NT3-chitosan elicits robust endogenous neurogenesis to enable functional recovery after spinal cord injury

Significance

Spinal cord injury (SCI) is a debilitating medical condition with no cure at present time. In this study we have discovered that a biodegradable material, chitosan, when loaded with, Neurotrophin-3 (NT3), allowed for slow release of this neural trophic factor, providing an optimal microenvironment for regeneration. NT3-chitosan, when inserted into a 5 mm gap of completely transected and excised rat thoracic spinal cord, elicited robust activation of endogenous neural stem cells forming functional neural networks, which interconnected the severed ascending and descending axons, resulting in sensory and motor behavioral recovery. Our study suggests that enhancing endogenous neurogenesis by NT3-chitosan could be a novel strategy for treatment of SCI.

Abstract

Neural stem cells (NSCs) in the adult mammalian central nervous system (CNS) hold the key to neural regeneration through proper activation, differentiation, and maturation, to establish nascent neural networks, which can be integrated into damaged neural circuits to repair function. However, the CNS injury microenvironment is often inhibitory and inflammatory, limiting the ability of activated NSCs to differentiate into neurons and form nascent circuits. Here we report that neurotrophin-3 (NT3)-coupled chitosan biomaterial, when inserted into a 5-mm gap of completely transected and excised rat thoracic spinal cord, elicited robust activation of endogenous NSCs in the injured spinal cord. Through slow release of NT3, the biomaterial attracted NSCs to migrate into the lesion area, differentiate into neurons, and form functional neural networks, which interconnected severed ascending and descending axons, resulting in sensory and motor behavioral recovery. Our study suggests that enhancing endogenous neurogenesis could be a novel strategy for treatment of spinal cord injury.

Regeneration of axons in the central nervous system (CNS) is dampened by a nonpermissive microenvironment after injury (1–6). Following spinal cord injury (SCI), a cascade of pathological processes occurs, including the breaking down of the vasculature system, edema, infiltration of immune cells, inflammation, initiation of wound healing processes, seizing of ongoing neurotransmission, gliosis/glial scar formation, cell death, demyelination, and remyelination (7), along with activation of neural stem/progenitor cells (NPCs) attempting to participate in neural repair (8). Often, prolonged activation of immune cells, including microglia, lymphocytes, and macrophages, leads to secondary lesions of the nervous system, creating a very harsh environment for already compromised CNS neurons to regenerate axons (9).

In the past, much of the attention in SCI research focused on how to promote the regrowth of severed corticospinal tract (CST) for fairly long distances to reconnect with their originally innervated alpha motor neurons in the ventral horn and restore function. In recent years, several lines of evidence have suggested that perhaps a more feasible alternative proregeneration strategy is to generate nascent local spinal interneurons to provide a relay neuronal network across the injury site, serving as a “bridge” for the severed axons to establish connections with their targeted motor neurons without long-distance axonal growth (10, 11).

It has been widely accepted that neural stem cells (NSCs) reside in many regions of the adult CNS (8, 12, 13). CNS injury frequently causes neuronal loss. Although new neurogenesis may occur to some extent spontaneously (14), the inflammatory injury environment does not allow for sufficient neurogenesis. Instead, it is progliosis, which is involved in glial scar formation. Researchers have explored various methods to promote effective neurogenesis (15, 16); however, all of those studies concerned only brain injury repair. Few if any studies have addressed the promotion of regeneration after SCI via activation of endogenous new neurogenesis in the spinal cord.

We have previously shown that neurotrophin-3 (NT3) facilitates proliferation of NSCs in vitro through activation of the TrkC receptor and promotes neurogenesis in the spinal cord (17, 18). NT3 has also been shown to facilitate axonal growth, including in the CST, in vivo (19). Using a biodegradable chitosan carrier in combination with NT3, we achieved prolonged slow release of NT3 over a period of 14 wk (18). In the present study, we aimed to determine whether slow release of NT3 using chitosan (Fig. S1) to replace a 5-mm thoracic segment of the spinal cord after complete transection and removal of the segment could create an optimized microenvironment to promote new endogenous neurogenesis. We further examined whether the subsequent formation of a nascent synaptic network within the NT3-chitosan tube could functionally bridge the two ends of the severed cord and achieve regeneration in severe cases of SCI.

Fig. S1.

A schematic of the NT3-chitosan tube, composed of an chitosan outer tube stuffed with NT3-coupled chitosan carriers. Slow release of soluble NT3 molecules by the carrier within the tube could be achieved.

Results

NT3-Chitosan Enables Nerve Regeneration and Functional Recovery.

Three major experimental groups were included in this study: the lesion control (LC) group, in which a 5-mm thoracic spinal cord segment at T7-8 was surgically removed after complete transection of the cord; the empty tube (ET) group, in which an empty (i.e., non–NT3-loaded) chitosan tube was inserted into the 5-mm gap following surgical removal of the thoracic segment; and the NT3-chitosan tube (NT3) group. The loading of NT3 onto the chitosan tube is described in detail in Fig. S1.

In all three groups, gross anatomic changes within areas of lesions were examined at various time points from 3 d to 1 y after injury. At the light dissecting microscopic level, at 3 d after the operation, a small amount of tissue was observed at the rostral and caudal ends of the lesion area in all groups (Fig. S2A). At 10 d after the operation, the lesion area was occupied with fibrous and cystic tissues in the LC group. This process continued and eventually formed scar-like tissues, anatomically “gluing” the two ends of severed cord together (Fig. S2A, Middle and Fig. 1A, Right). For the ET group, at day 10, tissue infiltration was no longer present, and the tube remained empty from day 10 and onward to 1 y after injury (Fig. S2A, Bottom, and Fig. 1A, Middle). In the NT3 group, tissues continued to grow into the tube from two ends, gradually narrowing the tissue gap, and by day 30, tissues from both the rostral and caudal ends of the cord fused (Fig. S2A, Top). The diameters of newly formed tissues were initially irregular and not uniform. New tissues continued to grow and remodel, and by day 90 formed a smooth nerve bundle-like structure that was thick at the two ends and thin in the middle (Fig. 1A, Left). This anatomic structure was stabilized from day 90 and onward, up to at least 1 y after injury (Fig. 1A and Fig. S2B).

Fig. S2.

