In contrast to neurons of the central nervous system, many adult peripheral neurons not only survive axotomy but also undergo axonal regeneration. Within just a few hours, injured axons grow collaterals ("sprouts") and their growth cones respond to guidance cues in order to find an environment supportive of regeneration [1–3]. As well as being regarded as the beginning of the regenerative response, many sprouts form close associations with neighbouring ganglion neurons. These have been best described for sympathetic ganglion and primary sensory (dorsal root ganglion) neurons [4–6]. Neurotrophic factors, released by target tissues and by glial cells in the vicinity of the injured neuron, are critical for this growth [3, 7]. In sympathetic ganglia the neurotrophins, especially nerve growth factor, play this role after injury. Neurotrophin-dependent regeneration is also associated with changes in expression of neurotrophin receptors, possibly caused by altered availability of ligand, and setting up a feedback loop in the regenerative process [8–10].

The regenerative abilities of injured parasympathetic ganglion neurons are very poorly understood. This is partly because of technical difficulties in selectively injuring their axons, given that most parasympathetic neurons lie in small, dispersed ganglia that are closely associated with or even within their target tissues [11]. Therefore, damage to many organs injures not only the axons but also the somata of parasympathetic ganglion neurons. Another major factor has been the delay in identifying the neurotrophic factors for parasympathetic neurons. Neurturin (NTN) belongs to a recently identified family of neurotrophic factors, which include glial cell line-derived neurotrophic factor (GDNF), NTN, artemin and persephin [12–14]. NTN and GDNF are important during the development (survival) and/or maintenance of many cranial parasympathetic neurons. Their actions vary with the stage of development and between different cranial ganglia [15–18]. The GDNF family ligands (GFLs) act via a two-component receptor complex, a common signalling component, the Ret receptor tyrosine kinase, and a glycosylphosphatidylinositol-anchored GFRα protein, which binds ligand with high affinity and determines specificity [14]. GDNF preferentially binds to GFRα1 and NTN to GFRα2, although in vitro studies have suggested that each ligand may be able signal through alternative GFR mechanisms at higher concentrations [13, 19].

NTN is a neurotrophic factor for many pelvic cholinergic ganglion neurons, which are the sole source of parasympathetic innervation of the urogenital organs and a major source of extrinsic autonomic nerves supplying the lower bowel [20]. Most or all of the pelvic organs express NTN [21–23] and most adult cholinergic pelvic neurons express GFRα2 [24]. Mice deficient in GFRα2 exhibit loss of cholinergic nerve terminals in some regions of the penis [22]. These mice, and neurturin knockout mice, also show a substantial decrease in cholinergic innervation of reproductive organ epithelia [24]. Both deficits may indicate a failure of survival, arborisation or maintenance of these projections. Recently, NTN has been shown to have multiple neurotrophic actions on adult rat pelvic parasympathetic ganglion neurons in vitro [25]. GDNF is also relevant to maintenance of penile innervation; many penile neurons express GFRα1 and retrogradely transport GDNF from the penis [22].

Pelvic autonomic neurons, especially parasympathetic neurons, are inadvertently axotomised during pelvic surgical procedures such as prostatectomy, hysterectomy, and bowel resections for cancer removal [26–28]. This commonly leads to post-operative urogenital problems such as erectile dysfunction and urinary incontinence. There is high demand for new regenerative strategies for these injured neurons. As an initial step in understanding the injury and regenerative processes in these neurons, we have investigated whether pelvic parasympathetic neurons undergo similar changes after axotomy as seen in adult sympathetic and dorsal root ganglia, by assessing structural changes in adult rat pelvic ganglia shortly after penile (cavernous) nerve transection or crush. To provide further insight into the role of growth factors in the regenerative response we investigated the effects of axotomy on the neuronal expression of the NTN receptor, GFRα2, and the GDNF receptor, GFRα1. We chose to use immunohistochemical analysis for this component of the study as it allows changes in neuronal expression to be distinguished from those in non-neuronal cells, such as glia, and also allows expression to be monitored in different chemical classes of neurons or neurons retrogradely labelled from the penis. Our results show profound structural changes in the pre- and postganglionic nerve pathways innervating the penis following penile nerve injury. We also identify an effect of axotomy on GFRα2 expression by injured neurons, suggesting an effect of injury on growth factor signalling.

