal root ganglion/ganglia; GDNF, glial cell line-derived neurotrophic factor; GFRα1, GDNF family receptor alpha 1; IB-4, isolectin B4; NF200, neurofilament 200; nNOS, neuronal nitric oxide synthase; NPY, neuropeptide Y; PG, pelvic ganglion/ganglia; PGP 9.5, protein gene product 9.5; TH, tyrosine hydroxylase. ⁎ Corresponding author. E-mail address: [email protected] (S.C. Payne).

http://dx.doi.org/10.1016/j.expneurol.2015.02.026 0014-4886/© 2015 Elsevier Inc. All rights reserved.

(the inferior hypogastric plexus) that are closely apposed to the surgical site. This plexus also forms a conduit for sensory nerves that innervate pelvic organs, such that injury often has widespread effects on neural regulation of urogenital function (Keast, 2006). Understanding the impact of injury on these visceral nerves is essential for developing treatments that are neuroprotective and promote repair. The anatomy of sensory and autonomic innervation of urogenital organs is well defined in rodents (Keast, 2006; Keast and de Groat, 1989). The major pelvic ganglia (PG), functionally equivalent to the inferior hypogastric plexus in humans, contain a mixture of parasympathetic and sympathetic neurons that provide motor innervation to the bladder and other urogenital organs. Sensory neurons that innervate the bladder have somata in lumbosacral dorsal root ganglia (DRG), and project via the pelvic ganglia to their targets. Therefore, damage to nerve bundles that project from the PG to the bladder invariably injures both motor and sensory axons. The first aim of our study was to establish an experimental model of visceral nerve injury that allows assessment of the response of both sensory and motor neurons to injury. We performed unilateral transection of bladder nerves that contain both sensory and motor axons, while axons that project from neurons

128

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

in spared ganglia remain uninjured. This model of bladder nerve injury therefore allows the assessment of the innate capacity of injured and spared sensory and motor axons to reinnervate bladder tissues. Recovery of both motor and sensory innervation is essential to restore normal voiding and continence behaviour. Following peripheral nerve injury, reinnervation may occur via the regeneration of injured axons or compensatory growth of spared axons (Cobianchi et al., 2014; Navarro et al., 2007). Most studies of pelvic visceral nerve injury focus on strategies to understand and improve the response of penis-projecting motor neurons after prostatectomy (Albersen et al., 2012; Canguven and Burnett, 2008). Rodent studies show that the cavernous (penile) nerve, which predominantly contains pro-erectile parasympathetic axons, has a limited capacity to regenerate following transection (Carrier et al., 1995; Kato et al., 2003), although axons regrow slowly after crush (Nangle and Keast, 2007). Following an incomplete (unilateral) injury, reinnervation of cavernosal tissue may also occur by the compensatory growth of spared axons (Carrier et al., 1995; Nangle and Keast, 2007). The growth of collaterals from neurons in the spared pelvic ganglion occurs in parallel with an increase in expression of growth-associated-protein 43 (GAP43), a marker of neuron regeneration and plasticity (Kato et al., 2003). This collateral growth from spared neurons likely contributes to the restoration of function (Nangle and Keast, 2007). Therefore, the second aim of this study was to determine how neural connections are restored to the bladder, whether by the regeneration of injured axons, the compensatory growth of spared axons, or both. This aim was addressed by applying retrograde tracer to the bladder ipsi- and contralateral to injury following unilateral transection of the bladder nerves. A separate analysis of injured and spared neurons that project to the bladder allowed the direct assessment of regenerative and compensatory axonal growth. This was combined with analysis of sensory and motor innervation within tissues on each side of the bladder. Injured peripheral neurons undergo numerous changes to switch to a regenerative phenotype. This involves upregulation of early response transcription factors and growth-associated genes, protein synthesis and cytoskeletal reassembly to form a growth cone, axonal elongation and the re-establishment of synaptic contact to target tissue (Bradke et al., 2012; Navarro et al., 2007; Scheib and Hoke, 2013). The basic leucine zipper transcription factor, activating transcription factor-3 (ATF-3), has regenerative and anti-apoptotic roles in axotomised neurons (Francis et al., 2004; Seijffers et al., 2006). Viral vector delivery of ATF-3 increases neurite outgrowth of cultured DRG neurons (Seijffers et al., 2006) and prevents kainic acid-induced apoptosis of hippocampal neurons in vivo (Francis et al., 2004). Furthermore, ATF-3 is widely regarded as a marker of neuronal injury as it is rapidly upregulated in all injured DRG neurons following sciatic nerve transection, and downregulated after axons reinnervate their target (Tsujino et al., 2000). However, the expression profile of this regenerative marker in injured bladder sensory and motor neurons is unknown. The early response transcription factor, c-Jun, is also recognized as a functional marker of axonal injury (Herdegen et al., 1997). Recent evidence suggests that cJun expression is also associated with injury-independent ‘sprouting,’ and is upregulated in spared neurons that are undergoing compensatory growth. For example, deafferentation of pelvic ganglion motor neurons stimulates the upregulation of c-Jun and extensive local sprouting of axon collaterals, even though these neurons themselves are not injured (Nangle and Keast, 2009; Uvelius and Kanje, 2010). Furthermore, genetic deletion of c-Jun in CNS neurons reduces the degree of perineuronal sprouting following facial nerve injury (Raivich et al., 2004). Therefore, as the regeneration and compensatory growth of axons is indicated by the expression of ATF-3 and c-Jun, the third aim of this study was to determine the expression of these injury markers within injured and spared motor and sensory neurons innervating the bladder following unilateral transection of bladder sensory and motor nerves.

Death of injured peripheral neurons that supply somatic targets is well described and recognized as a major contributor to poor functional recovery (Scheib and Hoke, 2013; Terenghi et al., 2011). The majority of axotomy-induced apoptosis occurs between 1 day and 2 weeks following injury, with a 10–40 % loss of DRG neurons detected by 2 weeks following sciatic nerve injury (Groves et al., 1997; McKay Hart et al., 2002; Vestergaard et al., 1997). The severity of loss depends on where the injury occurs, and the nature of the injury, in that proximal transection of the sciatic nerve invokes a 3 fold higher loss of DRG neurons compared to the loss seen following a distal injury (Ygge, 1989). Ventral root avulsion causes a severe 70 % loss of somatic motor and preganglionic parasympathetic neurons (Hoang et al., 2006). The impact of injury on survival of pelvic visceral sensory and motor neurons has not, to our knowledge, been directly examined. Preventing the loss of injured neurons is a key therapeutic strategy for improving functional outcome. Therefore the final aim of this study was to determine whether visceral neurons survive axotomy. Materials and methods Animals and surgical procedures All experiments were approved by the Animal Ethics Committee of the University of Melbourne, and complied with the Australian Code for the Care and Use of Animals for Scientific Purposes (National Health and Medical Research Council of Australia). Male Wistar rats (6 weeks; Animal Resource Centre, Perth, Australia) were housed in groups of 3, under a 12 hour light–dark cycle, with ad libitum access to water and standard chow. All surgical procedures were conducted under isoflurane anaesthesia (3 % induction, 2–2.5 maintenance in 1.5–2 % oxygen). For postoperative care, animals were administered Temgesic (buprenorphine 0.05 mg/kg, subcutaneous) and Terramycin (oxytetracycline 10 mg/kg, intramuscular) immediately following surgery, and another dose of Temgesic 8 h later. Pelvic ganglia (PG) in male rats lie on the dorsolateral aspect of the prostate gland. A bundle of nerves exit the ventromedial aspect of the major pelvic ganglion and project to the lower urinary tract and reproductive organs (Keast, 2006). A small cluster of microganglia (accessory ganglia) is embedded along some of these projections and are considered as an extension of the major PG (Keast et al., 1989). Therefore this study refers to these projections as “accessory nerves”. In order to study the effects of axotomy on bladder-projecting neurons, unilateral injury of the accessory nerves was performed under conditions described previously (Nangle and Keast, 2009). In brief, the lower abdominal cavity was opened and pelvic organs exposed. Accessory nerves that project ventrally towards the bladder were identified, isolated from underlying prostate tissue with fine forceps and cut with iris scissors at approximately 1 mm from the ganglion. Resection of the accessory nerve was not necessary as upon transection the proximal and distal nerve stumps retracted by approximately 2–3 mm. Transection of the accessory nerves was predicted to axotomise the majority of bladderprojecting neurons residing in lumbosacral DRG and PG. To standardise the injury and ensure that all branches of the accessory nerves were cut, injury was performed and retrograde tracer dye applied immediately to the same side of the bladder (n = 3). Ganglia were dissected 1 week following injury and numbers of retrogradely labelled neurons counted. Care was taken to avoid damaging blood vessels, although some bleeding could not be avoided because some micro-vessels are embedded with the accessory nerves. Abdominal muscle and skin were sutured and post-operative care administered. Retrograde tracing The neural connectivity of the bladder in naive animals (n = 4) was assessed by applying retrograde tracers FluoroGold (FG) and Fast Blue (FB) to opposite sides of the bladder under anaesthesia as described

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

above. In brief, the urinary bladder was exteriorised and the midline identified. FluoroGold (FG; total 30 μl, 4 % in sterile water; Fluorochrome, Denver CO) and Fast Blue FB; (total 30 μl, 2 % in sterile water; Dr Illing Plastics GmbH, Breuberg, Germany) were applied uniformly to each side of the bladder (8 injection sites/dye) using an insulin syringe (30 G). Each dye was injected, exclusively within one hemisphere of the bladder, away from the bladder midline, and care was taken to ensure that each dye did not spread across the midline. Specifically, injections of dye were made to both the bladder base (the primary target of sympathetic nerves innervating the detrusor) and body (primarily parasympathetic innervation). Care was taken to ensure that any leakage of dye from the injection site was immediately removed by blotting and extensive washing with sterile saline. To determine the impact of unilateral transection on the neural connectivity of neurons within injured and spared ganglia, retrograde tracers (FG and FB) were applied at 1 week (n = 4), 4 weeks (n = 5) and 12 weeks (n = 5) after surgery. Only neurons that successfully projected to the bladder tissues by the day of dye injection would be labelled so we have referred to the animal cohorts by this post-surgical time at which axons in the target were re-connected and exposed to dye, i.e. as 1 week, 4 weeks and 12 weeks post injury (Table 2; Figs. 2C, D, H–J and 3C–E). The dye was allowed 7 days to transport back to somata residing in DRG or PG (Keast et al., 1989) so animals were euthanised i.e. at 2 weeks, 5 weeks or 13 weeks after the initial injury. The second part of this study aimed to determine the upregulation of regenerative markers (c-Jun and ATF-3) in axotomised neurons. Therefore, in a separate cohort of experimental animals retrograde tracers (FG and FB) were applied to the bladder, and 7 days later the unilateral transection was performed. At 1 week (n = 4) and 3 weeks (n = 6) following injury, DRG and PG were removed (described below). Naive animals were used as controls (n = 3). Qualitative analysis of sensory and motor nerves in the bladder was performed in a separate cohort of animals that did not receive retrograde tracer dye. This was performed in naive controls and at 1 and 4 weeks following injury (n = 3/group). Tissue preparation At the end of the experiment, animals were deeply anaesthetised (ketamine: 100 mg/kg and xylazine: 10 mg/kg, intra-peritoneal) and intra-cardially perfused with saline solution (0.9 % sodium chloride, 1 % sodium nitrite and 5000 IU/ml heparin) and freshly made fixative (4 % paraformaldehyde in 0.1 M phosphate buffer, pH 7.4). PG were removed and pinned flat on silicone-lined dishes. DRG from spinal levels L1, L2, L6, and S1, known to contain sensory neurons innervating the bladder (Keast and De Groat, 1992), were dissected. Bladders were orientated using a suture to distinguish between left and right sides prior to removal. All tissues were post-fixed overnight, then washed in PBS (0.1 M phosphate buffer, pH 7.2) and stored at 4 °C in PBS containing 0.1 % azide. Tissue was cryoprotected in 30 % sucrose in PBS overnight

