Since ancient times it has been recognized that injury to the spinal cord can result in dramatic disability, including both negative symptoms (such as loss of voluntary movement and tactile sensibility) and positive symptoms (such as chronic pain and spasticity). Although many patients show some spontaneous recovery after injury, most patients with significant cord damage have permanent symptoms. Moreover, damage to the spinal cord, whether caused by injury or disease, cannot currently be repaired by any therapy. Yet there is now considerable optimism among neurobiologists that effective clinical therapies are within reach. This optimism is prompted on the one hand by a great increase in our understanding of the consequences of spinal cord injury (SCI) and the reasons for failure of spontaneous repair, and on the other hand by an increasing number of treatments that have promoted functional improvements in experimental models of SCI (with functional outcome being assessed in several ways) (Box 1). Yet, surprisingly perhaps, the mechanism(s) by which different experimental treatments promote functional improvement is in most cases undetermined. It is becoming clear that there are several distinct mechanisms that could underlie functional recovery, and that there is a considerable academic as well as practical value in understanding their range and scope.
Traumatic SCI leads to a series of reactive changes. First, the injury could directly damage the cell bodies and/or processes of neurons. The cells that are damaged might die and, as far as is known, they are not replaced (although the adult spinal cord does contain stem cells). However, the functional consequences of this neuronal loss are typically modest. Often, the damage to spinal axons is much more important1. The spinal cord is segmentally arranged and the sensory, motor and autonomic functions of each segment depend crucially on connections with supraspinal sites for all conscious or voluntary actions (Fig. 1). Damage to these connections leaves spinal segments caudal to the lesion site partially or totally isolated from the brain, which has debilitating consequences. The distal axon segment (the part isolated from the neuronal cell body) retracts from postsynaptic neurons and undergoes Wallerian degeneration2, and although the proximal segment typically survives, it does not spontaneously regenerate. The injured axon also encounters a series of inhibitory cues within the injured CNS neuropil, which further prevent a successful regenerative response (Box 2). An implicit assumption of much SCI research, at least until recently, has been that the major goal is to induce damaged axons to regrow, to reconnect to appropriate targets and thereby restore function (as is indeed possible in the PNS). Here, we examine the extent to which this has proved possible. In general, however, there is a lack of clear evidence that the regeneration of lesioned axons underlies significant functional improvements induced by experimental treatments. This has led to an examination of alternative putative repair mechanisms, evidence for which is discussed below.
Figure 1 | Structure of the spinal cord.
The spinal cord is responsible for transmitting signals between the brain and body. It is encased in vertebral bone and is segmentally arranged. The segments are grouped into 4 major divisions: cervical (C; 8 segments), thoracic (T; 12 segments), lumbar (L; 5 segments) and sacral (S; 5 segments). Each spinal segment makes connections with discrete body regions through projections running in the sensory and motor roots. The spinal segments are connected to the brain through a number of long axonal pathways that mostly run around the periphery of the cord. The main ascending pathways (which transmit sensory information to the brain) are indicated in red and the main descending pathways (which transmit motor information to the body) are indicated in blue. An injury to the spinal cord has devastating consequences owing to the disruption of these signals passing between brain and body, resulting in essentially no feeling or control of function below the level of the injury. Therefore, the higher the injury the more severe the debilitation. For example, injuries occurring at the lumbar level can result in paraplegia, as well as sexual and bladder dysfunction; cervical injuries are more debilitating and can result in quadriplegia; and high cervical injuries (such as the injury suffered by Christopher Reeve, level C2) can impair breathing function and lead to dependence on a ventilator. Anatomical image adapted, with permission, from Ref. 132 © (1996) Appleton and Lange.
Regeneration of damaged axons
A major focus of SCI research has been to overcome the failure of axons to regenerate (Fig. 2). The three main aims of this research are: to initiate and maintain axonal growth and elongation; to direct regenerating axons to reconnect with their target neurons; and to reconstitute original circuitry (or the equivalent), which will ultimately lead to functional restoration. Although many interventions have now been reported to promote functional recovery after experimental SCI, none has fulfilled all three of these aims. In fact, only the first aim has been extensively studied and convincingly demonstrated.
