The synaptic terminals' withdrawal from the somata and proximal dendrites of injured motoneuron by the processes of glial cells following facial nerve axotomy has been the subject of research for many years. This phenomenon is referred to as synaptic stripping, which is assumed to help survival and regeneration of neurons via reduction of synaptic inputs. Because there is no disruption of the blood–brain barrier or infiltration of macrophages, the axotomy paradigm has the advantage of being able to selectively investigate the roles of resident glial cells in the brain. Although there have been numerous studies of synaptic stripping, the detailed mechanisms are still under debate. Here we suggest that the species and strain differences that are often present in previous work might be related to the current controversies of axotomy studies. For instance, the survival ratios of axotomized neurons were generally found to be higher in rats than in mice. However, some studies have used the axotomy paradigm to follow the glial reactions and did not assess variations in neuronal viability. In the first part of this article, we summarize and discuss the current knowledge on species and strain differences in neuronal survival, glial augmentation and synaptic stripping. In the second part, we focus on our recent findings, which show the differential involvement of microglia and astrocytes in synaptic stripping and neuronal survival. This article suggests that the comparative study of the axotomy paradigm across various species and strains may provide many important and unexpected discoveries on the multifaceted roles of microglia and astrocytes in injury and repair.

Dorsal root ganglia (DRG) respond to peripheral nerve injury by up-regulating nitric oxide (NO) production by neurons and glia in addition to local fibroblasts, endothelium and macrophages. We hypothesise that NO produced from these cells has specific roles. We have shown that when neuronal NO synthase (nNOS) is blocked in axotomised DRG, neurons undergo degenerative changes (Thippeswamy et al., 2001, 2007a). Further, we demonstrated that increased neuronal NO production, in response to axotomy/growth factor-deprivation in vitro, signals glial cells to produce trophic factors to support neuronal survival (Thippeswamy et al., 2005a). Recently, we found that treating satellite glia–neuron co-cultures with nNOS inhibitor, 7-nitroindazole (7NI), decreases the number of nestin+ cells that show neuron-like morphology. Cultured/axotomised DRG also upregulate inducible NOS (iNOS) in non-neuronal cells. Therefore, it is plausible that degenerative changes following nNOS inhibition are also due to iNOS-mediated excessive NO production by non-neuronal cells, which indeed is cytotoxic. NG-nitro-l-arginine methylester (L-NAME), the pan NOS inhibitor did not significantly change nNOS+ neuron number in axotomised DRG compared to 7NI suggesting that iNOS-mediated NO contributes to the degenerative process. In this paper, these findings from our and others' past work on NO-mediated neuron–glia signalling in axotomised DRG are discussed.

Starburst amacrine cells in the rabbit retina were labeled following an intraocular injection of the fluorescent dye, 4, 6, diamidino-2-phenylindole (DAPI). From each eye a strip of retina was removed, mounted on a platform beneath an epifluorescence microscope, and superfused with a physiological solution. The tip of a tungsten microelectrode (for extracellular recording) was visually positioned near the cell body of a DAPI-labeled starburst amacrine cell that was located in the ganglion cell layer. Light-evoked responses from the displaced starburst amacrine cells were studied in normal retinas and in retinas that had received a small electrolytic lesion near the optic disk 5–9 months beforehand. In normal retinas, a small spot of light centered over the receptive field of a displaced starburst amacrine cell in nearly all cases evoked a brief burst of spikes only at light onset. When stimulated with a large spot or an annulus of light, many cells gave a small burst of spikes at light offset. In lesioned retinas, the light responses of displaced starburst amacrine cells were recorded in areas of the retina where ganglion cells had degenerated. All cells responded with a large burst of spikes at the onset and offset of a small, centered spot of light. Large spots and annuli of light also evoked robust ON/OFF responses from these cells. The results from this study show that the receptive-field properties of displaced starburst amacrine cells change following axotomy-induced degeneration of ganglion cells. This finding indicates that changes in either synaptic transmission or the membrane properties of neurons occur in the retina following degeneration of ganglion cells.