129

and frozen in OCT prior to sectioning on a cryostat. Longitudinal sections (14 μm) were taken of PG (ventral–dorsal axis) and DRG (rostral–caudal axis). The bladder base was chosen for sectioning. To ensure the equivalent region of the bladder base was analysed, ureters were located and a 2 mm slice of bladder (above the level of the ureters) was removed using a scalpel blade. A small cut was made in the slice of tissue to maintain orientation and transverse sections (14 μm) were taken. Serial sections of tissue were thaw-mounted and distributed across 8 slides (1 % gelatin-coated) so that adjacent sections were not placed on the same slide, thus avoiding double counting of neurons. Tissue was processed for immunohistochemistry on the same day as sectioning. Immunohistochemical processing Sections were blocked with 10 % normal horse serum in 0.1 % TritonX in PBS for 1 h, then incubated overnight (18–24 h) at room temperature with primary antibodies (Table 1). The next day, slides were thoroughly washed in PBS and incubated with species-appropriate secondary antibodies coupled to AlexaFluor® 488 or AlexaFluor® 594 (1:500 in hypertonic PBS, Life Technologies, CA, USA) for 2 h. Slides were washed and mounted with carbonate buffered glycerol (pH 8.6). Neuronal counts and characterisation To initially determine the neural connectivity of the bladder in naive animals, FluoroGold (FG) and Fast Blue (FB) were applied to opposite sides of the bladder. Experimental animals that received unilateral transection received both retrograde tracers after the injury. Retrogradely labelled neurons displaying a nucleus were counted manually by viewing under a fluorescence microscope (Zeiss Axioimager M1) in a total of 3 sections from each PG in naive animals (n = 3) and at 1 week (n = 3), 4 weeks (n = 3) and 12 weeks (n = 4) following unilateral accessory nerve transection. Similarly, retrogradely labelled neurons were quantified across 4 sections of DRG from L1, L2, L6 and S1 levels. Data from L1–L2 were combined and referred to as “lumbar” and L6–S1 ganglia were combined and referred to as “sacral”. To determine the proportion of neurons that project to ipsilateral, contralateral or bilateral hemispheres of the bladder, the prevalence of FG +, FB + or FG +/FB + neurons was calculated as a proportion of the total number of retrogradely labelled nucleated neuron profiles counted in that ganglion. The major chemical classes of retrogradely labelled motor neurons were evaluated by assessing their immunoreactivity for tyrosine hydroxylase (TH), neuronal nitric oxide synthase (nNOS), and neuropeptide Y (NPY) immunoreactivity, while sensory neurons were assessed for calcitonin gene-related peptide (CGRP), neurofilament 200 kD (NF200) and glial cell line-derived neurotrophic factor receptor, GFRα1. Fluorophore-tagged lectin IB4 was used to label unmyelinated non-peptidergic DRG neurons (Leclere et al., 2007). The proportion of retrogradely labelled neurons immunopositive for each

Table 1 Primary antibodies and tags. Antigen

Abbreviation

Host

Company

Supplier address

Dilution

Activating transcription factor-3 βIII-tubulin (Tuj 1) Calcitonin gene-related peptide Caspase-3 c-Jun GDNF family receptor α-1 Isolectin B4a Neurofilament - 200 Neuronal nitric oxide synthase Neuropeptide Y Protein gene product 9.5 Tyrosine hydroxylase

ATF3 βIII-tubulin CGRP Caspase-3 c-Jun GFRα1 IB4-FITCa NF200 nNOS NPY PGP 9.5 TH

Rabbit Rabbit Goat Rabbit Rabbit Goat N/A Mouse Rabbit Rabbit Rabbit Rabbit

Santa Cruz Covance AbD Serotec Cell Signaling Santa Cruz R&D Systems Vector Laboratories Sigma Aldrich Life Technologies Dia Sorin Merck Millipore Merck Millipore

Clayton, Vic North Ryde, NSW Meadowbrook, QLD Arundel, QLD Clayton, Vic Gymea, NSW Meadowbrook, QLD, Aus Castle Hill NSW Mulgrave, Vic Noble Park North, Vic Bayswater, Vic Bayswater, Vic

1:500 1:2000 1:2000 1:500 1:2000 1:400 1:500 1:4000 1:500 1:2000 1:2000 1:250

a

Indicates that the product is a fluorescent tag, not an antibody.

130

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

antigen was quantified. The expression of regenerative markers, ATF-3 and c-Jun, in retrogradely labelled neurons was similarly assessed in each chemical class of neurons in both naive and injured ganglia. Qualitative analysis of bladder nerves Monochrome 16-bit images were taken of bladder sections, keeping the camera settings and exposure time constant for each marker (Zeiss Axioimager M1 and MRm camera). In naive animals, left and right sides of the bladder were imaged. In experimental animals that received a unilateral transection, injured and spared sides of the bladder were imaged in at least 3 fields of view (using the 10× objective) in at least 3 sections per bladder. Statistics and figure preparation Data are presented as mean ± standard error (S.E.). Mean numbers of retrogradely-labelled neurons in naive, injured and spared ganglia were compared using paired two-tailed students t-tests. Data representing the proportions of neurons projecting to the bladder or expressing different markers were plotted as raw data, but statistical analyses were performed on arcsine-root transformed data. Multiple comparisons were analysed as appropriate by either a repeated measures one-way ANOVA with Dunnett's post-hoc test, or a one-way ANOVA (not repeated measures) with a Tukey's post hoc test. All data were analysed using GraphPad Prism 6 software (La Jolla, CA, USA) and significance was accepted at P b 0.05. For figure production, monochrome images of fluorescent samples were digitally colorized, and adjustments made where necessary in contrast and brightness to best represent the immunoreactivity as seen under the microscope (Adobe InDesign and Photoshop CS6; Adobe Systems, San Jose, CA, USA). Results Neural connectivity of the bladder in naive animals Retrograde tracing of visceral neurons innervating pelvic organs is a well-established technique (Keast et al., 1989; Vizzard et al., 1995). To initially determine the neural connectivity of the bladder in naive animals, FluoroGold (FG) and Fast Blue (FB) were applied to opposite sides of the bladder. This technique allowed us to distinguish bladderprojecting neurons as having ipsilateral, contralateral or bilateral connectivity, the latter being indicated by the presence of both dyes within a single neuron (example shown in inset of Fig. 1A–Ai). There was only a small variation between animals (Fig. 1B, G; n = 4), indicating that injecting two neurotracer dyes to opposite hemispheres of the bladder was a reliable and valid methodology for determining projection pathways of bladder-projecting neurons. Similar to previous studies, bladder-projecting neurons were not consistently localised to a specific area of the pelvic ganglion (Keast et al., 1989). This was the case for all three types of projections. Furthermore, FB and FG dyes were transported with similar efficiency, as there were no differences in the mean number of retrogradely labelled neurons in equivalent samples of left and right ganglia in naive animals (left: 87.1 ± 8.3 vs. right 111 ± 10.0 neurons/section; P N 0.05). Comparable to a previous report of pelvic ganglion connectivity (Ekstrom et al., 1986), the majority of bladder motor neurons project to the ipsilateral side of the bladder, although ~20 % projected contralaterally and ~5 % projected bilaterally (Fig. 1B). Bladder motor neurons are chemically diverse. Nearly all are cholinergic or noradrenergic, and also co-express neuropeptides and other neuroactive substances (Keast, 2006). These retrogradely labelled motor neurons were immunohistochemically identified as being noradrenergic (TH, 19.7 ± 2.5 %, Fig. 1C), which predominantly innervate the base of the bladder, nitrergic (nNOS, 12.1 + 2.6 %, Fig. 1D) or peptidergic (NPY, 55.6 ± 4.1 %, Fig. 1E; all groups n = 3). The presence of dye labelled neurons that were negative

for TH and positive for NPY (approximately 35 %) represents the cholinergic (parasympathetic) subpopulation of neurons that innervate both the body and base of the bladder. The phenotype of motor neurons that were categorised as projecting ipsi-, contra- and bilaterally was distinguished (Fig. 1F), but no differences were detected between types of projections (P N 0.05). The majority of bladder-projecting sensory neurons are located in L1–L2 (“lumbar”) and L6–S1 (“sacral”) DRG (Keast and De Groat, 1992). Analysis of retrogradely labelled sensory neurons from these same animals showed that FG and FB were transported with similar efficiency in naive animals (lumbar DRG left: 5.5 ± 1.1 vs. right: 5.0 ± 0.4 neurons/section, P N 0.05) and sacral DRG (left: 17.3 ± 3.1 neurons/ section, right: 13.4 ± 2.4 neurons/section; P N 0.05). As reported previously (Keast and De Groat, 1992; Vizzard et al., 1995), fewer bladder sensory neurons resided in lumbar compared with sacral DRGs (L1/L2: 5.2 ± 0.7 vs. L6/S1: 15.3 ± 2.5 dye labelled neurons/section, P = 0.0235). The projection pathways of bladder sensory neurons differed between spinal levels (Fig. 1G). Both contralateral (P = 0.0304) and bilateral (P = 0.0085) projections were more prevalent in lumbar than sacral DRG, i.e. ipsilateral sensory projections were more common at the lumbar level (P = 0.001; Fig. 1G). Previous studies indicate that afferent innervation of the bladder consists primarily of Aδ myelinated and C-fibres that are either peptidergic or bind IB4, and respond to specific growth factors (de Groat and Yoshimura, 2009; Forrest et al., 2013). Therefore, bladder sensory neurons were distinguished immunohistochemically, with the largest proportion of neurons being CGRP-positive (L1/L2: 59.7 ± 7.9 % vs. L6/S1: 31.9 ± 3.7 %; Fig. 1H–J) and NF200-positive (L1/L2: 57.6 ± 3.2 % vs. L6/S1: 42.4 ± 3.5 %; Fig. 1H, I, K) and with fewer binding IB4 (L1/L2: 24.7 ± 2.1 % vs. L6/S1: 23.6 ± 1.3 %; Fig. 1H, I, L) and GFRα1positive (L1/L2: 27.4 ± 3.0 % vs. L6/S1: 29.2 ± 2.2 %; Fig. 1H, I, M). When considered as a proportion of all retrogradely labelled neurons at that spinal level, CGRP-positive bladder neurons were more prevalent in lumbar than sacral DRG (55.2 ± 2.3 % vs. 31.9 ± 3.7 %, P = 0.0302, n = 4). These data are supported by a recent study (Forrest et al., 2013). No differences were detected between lumbar and sacral DRG in the chemical class of neurons projecting ipsi-, contra- and bilaterally to the bladder (Fig. 1H, I). However, there were distinct neuronal phenotypes at each spinal level, as reported previously (Forrest et al., 2013; Keast and De Groat, 1992). Injured axons of visceral motor neurons had a limited capacity to regenerate in contrast to the effective regeneration of visceral sensory axons Unilateral transection was made to sensory and autonomic nerves (known as the accessory nerves) that innervate the bladder. To ensure that the surgical technique reliably axotomised bladder-projecting neurons, unilateral transection was performed and retrograde tracer dyes applied immediately to the bladder, i.e., during the same period of anaesthesia (n = 3). After 1 week, a time point at which few if any axons were likely to have regenerated, ganglia were dissected and assessed for the presence of retrogradely labelled neurons. Very few retrogradely labelled neurons were present in sacral DRG (0.8 ± 0.3 neurons/section), while a number of labelled neurons were consistently seen in PG (23 ± 2.7 neurons/section). We did not see evidence of major changes in voiding patterns, such as urinary retention or incontinence in our animals, at any time after unilateral accessory nerve transection. This agrees with a previous study (Berggren et al., 1993) who noted no disruptions to cystometric recordings following unilateral pelvic ganglionectomy. Retrograde tracers were applied to opposite sides of the bladder at 1, 4 and 12 weeks after injury, and euthanised 1 week later. However, given that new reconnections to the bladder were assessed during the week in which tracer dyes were applied, we refer to this time period as 1, 4 and 12 weeks after injury (Table 2; Figs. 2C, D, H–J and 3C–E). The number of retrogradely labelled neurons within injured ganglia