Figure 2 | The major functional deficits associated with spinal cord injury arise from the interruption of long ascending and descending spinal tracts.
When the normal axons of long ascending and descending spinal tracts (a) are damaged (b), the distal segment undergoes Wallerian degeneration and the proximal segment is unable to successfully regenerate. Therefore, ascending (sensory) information no longer reaches supraspinal sites, and descending motor and autonomic systems are disconnected from caudal spinal circuitry, including central pattern generators (CPGs), which are responsible for coordinated locomotor function. Much spinal cord injury research has focused on the possibility of inducing these damaged projection systems to regenerate and recapitulate original circuitry (that is, to restore the normal arrangement, as illustrated in c; regenerating axons are depicted in red).
Axon regeneration after SCI. There is now a wealth of data demonstrating axon regeneration after experimental intervention. For example, strategies aimed at overcoming inhibitory environmental cues have resulted in significant regeneration of lesioned axons through and beyond sites of SCI. These strategies include the neutralization of myelin inhibitors3, 4 and the degradation of inhibitory components of the glial scar5, 6. Axon regeneration after SCI has also been demonstrated following the administration of growth-promoting molecules, which make the lesion environment more permissive to growth. This includes the provision of exogenous neurotrophic factors7, 8, 9 or the manipulation of pro-regenerative neuronal signalling pathways10, 11, as well as several cellular transplantation strategies such as genetically modified fibroblasts8, 12, Schwann cell bridges13, olfactory ensheathing cells (OECs)14 and stem cells15. Furthermore, combination therapies aimed at targeting multiple factors have also been shown to act synergistically to enhance the extent of axon regeneration after SCI (Box 3).
Although this is an impressive (and growing) list, there are at least three important caveats to these studies. First, the contribution of regeneration to functional recovery is in most cases unknown. Second, the proportion of fibres that regenerate is always small. Third, the distance regrown is modest, certainly in the context of human spinal cord anatomy.
Regarding the first point, in most cases the best evidence is simply that the degree of regeneration observed is correlated with a concurrent recovery of behavioural function. However, in individual experiments significant regenerative responses might be correlated with other adaptive mechanisms. In many cases it is also possible that what have been perceived as regenerating axons are, in fact, sprouting spared axons. Some studies have attempted to provide stronger evidence that regeneration is at least in part responsible for the observed functional recovery. For example, the functional improvements that were observed following treatment with IN-1 (an antibody that neutralizes the inhibitory myelin-associated protein Nogo-A) after SCI were abolished following removal of the sensorimotor cortex, suggesting that the recovery depended on the regrowth of corticospinal tract (CST) axons3. Even here, the possibility remains that plasticity of brainstem and/or sensory systems contributed to the observed recovery (see below). Perhaps the best evidence comes from re-lesion experiments. Liu et al.12 transplanted genetically modified fibroblasts to a lesion site to promote rubrospinal tract regeneration, and showed that a re-lesion rostral to the original lesion/transplantation site abolished the recovery of forelimb use in a vertical exploration task. However, although this indicates that recovery depended on the presence of the transplanted tissue, it still does not determine the mechanism(s) by which the presence of a transplant can enable functional repair. Interestingly, re-lesioning can produce a more severe functional impairment than a single lesion alone16. This illustrates some of the potential problems of re-lesion experiments — that the re-lesion could be more extensive than intended ('to be safe') and might thereby involve other pathways; or the abolition of function could be confused with oedema, spinal shock or inflammation-mediated conduction block that could occur as a result of the second lesion.
The number of regenerating axons following experimental intervention is typically low, and at best amounts to only a small percentage of the original fibre tract. The majority of studies have focused on CST regeneration and have estimated success using anterograde or retrograde tracing techniques. Typically, 2–10% of CST axons are reported to regenerate. For example, as few as 5% of CST axons were induced to regenerate following Nogo-A neutralization, yet an almost complete recovery was observed in contact-placing responses and some aspects of locomotor function in treated animals3. A similar proportion has been observed in studies investigating rubrospinal tract regeneration, with typical estimates amounting to 
7% (ref. 12). It is worth comparing these results with the far superior regenerative capacity of the adult PNS (in which 
90% of damaged axons regenerate under optimal conditions) or, of course, the targeted growth that occurs during development.