(A) Gross anatomical changes in lesion areas of various treatment groups and at various time points. For the NT3-chitosan tube-treated group (Top; both low-magnitude and high-magnitude images), red arrows indicated apparent regenerated tissues in the lesion area. At 30 d after the operation, regenerated tissues from both ends met in the middle, bridging the gap. In the LC group (Middle; both low-magnitude and high-magnitude images), the red arrows pointed to new scar tissue in the lesion area. In the ET group (Bottom; both low- and high-magnitude images), the red arrows pointed to initially regenerated tissues. At 10 d after the operation, the regenerated tissues had been retrieved. (B) Images of 10 cases of nerve regeneration in the NT3-chitosan group at 1 y after the operation, demonstrated the highly reproducible nature of the treatment.

Fig. 1.

The NT3-chitosan tube facilitated regeneration and functional recovery after complete SCI. (A) Dorsal view of representative spinal cords from three treatment groups—the NT3 group, ET group, and LC group—at 12 mo after the operation. Note the presence of nerve tissue-like regenerated cable (black arrows) in the NT3 group. No tissue growth was seen in the ET group. Arrowheads mark the location of undegraded chitosan tubes. The white arrow indicates apparent scar tissues. (B) NF staining of regenerated nerve fibers in the NT3 group at 12 mo after the operation. Note presence of the NF-positive regenerating fibers growing inside the NT3-chitosan tube, traversing the lesion site, and reaching the caudal end of the spinal cord. Almost no NF staining was found across the lesion site in the LC group, indicative of no regeneration. (C) BBB open-field walking scores of bilateral hindlimbs for all eight groups over time, up to 52 wk. At 4 wk after the operation and thereafter, the NT3 groups with or without DMSO had significantly higher BBB scores in both the right and left hindlimbs compared with the other groups (mean ± SE; n = 8; P < 0.05, one-way ANOVA with Bonferroni post hoc test). Retransection was performed at 52 wk after the initial lesion. In the middle of the lesion area (n = 4), a plastic diaphragm was inserted to prevent regeneration, with an additional 5 wk of BBB scoring (mean ± SE; n = 4; P < 0.05, paired Student t test comparing before and after recutting). Retransection dropped BBB scores back to complete SCI levels.

Using neural filament (NF), Nissl, and neuron-specific enolase (NSE) staining, we confirmed that regenerated tissues in the NT3 group were neural tissues composed of nerve fibers as well as neuronal cells (Fig. 1B and Fig. S3 A and B). NF-positive nerve fibers observed in the NT3-chitosan tube could be composed of regenerated ascending and descending fibers, as well as new fibers originating from newly regenerated neuronal cells. Quantitative analyses of the exact contribution from each of these three sources await future studies.

Fig. S3.

(A) Nissl staining of regenerated nerves in the NT3 group at 1 y after the operation. Serial magnifications of images at positions a, b, and c were shown as a1, a2, a3, b1, b2, b3, c1, c2, and c3. Both nerves and neuronal cells were present in regenerated tissues. (B) Staining of NSE of regenerated tissues in the NT3 group. Again, serial magnifications of images at positions a, b, and c were shown as a1, a2, a3, b1, b2, b3, c1, c2, and c3. Both nerves and cells were present in the regenerated tissue. The white dotted lines indicated the boundary between the host spinal cord and the lesion area.

To determine whether regenerated nerve tissues could elicit behavioral functional recovery, we used the Basso–Beattie–Bresnahan (BBB) open-field walking scale to measure hindlimb locomotor activity of rats after SCI. As expected, hindlimb locomotion was zero immediately after the operation for all of the experimental rats. Over the course of 1 y, the right and left hindlimb BBB scores were no higher than 1.5 in the LC and ET groups (Fig. 1C). In contrast, in the NT3 group, the mean BBB scores for right and left hindlimbs increased significantly at week 4, and continued to rise steadily up to 1 y, at which time scores were 10-fold higher than those at week 1 (11.4–12.2; Fig. 1C). To evaluate whether this result was related to a compensatory effect rather that spinal cord regeneration, at 52 wk after the first operation, regenerated nerve bundles were retransected and a plastic diaphragm was placed in the relesion site. As a result, the mean hindlimb BBB scores dropped from 11.4–12.2 to 0–1 (not significantly different from those in the ET and LC groups), and remained at this low level for the next 5 wk (Fig. 1C). The retransection results indicate that the restoration of locomotor function observed in the NT3 group was most likely related to reestablishment of synaptic transmission from motor axons crossing the lesion area to elicit motor behavior, but not due to a compensatory response below the lesion area.

We studied five additional control groups to confirm that NT3-chitosan enabled the nerve-like tissue growth and functional recovery through activation of the NT3/TrkC pathway. We infused TrkC tyrosine kinase (NT3 receptor) inhibitors, k-252a, k-252b, and an NT3-neutralizing antibody into NT3-chitosan tubes for 12 wk using osmotic pumps and found that, just like in the ET group, tissue growth was no longer present inside the NT3-chitosan tube with a blockade of NT3/Trk signaling and BBB scores were all <1.5 (Fig. 1C). Because the k-252a and k-252b were dissolved in dimethyl sulfoxide (DMSO), we also infused DMSO into both the NT3-chitosan and empty chitosan tubes to serve as controls, and found that DMSO did not influence the experimental outcome (Fig. 1C).

On EM analysis, the ultrastructure of the regenerated tissue in the NT3 group showed numerous myelinated axons with typical layered myelin sheets confined to endoneurium- or perineurium-like structures (Fig. S4A). Over time, the number of myelinated fibers at the caudal end of the NT3-chitosan tube increased and more or less stabilized from around 26 wk up to 1 y (Fig. S4B). Based on myelin morphologies and the number of myelinated axons associated with each myelinating cell, we determined that both Schwann cells (SCs) and oligodendrocytes (OLs) participated in the remyelination process (20) (Fig. S4A). In contrast, no regenerated myelinated axons were found in the middle or caudal ends of the lesion areas in the LC or ET group (Fig. S4). In the LC group, the bulk of the tissue in the rostral ends was composed of nonaxonal scar tissues and some degenerated axons (Fig. S4A). Moreover, blood vessels wrapped by endothelial cells were frequently found in the regenerating tissue under EM, indicative of good vascularization (Fig. S4A).

Fig. S4.

(A) Structural and ultrastructural images of regenerated tissues in cross-sections of the lesion area for the NT3-chitosan group (left four panels with serial magnifications of the white boxed region in the top left panel). The section was at the caudal end (“C” in the scheme) of the regenerating tissue at 6 mo after the lesion. Images of LC groups were presented in the two rightmost panels at different magnifications. The section was at the rostral end (“R” in the scheme) of the lesion area at 6 mo after the lesion. Red arrows pointed to endothelial cells wrapping around blood vessels. SC, labeled Schwann cell nucleus; OL, labeled olidogendrocyte nucleus. (B) Quantitative analyses of the numbers of myelinated axons in the lesion area (R, M, and C) at various time points for all treatment groups. No regenerated myelinated axons were observed in the middle (M) or the caudal end of the lesion area in the LC and ET groups. Data were presented as number of myelinated axons/mm² (mean ± SD; n = 4).