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

131

Fig. 1. Neural connectivity of the bladder of naive animals. Retrograde tracers FluoroGold (FG) and Fast Blue (FB) were applied to opposite bladder hemispheres allowing neurons to be distinguished as projecting to the ipsi-, contra- or bilateral sides of the bladder. (A–F) show data from pelvic ganglia (PG) and (G–M) from dorsal root ganglia (DRG). (A) A cryosection of PG showing that the predominant population of retrogradely labelled neurons projected ipsilaterally while (Ai) inset shows the boxed region at higher magnification and examples of neurons projecting ipsilaterally (FB+ single arrowhead), contralaterally (FG+, single arrow) and bilaterally (FB+/FG+, double arrowhead). (B) Most motor neurons projected ipsilaterally, however about 25 % projected contralaterally or bi-laterally. (C–E) Pelvic motor neurons were immunohistochemically identified with antibodies against TH (C), nNOS (D) and NPY (E). (F) No differences in the prevalence of each chemical class of ipsi-, contra- and bi-laterally projecting motor neurons were detected (P N 0.05). (G) Ipsilaterally-projecting sensory neurons were more common in sacral than lumbar DRG (P = 0.001), whereas contralaterally and bilaterally projecting neurons were more common in lumbar than sacral DRG (P = 0.0304 and P = 0.0085, respectively). (H–I) No differences (P N 0.05) in the prevalence of each chemical class of ipsi-, contra- and bilaterally projecting retrogradely labelled neurons were detected in lumbar (H) or sacral DRG (I). Sensory neurons were labelled for CGRP (J), NF200 (K), IB4 (L) and GFRα1 (M). Data show mean + SEM, n = 3–4 rats. Significant differences were identified using paired t-test (D) and repeated measures one-way ANOVA with a Tukey's post hoc test (C, E, F). Scale bars: (A) 200 μm, (Ai, C–E, J–M) 50 μm.

was assessed to determine how many neurons reconnected to either side of the bladder. The number of retrogradely labelled neurons within injured ganglia was assessed at 1, 4 or 12 weeks following injury by applying retrograde tracers (FB and FG) to opposite sides of the bladder

after the injury; this revealed how many axotomised neurons regrew axons to reinnervate the bladder. Retrogradely labelled neurons were counted within injured and spared PG (Fig. 2A–C). At 1 week following injury it is unlikely that

132

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

Table 2 Phenotypes of bladder-projecting motor neurons in naive, injured and spared pelvic ganglia.

Injured ganglia

Spared ganglia

TH nNOS NPY TH nNOS NPY

Naive

1 week

4 weeks

12 weeks

19.7 + 2.5 12.1 + 2.6 55.6 + 4.1 19.7 + 2.5 12.1 + 2.6 55.6 + 4.1

17.0 + 5.2 26.0 + 8.6 48.0 + 3.8 10.2 + 0.9 13.2 + 1.9 50.5 + 9.2

26.7 + 8.7 17.6 + 0.5 53.3 + 2.9 11.5 + 5.1 10.2 + 0.8 46.5 + 5.5

34.3 + 8.4 20.5 + 1.2 60.2 + 3.7 17.7 + 5.2 17.0 + 2.9 57.4 + 5.4

injured axons have regenerated and reconnected to the bladder (Scheib and Hoke, 2013). However, at this time point a number of retrogradely labelled neurons (37.8 ± 2.7 neurons/section; Fig. 2C) were still present within injured PG. This raised the possibility that transection of the

accessory nerve was incomplete. Regardless of this possibility, statistical analysis showed that there was a significant main effect of side (F1, 6 = 344.9, P b 0.0001), indicating that fewer retrogradely labelled neurons were present in injured ganglia compared to spared ganglia following injury. There were no differences in numbers of retrogradely labelled neurons in spared ganglia compared to the naive control (P N 0.05, Fig. 2C). The total number of retrogradely labelled neurons remained approximately 60 % less than spared control ganglia at 4 and 12 weeks (Fig. 2C), suggesting that injured motor neurons have a limited capacity to regenerate axons. At chronic stages of injury (N 4 weeks), changes were seen in ipsilateral and bilateral projection pathways of neurons in injured ganglia that had not been axotomised by transection surgery (F3, 9 = 10.61, P = 0.0026, Fig. 2D). Specifically, the proportion of neurons within injured PG that bilaterally innervated the bladder was significantly higher at

Fig. 2. Axotomy-induced response of motor and sensory neurons in injured pelvic and dorsal root ganglia. Retrograde tracers were applied to the bladder at 1, 4 or 12 weeks after unilateral accessory nerve transection. One week later, animals were euthanised and numbers of retrogradely labelled neurons within injured pelvic and dorsal root ganglia assessed. (A–D) show data from PG and (E–J) from DRG. (A–B) Images show injured (A) and spared (B) PG at 12 weeks following injury, illustrating the loss of connectivity (i.e. fewer dye + neurons) ipsilateral to the surgery. (C) Retrogradely labelled neurons were quantified as mean numbers of neurons per section and compared between injured and spared ganglia. Following injury, there was a main effect of side (F1, 6 = 344.9, P b 0.0001), indicating that there were fewer retrogradely labelled neurons in injured PG, i.e. less connectivity with the bladder. There was no significant difference between the spared and naive ganglia (P N 0.05). (D) Changes were seen in the projection pathways of retrogradely labelled PG neurons, which will include those that had not previously been axotomised (F3, 9 = 10.61, P = 0.0026), with a greater proportion of neurons bilaterally innervating the bladder at 4 weeks (P = 0.0015) and 12 weeks (P = 0.0062). (E– G) Images show retrogradely labelled sensory neurons in naive (E) and injured DRG at 1 week (F) and 4 weeks (G) after injury, showing a transient loss of connectivity. (H) Analysis of retrogradely labelled sensory neurons showed that following injury, there was a main effect of side (F1, 6 = 22.21, P = 0.003), indicating fewer labelled neurons in the injured side, and a significant interaction between side and time (F2, 6 = 8.357, P = 0.018), with the only effect occurring at 1 week (P = 0.0028). An overall difference between naive and spared ganglia was detected (F3, 10 = 4.037, P b 0.05; Dunnet's post hoc test P N 0.05). (I) Projection pathways of sensory neurons to the bladder were affected by injury (F2, 7 = 6.873, P = 0.0223), with more neurons in the injured ganglia having bilateral innervation at 4 weeks (P = 0.0176); this increase was not sustained at 12 weeks. (E) No effect of injury was detected in the proportion of bladder-projecting sensory neurons that were positive for CGRP, NF200, IB4 and GFRα1 (P N 0.05). Data show mean + SEM, n = 3–4. Significant differences were identified using a twoway ANOVA and Sidak's post hoc test (C, H) and repeated measures one-way ANOVA and Dunnet's post hoc test (D, I, J). (A, B, E–G) Scale bars represent 200 μm, and apply to all images in the same row.

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

133

Fig. 3. Axotomy-induced response of motor and sensory neurons in spared pelvic and dorsal root ganglia. Retrograde tracers were applied to the bladder at 1, 4 or 12 weeks after unilateral accessory nerve transection. One week later, animals were sacrificed and numbers of retrogradely labelled neurons within spared pelvic and dorsal root ganglia assessed. Compensatory growth of axon collaterals may be reflected in changes to the projection pathway of these uninjured neurons. (A–B) In naive PG (A), few motor neurons projected bilaterally (examples of FG/FB neurons indicated by arrows) compared to 4 weeks post-injury (B). (C) Injury affected the projection pathway of neurons in spared ganglia (ipsilateral projections: F3, 9 = 10.08, P = 0.0031; bilateral projections: F3, 9 = 18.77, P = 0.0003). At chronic stages of injury, fewer spared neurons projected ipsilaterally (4 weeks: P = 0.0261; 12 weeks: P = 0.0091), while a higher proportion of these neurons projected bilaterally (4 weeks: P = 0.002; 12 weeks: P = 0.006). (D) In spared sacral DRG, injury had an effect on the projection pathway of uninjured sensory neurons (ipsilateral projections: F3, 9 = 12.40, P = 0.0015; bilateral projections: F3, 9 = 8.335, P = 0.0058). Specifically, fewer uninjured sensory neurons projected ipsilaterally at chronic stages of injury (4 weeks: P = 0.0052; 12 weeks: P = 0.0316), while a higher proportion of these neurons projected to both sides of the bladder (4 weeks: P = 0.0102). (E) Changes were seen in the prevalence of different chemical classes of spared sensory neurons (NF200: F3, 9 = 12.22, P = 0.0016; GFRα1: F3, 9 = 6.719, P = 0.0113). A transient increase in the proportion of NF200 neurons occurred at 1 week (P = 0.005) and 4 weeks (P = 0.0023), while the proportion of GFRα1 neurons decreased at 12 weeks: P = 0.0399. (C–E) Data show mean + SEM, n = 3–4. Significant differences were identified using a repeated measures one-way ANOVA with a Dunnet's post hoc test. (A–B) Scale bar applies to both images and represents 200 μm.