An important question, therefore, is whether these modest numbers of regenerating axons can support the observed functional improvements. A priori, this might seem unlikely, but there is some evidence to the contrary. Raisman's group have used electrolytic lesions to interrupt CST axons in adult rats14, 17. They found that to achieve long-lasting deficits in a forepaw reaching task, ablation of the entire tract was required; when as little as 1–2% of the CST was spared, performance on this task recovered. Although in this case the lack of functional deficits was due to the sparing of a spinal projection (which makes appropriate connections) rather than the regeneration of lesioned axons (which might not), these observations nonetheless imply that regeneration of a small percentage of axons and restoration of their original circuitry could potentially lead to a significant recovery of function. Similarly, Wall et al.18 showed that more than 90% of dorsal column fibres needed to be lesioned for clear deficits in tactile sensory function to arise, and Windle et al.19 claimed that as little as 10% of spinal white matter tracts were sufficient to permit spontaneous walking without external support in adult cats. Furthermore, this phenomenon was classically demonstrated in human patients when large portions of spinal tracts were transected owing to intractable pain problems without causing disturbances in motility20.
The distance that central axons can regenerate is also limited. Most studies have focused on rodents, and, even with the most successful interventions, regenerated axons are most numerous in the few millimetres caudal to the lesion, and longer regenerating axons become progressively rare. In the rat spinal cord, each spinal segment is typically 2–3 mm in length. The proportion of lesioned axons induced to regenerate at least two spinal segments (that is, enough to be of clear functional use) is typically less than 10% (refs 3–5). Again, most of these data relate to the CST, although the more limited data available for other projections such as dorsal column sensory fibres7, 9, 10 or the rubrospinal tract12 are not dissimilar. Some studies have assessed regeneration in terms of the longest single regrown axon. In the rat, this is typically 10–15 mm. A crucial question — to which we do not have the answer — is whether these distances are absolute or relative. That is, if these treatments were applied to human spinal cord, would they induce growth of up to approximately one or three spinal segments?
Evidence for functional connectivity. So far, few studies have shown that regenerating axons actually make functional postsynaptic connections. Perhaps the most robust way of demonstrating such connectivity is to perform an electrophysiological analysis. For example, Ramer et al.21 showed that regenerating sensory axons made functional connections with dorsal horn neurons. Multiple rhizotomy of the cervical dorsal roots led to reduced sensory input to the spinal cord, and treatment with neurotrophic factors resulted in a re-emergence of postsynaptic activity in the dorsal horn in response to peripheral nerve stimulation. Re-injury of the dorsal root abolished dorsal horn activity. A concurrent recovery in sensitivity to mechanical and thermal noxious stimuli applied to the forepaw was observed, thereby providing convincing evidence that regenerated axons had made functional connections with target neurons, leading to a re-establishment of 'normal' circuitry and recovery of sensation in the denervated forepaw. Similarly, Steinmetz et al.22 demonstrated a recovery of the H-reflex response after rhizotomy and treatment with chondroitinase ABC (ChABC) combined with a preconditioning nerve injury. H-reflex recordings disappeared following acute resection of the dorsal root, indicating that functional connectivity of regenerated axons had occurred.
This type of analysis (electrophysiological assessment of functional connectivity then re-lesion) is, for practical reasons, more difficult to apply after the lesioning of CNS tracts, but there are already several examples of what might be done. For instance, Bradbury et al.5 electrically stimulated corticospinal neurons below the SCI level in lesioned animals treated with ChABC, and found a partial restoration of postsynaptic activity. These responses had longer latencies than in unlesioned animals, consistent with poorly myelinated regenerating axons. Furthermore, when the CST was acutely re-lesioned, this activity was abolished, providing evidence that functional reconnections had been formed by regenerating CST axons. Other approaches include the use of transcranial magnetic stimulation to activate corticospinal neurons, coupled with electromyographic recordings of, for example, hindlimb musculature23. Immunohistochemical techniques — such as assessing for the presence of the synaptic marker synaptophysin, or activity-dependent expression of the immediate-early gene c-fos in postsynaptic cells — might also provide strong supporting evidence of functional connectivity24, 25. With an increasing number of experimental treatments reporting pro-regenerative effects, it is vital that this question of functional connectivity is more comprehensively addressed.