To further confirm sensory and motor functional restorations, we performed electrophysiological analysis. Before SCI, the somatosensory evoked potentials (SEP) and motor evoked potentials (MEP) exhibited stable wave forms in all rats examined (Fig. 2 A and B). In all groups, at 1, 7, 14, and 21 d after the first operation, even with gradually increasing electric stimulation, no SEP could be evoked from the cerebral sensory cortex, and no MEP could be evoked from the cerebral motor cortex. In the NT3 group (with or without DMSO), however, SEP and MEP started to recover by 30 d after the operation. At 1 y after the operation, the latency and amplitude of SEP and MEP were partially restored, but not to the levels seen in nonlesion controls (Fig. 2 A and B). Again, after retransection, or with inhibition of the NT3-trkC pathway, SEP and MEP could no longer be induced in the NT3 group with or without DMSO (Fig. 2 A and B). Taken together, these data further confirm the achievement of partial recovery of nerve function in the NT3 group.

Fig. 2.

Electrophysiological analyses of regeneration elicited by NT3-chitosan. (A) Representative SEP and MEP traces for each group of experimental rats at 12 mo after the operation. Note that only the NT3 group showed partial recovery, and recutting eliminated the responses. (B) Quantitative analyses of the latency and amplitude of SEP and MEP of bilateral hindlimbs of all treatment groups, showing that the NT3 group had significantly longer SEP and MEP latency and lower SEP and MEP amplitude compared with uninjured rats (mean ± SD; n = 6; P < 0.05), indicative of partial functional recovery. All groups except uninjured rats and the NT3 with and without DMSO groups had no detectable signals.

New Neurons Are Found in NT3-Chitosan Tubes.

Analysis of tissues grown in NT3-chitosan tubes revealed, in addition to nerve fibers, numerous Nestin-positive cells, suggesting that endogenous NSCs had migrated into the tube (Fig. 3). Interestingly, 3 d later, Nestin-positive cells started to appear in both the rostral and caudal ends of the tube, and by day 30, these cells reached the middle part of the tube (Fig. 3 B and C). The number of Nestin-positive cells peaked at day 20 in both the rostral and caudal parts of the tube but not until day 30 in the middle part of the tube, suggesting bidirectional migration routes of NSCs, from both ends of the injured spinal cord (Fig. 3 B and C). In addition to NSCs, we identified NeuN-positive and/or β-tubulin 3-positive neuronal populations, which were quite abundant in the NT3-chitosan tube but not in the corresponding injury segment in the LC group (Fig. 3 A–C).

Fig. 3.

NSCs and neurons were found in the regenerating nerve tissues. (A) Immunographs of Nestin- and Tuj1-positive cells in the regenerating nerve tissues of the NT3 groups at 1 mo after the lesion. *Labeled undegraded chitosan biomaterials. Serial magnifications of framed regions were shown. Occasional Nestin- and Tuj1-double-positive cells were observed, indicative of newly differentiating neurons. Note that the LC group had no Nestin- or Tuj1-positive cells in the lesion area. The white dotted lines indicate the boundary between the host spinal cord and the lesion area. (B) Schematic diagram of the 5-mm manipulated/lesion area examined, which was divided into three segments, R, M, and C, depending on the rostral-caudal position. (C) Quantitative analyses of infiltrating cells into the lesion areas with all eight treatment groups at various time points after the operation and onward up to 90 d, except for NeuN labeling, which had one additional time point at day 120. Longitudinal sections of the lesion area, 8 μm thick, were examined, and average cell numbers per 0.5 mm² or 0.2 mm² area were measured. Data were presented as mean ± SD; n = 4. *P < 0.05 comparing the NT3 and NT3-DMSO groups to the other groups. Note that only the NT3 groups had a substantial amount of neural cells in the lesion area, with more cells in the R and C regions than in the M section at early time points, indicating bidirectional infiltration from both ends. Few to no cells were detected in the remaining control groups. More detailed information is provided in Table S1.

Interestingly, Nestin and neuron-specific β-tubulin (Tuj1) double-positive cells were occasionally observed in the regenerating sites, suggesting differentiation of progenitor cells into new neurons (Fig. 3A). The decline of Nestin-positive cells after day 20 was likely related to differentiation of NSCs. In contrast, the decrease in β-tubulin 3-positive cells was likely related to programmed cell death of excessively generated immature neurons that failed to integrate into neural circuits, a cardinal feature of nervous system development (21). NeuN labeled more mature neurons, which did not decline even at 120 d, long after NT3 had been exhausted (i.e., at 14 wk after the operation) (Fig. 3C).

To confirm that neurons detected within the NT3-chitosan tubes were newly born neurons from endogenous NSCs rather than migrated preexisting neurons, we administered daily i.p. injections of 5-bromo-2 -deoxyuridine (BrdU) to rats from day 1 to day 7 after the operation. This labeled proliferative cells and newly born neurons in vivo during this period. At 4 wk postsurgery, approximately 57% of the Nestin-positive cells were labeled with BrdU, and 64% of the Tuj1-positive and 67% of the NeuN-positive neurons were BrdU-positive, indicating that these neurons were generated after the operation (Fig. 4 A and B, Fig. S5, and Fig. S6 A–C).

Fig. 4.

NT3-chitosan promoted new endogenous neurogenesis in the lesion area after SCI. BrdU was injected i.p. at 24 h postsurgery and then daily for the first 7 d after the operation. Stars indicated undegraded chitosan biomaterials. The white dotted lines indicated the boundary between the host spinal cord and the lesion area. (A) Immunographs showing BrdU-labeled Nestin- and Tuj1-expressing neural lineage cells in the lesion area at 1 mo after lesion. Serially amplified images of highlighted regions, as well as single optic sections of confocal images with Z-stack demonstrated colabeling. (B) Quantitative analyses of immunohistochemistry shown in A. Data were presented as mean ± SD; n = 4. *P < 0.05, comparing the NT3 group to the LC group.

Fig. S5.