4 weeks (P = 0.0015, Fig. 2D). This indicates that by this time an additional subpopulation of motor neurons had axons that extended into both the denervated and uninjured hemispheres of the bladder; as such, they became bilaterally projecting neurons. This occurred even though the tissue within the injured bladder hemisphere was not completely denervated following unilateral transection (see above). This altered connectivity was sustained until at least 12 weeks (P = 0.0062, Fig. 2D), a time at which no additional motor axon regeneration had occurred. On assessing the phenotypes of retrogradely labelled neurons in injured PG, no changes were seen in the proportion of motor neurons that were noradrenergic (TH), nitrergic (nNOS) and peptidergic (NPY) following injury (P N 0.05, Table 2). Values represent proportions (%) of retrogradely labelled neurons immunoreactive for each neuronal marker, expressed as mean + SEM. The main effects of injury were identified by one-way ANOVA and Dunnett's post hoc test, n = 3–4, with statistical significance accepted as P b 0.05. Analysis of DRG from the same animals showed that the regenerative response of bladder sensory neurons differed from that of motor neurons. The total number of retrogradely labelled neurons was quantified within naive, injured and spared sacral DRG (Fig. 2E–G). Because of the small number of neurons labelled in L1–L2 DRGs (see above), these were not assessed in injured animals. At 1 week following injury, very few retrogradely labelled sacral sensory neurons were present (1.5 ± 0.6 neurons/section), indicating that transection of the accessory nerves axotomised approximately 90 % of sacral sensory neurons that innervate the bladder (Fig. 2H). This also reassured us that the surgical procedure was successful in transecting bladder-projecting sensory axons. Statistical analysis showed that there was a main effect of side (F1, 6 = 22.21, P = 0.003), indicating that there were less retrogradely

labelled neurons within injured than spared ganglia. This reduction varied with time, however (side ∗ time F2, 6 = 8.357, P = 0.018), with the reduction evident only at 1 week following injury (P = 0.0028, Sidak's multiple comparison test). This data suggests that by 4 weeks some sensory neurons have regenerated axons to reconnect with the bladder, as indicated by their ability to take up tracer. There was an overall group difference in the number of retrogradely labelled neurons in spared compared to naive ganglia (F3, 10 = 4.037, P b 0.05), however no differences were detected at individual time points (Dunnett's test; P N 0.05, Fig. 2I). During the post-surgical period, a significant change was also seen in the bilateral projection pathway of retrogradely labelled bladder sensory neurons (F2, 7 = 6.873, P = 0.0223, Fig. 2I). In injured ganglia, the proportion of neurons that bilaterally innervated the bladder increased at 4 weeks following injury (P = 0.0176), but in contrast to that occurring with motor neurons (see above), this did not remain significantly elevated by 12 weeks (P N 0.05, Fig. 2I). The transient nature of the bilateral projection from sensory neurons may suggest that this new connectivity is not sustained once the bladder becomes reinnervated from its original side. The proportion of bladderprojecting sensory neurons that are CGRP-, NF200-, IB4- and GFRα1positive remained similar to naive ganglia (P N 0.05, Fig. 2J), suggesting that each chemical class of neuron is able to regenerate axons with similar capability and efficiency. Axons of spared motor and sensory neurons underwent compensatory growth As mentioned above, different tracer dyes were applied to opposite hemispheres of the bladder after the unilateral transection injury was

134

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

performed. Here we discuss the impact of accessory nerve surgery on the spared pelvic and dorsal root ganglia, where compensatory growth may be reflected in the changes to projection pathways of these uninjured neurons. The total number of retrogradely labelled neurons was quantified within PG of naive animals and 1, 4 and 12 weeks post-injury (Fig. 2C), but no differences between groups was detected (P N 0.05). However, significant changes in the ipsilateral and bilateral projection pathways of motor axons were detected during chronic stages of injury (ipsilateral projections: F3, 9 = 10.08, P = 0.0031; bilateral projections: F3, 9 = 18.77, P = 0.0003) (Fig. 3A–C). One explanation of this data might be that a proportion of spared motor neurons undergo compensatory growth to extend axon collaterals into the denervated bladder hemisphere, thereby bilaterally innervating the bladder. Images of PG show examples of the increase in the prevalence of FG/FB neurons at 12 weeks following injury (Fig. 3A, B, arrowheads). This increase in bilateral innervation by spared neurons is supported by a decreased prevalence of spared neurons that projected only ipsilaterally at 4 weeks (P = 0.0261) and sustained until 12 weeks (P = 0.0091); concurrently the proportion of motor neurons that projected axons bilaterally increased at 4 weeks (P = 0.0020) and 12 weeks (P = 0.0006, Fig. 3C). Together this suggests that a proportion of spared motor neurons that originally innervate the uninjured side of the bladder underwent compensatory growth to extend axons into the area of the bladder that was denervated. No changes in the phenotypes of motor neurons in spared ganglia were detected following injury (P N 0.05, Table 2). In spared sacral DRG, there was a main effect of injury on the numbers of retrogradely labelled neurons (F3, 10 = 4.037, P = 0.0404), however Dunnett's post hoc analysis detected no difference from the naive control (P N 0.05, Fig. 2H). As described in detail below, no death of sensory neurons was observed in spared ganglia (Fig. 7). Furthermore, no such decrease was seen in spared PG (P N 0.05, Fig. 2C). Given that the sacral DRG and PG were taken from the same animals, the result cannot be explained by variation in the amount of dye administered between subjects. Similar changes were seen in the projection pathways of sensory axons as seen for motor pathways (ipsilateral projections: F3, 9 = 12.40, P = 0.0015; bilateral projections: F3, 9 = 8.335, P = 0.0058). The effect of injury on the projection pathways of spared sacral sensory neurons was similar to that seen with motor neurons. Specifically, the proportion of spared sensory neurons that only projected axons ipsilaterally decreased at 4 weeks (P = 0.0052) and 12 weeks (P = 0.0316). Concurrently, there was an increase in bilateral projection of spared sensory neurons at 4 weeks (P = 0.0102), but at 12 weeks this did not achieve significance (P = 0.0621). Taken together, one explanation of the data is that spared visceral neurons undergo compensatory growth following unilateral injury. Changes were seen in the prevalence of different chemical classes of spared sensory neurons following injury (NF200: F3, 9 = 12.22, P = 0.0016; GFRα1 F3, 9 = 6.719, P = 0.0113). A transient increase in the proportion of NF200 neurons was seen at 1 week (P = 0.005) and 4 weeks (P = 0.0023), with no detectable difference from naive controls at 12 weeks (P N 0.05, Fig. 3E). The proportion of GFRα1 neurons decreased at 12 weeks (P = 0.0399). Despite differences in regenerative capacities, original distributions of both motor and sensory innervation were re-established in the denervated bladder Bladder innervation was qualitatively assessed in transverse sections of the bladder base taken from naive controls and at 1 and 4 weeks following injury. Images show innervation of naive controls (Fig. 4A/Ai, F, K), as well as regions of the bladder that were ipsilateral (Fig. 4B, D, G, I, L, N) and contralateral (Fig. 4C, E, H, J, M, O) to the injury. Neuropeptide Y was expressed in many retrogradely labelled pelvic motor neurons (Fig. 1C), most of which are cholinergic (Keast, 2006). We therefore used NPY-immunolabelling as a marker of motor

innervation in the bladder (Fig. 4A–E). In the naive bladder, NPY axons were predominantly seen within the detrusor muscle, with a few innervating the sub-urothelium (Fig. 4A). Blood vessels were heavily innervated with NPY axons (indicated with arrowheads, Fig. 4Ai); these are known to be sympathetic. At 1 week following injury, there was an obvious decrease in NPY axons in the detrusor muscle and suburothelium ipsilateral to the injury (Fig. 4B), compared to the contralateral side or naive tissue (Fig. 4A, C). However, a sparse population of NPY axons was still present within these tissues, which was in agreement with our retrograde labelling data that indicated that approximately 40 % of bladder-projecting motor neurons were not axotomised following accessory nerve transection. Some of the residual axons may also arise from the contralateral spared pelvic ganglion. Interestingly, a majority of blood vessels within bladder tissue ipsilateral to accessory nerve transection remained heavily innervated with NPY axons (indicated with arrowheads, Fig. 4Bi). At 4 weeks following injury, NPY innervation was re-established throughout the detrusor muscle and sub-urothelial layers (Fig. 4D), although still appearing less dense than contralateral to the injury (Fig. 4E). Sensory innervation of the bladder was indicated using antibodies to CGRP. CGRP fibres innervated the sub-urothelium, detrusor muscle and vasculature (Fig. 4F), although sensory innervation of the muscle and vasculature was less dense than the motor innervation by NPY axons (Fig. 4A). Conversely, there were more CGRP than NPY axons in the sub-urothelial tissue. One week after unilateral transection of the accessory nerves, CGRP axons were sparse in all bladder tissues ipsilateral to injury (Fig. 4G), but tissues within the contralateral hemisphere of the bladder appeared normal (Fig. 4H). Given that accessory nerve transection resulted in the axotomy of approximately 90 % of bladderprojecting sensory neurons, these residual CGRP axons are likely projections from uninjured neurons within spared ganglia. There was a substantial degree of reinnervation by CGRP axons into the denervated bladder wall at 1 month, however the innervation of the detrusor and sub-urothelial tissue did not reach the original density and appeared less regular compared to that of naive bladder or the spared side (Fig. 4J). CGRP innervation of blood vessels was restored by 1 month. Our observations using specific markers of major motor and sensory axons populations were supported by use of the pan-axonal marker, βIIItubulin (Tuj1). In normal bladder tissue (Fig. 4K), βIII-tubulin fibres heavily innervated the detrusor muscle, vasculature and suburothelium. The innervation of the detrusor and sub-urothelium was vastly reduced at 1 week (Fig. 4L) but largely reinstated at 4 weeks (Fig. 4N). Tissues within the bladder hemisphere contralateral to injury (Fig. 4M, O) appeared comparable to controls. Vascular innervation by β-IIItubulin fibres was not significantly affected by unilateral transection (not shown). Differential upregulation of ATF-3 occurred in a sub-population of axotomised motor and sensory neurons The expression of ATF-3 was examined, as this transcription factor is exclusively upregulated in neurons with axonal injury (Peddie and Keast, 2011; Takeda et al., 2000; Tsujino et al., 2000). For this set of experiments, bladder-projecting neurons were labelled with retrograde tracer dyes (FG and FB) 1 week prior to nerve injury, thereby enabling the identification of ATF-3 expression in axotomised neurons. The expression of regenerative markers, ATF-3 and c-Jun, was assessed within injured and spared PG at 1 and 3 weeks following unilateral accessory nerve transection. No basal expression of ATF-3 was observed in naive (Fig. 5A) or spared PG. Following injury however, nuclear expression of ATF-3 was upregulated in a subpopulation of retrogradely labelled motor neurons at 1 week (34.0 + 5.6 %; n = 4 Fig. 5B, C) and 3 weeks (22.4 ± 2.9 %, n = 6, P b 0.0001; paired t-test with Bonferroni correction for multiple comparisons, Fig. 5C). The sustained upregulation of ATF-3 indicates that motor neurons have not regenerated axons to target tissue and are still in a regenerative phenotypic state

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

135

Fig. 4. Reinnervation of the bladder by motor and sensory axons following unilateral accessory nerve transection. Innervation of the bladder wall was qualitatively assessed in transverse sections of the bladder base. Each image is oriented with the urothelium at the top. (A–E) Images show NPY (motor) innervation of the detrusor muscle, suburothelial tissue and blood vessels; higher magnification of vascular innervation is indicated by arrowheads in Ai and Bi. (A) Naive: Numerous axons are distributed throughout the detrusor and closely associated with many blood vessels. A small number of axons lie close to the basal surface of the urothelium. One week after injury, NPY innervation of the detrusor and suburothelial tissue is greatly reduced ipsilateral to the injury, although blood vessel innervation is retained (B, Bi); contralateral to the injury, innervation of all bladder tissues appears unaffected (C). Four weeks after injury, NPY innervation is partially restored to both the detrusor and suburothelial tissues, ipsilateral to injury (D), while the contralateral side appears unchanged (E). (F–J) Images show CGRP (sensory) innervation of the detrusor muscle and within the suburothelial tissue. (F) Naive: CGRP axons are present in the detrusor and suburothelial tissue. Ipsilateral to injury, CGRP axons are vastly reduced at 1 week (G) but restored at 4 weeks (I). No effect of injury was seen on CGRP innervation in the spared side of the bladder at 1 week (H) or 4 (J) weeks. A similar pattern of innervation was demonstrated by the pan-axonal marker, βIII-tubulin in naive bladder (K) and after injury (L–O). Loss of βIII-tubulin axons are evident ipsilateral to injury at 1 week (L) but partially restored by 4 weeks (N). Contralateral to injury, βIII-tubulin innervation appeared no different from that of naive (M, 1 week; O, 4 weeks). Scale bars: 50 μm. Bar in A applies to all images of NPY except Ai and Bi. Bars in F and K apply to all images of CGRP and βIII-tubulin, respectively.