Anatomical plasticity
The term 'neuronal plasticity' is used to describe a wide range of phenomena. One example relates to brain development, where multiple synaptic contacts are formed and experience determines which connections will be strengthened and which will be pruned. Another form of plasticity, which we consider here, is an adaptive reorganization of the neural pathways occurring after injury that acts to restore some of the lost function. We examine the hypothesis that functional recovery after SCI arises from the collateral sprouting of either damaged or spared pathways, which leads to the creation of novel neuronal circuits (Fig. 3). We first discuss evidence for spontaneous neuronal plasticity after SCI. We then consider the growing evidence that several treatments for experimental SCI act to enhance this process.
Figure 3 | Restoration of function after spinal cord injury might arise from anatomical plasticity of damaged or spared connections.
a | Collateral sprouts (illustrated in red) might form from long descending pathways that bypass the lesion site and activate spinal circuits more effectively or activate novel circuits, or from damaged or intact descending projections that activate local propriospinal neurons that bypass the injury site and form a novel descending system. b | Ascending sensory projections might also form collateral sprouts, which activate spinal circuits more effectively. CPG, central pattern generator.
Spontaneous neuronal plasticity after SCI. Although many SCIs lead to permanent disabilities, when the SCI is incomplete there is frequently a period of significant functional improvement. This is seen in many species, albeit with different time courses26, 27, 28. Although this recovery was originally described as recovery from 'spinal shock', it is perhaps more useful to describe it in terms that have mechanistic meaning. One contributing mechanism appears to be anatomical plasticity29. The spinal shock phase in humans is well defined, and lasts for several weeks or more. In animals, it is typically shorter, lasting for hours or days depending on the species30. It might be due in part to the loss of tonic descending excitation (for example, by the serotonin-mediated system), as shown by the fact that even spinal monosynaptic reflexes are affected. A significant part of spontaneous functional recovery in SCI patients occurs 2–6 months after injury, and is closely linked to intensive rehabilitative treatment. Here, the underlying mechanisms are likely to include neuronal plasticity and rearrangement.
Early studies demonstrated the capacity for functional compensation of descending pathways following injury. For example, compensatory plasticity of rubrospinal and bulbospinal tracts was found to mediate the recovery of forepaw function following CST lesion in cats31. Similarly, following a dorsal column lesion in rats, the spontaneous recovery of skilled reaching ability was found to be due to compensation by other sensorimotor pathways32. The importance of CST plasticity as a mechanism underlying such recovery has also been demonstrated. For example, Weidner et al.33 demonstrated that transection of the dorsally projecting CST led to spontaneous sprouting of the ventral CST projection, which was paralleled by a recovery of skilled forelimb use. This recovery was abolished by subsequent lesioning of the ventral CST. Remodelling of minor CST components (dorsolateral and ventral CST) after dorsal CST lesioning has also been recently demonstrated in a transgenic mouse in which the CST is exclusively labelled34. Furthermore, Bareyre et al.35 demonstrated that spontaneous recovery of function after a dorsal hemisection injury involves extensive reorganization of spinal circuits. They found that some of the transected CST axons (that would normally innervate lumbar segments) sprouted into the cervical cord to innervate propriospinal neurons. Some of these connections were transient (those with short propriospinal neurons that did not bridge the lesion), whereas others were maintained in the long term (those with long propriospinal neurons that circumvented the injury). This led to a novel, indirect motor pathway to lumbar motor circuits. Importantly, the functionality of these connections was confirmed electrophysiologically and corresponded with anatomical evidence of transynaptic connectivity as well as improvements in hindlimb function. Unilateral transection of the CST in the brainstem significantly reduced the improvements in hindlimb function, indicating that recovery depended on the new CST projections. The molecular events that direct this remarkably specific and novel change are not clear. Therefore, in several cases of experimental SCI, functional improvements occur over time in lesioned animals without any treatment intervention. Is it possible that some treatment interventions enhance this type of spontaneous recovery by increasing the plasticity of intact (and injured) systems, rather than by frank regeneration? Evidence that this is indeed the case is discussed below.