A series of a total of 36 optic confocal images were shown, combination of which composed the image shown within the magenta frame in Fig. 4A. However only the 16th image was displayed in Fig. 4A bottom right panel. Obviously, the merged image should be slightly different from the image of only one optic section.

Fig. S6.

(A) Immunographs showing BrdU (green)-labeled NeuN (red) expressing neural lineage cells in the lesion area at 4 wk after the lesion. (B) Quantitative analyses of immunohistochemistry. Dapi (blue) was used to mark nuclei. Serial magnifications of the color-boxed area were shown. A confocal Z-stack of white-boxed cells was presented in the bottom-right corner, indicative of a double-labeled cell. The percentage of NeuN cells that were also BrdU-positive is shown (mean ± SD, n = 4). *P < 0.05. R, M, and C represented areas for quantification shown in D. (C) Immunostaining revealing that at 4 wk after the SCI operation, there were large numbers of BrdU (green) and neuronal-specific β-tubulin–positive (red) cells and fibers in the regenerated tissue of the NT3-chitosan tube group, indicating the appearance of active endogenous neurogenesis. Nuclei were stained with DAPI (blue). As in A, serial magnifications of color-boxed areas were shown with confocal Z-stack images, demonstrating the presence of the two labels in the same cell. *Not-yet degraded NT3-chitosan debris. The white dotted lines indicated the boundary between the host spinal cord and the lesion area. (D) Schematic diagram of the 5-mm manipulated/lesion area examined, which is divided into three segments, R, M, and C, depending on the rostral-caudal position. (E) From day 3 to day 30, large numbers of BrdU cells appeared in the NT3-chitosan tube, some of which were dividing Nestin-positive NPCs or their downstream lineage cells. Note that BrdU injections were done for only 7 d postlesion. In the LC group, BrdU-positive cells at early time points were likely nonneural but infiltrating immune cells and endothelial cells. The decline in BrdU-positive cells at 60–90 d might result from cell division and thus dilution of the BrdU label, and/or some degree of elimination of BrdU-labeled but no longer dividing postmitotic cells, such as neurons and oligodendrocytes, which underwent apoptosis. All data were presented as mean ± SD. *P < 0.05, compared with the ET or LC group. (F) Immunographs showing BDA-labeled CST forming morphological synaptic connections with MAP₂/BrdU double labeled newly born neurons in the NT3-chitosan tube at 4 wk after the lesion. *Undegraded NT3-chitosan materials. The white dotted lines indicated the boundary between the host spinal cord and the lesion area. Serial magnifications of color-boxed areas were shown. Z-stacks of confocal images were shown in the bottom-right corner, demonstrating the copresence of quadruple labels: DBA (green), BrdU (red), MAP2 (white), and DAPI/nuclei (blue). These images showed regenerating CSTs making synaptic contact with newly born neurons in the regenerating tissues.

The BrdU-labeling method detected many cell types in addition to neural progenitors, particularly after SCI. In fact, in the LC group, but not in the ET or the NT3 group, some BrdU-positive cells appeared earlier in the middle part of the lesion area (Fig. S6E). Since these BrdU cells in the LC group were not labeled with Nestin or Tuj1, they were most likely infiltrating inflammatory immune cells (22). Moreover, the fact that those cells did not infiltrate into the middle part of the chitosan tubes at such early time points in the ET and NT3 groups suggests that the chitosan tubes with or without NT3 likely created an anti-inflammatory environment early after injury and, when coupled with NT3, supported new neurogenesis and axonal regeneration (22).

Newly Born Neurons Form a Nascent Functional Neural Network.

To explore the underlying mechanisms by which newly generated neurons may participate in locomotor functional recovery, we hypothesized that newly born neurons may form nascent synaptic networks that serve as information relay stations, bridging ascending and descending neural transmission signals across the lesion. Using BDA (dextran, fluorescein, and biotin) to label cortical motor axons and BrdU/MAP₂ to label newly born neurons in the NT3-chitosan tube, we found evidence that CST axons make contact (likely synaptic) with newly born neurons (Fig. S6F).

We used immunoelectron microscopy technology to investigate whether morphological synaptic connections were also established among newly generated neurons. At 4 wk after the operation, immunogold-labeled Tuj1-positive axons and dendrites formed mature symmetrical or nonsymmetrical synapses in the middle of the NT3-chitosan tube (Fig. 5A).

Fig. 5.

Formation of functional synaptic connections among newly born neurons in the NT3-chitosan tube and the host spinal cord caudal to the lesion. (A) Immunogold-labeled β-tubulin 3–positive axons and dendrites forming symmetrical (Upper) or asymmetrical (Lower) mature synapses (arrows). At, axonal terminal; Den, dendritic side of the synapse as shown by postsynaptic density. *Undegraded chitosan materials. (B) Photograph of typical MED64 with 8 × 8 arrays (interelectrode distance, 150 μm) of probe recordings on the spinal cord slice, including the lesion/regenerated area from the NT3-chitosan group at 4 wk after operation. The narrow “cable” represented regenerated tissues at the top. While stimulating one site (yellow star) in the caudal part of the regenerated spinal cord, the field excitatory postsynaptic potential (fEPSP) was recorded at multisite in the lesion area and in the caudal of host spinal cord, which could be suppressed by CNQX (10 μM), suggesting that the nature of the synaptic transmission was glutamatergic. The red arrowhead pointed to an fEPSP recorded at electrode 58. The red dotted line indicated the boundary between the residual host caudal spinal cord and the lesion/regenerated area.

To determine whether the morphological synapses detected were electrophysiologically active, we used a planar multielectrode dish system (MED64; Alpha MED Scientific) to reveal features of the neuronal network formed in the NT3-chitosan tubes. In addition, we used pharmacologic compounds to determine whether the field potentials recorded by the MED64 system in the lesion area were transmitted by synapses formed among newly born neurons. Stimulation of the caudal part of the regenerated spinal cord tissue in the lesion area produced multisite neural responses within the lesion area, as well as in the host spinal cord caudal to the lesion (Fig. 5B). The amplitude of field excitatory postsynaptic potentials (fEPSP) recorded at multiple sites in the lesion area as well as in the host spinal cord caudal to the lesion could be significantly suppressed by 6-cyano-7-nitroquinoxaline-2, 3-dione (CNQX; 10 μM), suggesting that major excitatory neurotransmissions were glutamatergic. All of the effects of the pharmacologic agents on fEPSP were reversible on reperfusion (i.e., washout of CNQX), as expected. Taken together, these data indicate that functional synaptic connections can be established among newly generated neurons in the lesion area as well as with the host spinal cord, and that these new synaptic connections likely participate in the locomotor functional recovery.