(Tsujino et al., 2000). In this part of the study we did not distinguish the projection pathways of retrogradely-labelled neurons in injured ganglia that expressed ATF-3. Because only a sub-population of motor neurons upregulated ATF-3 following injury, we then asked if this correlated with a particular chemical subclass. ATF-3 upregulation was seen in both cholinergic and noradrenergic neurons. However, the majority of ATF-3-positive retrogradely labelled neurons were negative for TH and nNOS (1 week: 81.3 ± 3.5 %; 3 weeks: 75.8 ± 3.8 %; n = 3), i.e. they were likely cholinergic NPY neurons (Keast and de Groat, 1989). A small population of ATF-3-positive retrogradely labelled

neurons was TH-positive (1 week: 7.3 ± 4.2 %, 3 weeks: 7.3 ± 1.0 %) or nNOS-positive (1 week: 11.4 ± 0.8 %, 3 weeks: 16.7 ± 4.3 %). There were no changes in the proportion of peptidergic, noradrenergic and nitrergic neurons that expressed ATF-3 at 1 vs. 3 weeks (P N 0.05). In spared PG, there was no detectable of ATF-3 expression following injury. No ATF-3 expression was observed in perisomatic glia of PG in any animal group. Within injured sacral DRG, the expression profile of ATF-3 in bladder-projecting sensory neurons differed from that of motor neurons. In naive and spared sacral DRGs, ATF-3 expression was absent

136

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

(Fig. 5D). Following accessory nerve transection, although 90 % of bladder-projecting sensory neurons ipsilateral to injury were axotomised (Fig. 2F), ATF-3 was upregulated at 1 week in only 26.5 + 4.0 % of retrogradely labelled neurons (n = 3; P b 0.0001; Fig. 5E, G). By 3 weeks however, ATF-3 was downregulated to basal levels (0.9 ± 0.9 %, n = 5, P N 0.05, Fig. 5F, G). Given that ATF-3 is downregulated in neurons whose axons have re-established contact with target tissue (Tsujino et al., 2000), the downregulation of ATF-3 in sensory neurons here likely indicates that axons of these neurons have reinnervated the bladder (see also Fig. 2E). Again, no attempt was made to distinguish the projection pathways of retrogradely-labelled neurons expressing ATF-3. All chemical classes of bladder-projecting sensory neurons upregulated ATF-3 at 1 week following injury, although ATF-3 was preferentially expressed in peptidergic and myelinated neurons. That is, out of the total population of ATF-3-positive retrogradely labelled neurons 65.9 ± 3.4 % were CGRP-positive, 36.4 + 6.1 % NF200-positive and only 2.5 ± 2.5 % IB4positive (n = 3). No ATF-3 expression was observed in perisomatic glia of naive, injured, spared or sacral DRG. Sustained upregulation of c-Jun occurred in a sub-population of axotomised motor and sensory neurons The transcription factor c-Jun is upregulated in nuclei of injured neurons (Herdegen et al., 1997). c-Jun is distinguished from ATF-3 in that it is also upregulated in neurons that have not received axonal injury but are undergoing compensatory growth (Nangle and Keast, 2009; Uvelius and Kanje, 2010). Therefore, we considered the possibility that c-Jun was upregulated in spared ganglia following injury. Very little basal expression of c-Jun was detected in neurons or glia of naive PG (Fig. 6A) and DRG (Fig. 6D). The expression of c-Jun following injury was then compared between injured and spared ganglia from the same animals, using paired t-tests with Bonferroni correction for multiple comparisons. One week after injury, c-Jun was upregulated within a subpopulation (37.0 ± 4.6 %; P = 0.0002; n = 4) of retrogradely labelled neurons in injured PG (Fig. 6C). This elevation within retrogradely labelled neurons was sustained at 3 weeks (35.7 ± 3.3 %; P = 0.0002, n = 6; Fig. 6B: indicated with arrows, C). We then examined which classes of injured motor neurons upregulated c-Jun and by immunofluorescence showed that although both noradrenergic and cholinergic motor neurons expressed c-Jun after injury, c-Jun was primarily upregulated in neurons that were TH- and nNOS-negative (1 week: 81.8 ± 1.1 %, 3 weeks: 77.2 ± 2.3 %;). Fewer c-Jun neurons were TH-positive (1 week: 7.8 ± 1.1 %, 3 weeks: 6.2 ± 1.0 %) or nNOS-positive (1 week: 10.5 ± 1.2 %, 3 weeks: 16.5 ± 2.7 %, n = 4–6). Despite the response in injured neurons, little c-Jun was seen in retrogradely labelled neurons within spared PG following injury (1 week: 4.5 ± 3.0 %, n = 4; 3 weeks: 8.3 ± 4.0 %, n = 6; grouped unpaired t-test comparing to a hypothetical mean of 0, P N 0.05, Fig. 6C). It should be noted that there was a degree of variability in c-Jun expression between spared PG of different animals following injury. Surprisingly however, many neuronal nuclei within spared PG that did not contain retrograde tracer upregulated c-Jun (data not quantified, Fig. 6C indicated by double arrowhead). In naive sacral DRG, there was no basal expression of c-Jun (Fig. 6F). One week following injury, c-Jun was upregulated in 41.5 ± 6.6 % of retrogradely labelled sacral sensory neurons (P b 0.0001, n = 3, Fig. 6G, J) and 18.0 ± 2.3 % at 3 weeks (P = 0.0002, n = 4, Fig. 6H, J).

137

The chemical phenotype of sensory neurons expressing c-Jun was assessed at 1 week following injury but not at 3 weeks as there were too few neurons expressing c-Jun for a valid analysis. c-Jun was predominantly expressed in peptidergic (CGRP: 69.6 ± 6.2 %) neurons, myelinated (NF200: 48.8 ± 4.9 %) and non-peptidergic unmyelinated neurons (IB4: 15.7 ± 2.9 %). In spared sacral DRG, no significant upregulation of c-Jun was seen in retrogradely labelled sensory neurons following injury (1 week: 0.8 ± 0.3 %, n = 3; 3 weeks: 1.2 ± 0.6 % n = 5, Fig. 6I, J; grouped unpaired t-test comparing to a hypothetical mean of 0, P N 0.05). However, as seen in spared PG from the same animals, there was a substantial upregulation of c-Jun in these ganglia, albeit primarily in neurons that did not contain retrograde tracer (Fig. 6I). Glial cells expressing c-Jun were observed in injured PG and DRG (Fig. 6D). These were morphologically distinguished from neurons as having small, elliptical nuclei that formed clusters around neuron somata, as well as lacking immunoreactivity for sensory or motor neuronal markers (Fig. 6D). c-Jun-positive glia did not appear to be clustered or preferentially associated with retrogradely labelled neurons, but were distributed uniformly across ganglia. There were no changes in the prevalence or distribution of glia at 1 and 3 weeks following injury. Axotomy of visceral motor and sensory neurons does not cause significant neuronal death To determine the impact of axotomy on visceral neuron survival, cell death was directly examined using immunoreactivity for cleaved caspase-3, a key mediator of caspase-dependent apoptosis (Elmore, 2007). No caspase-3-positive neurons were detected in naive pelvic or dorsal root ganglia. Peripheral nerve injury studies show that neurons dying via apoptosis appear within the first 2 weeks (Groves et al., 1997; McKay Hart et al., 2002). Therefore, we chose to investigate neuronal apoptosis within injured pelvic and dorsal root ganglia at 1, 2 and 3 weeks following unilateral accessory nerve transection. Quantification of neuronal apoptosis was performed at 1 week, as this was the expected time in which most death might occur. Within injured PG and DRG, very few or no retrogradely labelled caspase-3-positive neurons were observed at 3 days (data not shown, n = 1, 0 %), 1 week (PG: 1.2 ± 0.5 %; DRGs: 0.13 ± 0.1 %, n = 3) or 3 weeks (PGs: 0%; DRGs: 0%, n = 3). As retrogradely-labelled neurons only represented a proportion of bladder-projecting neurons, the quantification of dying neurons was extended to encompass the entire injured pelvic (Fig. 7A–C) and sacral DRG (Fig. 7D–F). Again, no caspase-3-positive neurons were observed 3 days or 3 weeks following injury (n = 1/group, Fig. 7E, F). A few caspase-positive neurons were seen at 1 week (1.4 + 0.7 neurons/section, n = 3). Apoptosis is characterised by pyknosis, in which nuclear chromatin condenses into dense spheres (Jin and El-Deiry, 2005). Therefore, some sections were labelled with the nuclear acid stain Hoechst in order to examine the morphological integrity of injured pelvic and dorsal root ganglion neuronal nuclei. However, virtually no pyknotic neuronal nuclei were observed following injury. In spared pelvic and sacral ganglia very few caspase-3-positive profiles were detected. Caspase-3-positive glial-like profiles were detected in naive, injured and spared pelvic and dorsal root ganglia (Figs. 7A–F). Glial cells were morphologically distinguished from neurons as having small, ovoid shaped nuclei (Fig. 7G). Caspase-3-positive glia were distributed

Fig. 5. Expression of ATF-3 in bladder-projecting motor and sensory neurons following unilateral accessory nerve transection. Bladder-projecting neurons were labelled with retrograde tracers 1 week prior to unilateral injury. Motor neurons in PG are shown in A and B, while sensory neurons from sacral DRG are shown in C–E. (A) Images of naive PG show no basal ATF-3. (B) Images of injured PG one week after injury show examples of retrogradely labelled motor neurons with nuclear ATF-3 (examples indicated by arrows), although some still showed no ATF-3 expression (arrowheads). (C) ATF-3 was upregulated in a subpopulation of retrogradely-labelled motor neurons in injured pelvic ganglia following injury (1 and 3 weeks: P b 0.0001, n = 4–6), while no ATF-3 expression was seen in and spared ganglia. (D) No ATF-3 was expressed in naive sacral DRG. (E) Images of injured sacral DRG at 1 week show that some bladder sensory neurons express ATF-3 (indicated by arrows), while others have no ATF-3 labelling (arrowheads). (F) At 3 weeks after injury, ATF-3 expression in sensory neurons is downregulated to basal levels (arrowheads). (G) ATF-3 was transiently upregulated in a sub-population of retrogradely-labelled sensory neurons in injured DRG at 1 week following injury (P b 0.0001, n = 3), while no ATF-3 expression was seen in and spared ganglia. Data show mean proportions of retrogradely-labelled neurons expressing ATF-3 + SEM. Significant differences were identified using a paired t-test with Bonferroni correction for multiple comparisons. Scale bar in A applies to all images and represents 50 μm.