Plasticity induced by experimental treatments for SCI. There is now strong evidence that two types of treatment that lead to functional improvements in experimental SCI can also induce significant anatomical plasticity. The first treatment type targets the inhibitory components of CNS myelin. For example, collateral sprouting of the unlesioned CST into the denervated spinal cord has been demonstrated following unilateral pyramidotomy and IN-1 treatment. This compensatory sprouting was accompanied by the recovery of forelimb motor behaviours on the denervated side36, 37. Furthermore, after bilateral pyramidotomy, reorganization of the rubrospinal tract has been observed following IN-1 treatment, with increased innervation of ventral horn motor neurons that normally receive CST input; this too was associated with recovery of precision movements of the forepaw38. Functional connectivity of these novel rubrospinal projections was confirmed electrophysiologically following red nucleus microstimulation39.
Further evidence comes from observations of corticospinal and raphespinal fibre sprouting after dorsal over-hemisection injury and blockade of Nogo receptor (NgR) function. This sprouting was associated with a recovery in locomotor performance23, 40, 41. In an earlier study, delivery of the peptide fragment NEP1–40 (to block NgR signalling) led to functional recovery in hemisected rats and some regeneration of CST axons4. However, the functional recovery was observed as early as 2 days post-lesion, and it is unlikely that regenerating axons could have reformed functional postsynaptic connections in that time. It is therefore probable that the early recovery was mediated by plasticity mechanisms.
Corollary evidence that inhibitory components of CNS myelin restrict plasticity has come from stroke models. Partial recovery of function frequently occurs after cortical ischaemic injury, and this is largely attributed to the plasticity of axonal connections. Anti-Nogo therapy has been shown to improve functional recovery in experimental stroke models, and this too has been attributed to the reorganization and plasticity of unlesioned cortical areas42, 43. Therefore, manipulation of myelin inhibitors after injury can unmask the capacity of the adult CNS for reorganization. However, the extent to which this plasticity is responsible for functional improvement has not been established in most studies.
Chondroitin sulphate proteoglycans (CSPGs) also contribute to the inhibitory environment of the CNS, particularly after injury44. CSPGs are increasingly being seen as molecules that regulate plasticity in the adult CNS (for a review, see Ref. 45). Key evidence first came from studies in the brain; for example, high levels of CSPGs are present in perineuronal nets46, and disruption of these using ChABC can restore synaptic plasticity in the visual cortex of adult rats to a level normally seen only during development47. Degradation of CSPGs by ChABC treatment also leads to sprouting of intact Purkinje axons in the cerebellum48 and promotes sprouting of undamaged retinal afferents into the denervated superior colliculus following a partial retinal lesion injury49.
Recent evidence shows that CSPGs can also regulate plasticity in the spinal cord. In recent work from our laboratories, we have observed sprouting of both intact and injured descending pathways as well as intact sensory afferents following SCI and treatment with ChABC (A. W. Barritt et al., unpublished observations). In all cases, sprouting fibres were observed in aberrant locations close to the SCI (within one or two segments) in areas of degeneration and CSPG degradation. Although these data provide evidence that CSPGs are important for restricting the plasticity of adult spinal systems, the functional role of these sprouting fibres is less clear. However, a recent study has demonstrated functional collateral sprouting of forelimb sensory afferents into partially denervated brainstem nuclei following ChABC injections50. This study combined anatomical tracing techniques with receptive field mapping to demonstrate sprouting of intact afferents and an expansion of receptive fields after CSPG degradation. However, it did not determine whether such sprouting could affect forelimb function. By contrast, we have found that terminal field expansion within the spinal cord following ChABC treatment can result in a remarkable recovery of forelimb function following denervation51. In a spared-root injury model (in which all the dorsal roots from C5–T1 are rhizotomized, with the exception of C7), little activity was recorded in the C7 spinal cord following stimulation of the C7 dorsal root (determined by recording cord dorsum potentials, focal field potentials and single units in the dorsal horn). A single intraspinal injection of ChABC restored postsynaptic responses to levels that were indistinguishable from those observed in uninjured control animals and led to a complete recovery on several tests of sensory function of the denervated forelimb. These data provide robust evidence that degrading CSPGs within the spinal cord can lead to a functional reorganization of dorsal horn input, and that novel connections are organized in a functionally meaningful way.