We have examined the effects of axotomy on pelvic parasympathetic ganglion neurons and possible changes in growth factor receptor expression initiated by the injury. The penile nerve was chosen as it is a well-characterised projection from the pelvic ganglion, and is both easily identifiable and readily accessible for experimental intervention. It is also relevant to clinical problems associated with pelvic surgery (e.g. prostatectomy) where penile axons are injured and the demand for regenerative therapies is high. Our results show that axotomy causes sprouting within the pelvic ganglion, most likely from the axons of injured parasympathetic neurons. Moreover, axotomy may trigger structural or chemical changes in preganglionic terminals innervating the damaged neurons. Our results showed that following axotomy there is a decrease in GFRα2 but not GFRα1 expression in pelvic parasympathetic neurons, which may indicate a change in GFL signalling mechanisms at an early phase of the regenerative state.

The pelvic ganglion is unusual, not only because it contains intermingled populations of sympathetic and parasympathetic neurons, but also because many of its parasympathetic neurons lie distant from their target organs [20]. Moreover, a minority of cholinergic pelvic neurons receive synaptic inputs from the lumbar cord, so strictly speaking should be referred to as sympathetic [29]. This is also the case for a minority of penis-projecting pelvic cholinergic neurons [38], however in the present study we did not determine the source of spinal inputs for FG-labelled neurons. Because most of these are likely to be parasympathetic (i.e. sacral inputs) and in recognition of their similar chemical phenotype and likely growth factor dependency, for simplicity we have referred to them collectively as penile parasympathetic neurons.

In the present study we injured only one population of pelvic parasympathetic neurons but assessed the effects of injury on the broader population of parasympathetic neurons and on the neighbouring uninjured noradrenergic sympathetic populations. Most of our results concur with structural and chemical changes being restricted to the injured parasympathetic population, although a small decrease in GFRα1 expression in sympathetic neurons after penile nerve transection suggests that some changes may occur more broadly in the ganglion. One possible mechanism for this is if axotomy alters glial secretions or glial coupling within the ganglion, allowing signalling between pelvic ganglion neurons. This has been described in sensory ganglia [39].

Penile nerve injury caused sprouting of VIP neurons in the region of the ganglion closest to the injury site, resembling the sprouting seen in sympathetic ganglia after axotomy, and regarded as an early phase of the regenerative response [1, 2]. The vast majority of rat pelvic ganglion neurons are adendritic or have only a few short fine dendrites [40]. Therefore potential cellular targets of growing fibres are relatively easy to identify. Following penile nerve lesion, VIP fibres formed basket-like structures around pelvic ganglion neurons, many of which were injured (i.e. FG-labelled). It is not known if these fibres are capable of forming functional synapses or are likely to be permanent, although eight days after injury did not yet express immunohistochemically detectable levels of synapsin or synaptophysin. Aberrant connections are made within other injured ganglia and are thought to underlie prolonged problems with some visceral reflexes [5, 6, 36]. In the pelvic ganglion these changes could lead to profoundly altered connectivity and functionality in the penile erection reflex circuitry following axotomy. Therefore, even if regeneration to the target occurs, other problems with reflexes may continue.

Our immunohistochemical studies raise the possibility of axotomy triggering dramatic changes within preganglionic axons that synapse on penis-projecting neurons. The loss of synaptic protein immunoreactivity within these terminals may represent a chemical or structural change, or both. Ultrastructural or electrophysiological methods are necessary to determine the functional nature of the deficit. There are various precedents for such a retrograde effect of neuronal injury. Previous studies of sympathetic ganglia show retraction of dendrites and of preganglionic inputs after postganglionic axotomy, both of which reverse if postganglionic regeneration occurs [41, 42]. There are also profound and at least partially reversible changes in synaptic transmission, nicotinic receptor expression and postganglionic membrane properties after axotomy [43–46]. It is important to determine the longer-term impact of penile nerve injury on lumbosacral spinal neurons. The reversibility of this change with penile nerve regeneration or compensatory growth from ipsilateral pathways should also be determined in order to understand the possible extent of recovery or manipulation required to facilitate recovery in this system.