NPC Activity and Functional Recovery.

To address whether mitotic activation of endogenous NSCs and subsequent neurogenesis were important for NT3-chitosan–induced functional recovery after SCI, we used an antimitotic drug, cytosine β-d-arabinofuranoside (Ara-C) (23) to ablate proliferative progenitors. In the NT-3 group, Ara-C was infused into the rat lateral ventricle via a micropump for 14 d. As shown in Fig. S7A, Ara-C treatment significantly decreased the number of Nestin-positive neural progenitors and NeuN-positive neurons within the NT3-chitosan tubes at 4 wk after the operation. BBB scores in the bilateral hindlimbs at 1 y decreased by 30–40% after Ara-C treatment compared with no Ara-C treatment (Fig. 6A).

Fig. S7.

(A) An antimitotic agent, Ara-C, was injected daily in the animals in the NT3 group for 14 d after the operation, to substantially eliminate dividing cells, including activated NPCs. As expected, both Nestin-positive NPCs and NeuN-positive neurons were significantly reduced in the entire NT3-chitosan tube (R, M, and C regions included) with Ara-C treatment. *P < 0.05 (n ≥ 4), comparing Ara-C treated and untreated samples. (B) The number of BDA-positive CST axons in the lesion area did not change after Ara-C treatment, suggesting that new neurogenesis contributed to the full range of functional recovery elicited by the NT3-chitosan tube.

Fig. 6.

Blockade of stem cell activity and new neurogenesis led to worsened functional recovery. (A) BBB scores of animals belonging to the NT3 group, untreated or treated with Ara-C. *P < 0.05 (n ≥ 5), comparing Ara-C–treated and untreated groups. (B) SEP/MEP latencies and amplitudes in uninjured, NT3-chitosan, Ara-C–treated, and untreated groups. *P < 0.05 (n ≥ 5), comparing uninjured and NT3-chitosan groups and Ara-C–treated and untreated groups based on labeling.

Interestingly, there was no significant decrease in the latency of SEP and MEP, but an approximate 30% decrease in amplitude was apparent (Fig. 6B). A partial CST tracking experiment with BDA labeling (with eight 0.4-uL injections of 10% BDA) indicated that Ara-C did not change the number of regenerating axons, which likely contributed to the maintenance of SEP and MEP latencies (Figs. S7B and S8). Changes in the amplitude, on the other hand, suggested that in the NT3 group, large amount of regenerated neurons in the lesion area likely participated in locomotor functional recovery.

Fig. S8.

(A) BDA-FITC (green) anterograde tracing of the CST at 1 mo after the operation. In the NT3 groups, BDA-positive fibers entered the lesion area from the rostral end, traversed the 5-mm gap, and reentered the host caudal spinal cord. The white dotted lines indicated the boundary between the host spinal cord and the lesion area. Small white arrows marked regenerated BDA-positive fibers. *Undegraded NT3-chitosan materials. (B) Diagram of four segments selected for quantitative analyses across the lesion area: R1 (5-mm segment rostral to the lesion), R2 and C1 (2-mm segments within the lesion area), and C2 (5-mm segment caudal to the lesion). (C) Quantitative analyses of the number of BDA-positive fibers rostral and caudal to the lesion area, in segments R1, R2, C1, and C2, at various time points for all eight treatment groups. Note that only the NT3 groups with or without DMSO showed BDA-labeled fibers in the R2, C1, and C2 segments over time, indicative of regeneration. All data are presented as mean ± SD (n = 4). *P < 0.05, NT3 or NT3-DMSO compared with the other groups.

Discussion

An Excellent Microenvironment Ensured Regeneration and Functional Recovery.

Through this series of studies, we have demonstrated that endogenous NSCs have significant potential to regenerate and repair CNS injuries. A major obstacle limiting the ability of endogenous NSCs to achieve neural repair is the hostile microenvironment in the injured CNS resulting from myelin-associated axonal growth inhibitory agents and the inflammatory immune environment, which prohibit neurogenesis from NSCs and promote scar-forming gliosis. Therefore, the establishment of an excellent growth-promoting microenvironment is key to enable endogenous and, when applicable, exogenous NSCs to generate new neurons and form nascent neural circuits that can be used for neural repair.

In the present study, prolonged (14 wk) slow release of NT3 was achieved by coupling it with chitosan (18), a biodegradable material, to create a regeneration-promoting microenvironment (Fig. S1). With NT3-chitosan, endogenous NSCs and their progenies were effectively activated and attracted to migrate into the NT3-chitosan tubes and subsequently differentiate into neurons, which matured to form a nascent functional synaptic network. We have shown that such nascent local neural networks can be used by regenerating descending and ascending axons through synaptic connections without the need for long-distance growth, providing a unique regeneration strategy using endogenous NSCs (Fig. 7). Previously, we reported that chitosan loaded with collagen also promoted axonal regeneration in semisectioned spinal cord (24); however, there were very few neurons in the regenerated tissues, suggesting that NT3 is much more neurogenic than collagen.

Fig. 7.

Working model. The diagram showed that NT3-chitosan, by providing an excellent microenvironment, enhanced local neurogenesis from endogenous NSCs and facilitated the formation of nascent neural synaptic networks, serving as a relay station to transmit nerve impulses from severed nerves.

Three Mechanisms Underlying the Regenerative Power of NT3-Chitosan.

Previous SCI research has largely focused on ways to overcome the inhibitory CNS environment as well as enhance CNS neuronal intrinsic regenerative potentials for nerve fibers to reextend from the CST. In recent years, the Sofroniew and Egerton groups have shown that spinal proprial interneurons are crucial for motor function recovery after partial SCI (25). Moreover the Nakashima group has demonstrated that functional recovery from SCI can be facilitated by transplantation of NPCs and that the subsequent neuronal differentiation from NPCs is crucial (11). Our present findings show that after complete SCI, the use of the NT3-chitosan tube, through activation of NT3-TrkC signaling, elicits two beneficial events: enabling of long-distance nerve fiber growth, and enhancement of new neurogenesis and subsequent formation of a functional new neural synaptic network. Through inhibition of new endogenous neurogenesis, behavioral recovery was partially impaired, as were MEP and SEP amplitudes, demonstrating the contribution of nascent relay neuronal networks created via activation of endogenous NSCs and subsequent neurogenesis. Moreover, we also found that the NT3-chitosan tube not only acted on neural cells, but also elicited inhibition of the inflammatory immune process (22), both of which could contribute to better nerve regeneration and adult neurogenesis.