The results of these studies suggest that CSPGs have a similar role to myelin inhibitors in suppressing sprouting responses of injured and intact systems after injury. Targeting both myelin and CSPG inhibitors might be a future strategy to consider for maximizing the potential for sprouting or plasticity after SCI.
It should be noted that in addition to having beneficial effects, increased sprouting can also lead to abnormal connections being made, with potentially detrimental consequences. For example, increased primary afferent sprouting in the spinal cord of experimental animals has been associated with pain and autonomic dysreflexia52, 53. In one study, the beneficial effects on motor function observed following neural stem cell transplants were limited by the development of allodynia, which was associated with aberrant sprouting54. The outcome was improved by differentiating the cells prior to transplantation, but this study highlights the importance of balancing potentially opposing influences. Pain, autonomic dysreflexia and spasticity are common symptoms in patients with SCI55, 56, 57, 58, 59 that could potentially be increased if abnormal connections were formed. Therefore, targeted treatments aimed at minimizing detrimental sprouting and optimizing beneficial sprouting will be a future goal for the treatment of patients with SCI.
Other mechanisms
Although there is some (albeit limited) evidence that regeneration and anatomical plasticity of connections contribute to functional recovery after SCI, there are several less well-explored, but entirely plausible, putative processes that might also contribute to repair. If these, or indeed any other mechanisms, can contribute to functional recovery from SCI, there is a self-evident value in optimizing their effects. Establishing that some or all of them do not contribute to functional improvement might also be important in removing experimental confounds. Some of these alternative mechanisms are discussed below and illustrated in Fig. 4.
Figure 4 | Various well-established phenomena might contribute to functional recovery from spinal cord injury after experimental interventions.
a | Rapid changes in synaptic efficacy can be triggered in the spinal cord by activity, neurotrophic factor treatment and manipulation of neurotransmitter systems, among others. Some of these changes are known to be associated with altered functional responses such as locomotion. b | Spinal cord injuries (SCIs) are known to be associated with demyelination of some intact axons. Several treatments aimed at promoting regeneration might in fact reduce the degree of demyelination or promote recovery from it, thereby accounting for some of the improved function. c | SCI is frequently associated with a slowly increasing area of secondary damage. Several experimental interventions could interfere with this secondary damage and so promote an apparent recovery. d | Another speculative mechanism of functional recovery relates to an altered motivational state. One example is putative analgesic actions. Many SCIs lead to chronic pain, triggered at least in part by local inflammatory processes. Many experimental treatments for SCI could affect these inflammatory or other pain-producing processes and in so doing increase functional capabilities, illustrated here as improvements on the Basso, Beattie and Bresnahan (BBB) Locomotor Rating Scale89. The mechanisms illustrated are not comprehensive, but they represent plausible options that have rarely been considered. CPG, central pattern generator.
Functional reorganization of spinal circuits. After complete spinal cord transection at thoracic levels, all voluntary control of hindlimb locomotion is lost. However, there is a considerable body of work that shows that pharmacological manipulations of such isolated spinal segments can enhance locomotor activity in various ways and species. For example, NMDA (N-methyl-D-aspartate) receptor agonists and manipulations that enhance serotonin or noradrenaline levels can act synergistically to induce treadmill walking (in which body weight is supported) in spinalized cats (for a review, see Ref. 60). These effects are present almost immediately after treatment and so cannot represent anatomical regeneration or sprouting. Rather, they unmask or enhance existing neuronal mechanisms. But in so doing they reveal a major shift in the functional capabilities of the spinal segments caudal to injury. Following partial, rather than complete, SCIs this enhancement of locomotor function might greatly increase voluntary control of locomotion, but the mechanism would have nothing to do with repairing the original damage. Treadmill training, which is used in many SCI clinics with great success61, 62, might promote this form of plasticity.
Furthermore, many of the experimental treatments that are reported to improve function after SCI (such as neurotrophic factor treatment) are likely to alter neurotransmitter levels in the spinal cord63, or affect synaptic strengths in other ways64. Another example is the powerful heterosynaptic facilitation of dorsal horn neuron responsiveness that is induced in seconds by simple repetitive activity in small diameter sensory neurons (for a review, see Ref. 65). This phenomenon potentiates spinal reflex excitability, thereby potentially confounding the analysis of sensorimotor or sensorivisceral function. The spinal facilitation depends in part on glutamate acting at NMDA receptors and brain-derived neurotrophic factor acting at TrkB receptors, and therefore could plausibly be modulated by several experimental 'treatments' for SCI.