A number of groups have shown that penile erection can return after injury, although after bilateral transection the success of reinnervation is low [47–49]. Regeneration of the erectile response is much more successful after unilateral penile nerve lesion, indicating that it is largely due to compensatory sprouting of uninjured penile-projecting fibres. Both types of growth, regeneration and compensatory, may be driven by and depend on neurotrophic factors, the most likely candidates being NTN and GDNF. NTN is synthesised by many urogenital tissues, including the penis [22, 50, 51] and radioactively labelled NTN is transported retrogradely along penile axons [22]. Indeed, axotomy of penis-projecting neurons causes a decrease in soma size [25], consistent with reduced access to neurotrophic factors. However, NTN may also be secreted by glial cells within the ganglion or glia associated with the injured axon [52–54]. These would be the most likely sources of trophic factor after injuries where axons are disconnected from their target organs. Another interesting possibility has been raised by a recent report of NTN expression by immune cells [55].

GDNF [56] and neurotrophin-3 [57] are additional possible neurotrophic factors for penile parasympathetic neurons. In the present study all of the VIP-immunoreactive and FG-labelled neurons expressed GFRα1. GDNF can also prevent some of the axotomy-induced changes, including sprouting, in sensory neurons [58], and has additive effects with NTN to rescue retinal ganglion neurons from axotomy-induced death [59]. It is possible that a combination of these growth factors acts in concert to promote regenerative responses in the penis-projecting neurons.

The present study showed that penile nerve axotomy consistently caused a decrease in GFRα2 expression in injured neurons but had no effect on GFRα1, however our experiments do not exclude an effect on uninjured parasympathetic neurons. Previous experiments on dorsal root ganglion neurons showed that after sciatic nerve injury there was a decrease in GFRα2 but an increase in GFRα1 expression, and that GDNF was able to prevent the change in GFRα2 [60, 61]. The greater effect of injury on GFRα1 shown in these earlier studies may indicate a fundamental difference between sensory and parasympathetic ganglia in their injury responses or in their GFL availability within injured ganglia. It is also possible that in our experiments subtle changes occurred in GFRα1 expression that were not detected with our technique. Our observation of a decrease in GFRα2 expression after axotomy does not necessarily preclude a role of NTN in axonal regeneration. For example, it is possible that NTN is involved in the initial extension of axons or collatorals following injury but then further axonal extension is NTN-independent. Alternatively, even after down-regulation of GFRα2 there may still be sufficient receptor expression to mediate physiological maintenance and growth responses. Ideally the expression of NTN and GFRα2 should be monitored or manipulated at various time points following injury. In addition, our studies relied on immunohistochemistry to assess changes in GFR expression. The strength of this approach is that it allows changes in neuronal expression to be determined (unlike Western blotting and RT-PCR methods, which require extracts of entire ganglia and so will include glia and other non-neuronal cells). Moreover, this approach has allowed us to selectively analyse retrogradely labelled and TH-positive neurons. We acknowledge that there is unlikely to be a linear relationship between fluorescence intensity measurements and protein levels, nevertheless expression changes and the direction of those changes can be clearly identified following injury.

In contrast to GFRα2, which was selectively expressed by parasympathetic neurons, GFRα1 was expressed similarly in sympathetic and parasympathetic neurons. Our results indicate that axotomy had no effect on GFRα1 expression by pelvic parasympathetic neurons, although subtle changes may have occurred that were not measurable with our technique. However GFRα1 was down-regulated in sympathetic (TH-positive) neurons, even though these neurons do not project to the penis and therefore are unlikely to be injured. Recent studies on regeneration of injured penile axons show that GAP-43 expression is increased in some TH-positive neurons on the contralateral (uninjured) side [48], suggesting a compensatory response. This may be related to their decrease in GFRα1 expression.

These studies show that there are profound changes within the pelvic ganglion after penile nerve injury. Sprouting of injured postganglionic axons occurs concurrently with structural or chemical changes in preganglionic terminals. New growth of postganglionic axon collaterals within the ganglion raises the possibility of the formation of aberrant synaptic connections between injured and un-injured ganglion neurons. Together these changes demonstrate a broader effect on the pelvic autonomic circuitry than simply loss of neuroeffector connections. These structural changes are accompanied by potential changes in neurotrophic factor signalling due to altered expression of receptors for members of the GDNF family. Together our results advance understanding of the responses of pelvic autonomic nerve circuits to injury and may assist in designing strategies for promoting regeneration.