138

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

Fig. 6. Regulation of c-Jun in motor and sensory neurons within injured and spared pelvic and dorsal root ganglia. c-Jun immunoreactivity was assessed in naive controls and in injured and spared ganglia at 1 and 3 weeks after injury. Motor neurons in PG are shown in A–E, while sensory neurons from sacral DRG are shown in F–J. (A) Images of naive PG show no basal c-Jun. (B, C) Images of injured (B) and spared (C) PG at 3 weeks following injury show examples of retrogradely labelled neurons that express c-Jun (indicated by arrows) or are c-Jun-negative (indicated by arrowheads); c-Jun expression in both injured and spared ganglia was also seen in neurons that did not contain retrograde tracer. (C) In spared ganglia, very few retrogradely labelled neurons expressed c-Jun, even though c-Jun expression was observed in many neurons that were not retrogradely labelled. (D) c-Jun is expressed in nNOS+ neurons and nNOSglia. (E) c-Jun was upregulated in a subpopulation of retrogradely-labelled motor neurons in injured pelvic ganglia following injury (1 and 3 weeks: P = 0.0002, n = 6). No significant upregulation of c-Jun was detected in retrogradely labelled neurons of spared ganglia. (F) No expression of c-Jun was seen in naive DRG. (G, H) Images of injured sacral DRG at 1 (G) and 3 (H) weeks following injury show examples of c-Jun-positive (indicated by arrows) and c-Jun-negative (indicated by arrowheads) retrogradely labelled neurons. (I) In spared ganglia, the majority of retrogradely labelled neurons showed no labelling for c-Jun (indicated by arrowheads), despite an upregulation of c-Jun seen in many neurons that did not contain retrograde tracer (indicated by double arrowheads). (J) c-Jun was upregulated in a subpopulation of retrogradely-labelled sensory neurons in injured DRG at 1 week (P b 0.0001, n = 3) and 3 weeks (P = 0.0002, n = 4) following injury. No significant upregulation of c-Jun was detected in retrogradely-labelled neurons of spared ganglia. Data show mean proportion of retrogradely-labelled neurons expressing c-Jun + SEM. Significant differences were identified using a paired t-test with Bonferroni correction for multiple comparisons. Scale bar in A applies to all images and represents 50 μm.

uniformly across ganglia, with no clustering of dying glia seen. No changes to the distribution of glia were observed within injured and spared ganglia at 1 and 3 weeks after injury. Discussion In this study we established a novel model of visceral nerve injury that allowed analysis of the capacity of bladder-projecting sensory and motor nerves to recover from injury. By performing a unilateral injury and applying retrograde tracer to both denervated and spared hemispheres

of the bladder, we determined the contribution of regenerative and compensatory axon growth to restore bladder innervation. This technique provided new insights into the connectivity of the intact bladder nerve supply. Furthermore, our results showed that injured visceral sensory neurons have a greater capacity to regenerate axons following axotomy than motor neurons, even though both un-injured sensory and motor neurons readily extend axons into denervated bladder tissue, likely reflecting compensatory growth. Our results demonstrate both the remarkable capacity of the bladder's sensory neuronal circuitry to recover, as well as revealing the limitations of this recovery. The development of

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

139

Fig. 7. Assessment of cell death in pelvic and dorsal root ganglia following unilateral accessory nerve transection. The survival of bladder-projecting motor and sensory neurons was evaluated within injured pelvic and dorsal root ganglia using cleaved caspase-3 immunofluorescence with co-labelling for motor and sensory neuron markers. (A–C) No caspase-3+ neuronal nuclei were detected in the ganglia of naive controls (A). In injured pelvic ganglia, very few caspase-3 neurons were seen at 1 week (B) and 3 weeks (C) following unilateral accessory nerve transection. (D–E) In sacral DRGs very few caspase-3+ neuronal profiles were detected in naive control ganglia (D) or at 1 week (E) and 3 weeks (F) following injury. (G) Gliallike profiles co-localised with nuclear stain Hoechst (Hcst) and caspase-3. (A–G) Scale bars: 50 μm.

new therapies should focus on promoting motor innervation for a full repair of the bladder circuitry system. A population of neurons that bilaterally innervate the bladder was identified for the first time by applying retrograde tracer to opposite hemispheres of the bladder. Such bilaterally projecting neurons, indicated by having taken up both tracer dyes, were present in motor and sensory pathways of naive tissue. This double labelling approach has previously revealed a population of bladder sensory neurons that also innervates the prostate gland (Chen et al., 2010). Our data shows that the majority of sensory and motor neurons innervating the bladder project ipsilaterally, 20 % project contralaterally and 5 % bilaterally; this contribution from contralaterally located neurons is similar to that described previously (Ekstrom et al., 1986). Therefore, some degree of innervation remains following unilateral transection, such that basic voiding reflexes are retained (Berggren et al., 1993; Berggren and Uvelius, 1996). Following unilateral accessory nerve transection, bilaterally projecting neurons were more prevalent in spared ganglia. This suggests that spared neurons grew axon collaterals to restore innervation to denervated bladder tissues while retaining their original connections. Therefore, the novel identification of neurons that bilaterally innervate the bladder provides a deeper understanding of the changes in neural connectivity within the bladder and compensatory growth following injury. This study was underpinned by establishment of a novel model of pelvic visceral nerve injury that focused on bladder innervation. By transecting the accessory nerve we were able to axotomise most bladder-projecting neurons, although our results showed that an alternative route of innervation exists for some motor neurons. The accessory nerve also contains axons that supply the reproductive organs, but assessing the impact of injury on these circuits was outside the scope of the present study. Analyses of bladder tissues following unilateral transection further verified the injury and strengthened our interpretation of neuronal connectivity obtained from retrograde labelling studies. Transection of the accessory nerves axotomised most bladder sensory neurons, resulting in substantial loss of CGRP axons from all of the bladder tissues in the denervated hemisphere at 1 week after injury. Sensory innervation to urothelial and muscle tissue was restored at 4 weeks, although the source of these fibres i.e. from regenerated axons or collateral growth of spared axons, could not be distinguished. Accessory nerve transection led to a significant motor denervation of bladder tissues, with the exception of the vasculature, where no loss was detected. This suggests that vascular innervation reaches the bladder via a different route, entering at the bladder base (Andersson et al., 2000). Like other vascular innervation, it is likely to originate primarily from paravertebral sympathetic ganglia (Janig and McLachlan, 1987). Some return of motor axons is seen within detrusor and suburothelial tissue, and, given our retrograde tracing results, it is likely

that this innervation originated from the compensatory growth of spared motor neurons. Our study of bladder tissues did not incorporate assessment of all nerve types (e.g. non-peptidergic sensory axons are difficult to label and therefore not assessed), but our retrograde labelling studies do not indicate that different responses occur in other sensory or motor neuron populations (see below). Our primary conclusions have been made on the basis of a range of approaches, including retrograde labelling of projections, visualisation of axon populations within the bladder, and chemical characterisation of sensory and motor neurons projecting to the bladder via different routes at different time points after injury to them or their neighbours. While our results are consistent between the different approaches, it should be noted that our quantitative studies on retrogradely-labelled neurons did not use non-biased methods such as the Abercrombie's correction factor. For example, even though we did not notice any substantial change in neuronal size during the course of our analyses, it is possible that small changes in soma size occurred in a number of populations of neurons included in our study — either a decrease in soma size of axotomised neurons or an increase in soma size of uninjured neurons undergoing compensatory sprouting and therefore increasing their target field and activity. These changes in soma size would influence their representation in each field quantified. While we consider this aspect unlikely to substantively affect our primary results, we recognise that our study would have been stronger if more sophisticated analysis methods had been employed for this component of the study. Regrowth of axons from damaged pelvic motor neurons was poor, indicated by our retrograde labelling studies and the sustained upregulation of ATF-3 and c-Jun. In support, no regeneration of pro-erectile parasympathetic axons occurs following bilateral transection of the penile nerve (Carrier et al., 1995; Kato et al., 2003). The regeneration of injured parasympathetic axons is improved by neurotrophic factors and structural support from tissue grafts (Bond et al., 2013; Carrier et al., 1995; Kato et al., 2003; May et al., 2013). Furthermore, slow and incomplete regeneration of penis-projecting axons is described after penile nerve crush (Nangle and Keast, 2007) suggesting that with the appropriate environmental and structural support, parasympathetic neurons are able to regenerate axons. In contrast to the response of motor neurons, injured axons of sensory neurons regenerate and reconnect with the bladder. We did not predict the significant regenerative capacity of bladder sensory axons, given that recovery from transection injuries is more challenging than incomplete (e.g. crush or freeze) injuries that leave the Schwann cell sheath intact. Each population of sensory neurons was represented similarly following injury, indicating a comparable regenerative ability. This contrasts with previous studies that indicate a reduced growth ability of unmyelinated, non-peptidergic sensory neurons (visualised by binding IB4)

140

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

(Leclere et al., 2007). Further support that axons have reconnected to bladder tissues was provided by the down-regulation of ATF-3 (by 3 weeks), which likely demonstrates a successful regeneration of axons (Tsujino et al., 2000). It is proposed therefore that tissues near the nerve injury site provide a sufficiently conducive environment for bladder sensory axons to navigate to their target. Understanding the cues guiding this projection is a high priority. A key finding of this study is the differential regenerative ability of visceral sensory and motor neurons. This data parallels with vagal nerve transection studies that show sensory axons regenerate to restore connectivity to the gastrointestinal tract, while also re-establishing topographical organisation within organs (Phillips et al., 2000). However, as with the present study injured autonomic neurons consistently fail to reinnervate the gastrointestinal tract, lungs or heart (Phillips et al., 2003; Bregeon et al., 2007). Therefore, a key question asks what the biological basis of this distinct regenerative capacity is. First, endogenous differences between motor and sensory neurons may exist that cause sensory axons to regenerate more successfully. For example, superior cervical ganglion 10 (SCG10), a stathmin family protein that regulates microtubule dynamics and protein trafficking and is essential for axonal outgrowth during development, is preferentially expressed in regenerating sensory axons rather than motor axons following sciatic nerve injury (Shin et al., 2014). Several endogenous reasons may exist that account for differences in regenerative capacities of sensory and motor neurons. One possibility is that the tissue injury site expresses appropriate guidance cues for sensory but not autonomic axons. Alternatively, sensory nerves may be more successful at competing for a limited supply of trophic factors secreted from target tissue. Following superior cervical ganglionectomy sympathetic reinnervation of the tarsal muscle is impeded in the presence of sensory fibres, while capsaicin-induced elimination of sensory axons improves sympathetic reinnervation (Fike et al., 1992). Furthermore, afferent sprouting and hyper-innervation of peptidergic sensory fibres occur in sweat glands after sympathetic axons are chemically eliminated (Aberdeen et al., 1990; Yodlowski et al., 1984). A third possibility may lie in different axon regeneration rates between sensory and autonomic neurons. Axons must reconnect to target tissue during a critical time window after injury, as prolonged axotomy reduces the capacity of axons to regenerate (Fu and Gordon, 1995). The peripheral branch of sensory neurons regenerates at a rate of 4–5 mm/day (Wujek and Lasek, 1983), however vagal motor axons regrow at 1.4 mm/day (Kanje, 1991). Furthermore, return of sensory function precedes that of motor by several weeks following a sciatic nerve crush (Painter et al., 2014). This is also supported by the observation that axotomised sensory axons preferentially express SCG10, a protein essential for axonal outgrowth (Shin et al., 2014). The difference in regenerative ability may also relate to the site of axotomy, which occurs closer to the cell body of bladder-projecting motor neurons that lie in pelvic ganglia, compared with sensory neuron soma that lie in distant sacral DRG, as some studies in other parts of the nervous system have shown that injury close to the soma can induce neuronal death more readily than distal axotomy, potentially due to the impact of calcium influx (Hart et al., 2008; Mattsson et al., 1999). Conversely, other studies show that regeneration is more successful in neurons that receive axotomy proximal to the soma than those receiving a distal injury (Fenrich et al., 2007; Lu and Richardson, 1991). Taken together, it appears difficult to predict the effects of injury from studies carried out in different parts of the nervous system and it is likely that both exogenous and endogenous factors affect the regenerative capacity of sensory and motor neurons. ATF-3 promotes survival and promotes neurite outgrowth (Herdegen et al., 1997; Seijffers et al., 2006). Injury of peripheral somatic nerves shows that the ATF-3 expression is rapidly upregulated in the majority of injured sensory and motor neurons (Seijffers et al., 2006; Tsujino et al., 2000). In contrast, injury of the central axon of sensory neurons induces ATF-3 in only a small sub-population of the injured neurons (Seijffers et al., 2006), and only half of injured retinal ganglion cells

