In terms of models for spinal cord repair a localized dorsal root injury in the PNS compartment of the dorsal root primarily addresses the issue of axon regeneration from a permissive (PNS) into a non-permissive (CNS) environment. In this case, injury-induced axon elongation is initiated and the behavior of axons is studied when they reach the interface between the PNS and CNS and attempt to enter the spinal cord. Mechanisms underlying failure of dorsal root axons to re-enter the spinal cord can be explored and approaches designed to attempt to overcome this regeneration failure. For injury reproducibility, sectioning the dorsal root is the best. However, this approach has shortages for studying regeneration of dorsal root axons—proximal and distal stumps of the dorsal root have to be re-apposed, and fixed to each other, usually by biological material such as fibrin glue. If the repair site is not prepared in an optimal way it may present an obstacle for axons to cross the injury site. Crushing the dorsal root eliminates this constraint, but is more prone to injury variability and raises the issue of how to ensure completeness, in particular as it comes to the non-myelinated axons. Since the connective tissues of dorsal root are more delicate than those of peripheral nerves forceful or repeated crushes may result in a complete separation of proximal and distal root stumps. To establish crush injury of the dorsal root as the model, requires control experiments with careful anatomical examination of the injury site and the portion of the nerve isolated from the nerve cell body. Physiological verification that impulse propagation completely fails across the injury site is an additional control, although a transient conduction block without interruption of axonal continuity may be a potential pitfall (cf. (28)).

Interrupting dorsal root axons at the site of their entry into the spinal cord is a useful injury model for probing interventional strategies which may promote functionally useful growth of dorsal root axons into the spinal cord. The anatomy at the junction between dorsal root and spinal cord limits the type of injury to sectioning the individual rootlets, taking care to minimize damage to the spinal cord itself. The procedure involves exposing target dorsal root(s) via a partial laminectomy, carefully lifting the roots close to their entry site, and thereafter cutting the root filaments. Recent examples using this model are studies in which multiple dorsal roots were sectioned, re-attached, and olfactory ensheathing cells implanted at the injury site to bridge growing dorsal root axons (29, 30). This injury model is in principle a CNS type injury, and more closely related to clinical root avulsion injury than the injury models described above. Still, this injury model does not mimic the abrupt and powerful longitudinal forces subjected to the dorsal roots and their entry sites in clinical root avulsion.

The models that most closely reflect clinical dorsal root avulsion injury involve pulling dorsal roots away from their attachment with the spinal cord. This type of injury results in damage to the DRTZ and spinal cord which is significantly more extensive than after just cutting or crushing the dorsal roots. This injury can be achieved either by forceful pulling of one or more spinal nerves, e.g., the brachial plexus, outside the vertebral column (31–35), or by exposing dorsal root(s) at their entry into the spinal cord and thereafter pulling them away from the spinal cord (36–38). Pulling spinal nerve(s) breaks both ventral and dorsal roots without visual observation of the actual injury, which may increase its variability. Avulsing only dorsal roots will obviously not reflect the entire pattern of pathophysiological and functional effects of clinical plexus avulsion injury, but allows better reproducibility. Both models have so far been explored mainly for spinal cord pathophysiology and neuropathic pain induced by deafferentation, rather than for the purpose of studying regeneration of injured dorsal root axons.