We do acknowledge that the initial application of Ara-C for 2 wk not only suppressed NSC activation, but also inhibited proliferation of other cells, including infiltrating immune cells and endothelial cells. The fact that neurogenesis was impaired even long after Ara-C was no longer present suggests that the initial phase of NSC activation remained important for proper production of new neurons at the injury site (Fig. S7). Whether the initial blockade of proliferation of cells of the other systems, including immune and vasculature systems, might also elicit long-lasting effects remains to be determined.

Clearly there are advantages to engaging endogenous stem cells for neural repair as opposed to transplantation of exogenous stem cells, where immune rejection as well as safety issues related to tumor formation from transplanted stem cells must be considered. In summary, our findings suggest that CNS regeneration can be possible with new strategies targeting three areas: reduction of inflammation, promotion of axonal regrowth, and enhancement of new local neurogenesis and subsequent synaptic network formation. It is likely that improving these three areas will potentially lead to novel therapeutic approaches to treat SCI.

Materials and Methods

All experimental procedures were approved by and performed in accordance with the standards of the Experimental Animal Center of Capital Medical University and the Beijing Experimental Animal Association.

NT3-Chitosan Tube Preparation.

The chitosan tubes and NT3-chitosan carriers were prepared as described previously (17, 18, 24), with modifications. Detailed information is provided in SI Materials and Methods.

Surgical Procedure and Animal Care.

Female Wistar rats weighing 200–220 g were anesthetized by i.p. injection of 6% chloral hydrate (0.6 mL/kg body weight), to prepare rat models of thoracic spinal cord transection (SI Materials and Methods). The amounts of k-252a and k-252b used in these experiments were based on previously published studies (26).

Immunohistochemistry.

Morphological analysis and quantification were performed as described previously (24), with modifications, as detailed in SI Materials and Methods.

Nerve Tract Tracing and Quadruple Immune Fluorescent Labeling.

Biotinylated dextran amine conjugated with fluorescein [BDA-fluorescein, 10% (wt/vol) solution; Molecular Probes] was injected into eight different sites within the motor cortex to label the CST (24). Only fibers longer than 40 µm were counted (27). Details are provided in SI Materials and Methods.

Electrophysiological Studies.

Indices measured included SEP and MEP. Electrophysiological assays were performed as described previously (28, 29), with modifications (SI Materials and Methods).

Behavioral Assessment.

Observers blinded to the treatment methods and groupings performed BBB scoring in an open field (24, 30). More details are provided in SI Materials and Methods.

Labeling of Endogenous Stem Cells.

BrdU (50 mg/kg body weight; Sigma-Aldrich) was injected i.p. every 24 h for 1 wk after the operation (SI Materials and Methods).

Immunoelectron Microscopy Data Analysis.

We identified mature synapses on serial sections and defined them as described previously (31). Details are provided in SI Materials and Methods.

Neural Circuit Test with the Multielectrode Dish.

Preparation of the multielectrode dish system (Alpha MED Scientific) was essentially as described previously (32) (SI Materials and Methods).

Statistical Analysis.

Unless stated otherwise, all values are presented as mean ± SD. The Shapiro–Wilk method was used for data normality analysis, and the Levene test was used to test for homogeneity of variance. One-way ANOVA and Bonferroni analysis (multiple comparison for three groups), or the Student t test or Mann–Whitney U test, were used to determine statistical differences between two groups. P < 0.05 was taken to indicate a statistically significant difference.

SI Materials and Methods

Preparation of the NT3-Chitosan Tube.

In a modified method (17, 18, 24), under sterile conditions, 2% (wt/vol) solution of poly-N-acetyl glucosamine derived from 85% (wt/vol) deamidized chitosan (Sigma-Aldrich) in 100 mL of water containing 2% (wt/vol) acetic acid was plasticized by treatment with 1 g of di(hydroxyethyl) sulfoxide, which had a melting point of 112–113 °C, and 1 g of lithium chloride. This mixture was thoroughly stirred. A 2.0-mm-diameter glass capillary was washed, sterilized at high pressure, dried, vertically immersed in the foregoing chitosan solution, withdrawn slowly, and dried while keeping the tube vertical. This process was repeated until the inner and outer diameters reached 2.0 mm and 2.2 mm, respectively. The dried glass capillary with the chitosan tube was immersed in NaOH solution for 1 h, and then in distilled water. The distilled water was changed frequently until it became nonalkaline. The glass capillary was discarded, leaving a transparent chitosan tube. The tube was cut into a 5.0-mm length, immersed in 75% (vol/vol) alcohol for sterilization, and washed with PBS.

NT3-chitosan carriers were prepared according to a modified published method (17, 18). Under sterile conditions, 10 mg of 85% deacetylated chitosan particles (Sigma-Aldrich) were dissolved in 10 mL of sterile deionized water at pH 7.2, allowed to swell for 6 h, and then centrifuged. The supernatant was then discarded. The swollen chitosan particles were frozen at −20 °C for 24 h, and then at 4 °C for 10 h. NT3 (Sigma-Aldrich) was reconstituted to 100 µg/mL in sterile cold deionized water, and 100 ng of NT3 was mixed with the chitosan particles in solution at 4 °C. After stirring at 4 °C for 6 h, the NT3-loaded chitosan carrier mixture was vacuum-cooled and dried. The dried chitosan particles loaded with NT3 were added to a type I collagen solution, stirred for 30 min, centrifuged, collected, and stored at 4 °C. Then 10 mg of chitosan carriers loaded with NT3 (100 ng) was injected into the middle part of the 5-mm chitosan tube and kept at 4 °C.

Surgical Procedure and Animal Care.

All experimental procedures were approved by and performed in accordance with the standards of the Experimental Animal Center of Capital Medical University and the Beijing Experimental Animal Association. Wistar rats weighing 250–300 g were anesthetized by i.p. injections of 6% chloral hydrate (0.6 mL/kg body weight). To prepare a rat model of thoracic spinal cord transection, we created a laminectomy under an operating microscope, followed by transection at the T7-8 level, with removal of a 5.0-mm spinal cord segment. The blade was scraped repeatedly along the ventral surface of the spinal canal, and any residual fibers at the lesion site were removed by aspiration.

The animals were divided into eight groups. In the NT3-chitosan tube (NT3) group, n = 124, a 5-mm-long, 2.2 mm o.d., 2.0 mm i.d. chitosan tube seeded with the NT3-chitosan carrier was implanted into the lesion area. The ET group, n = 98, had only an empty chitosan tube implanted. The LC group, n = 120, received no treatment after the operation. Muscles and skin were closed in layers.