upregulate ATF-3 after a complete optic nerve crush (Takeda et al., 2000). Similarly, only half of axotomised sacral preganglionic neurons upregulate ATF-3 (Peddie and Keast, 2011). Likewise, in the current study ATF-3 was upregulated in only a fraction of the injured sensory and motor neurons, one week after axotomy. The reason for this heterogeneity is not known, or whether it correlates with an improved ability for those particular neurons to regenerate axons. However, it does indicate that in this system ATF-3 is not a universal marker of injury. A second mechanism exists for restoring neural control to an organ, namely compensatory growth of axon collaterals from spared neurons (Navarro et al., 2007). In the somatic nervous system, this leads to the sensory reinnervation of skin after sciatic nerve injury (Cobianchi et al., 2014; Kingery and Vallin, 1989). This has also been observed in the visceral nervous system, where growth of spared parasympathetic axons into denervated cavernosal tissue is associated with the slow return of erectile function (Carrier et al., 1995; Kato et al., 2003; Nangle and Keast, 2007; Palma and Keast, 2006). Plasticity of cholinergic bladder nerves has also been demonstrated previously (Uvelius and Gabella, 1998). In the current study a subpopulation of spared sensory and motor axons forms bilateral connections between the two bladder hemispheres, which suggests that these axons undergo compensatory growth into denervated tissue. However it is not known if these new connections are detrimental to function. For example, following dorsal root injury the disorganisation of the topographical arrangement of regenerating afferents within the dorsal horn is associated with the development of hyperalgesia and chronic pain conditions (Harvey et al., 2010). Therefore, although neural connections are restored to the bladder following injury, the question remains whether these newly formed synapses are completely functional. Upregulation of c-Jun expression is associated with axonal regeneration (Herdegen et al., 1997). In support, we see an upregulation of c-Jun in a subpopulation of neurons that received an axonal injury. Furthermore, c-Jun upregulation was maintained in approximately 18 % of sensory neurons at 3 weeks, after these same neurons have downregulated ATF-3 to basal levels. Recent studies show that c-Jun upregulation is associated with aberrant intra-ganglionic growth initiated by deafferentation rather than by direct injury (Nangle and Keast, 2009; Tsujino et al., 2000). In contrast however, the present study showed that compensatory growth of bladder axons is unlikely to be regulated by c-Jun, as very few uninjured bladder-projecting neurons in spared ganglia expressed this transcription factor. These c-Jun+ neurons were not seen consistently, but it is possible that a role of c-Jun could be identified with a greater number of replicates, examination of additional time points or assessment of c-Jun mRNA. Our observations are supported however by the observation that no upregulation of c-Jun occurs in uninjured penis-projecting neurons within spared pelvic ganglia after unilateral cavernous nerve transection (Nangle and Keast, 2009). Together, these observations fail to support our original hypothesis and suggest that compensatory growth occurs by a different mechanism than that seen after deafferentation- or axotomy-induced growth. Interestingly, many c-Jun-positive neurons that did not contain retrograde tracer were observed in spared ganglia. This parallels a previous study of unilateral cavernous nerve injury where c-Jun upregulation in spared pelvic ganglia did occur but was not found in penis-projecting neurons (Nangle and Keast, 2009). This was assessed by the absence of NOS in the c-Jun-positive neurons, as nearly all penis-projecting neurons express NOS (Hedlund et al., 1999; Mizusawa et al., 2001). Although we do not expect all bladder-projecting neurons to contain retrograde tracer, the weak correlation between c-Jun and retrograde tracer is more consistent with c-Jun upregulation in neurons that do not project to the bladder. It will be of interest to determine the function of these neurons. No neuronal death was detected within injured pelvic or dorsal root ganglia using caspase-3, sampled at a number of time points after injury. Furthermore, there was no significant effect of injury on the number of retrogradely labelled motor neurons with injured pelvic ganglia, which

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

is especially surprising given that these motor neurons remained chronically axotomised after transection. In support, other studies on injured visceral circuits support our findings by failing to demonstrate substantial loss of neurons after axotomy of autonomic preganglionic neurons or penis-projecting motor neurons (Coulibaly et al., 2013; Nangle and Keast, 2009; Palma and Keast, 2006; Peddie and Keast, 2011; Vizzard et al., 1995). The ability of pelvic motor neurons to survive axotomy is further indicated by their high survival rate following transplantation into the bladder wall (Gabella and Uvelius, 1998). This extended survival of chronically axotomised autonomic neurons provides a significant window of opportunity in which therapeutic strategies may be applied to promote axon regrowth. In addition to our analysis of neurons following accessory nerve injury, we observed changes occurring in glia of pelvic and dorsal root ganglia. In particular, upregulation in glial expression of c-Jun within injured and spared ganglia suggests an active role of glia in both the regenerative and compensatory types of axonal growth after injury. A similar observation was made within injured pelvic ganglia following cavernous nerve injury (Nangle and Keast, 2007). Furthermore, expression of caspase-3 is seen in glia within naive, injured and spared ganglia, which provides evidence of glial turnover. The direction of this communication and the nature of glial-neuronal signalling in pelvic nerve circuits are not known but have been investigated extensively elsewhere (Hanani, 2010, 2012). Conclusion By developing a model of combined pelvic sensory and motor visceral nerve injury, we identified distinct regenerative capacities of sensory and motor axons innervating the bladder. Sensory axons showed much greater regeneration following axotomy, but non-regenerating motor neurons in pelvic ganglia survived the injury, thereby allowing an extended opportunity for therapeutic intervention. Sensory and motor neurons in spared ganglia likely underwent compensatory axon growth to supply the denervated tissue, providing another opportunity for restoring function. Understanding the environmental and cellular mechanisms that promote regeneration of bladder sensory axons and inhibit axonal regeneration from pelvic ganglion neurons may provide insight to new therapeutic strategies for recovering bladder function after surgically-induced injury. This may also suggest strategies for promoting innervation of regenerated bladder tissue following grafts or transplants. Acknowledgments This study was supported by Project Grant 1022941 from the National Health and Medical Research Council of Australia. References Aberdeen, J., Corr, L., Milner, P., Lincoln, J., Burnstock, G., 1990. Marked increases in calcitonin gene-related peptide-containing nerves in the developing rat following long-term sympathectomy with guanethidine. Neuroscience 35, 175–184. Albersen, M., Kendirci, M., Van der Aa, F., Hellstrom, W.J., Lue, T.F., Spees, J.L., 2012. Multipotent stromal cell therapy for cavernous nerve injury-induced erectile dysfunction. J. Sex. Med. 9, 385–403. Andersson, K.E., Hedlund, P., Alm, P., 2000. Sympathetic pathways and adrenergic innervation of the penis. Int. J. Impot. Res. 12 (Suppl. 1), S5–S12. Berggren, T., Uvelius, B., 1996. Acute effects of unilateral pelvic ganglionectomy on urinary bladder function in vivo in the male rat. Scand. J. Urol. Nephrol. 30, 179–184. Berggren, T., Gabella, G., Malmgren, A., Uvelius, B., 1993. Effects of unilateral pelvic ganglionectomy on urinary bladder function in the male rat. Scand. J. Urol. Nephrol. 27, 181–188. Bloechlinger, S., Karchewski, L.A., Woolf, C.J., 2004. Dynamic changes in glypican-1 expression in dorsal root ganglion neurons after peripheral and central axonal injury. Eur. J. Neurosci. 19, 1119–1132. Bond, C.W., Angeloni, N., Harrington, D., Stupp, S., Podlasek, C.A., 2013. Sonic Hedgehog regulates brain-derived neurotrophic factor in normal and regenerating cavernous nerves. J. Sex. Med. 10, 730–737. Bradke, F., Fawcett, J.W., Spira, M.E., 2012. Assembly of a new growth cone after axotomy: the precursor to axon regeneration. Nat. Rev. Neurosci. 13, 183–193.