Axotomized CNS neurons appear to be intrinsically deficient in their capacity to mount and sustain an efficient program for long distance axonal growth. Furthermore, this capacity shows a significant decline with age. Dorsal root ganglion neurons respond in a similar way after dorsal root injury. Axotomy of the central processes of dorsal root ganglion neurons elicits a much weaker nerve cell body growth response than axotomy of their peripheral processes. As a consequence, the rate of regeneration of dorsal root axons is significantly less than that of regenerating peripheral sensory axons, ca. 1 and 3 mm, respectively in the rat (2), regenerating dorsal root axons have limited ability to grow for long distances in a peripheral nerve environment (52, 53), and are practically unable to enter the adult spinal cord. The dorsal root injury model has provided important information on interventions that enhance long distance axonal outgrowth, either by directly reactivating the intrinsic growth potential or by extracellular stimuli. The conditioning lesion paradigm has been paramount in this respect, demonstrating that injury of the peripheral processes prior to an injury of the central processes of the same sensory neurons mobilizes growth-associated molecules in dorsal root ganglion cells and enhances their capacity for elongation of injured dorsal root axons (54–59). This action allows limited number of regenerating axons to enter the spinal cord (60), but these presumably belong to a subpopulation of dorsal root axons, since a conditioning lesion appears to selectively stimulate the growth potential of dorsal root ganglion neurons not expressing calcitonin gene-related peptide or binding isolectin B4 (61). The conditioning lesion paradigm has also provided information on conditions that allow elongation of injured axons in the CNS. Thus, injury to peripheral sensory axons results in significant growth of injured primary sensory axons in the dorsal column (62), an effect that can be induced even a considerable time after the central lesion has occurred (63).

Extracellular delivery of growth factors enhances the potential of dorsal root regeneration into the spinal cord. The most striking effects have been shown with intrathecal delivery of neurotrophin (NT)3, or glial cell line-derived neurotrophic factor (GDNF) (64, 65), or systemic administration of Artemin (66, 67), a member of the GDNF family. These effects may be mediated by intracellular pathways which are at least in part similar as in the conditioning lesion situation (57), but concomitant effects on the properties of non-neuronal cells at the DRTZ may also occur (68). Notably, the time window for neurotrophin-mediated entry of regenerating dorsal root axons is restricted, since delayed treatment failed to induce spinal cord ingrowth (69).

Recently new mechanisms regulating the intrinsic growth state of neurons during postnatal development and after axon injury have been explored. Members of the Krüppel-like family of transcription factors regulate axonal growth development and their re-activation postnatally promotes axon regeneration in vivo (70, 71). PTEN (phosphatase and tensin homolog deleted on chromosome 10), inhibits phosphoinositide 3-kinase (PI3-K)/Akt signaling, a pathway critical for sustained neurite outgrowth. PTEN activation also appears to participate in mediating the growth inhibitory effect of myelin-associated glycoprotein (MAG) (72). Blocking the action of PTEN significantly promotes regenerative outgrowth in vivo from retinal ganglion cells (73, 74), corticospinal neurons (75), and even from peripheral nerve axons (76). Another intracellular signaling system which regulates the capacity for neural regeneration is the Jak-STAT (signal transducer and activator of transcription) pathway. SOCS3 (suppressor of cytokine signaling 3) inhibits this pathway, and relieving this inhibition markedly promotes regeneration of optic nerve axons (77), which is further enhanced by co-inactivation of PTEN (78). Combining PTEN inactivation with cAMP and the macrophage derived factor oncomodulin also enhanced growth of injured optic nerve axons (79), although oncomodulin by itself appears to have little effect on the regenerative capacity of injured dorsal root axons (80).

Another line of exploration has highlighted the fundamental role of microtubule stability for axonal regeneration in the CNS (81). Studies in non-mammalian species demonstrated efficient microtubule-based axon regeneration without the formation of growth cones (82). Treatment with the microtubule stabilizing agent taxol promoted axon regeneration in the mammalian spinal cord (83) and optic nerve (84). Furthermore, inhibition of kinesin-5, a motor protein that counteracts microtubule advancement, promotes sensory neurite extension on a growth inhibitory substrate in vitro (85). The dorsal root injury model is undoubtedly a useful model for exploring the intracellular pathways that are implicated in these effects, how these pathways interact (86) and in the identification of additional, hitherto unknown molecular mechanisms that could be targeted to optimize conditions for regeneration in the spinal cord.