Remyelination of fibre tracts. SCI results not just in Wallerian degeneration of damaged axons, but also in demyelination of some intact axons66, 67, and the functional impairments following experimental SCI have been correlated with the degree of demyelination68. These demyelinated axons have less secure transmission of action potentials and complete failure when even moderate lengths of axon are demyelinated. Caudal spinal segments are therefore functionally disconnected from supraspinal centres. The loss of oligodendrocytes that leads to demyelination might itself be modified by some treatments for SCI, such as neurotrophic factor delivery; moreover, established demyelination might be reversed by these (or other) treatments69. There is also evidence that enhancing nerve conduction through areas of demyelination by increasing the duration and safety of action potential transmission could improve function after SCI70, 71. Therefore, some treatments might produce a marked improvement in function without actually promoting any regeneration. Again, several of the treatments for SCI (most obviously the transplantation of stem cells, OECs or other glia17, 72, 73, but conceivably also some treatments targeting putative myelin inhibitors) could affect either demyelination or remyelination.
Amelioration of secondary damage. In humans and other species, many SCIs are compounded by secondary damage in the form of cavity formation or extension74, 75, 76 and a relatively slow-developing penumbra of cell death spreading from the injury site77, 78. A pronounced inflammatory reaction also contributes to secondary pathology after SCI77, 79. The causes of this secondary damage are not well understood, and it is therefore easy to speculate that multiple treatments for SCI might interfere with this phenomenon. Although several experimental treatments have been specifically targeted to minimize secondary damage, in particular by modifying inflammatory responses76, 80, 81, 82, it is also possible that other experimental treatments indirectly affect secondary pathology. Clearly, several factors (including neuroprotective factors, cellular transplantation and indirect effects on vascularization) are all potential modifiers of this process. It would be useful to develop well-validated methods to assess such secondary processes.
Altered motivational state. A more speculative explanation for functional recovery is that some treatments affect motivational or mood states. Such indirect actions on brain function are always possible if the treatment has systemic effects. However, one example of a local effect at or close to the lesion site that might affect motivation is a potential analgesic action. SCI leads to chronic pain problems in a large proportion of patients with SCI55. Experimental SCI in animals also produces altered pain sensitivity at and below the lesion site83. The molecular mechanisms underlying SCI-associated pain are poorly understood, but there is increasing interest in understanding them84, 85, 86. For example, pro-inflammatory cytokines released in the injured spinal cord from various immune and glial cell types are likely to be contributory factors87. Therefore, treatments that affect these processes might plausibly affect pain secondary to SCI. Indeed, treatment with the microglial inhibitor minocycline has recently been shown to attenuate the hyper-responsiveness of lumbar dorsal horn neurons and hypersensitivity to painful stimuli that occurs following SCI88. Interestingly, the effects on locomotor performance were less clear, with a partial recovery reported (using Basso, Beattie and Bresnahan (BBB) Locomotor Rating Scale scores). Further exploration is needed to determine whether analgesic treatment (putatively a side effect of some SCI treatment interventions) affects the locomotor or other functional outcomes of animals with SCI. We speculate that many putative treatment interventions might have unexpected nonspecific effects that could be responsible for the observed functional recovery.
Conclusions
The literature on SCI suggests that several treatment interventions can promote regeneration of damaged axons. The degree of such regeneration remains modest, but might be sufficient to account for functional recovery. Compensatory collateral sprouting has also been observed to occur spontaneously after SCI, and this process can be enhanced considerably by several interventions. Unfortunately, the extent to which these two mechanisms account for the functional effects of a known treatment is still not fully established. There are other eminently plausible mechanisms that could contribute to functional recovery. These remain mostly untested at present, and those discussed here do not form an exhaustive list. The diversity of mechanisms that might promote recovery from SCI could increase the options for developing novel therapies, but it also makes it more important to identify the mechanisms that are activated by any one treatment.
2 promotes functional regeneration of sensory axons in the spinal cord
