These findings indicate that bridging axonal regeneration in the adult CNS can be induced when experimental treatments are administered at unprecedented delays of more than a year after SCI. Regeneration after established injury requires modification of both the intrinsic growth state of the neuron, achieved by CLs, and modification of the non-permissive injury environment, accomplished by cell grafting and placement of growth factor gradients beyond the lesion site. Manipulation of the intrinsic neuronal growth state alone, or the environment alone, are insufficient to support axonal bridging beyond the lesion. Delayed conditioning of the injured neuron elicits modulation of broads set of genes remarkably similar in profile to CLs placed before central injury, suggesting extensive recruitment of intrinsic molecular mechanisms that contribute to axonal regeneration. Thus, neuron-intrinsic mechanisms and the injured environment can both be modified at extended delays after injury to successfully elicit axonal bridging into, and beyond, sites of SCI.
Few studies have examined axonal regeneration when therapies are administered at delays after injury, and most of these have examined relatively brief delays of 1–8 weeks (Houle and Tessler, 2003
). Further, most delay treatment paradigms resect the “chronic scar” (Coumans et al., 2001
; Jin et al., 2000
), risking spinal cord re-injury (Tuszynski et al., 2003
). Of these studies, only one to our knowledge has reported axonal growth beyond the lesion, using a combination of a fetal graft and growth factors applied to a re-lesioned spinal cord 4 weeks after the original injury (Coumans et al., 2001
). The latter study reported regeneration through both gray matter and inhibitory white matter below the lesion resulting purely from cell-extrinsic experimental manipulations within the lesion site; however, examples of specific axons leaving the graft site were not provided. In contrast, axonal bridging in the present study is observed when experimental therapies are provided at exceptionally prolonged delays after the original injury, and only when both the intrinsic neuronal growth state and the surrounding, inhibitory environment are modified, consistent with findings in acute injury models (Lu et al., 2004
). Re-lesioning the spinal cord was not necessary to achieve axonal growth either into or beyond the chronic lesion site in this study, enhancing clinical relevance: re-lesioning the spinal cord risks further deterioration of function, particularly in cervically injured subjects that are critically dependent of neural systems spared immediately above the lesion site to support residual function.
We examined a long-projecting axonal system, the dorsal column sensory tract, as a model system because the projection is well defined anatomically, it normally fails to regenerate after central injury, and lesion completeness is readily verifiable by examination of the gracilar target of this projection (Lu et al., 2004
; Taylor et al., 2006
). Five observations suggest that axons observed beyond the lesion actually regenerated, and were not simply spared. First, the morphology of axons extending beyond the lesion site was circuitous and their course was nonlinear, unlike intact axons (, ). Second, CTB-labeled axons emerged across the lesion-host interface at different dorsoventral levels of the graft, not simply the most ventral or dorsal aspects of the graft/lesion site where spared axons might be located (–). Third, CTB-labeled axons were not observed ventral to the lesion site or in lateral, unlesioned portions of the spinal cord. Fourth, the lesion site was devoid of GFAP immunoreactivity, which would otherwise be present in strands of spared spinal cord parenchyma (). Fifth, sectioning of the medulla through the entire extent of the nucleus gracilis revealed an absence of CTB-labeled axons in all lesioned subjects, indicating that lesions were complete, whereas dense labeling was readily detected in unlesioned subjects (Suppl. Fig. 1
). The latter observation also indicates that axons did not regenerate over extended distances to re-enter the denervated nucleus.
Having established that bridging axonal regeneration is feasible in the sensory model at remarkably extended delays after injury, findings can be adapted to descending motor projections of the spinal cord. While pools of axons recruitable for regeneration persist at prolonged time points post-injury, the number of sensory axons regenerating beyond the lesion site is lower when therapy is applied 15-months post-injury compared to 6 weeks post-injury, and the distance that axons regenerate beyond the lesion is also reduced after prolonged treatment times (, ). For these reasons, attempts to achieve chronic functional recovery should likely target cervical level lesions, because potential neuronal targets of regenerating axons can be contacted immediately below the lesion site. If reinnervation of denervated motor neuron pools even one level below a lesion were achieved, the impact on quality of life in spinal cord-injured subjects could be substantial, potentially providing movement of a wrist or hand that could support greater independence. Incremental improvements in function are a more tractable goal for human translation, and may be achievable with rational, safe therapies that target neuron-intrinsic and environmental mechanisms.
Surgery and Tissue Processing
Female Fisher 344 rats (n=182) weighing 160–200g were experimental subjects. C3 dorsal columns were completely transected bilaterally using a tungsten wire knife (Lu et al., 2004
). Peripheral CLs were made by firm compression of the exposed nerve at mid-thigh level using jeweler’s forceps for 15sec._ MSCs were isolated as described previously (Azizi et al., 1998
). The chronic lesion site was re-exposed and 2µl (75,000 cells/µl) of passage 4 MSCs, mixed with fibrin glue, were grafted into the lesion cavity. Lenti-NT-3 vector and Lenti-GFP vector were generated as previously described (Taylor et al., 2006
) and 2.5µl were injected 2.5mm rostral to the lesion site in dorsal column white matter (titer 100 µg/ml p24, ~1×108
Infectious Units/ml). Dorsal-column sensory axons were labeled transganglionically by CTB injection into the sciatic nerve proximal to the CL site (2µl 1% solution) 3 days before perfusion. Animals were perfused with 4%PFA and subjected to CTB, GFP and GFAP immunocytochemistry on 30µm-thick spinal cord sections.
Axon number was quantified in 1-in-6 sections within and beyond the lesion site. Lesion margins were determined using GFAP immunoreactivity (Taylor et al., 2006
), and axons crossing a vertical line placed in sagittal sections within the graft midpoint, and 0, 50, 100, 250, 500, 1000, 2000 and 2500 µm rostral to the rostral lesion border, were counted. Total axon number/subject was calculated by multiplying the counted number by 6. The longest distance of regenerating axons from the rostral lesion border was also measured. Observers were blinded to group identity.
DRG Isolation, Culture and Labeling
Dissected adult L4-6 DRGs were digested for 1h in 0.25% collagenase type XI, triturated, and resuspended on 35mm cell culture dishes coated with myelin (18 µg/ml/ per well). Cells were fixed 48h later with 4%PFA and labeled for NF200. Longest neurite length/cell was measured from a minimum of 150 cells/animal (n=4 DRGs/group). 35μm sections of L4-6 adult DRG were immunolabeled for GAP43 and c-Jun and counterstained with Hoechst 33342. %GAP43- or c-Jun-labeled neurons was quantified as described, n=3 DRGs/group (Qiu et al., 2005
L4-6 DRGs were dissected, homogenized and subjected to cAMP ELISA (Assay designs), n=7 DRGs/group.
L4-6 DRGs were dissected, frozen at −80°C, and RNA extracted. RNA samples were reverse transcribed and labeled per manufacturer’s instructions, and hybridized to Affymetrix high-density oligonucleotide GeneChip Rat Genome 230 2.0 Arrays (Affymetrix).
All experiments and analyses were conducted under blinded conditions. Quantification of axon number beyond lesion sites and axonal length was assessed by Kruskall Wallis followed by Dunn post-hoc and Bonferroni adjustment. In all other quantifications, multiple group comparisons were made by ANOVA, with post-hoc Fisher’s. A significance criterion of p<0.05 was used. Data presented as mean ± SEM.