Within days after dorsal root injury the DRTZ rapidly presents a virtually full-proof barrier for sensory axons to re-enter the spinal cord. A reasonable explanation for this situation would be the presence of a non-permissive environment for axon growth at the DRTZ, more or less analogous with the inability of adult degenerating CNS tissue in general to support axon growth. After dorsal root injury, DRTZ astrocytes proliferate (87), undergo a marked hypertrophy (88, 89) (Fig. 2), and growth inhibitory proteoglycans as well as myelin-associated growth inhibitors, including Nogo, myelin-associated glycoprotein (MAG) and oligodendrocytes myelin glycoprotein (OMgp) accumulate at this site (90, 91). Blockade of the Nogo-receptor (NgR) using soluble NgR peptide allows regeneration of myelinated axons into the spinal cord, restores connections with postsynaptic neurons and results in some degree of functional recovery (92). However, in other situations when the effect of growth inhibitors are blocked, this needs to be combined with a conditioning lesion of the dorsal root ganglion neurons (93, 94) or combined with potent immunostimulatory agents (95) in order to achieve significant regeneration into the spinal cord. One intriguing feature of the axon-glial interactions at the DRTZ after dorsal root injury is the formation of synapse-like contacts with DRTZ glial cell processes (96–98). The mechanism and significance of this process is still unclear, but its further studies may reveal novel information on the mechanisms preventing regenerating dorsal root axons from entering the spinal cord. Furthermore, exploring the interaction between regenerating dorsal root axons and glial processes my help to reveal novel processes that operate within the spinal cord to prevent long distance extension and/or plasticity of axons that attempt to grow or sprout after injury. In addition, the impact of pericytes and leptomeningeal cells in the emergence of a growth inhibitory environment at the DRTZ warrants exploration in view of their recently described roles in the formation of a scar after spinal cord injury (99, 100).

Section of one or more dorsal roots at the lumbosacral level is usually the best ones for studies on neuronal and non-neuronal processes associated with degeneration of the ascending axons in the dorsal column. An obvious advantage of this model is its absence of direct damage to the spinal cord, implying that its vascular supply and tissue architecture will be essentially untouched.

Axonal degeneration and secondary disintegration of myelin proceed from the injury area toward the spinal cord and along the axonal projections within the spinal cord although terminal parts of the axon appear to undergo degeneration even before their stem axon has disintegrated (101). Astrocytes and microglia undergo marked changes, and interact with infiltrating hematogenous cells in the removal of disintegrating axons and myelin, as well as in the formation of a tissue enriched in filamentous astroglial processes. Whereas axonal fragments are rapidly removed, myelin debris remains for long periods of time (102–104). In this context, important species variations exist. Remnants of myelin from degenerating heavily myelinated axons remain for decades in humans, and at least for years in, e.g., cats and rabbits. Astroglial processes may even be surrounded by myelin profiles in late stages of Wallerian degeneration in the cat spinal cord (105). One reason for this protracted presence of myelin debris may be an incomplete activation of the molecular mechanisms required for rapid myelin phagocytosis (106–108). The long-term presence of myelin-associated growth inhibitory molecules is likely to be a factor which counteracts growth of axons in the deafferented spinal cord (109). Thus, the dorsal root injury model can help to clarify the impact of the protracted Wallerian degeneration on axon growth in the spinal cord regeneration and possible ways to accelerate the elimination of myelin residues should this be beneficial. Recent studies, using dorsal root injury as experimental system, have also identified mechanisms involved in microglial-mediated clearance of axonal debris, a process which may also influence the conditions for regeneration in the spinal cord (110).

During Wallerian degeneration in the spinal cord white matter of monkeys abnormal profiles termed “myelinoclasts” appear (111). Similar structures were also observed in the cat (112, 113), but not in the rat (103). These profiles were characterized by layers of myelin wrapped around non-neuronal cells, presumably in most cases microglia, which at some later point in time appear themselves to undergo degeneration (113). Thus, the time course of myelin degradation and its associated microglial responses are distinctly different in small (e.g., rat, mouse) compared to large (e.g., cat, monkey, human) mammals, an observation which needs attention in translating findings from experimental models of spinal cord injury repair to humans.