In the fourth group (k-252a; n = 90) an NT3-chitosan tube was implanted into the lesion area, and then k-252a (Abcam) dissolved in DMSO was injected at a flow rate of 0.15 µL/h into the lesion spinal area via an osmotic minipump (model 1003D; Alzet), resulting in the delivery of 34 µg k-252a/100 g body weight/day via the spinal cord for 12 wk. In the fifth group (k-252b; n = 90) an NT3-chitosan tube was implanted into the lesioned area, followed by injection of k-252b (Abcam) dissolved in DMSO at a flow rate of 0.15 µL/h into the lesioned spinal area via an osmotic minipump (model 1003D; Alzet), resulting in delivery of 34 µg k-252b/100 g body weight/day through the spinal cord for 12 wk. In the sixth group (NT3 antibody; n = 90), an NT3-chitosan tube was implanted into the lesion area, followed by injection of NT3 antibody (300 ng/200 μL in PBS; Santa Cruz Biotechnology) at a flow rate of 0.15 µL/h into the lesion spinal area via an osmotic minipump (model 1003D; Alzet), resulting in a delivery of 5.4 ng of NT3 antibody per day through the spinal cord for 12 wk. In the seventh group (NT3 tube+DMSO; n = 90) an NT3-chitosan tube was implanted into the lesion area, followed by injection of 0.01% DMSO vehicle at a flow rate of 0.15 µL/h into the lesion spinal area via an osmotic minipump (model 1003D; Alzet), resulting in a delivery of 3.6 μL of DMSO per day via the spinal cord for 12 wk.

In the eighth group (tube+DMSO; n = 90) an empty chitosan tube (not loaded with NT3) was implanted into the lesion area, followed by injection of 0.01% DMSO vehicle at a flow rate of 0.15 µL/h into the lesion spinal area via an osmotic minipump (model 1003D; Alzet), resulting in a delivery of 3.6 μL of DMSO per day in the spinal cord for 12 wk. In the groups in which an osmotic minipump was used, the pump was changed at the end of week 6, and the new pump provided the injection at the same concentration and rate for the next 6 wk. The amounts of k-252a and k-252b used in this study were based on previously published studies (26).

After the operation, the rats were kept warm and placed on beds of sawdust. The rat bladders were massaged three or four times daily, and i.m. injections of ampicillin were administered (50 mg once daily up to 1 wk after the operation) to prevent infections. In some rats, at 12 mo after the first operation, the lesion site was reresected, and plastic diaphragms were placed between the two ends of the lesion sites (n = 4). In some other rats in the NT3-chitosan tube and LC groups (n ≥ 5), Ara-C was successively injected into the lateral ventricle via a micropump for 14 d. To prevent dehydration, rats were hydrated with up to 20 mL/d with lactated Ringer’s solution or normal saline injected i.p. Food and water were provided ad libitum, and supplemental oral feedings were given as necessary. The experimental rats were kept at a temperature of 24–26 °C and relative humidity of 35–45% on a 12-hour light/dark cycle.

Immunohistochemistry.

The primary antibodies included mouse or rabbit anti-Nestin (Chemicon/Cell Signaling Technology, diluted 1:200), used to label NPCs; mouse monoclonal anti–β-tubulin 3 (Chemicon, diluted 1:300), to label immature neurons; polyclonal rabbit anti-GFAP (Zymed, diluted 1:300), to label astrocytes; polyclonal rabbit anti-NeuN (Chemicon, diluted 1:200), to label mature neurons, and mouse anti-BrdU, to label proliferative cells.

At each time point after the operation, four or five rats were selected at random from each group and sacrificed by an overdose of anesthesia. After transcardial perfusion with 4% paraformaldehyde, the brain and spinal cord were excised, fixed at 4 °C in fixing solution for 6–8 h, and kept in 30% sucrose solution (with 0.1 M PBS) overnight. The lesion area was examined under a dissecting microscope. The spinal cord tissue including the lesion area was embedded in OCT compound (Sakura Finetek) for frozen tissue specimens to ensure optimal cutting temperature and sliced longitudinally or transversely with a cryostat microtome to produce 8-µm sections. All sections were divided into three groups: group 1 for H&E staining, group 2 for immunohistochemical staining, and group 3 as the control for the immunohistochemical staining.

The group 2 sections were washed three times with 0.01 M PBS and then incubated with the primary antibodies at 4 °C overnight. The sections were incubated with biotinylated secondary antibodies for 4 h and washed with 0.01 M PBS three times. They were finally reacted with the avidin-biotin-peroxidase complex, and antigen localization was visualized by reaction with 3,3′-diaminobenzidine and examined under a light microscope. Alternatively, after primary antibody incubation, the sections were incubated with appropriate secondary antibodies conjugated to various fluorescent labels, such as Texas red-conjugated Affinipure goat anti-mouse IgG and CyTm2-conjugated Affinipure goat anti-rabbit IgG (Jackson Laboratory, diluted 1:300), at room temperature for 3 h in the dark. The sections were covered with coverslips and Vectashield-mounting medium containing DAPI (Vector Laboratories), and examined under a fluorescence microscope (BX-51; Olympus).

As a control, the normal mouse or rabbit serum was used to replace specific mouse or rabbit primary antibodies and treated identically to the test samples.

Quantification for Immunohistochemistry.

Between 10 and 15 longitudinal sections including the lesion area were selected based on the odd-number set or even-number set via serial sectioning. The numbers of immunopositive cells in the rostral, middle, and caudal sections of the lesion area were analyzed. The numbers of cells expressing various markers were determined by counting immunopositive cells in defined areas in the lesioned area under high magnification using a counting frame (25 μm × 25 μm).

Observation with Light Microscopy and Electron Microscopy and Quantitative Analysis.

At each time point after the operation, four rats were selected at random from each group and sacrificed by an overdose of anesthesia. After transcardial perfusion with 4% paraformaldehyde plus 2% (wt/vol) glutaraldehyde, the brain and spinal cord were excised and then fixed in 4% paraformaldehyde plus 2% glutaraldehyde at 4 °C. The regenerated tissues in the rostral, middle, and caudal sections of the lesion area were immersed in 1% osmium tetroxide for 2 h, then washed several times with 0.075 M PBS, dehydrated with gradient alcohol and acetone, and embedded in epoxy resin. The semithin resin sections were sectioned at 1 µm, stained with hematoxylin, and observed under a light microscope. Between 10 and 15 semithin resin sections stained with hematoxylin were selected based on the odd-number set or even-number set. The myelinated axons in the rostral, middle, and caudal sections of the lesion area were counted under an oil lens with a light microscope. To observe the ultrastructure of the regenerated tissue, the ultrathin sections were stained with uranyl acetate and lead citrate and observed under an electron microscope.