141

Bregeon, F., Alliez, J.R., Hery, G., Marqueste, T., Ravailhe, S., Jammes, Y., 2007. Motor and sensory re-innervation of the lung and heart after re-anastomosis of the cervical vagus nerve in rats. J. Physiol. 581, 1333–1340. Canguven, O., Burnett, A., 2008. Cavernous nerve injury using rodent animal models. J. Sex. Med. 5, 1776–1785. Carrier, S., Zvara, P., Nunes, L., Kour, N.W., Rehman, J., Lue, T.F., 1995. Regeneration of nitric oxide synthase-containing nerves after cavernous nerve neurotomy in the rat. J. Urol. 153, 1722–1727. Chen, Y., Wu, X., Liu, J., Tang, W., Zhao, T., Zhang, J., 2010. Distribution of convergent afferents innervating bladder and prostate at dorsal root ganglia in rats. Urology 76 (764), e761–e766. Cobianchi, S., de Cruz, J., Navarro, X., 2014. Assessment of sensory thresholds and nociceptive fiber growth after sciatic nerve injury reveals the differential contribution of collateral reinnervation and nerve regeneration to neuropathic pain. Exp. Neurol. 255, 1–11. Coulibaly, A.P., Gannon, S.M., Hawk, K., Walsh, B.F., Isaacson, L.G., 2013. Transection of preganglionic axons leads to CNS neuronal plasticity followed by survival and target reinnervation. Auton. Neurosci. 179, 49–59. de Groat, W.C., Yoshimura, N., 2009. Afferent nerve regulation of bladder function in health and disease. Handb. Exp. Pharmacol. 91–138. Ekstrom, J., Malmberg, L., Oberg, S., 1986. Unilateral denervation of the rat urinary bladder and reinnervation: a predominance for ipsilateral changes. Acta Physiol. Scand. 127, 223–231. Elmore, S., 2007. Apoptosis: a review of programmed cell death. Toxicol. Pathol. 35, 495–516. Fenrich, K.K., Skelton, N., MacDermid, V.E., Meehan, C.F., Armstrong, S., Neuber-Hess, M.S., Rose, P.K., 2007. Axonal regeneration and development of de novo axons from distal dendrites of adult feline commissural interneurons after a proximal axotomy. J. Comp. Neurol. 502, 1079–1097. Fike, E.A., Simons, E., Boswell, C., Smith, P.G., 1992. Sensory nerves impair sympathetic reinnervation and recovery of smooth muscle function. Exp. Neurol. 118, 85–94. Forrest, S.L., Osborne, P.B., Keast, J.R., 2013. Characterization of bladder sensory neurons in the context of myelination, receptors for pain modulators, and acute responses to bladder inflammation. Front. Neurosci. 7, 206. Francis, J.S., Dragunow, M., During, M.J., 2004. Over expression of ATF-3 protects rat hippocampal neurons from in vivo injection of kainic acid. Brain Res. Mol. Brain Res. 124, 199–203. Fu, S.Y., Gordon, T., 1995. Contributing factors to poor functional recovery after delayed nerve repair: prolonged axotomy. J. Neurosci. 15, 3876–3885. Gabella, G., Uvelius, B., 1998. Homotransplant of pelvic ganglion into bladder wall in adult rats. Neuroscience 83, 645–653. Groves, M.J., Christopherson, T., Giometto, B., Scaravilli, F., 1997. Axotomy-induced apoptosis in adult rat primary sensory neurons. J. Neurocytol. 26, 615–624. Hanani, M., 2010. Satellite glial cells in sympathetic and parasympathetic ganglia: in search of function. Brain Res. Rev. 64, 304–327. Hanani, M., 2012. Intercellular communication in sensory ganglia by purinergic receptors and gap junctions: implications for chronic pain. Brain Res. 1487, 183–191. Hart, A.M., Terenghi, G., Wiberg, M., 2008. Neuronal death after peripheral nerve injury and experimental strategies for neuroprotection. Neurol. Res. 30, 999–1011. Harvey, P., Gong, B., Rossomando, A.J., Frank, E., 2010. Topographically specific regeneration of sensory axons in the spinal cord. Proc. Natl. Acad. Sci. U. S. A. 107, 11585–11590. Hedlund, P., Alm, P., Andersson, K.E., 1999. NO synthase in cholinergic nerves and NOinduced relaxation in the rat isolated corpus cavernosum. Br. J. Pharmacol. 127, 349–360. Herdegen, T., Skene, P., Bahr, M., 1997. The c-Jun transcription factor-bipotential mediator of neuronal death, survival and regeneration. Trends Neurosci. 20, 227–231. Hoang, T.X., Pikov, V., Havton, L.A., 2006. Functional reinnervation of the rat lower urinary tract after cauda equina injury and repair. J. Neurosci. 26, 8672–8679. Janig, W., McLachlan, E.M., 1987. Organization of lumbar spinal outflow to distal colon and pelvic organs. Physiol. Rev. 67, 1332–1404. Jin, Z., El-Deiry, W.S., 2005. Overview of cell death signaling pathways. Cancer Biol. Ther. 4, 139–163. Kanje, M., 1991. Survival and regeneration of the adult rat vagus nerve in culture. Brain Res. 550, 340–342. Kato, R., Kiryu-Seo, S., Sato, Y., Hisasue, S., Tsukamoto, T., Kiyama, H., 2003. Cavernous nerve injury elicits GAP-43 mRNA expression but not regeneration of injured pelvic ganglion neurons. Brain Res. 986, 166–173. Keast, J.R., 2006. Plasticity of pelvic autonomic ganglia and urogenital innervation. Int. Rev. Cytol. 248, 141–208. Keast, J.R., de Groat, W.C., 1989. Immunohistochemical characterization of pelvic neurons which project to the bladder, colon, or penis in rats. J. Comp. Neurol. 288, 387–400. Keast, J.R., De Groat, W.C., 1992. Segmental distribution and peptide content of primary afferent neurons innervating the urogenital organs and colon of male rats. J. Comp. Neurol. 319, 615–623. Keast, J.R., Booth, A.M., de Groat, W.C., 1989. Distribution of neurons in the major pelvic ganglion of the rat which supply the bladder, colon or penis. Cell Tissue Res. 256, 105–112. Kingery, W.S., Vallin, J.A., 1989. The development of chronic mechanical hyperalgesia, autotomy and collateral sprouting following sciatic nerve section in rat. Pain 38, 321–332. Lange, M.M., Maas, C.P., Marijnen, C.A., Wiggers, T., Rutten, H.J., Kranenbarg, E.K., van de Velde, C.J., Cooperative Clinical Investigators of the Dutch Total Mesorectal Excision, T., 2008. Urinary dysfunction after rectal cancer treatment is mainly caused by surgery. Br. J. Surg. 95, 1020–1028. Leclere, P.G., Norman, E., Groutsi, F., Coffin, R., Mayer, U., Pizzey, J., Tonge, D., 2007. Impaired axonal regeneration by isolectin B4-binding dorsal root ganglion neurons in vitro. J. Neurosci. 27, 1190–1199.

142

S.C. Payne et al. / Experimental Neurology 266 (2015) 127–142

Lu, X., Richardson, P.M., 1991. Inflammation near the nerve cell body enhances axonal regeneration. J. Neurosci. 11, 972–978. Mattsson, P., Meijer, B., Svensson, M., 1999. Extensive neuronal cell death following intracranial transection of the facial nerve in the adult rat. Brain Res. Bull. 49, 333–341. May, F., Buchner, A., Schlenker, B., Gratzke, C., Arndt, C., Stief, C., Weidner, N., Matiasek, K., 2013. Schwann cell-mediated delivery of glial cell line-derived neurotrophic factor restores erectile function after cavernous nerve injury. Int. J. Urol. 20, 344–348. McKay Hart, A., Brannstrom, T., Wiberg, M., Terenghi, G., 2002. Primary sensory neurons and satellite cells after peripheral axotomy in the adult rat: timecourse of cell death and elimination. Exp. Brain Res. 142, 308–318. Mizusawa, H., Hedlund, P., Hakansson, A., Alm, P., Andersson, K.E., 2001. Morphological and functional in vitro and in vivo characterization of the mouse corpus cavernosum. Br. J. Pharmacol. 132, 1333–1341. Nangle, M.R., Keast, J.R., 2007. Reduced efficacy of nitrergic neurotransmission exacerbates erectile dysfunction after penile nerve injury despite axonal regeneration. Exp. Neurol. 207, 30–41. Nangle, M.R., Keast, J.R., 2009. Deafferentation and axotomy each cause neurturinindependent upregulation of c-Jun in rodent pelvic ganglia. Exp. Neurol. 215, 271–280. Navarro, X., Vivo, M., Valero-Cabre, A., 2007. Neural plasticity after peripheral nerve injury and regeneration. Prog. Neurobiol. 82, 163–201. Nishizawa, Y., Ito, M., Saito, N., Suzuki, T., Sugito, M., Tanaka, T., 2011. Male sexual dysfunction after rectal cancer surgery. Int. J. Color. Dis. 26, 1541–1548. Painter, M.W., Brosius Lutz, A., Cheng, Y.C., Latremoliere, A., Duong, K., Miller, C.M., Posada, S., Cobos, E.J., Zhang, A.X., Wagers, A.J., Havton, L.A., Barres, B., Omura, T., Woolf, C.J., 2014. Diminished Schwann cell repair responses underlie age-associated impaired axonal regeneration. Neuron 83, 331–343. Palma, C.A., Keast, J.R., 2006. Structural effects and potential changes in growth factor signalling in penis-projecting autonomic neurons after axotomy. BMC Neurosci. 7, 41. 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. Front. Neurosci. 5, 6. Phillips, R.J., Baronowsky, E.A., Powley, T.L., 2000. Regenerating vagal afferents reinnervate gastrointestinal tract smooth muscle of the rat. J. Comp. Neurol. 421, 325–346. Phillips, R.J., Baronowsky, E.A., Powley, T.L., 2003. Long-term regeneration of abdominal vagus: efferents fail while afferents succeed. J. Comp. Neurol. 455, 222–237. Raivich, G., Bohatschek, M., Da Costa, C., Iwata, O., Galiano, M., Hristova, M., Nateri, A.S., Makwana, M., Riera-Sans, L., Wolfer, D.P., Lipp, H.P., Aguzzi, A., Wagner, E.F., Behrens, A., 2004. The AP-1 transcription factor c-Jun is required for efficient axonal regeneration. Neuron 43, 57–67.

Scheib, J., Hoke, A., 2013. Advances in peripheral nerve regeneration. Nat. Rev. Neurol. 9, 668–676. Seijffers, R., Allchorne, A.J., Woolf, C.J., 2006. The transcription factor ATF-3 promotes neurite outgrowth. Mol. Cell. Neurosci. 32, 143–154. Shin, J.E., Geisler, S., Diantonio, A., 2014. Dynamic regulation of SCG10 in regenerating axons after injury. Exp. Neurol. 252, 1–11. Takeda, M., Kato, H., Takamiya, A., Yoshida, A., Kiyama, H., 2000. Injury-specific expression of activating transcription factor-3 in retinal ganglion cells and its colocalized expression with phosphorylated c-Jun. Invest. Ophthalmol. Vis. Sci. 41, 2412–2421. Tandrup, T., Woolf, C.J., Coggeshall, R.E., 2000. Delayed loss of small dorsal root ganglion cells after transection of the rat sciatic nerve. J. Comp. Neurol. 422, 172–180. Terenghi, G., Hart, A., Wiberg, M., 2011. The nerve injury and the dying neurons: diagnosis and prevention. J. Hand Surg. Eur. Vol. 36, 730–734. 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. Mol. Cell. Neurosci. 15, 170–182. Uvelius, B., Gabella, G., 1998. The distribution of intramural nerves in urinary bladder after partial denervation in the female rat. Urol. Res. 26, 291–297. Uvelius, B., Kanje, M., 2010. Glial cell activation in pelvic ganglia after preganglionic but not postganglionic lesions. UroToday Int. J. 3. Vestergaard, S., Tandrup, T., Jakobsen, J., 1997. Effect of permanent axotomy on number and volume of dorsal root ganglion cell bodies. J. Comp. Neurol. 388, 307–312. Vizzard, M.A., Erdman, S.L., de Groat, W.C., 1995. Increased expression of neuronal nitric oxide synthase (NOS) in visceral neurons after nerve injury. J. Neurosci. 15, 4033–4045. Wallner, C., Lange, M.M., Bonsing, B.A., Maas, C.P., Wallace, C.N., Dabhoiwala, N.F., Rutten, H.J., Lamers, W.H., Deruiter, M.C., van de Velde, C.J., 2008. Causes of fecal and urinary incontinence after total mesorectal excision for rectal cancer based on cadaveric surgery: a study from the Cooperative Clinical Investigators of the Dutch total mesorectal excision trial. J. Clin. Oncol. 26, 4466–4472. Wujek, J.R., Lasek, R.J., 1983. Correlation of axonal regeneration and slow component B in two branches of a single axon. J. Neurosci. 3, 243–251. Ygge, J., 1989. Neuronal loss in lumbar dorsal root ganglia after proximal compared to distal sciatic nerve resection: a quantitative study in the rat. Brain Res. 478, 193–195. Yodlowski, M.L., Fredieu, J.R., Landis, S.C., 1984. Neonatal 6-hydroxydopamine treatment eliminates cholinergic sympathetic innervation and induces sensory sprouting in rat sweat glands. J. Neurosci. 4, 1535–1548.