The present study has examined the effects of early ganglion cell elimination upon the organization of the inner retina in the ferret. The population of retinal ganglion cells was removed by optic nerve transection on the second postnatal day, and retinas were subsequently studied in adulthood. Numbers of amacrine and bipolar cells were compared in the nerve-transected and nerve-intact retinas of operated ferrets, while stratification patterns within the inner plexiform layer were compared in these and in normal ferret retinas. Early ganglion cell elimination was found to produce a 25% reduction in the population of glycine transporter-immunoreactive amacrine cells, and 18 and 15% reductions in the populations of parvalbumin and calbindin-immunoreactive amacrine cells, respectively. GABAergic amacrine cells were also reduced by 34%. The number of calbindin-immunoreactive displaced amacrine cells, by contrast, had increased in the ganglion cell-depleted retina, being three times their normal number. Other amacrine and bipolar cell types were unaffected. Despite these changes, the stratification patterns associated with these cell types remained largely intact within the inner plexiform layer. The present results demonstrate a class-specific dependency of inner retinal neurons upon the ganglion cell population in early postnatal life, but the ganglion cells do not appear to provide any critical signals for stratification within the inner plexiform layer, at least not after birth. Since they themselves do not produce stratified dendritic arbors until well after birth, the signals for stratification of the bipolar and amacrine cell processes should arise from other sources.

We have previously reported that a small number of retinal ganglion cells (RGCs) of adult cats survive 2 months after transection of the optic nerve (ON) and that α cells have the greatest ability to survive among different types of RGCs (Watanabe et al., 1995). Here we report the time course of RGC survival within 15 days after ON transection using retrograde labeling with DiI injected into the bilateral lateral geniculate nuclei of cats. The density of DiI-labeled RGCs in the central retina as well as in the periphery did not change until day 3 after ON transection, then decreased rapidly, to 43% of the original density on day 7, and falling to 19% by day 14. We then intracellularly injected Lucifer yellow into the DiI-labeled RGCs to examine the difference in the time course between surviving α and β cells. Similar to the density change in total surviving RGCs, the proportion of surviving β cells did not change until day 3, then decreased rapidly to 65% of the original density on day 4, falling to 12% by day 14. By contrast, 64% of α cells survived for 14 days after axotomy. Analysis of regression lines for survival time courses indicated that death of β cells was characterized with a rapid period phase from day 3 to day 7 after axotomy whereas that of α cells lacked it. Axon-like sprouting from surviving β cells was first recognized in the nerve fiber layer on day 3, and were later more conspicuous.

The effect of systemic NGF treatment on loss and shrinkage of dorsal root ganglion cells was studied in adult male rats after permanent axotomy. Nineteen 16 to 18-wk-old rats had their right 5th lumbar spinal nerve ligated and cut approximately 7 mm peripheral to the ganglion. Two days before the operation, treatment with subcutaneous injections of human recombinant NGF (1.0–0.5 mg/kg/day) was started in 9 test rats; 10 controls were given saline injections. After 1 mo the levels of substance P (SP) and calcitonin gene related peptide (CGRP) were significantly increased in intact sciatic nerve. The number and mean volume of perikarya were estimated using assumption-free stereological techniques including vertical sections, the Cavalieri principle, optical disectors, the planar rotator and systematic sampling techniques. Systemic NGF administration had no influence on survival of primary sensory neurons after axotomy. The number of perikarya was 14300 (S.D.=1800) in axotomised ganglia in control rats versus 14700 (S.D.=2100) in axotomised ganglia of NGF treated rats. The reduction of perikarya volume after axotomy was significantly less after NGF treatment (11600 μm3 in the control group versus 8000 μm3 in the NGF treated group). However, the apparent protection of NGF-treatment on perikaryal volume is explained by a hitherto unrecognised size effect on nonaxotomised dorsal root ganglion cells. The untreated rats had a mean volume of 24700 μm3 (S.D.=2700 μm 3) whereas rats treated with NGF had a volume of 20400 μm3 (S.D.=1700 μm3) on the nonaxotomised side. In conclusion, systemic NGF treatment in adult rats has no effect on dorsal root ganglion cell loss in permanent axotomy whereas perikaryal size of intact nonaxotomised cells is reduced.

After being severed, optic axons in goldfish regenerate and eventually restore the retinotectal map; refinement of the map depends upon impulse activity generated by the ganglion cells. Because little is known about the changes in activity and receptive-field properties of ganglion cells during regeneration, we made extracellular recordings from them in the intact eye up to 95 days after sectioning their axons in the optic tract. Their receptive fields were classified as OFF-, ON–OFF-, or ON-centers, and their axonal conduction velocities measured by antidromic activation. The rate of encountering single units dropped drastically at 4–8 days postsection when only a few OFF-center units could be recorded, recovering to normal between 42 and 63 days. Receptive-field centers were normal in size, except for the few OFF-centers at 4–8 days which were abnormally large. Maintained discharge rates of all types were depressed up to 42 days, but ON–OFF-center units were more spontaneously active than normal around 42 days. Light-evoked responses in OFF-center units were subnormal at 4–8 days, becoming supernormal at 16 days and normal thereafter. ON–OFF- and ON-center units started to regain responsiveness at 16 days, and became supernormal at 42 days, before returning to normal. Conduction velocities of all fiber groups dropped to a minimum at 8 days, the fastest being affected most. There was a gradual recovery to normal conduction velocity by 63 days. The conduction latencies of OFF- and ON–OFF-center units recovered to normal by 42 days, and ON-center units by 63 days. Recovery of ganglion cell responsiveness correlates with functional recovery in the retinotectal system: OFF-center units recover light-evoked responses at about the time OFF activity first reappears in the tectum. ON- and ON–OFF-center units recover later, exhibiting supernormal spiking activity around the time that ON responses reappear in the tectum.

Injury to the central or peripheral nervous system is often associated with persistent pain. After ischemic injury to the spinal cord, rats develop severe mechanical allodynia-like symptoms, expressed as a pain-like response to innocuous stimuli. In its short-lasting phase the allodynia can be relieved with the γ-aminobutyric acid (GABA)-B receptor agonist baclofen, which also reverses the hyperexcitability of dorsal horn interneurons to mechanical stimuli. Furthermore, there is a reduction in GABA immunoreactivity in the dorsal horn of allodynic rats. Clinical neuropathic pain of peripheral and central origin often cannot be relieved by opiates at doses that do not cause side effects. The loss of sensitivity to opiates may be associated with the up-regulation of endogenous antiopioid substances, such as the neuropeptide cholecystokinin (CCK). CCK and its receptor (CCK-R) protein is normally not detectable in rat dorsal root ganglion cells. After peripheral nerve section, both CCK and CCK-R are up-regulated in the dorsal root ganglia. Furthermore, CI 988, an antagonist of the CCK-B receptor, chronically coadministered with morphine, reduces autotomy, a behavior that may be a sign of neuropathic pain following peripheral nerve section. Thus, opiate insensitivity may be due to the release of CCK from injured primary afferents. Similarly, in the chronic phase of the spinal ischemic model of central pain, the allodynia-like symptom is not relieved by systemic morphine, but is significantly reversed by the CCK-B antagonist. Consequently, up-regulation of CCK and CCK-R in the CNS may also underlie opiate drug insensitivity following CNS injury. Thus, dysfunction of central inhibition involving GABA and endogenous opioids may be a factor underlying the development of sensory abnormalities and/or pain following injury to neural tissue.

During development of the central nervous system (CNS) both deafferentation and axotomy induce increased neuronal death and result in a smaller brain with diminished function at maturity. Unilateral cerebellar pedunculotomy has been used as a model to study the relative importance of these 2 types of lesion on the survival of developing CNS neurons. Within the cerebellum, unilateral pedunculotomy causes deafferentation of the hemicerebellum and axotomy in the efferent pathway from the ipsilateral deep cerebellar nuclei. This results in a smaller hemicerebellum with normal cortical laminae but no extracerebellar outflow. In order to identify the sequence of events which leads to this altered structure and therefore to understand the relative importance of afferent versus target-derived trophic support, unilateral cerebellar pedunculotomy was performed on neonatal rat pups, aged between 1 and 3 days. The cerebella were analysed for histological and vascular changes after survival times of 0, 3, 6, 9, 12, 18, 21, 24 and 48 h. The results show that the effects of axotomy on the deep cerebellar nuclear neurons begin within 3 h of the lesion and apoptotic neuronal degeneration occurs within 48 h. However, the cerebellar cortical neurons continue to undergo normal histological development for at least 48 h after deafferentation. In addition, since ischaemia induces similar effects, a study of the vascular tree was made. The results indicate that the pedunculotomy does not alter the blood supply to the cerebellum, nor induce ischaemia of the cerebellar neurons. From this it may be hypothesised that target-derived trophic support is more crucial for the survival of immature neurons than is the trophic effect of afferent input.