Nerve Tract Tracing and Quadruple Immune Fluorescent Labeling.

At each time point after the operation, four or five rats were selected at random from each group, anesthetized with an i.p. injection of 6% (wt/vol) chloral hydrate (0.6 mL/kg body weight) and fixed on a stereotaxic apparatus. Biotinylated dextran amine conjugated with fluorescein (BDA-fluorescein, 10% wt/vol solution; Molecular Probes) was injected into eight different sites within the motor cortex to label the CST (24).

At the end of the 3- to 4-wk survival period, the rats were perfused transcardially. The brain and whole spinal cord were excised and fixed in 4% paraformaldehyde at 4 °C for 7–8 h, then placed in 30% sucrose PB solution at 4 °C overnight. The brain was coronally sectioned and the spinal cord longitudinally sectioned on a cryostat microtome, and observed under a fluorescent microscope. Between 10 and 15 longitudinal sections including the lesion area were selected based on the odd-number set or even-number set (24). BDA-FITC–positive fibers were counted in the 5-mm segment rostral and caudal to the lesion area, as well as in the 2-mm segment rostral and caudal within the lesion area. Only fibers longer than 40 µm were counted (27).

In another experiment, at 7 d after the spinal cord operations, BDA-fluorescein was injected bilaterally (0.4 μL per point) in a total of five points into the motor cortex of both hemispheres. Four rats were used in each group. At the end of the additional survival period after injections, the animals were perfused, and spinal cords encompassing the lesion site were collected for anatomic and immunohistochemical analyses.

Sections with CST-fluorescein labels were washed three times with 0.01 M PBS, shielded from light, and incubated with the primary antibodies (mouse anti-BrdU and chicken anti-MAP₂; Abcam) at 4 °C overnight. The sections were incubated with biotinylated secondary antibodies (goat anti-mouse IgG Alexa Fluor 594, from Jackson Laboratory; donkey anti-chicken IgG Alexa Fluor 647, from Abcam) for 4 h and washed three times with 0.01 M PBS. The sections were mounted with coverslips using Vectashield-mounting medium containing DAPI (Vector Laboratories).

Electrophysiological Studies.

Electrophysiological assays were performed for each group (n = 6) before the operation and at various time points after the operation. The electrophysiological data obtained before the operation served as a normal control (n = 6). The indices measured included SEP and transcranial electric stimulation MEP, which could comprehensively reflect the sensory and locomotor function in the healthy state, after SCI, and during the recovery process (28).

The rats were anesthetized by an i.m. injection of ketamine (50 mg/kg body weight), and their limbs were abducted and fixed on a board by cloth bands. The room temperature was kept at 25–28 °C. A Keypoint-II bichannel evoked potential/electromyograph (Dantech) was used to test the MEP and SEP of the experimental rats. For MEP, the stimulating electrodes included a positive electrode and a negative electrode. The positive electrode (a 2-mm ball) was placed on the skull surface at the midline of the motor area of cerebral cortex, 2.5 mm behind the anterior fontanel and 2 mm on the left or right side of the midline. The negative electrode (a 4-mm disk) was placed on the skull surface of the hard palate. The recording electrodes (i.e., the needle electrodes), were inserted into the tibialis anterior muscle of the bilateral hindlimbs at the depth of 1.5 mm. The reference electrode was placed 2 cm away from the distant end of the record electrode, and the ground line was placed between the stimulating and record electrodes. A single square wave was used to stimulate the motor area of the cerebral cortex through the skull, at an intensity of 5–12 mA (29), duration of 0.2 ms, stimulating frequency of 1 Hz, a bandpass filter from 2 Hz to 10 kHz, and an amplifier sensitivity of 0.1 mV/D. MEP was recorded at the tibialis anterior muscle of the bilateral hindlimbs, i.e., the latency and the amplitude from the negative peak to the adjacent positive peak. The distance between the stimulating electrode and the recording electrode was measured as well. Before the MEP measurements, the motor threshold intensity was set as described previously (29).

We chose another four rats and cut their spinal cords at C7. These rats served as controls to show that the MEP was specifically induced from the motor cortex, not volume conduction from the stimulating point (29).

SEP measurements were also done with a Keypoint-II bichannel evoked potential/electromyograph. The positive electrodes (i.e., needle electrodes) were inserted into the tibialis anterior muscle of the bilateral hindlimbs. This muscle was successively stimulated with an average of 200 pulses at a stimulating intensity of 3–5 mA (to make the toes of the hindlimbs move slightly) (29), a duration of 0.2 ms, an amplifier sensitivity of 10 μV/D, a bandpass filter from 20 Hz to 3 kHz, and a sweep length of 80 ms. SEP were recorded on the skull surface of the sensory area of cerebral cortex, including P1 and P1-N1. The distance between the stimulating electrode and the recording electrode was measured as well.

At 12 mo after the operation, when all of the foregoing tests were completed, we performed re-resection in the lesion area/or regenerated area in all groups (n = 4), and inserted a plastic diaphragm between the two ends of the lesion area. At 30 d and 60 d after the re-resection, we remeasured MEP and SEP of the bilateral hindlimbs of the experimental rats with the same parameters as in the previous tests. The latency and amplitude measurements induced at 10 mA for MEP and induced at 3 mA for SEP were recorded for the statistical analysis.

Behavioral Assessment.

Before the operation, and at 1 d and each week after the operation, observers who were blinded to the treatment methods and groups performed BBB scoring in an open field in eight animals per group, to evaluate restoration of hindlimb locomotor function after SCI (24, 30). At 52 wk after the operation, four rats were selected at random from each group for re-resection at the original lesion site, and BBB scoring was assessed 5 wk later. Repeated-measures ANOVA with Bonferroni post hoc analysis were used to assay the statistical differences among the groups (n = 8) (24).

Labeling of Endogenous Stem Cells.

BrdU (50 mg/kg body weight; Sigma-Aldrich) was injected i.p. every 24 h for 1 wk after the operation. The injected BrdU was used to label the dividing cells, to allow observation of the dynamic process of cell division. The animals were sacrificed at different time points to examine the cell fate of BrdU-labeled cells.

Immunoelectron Microscopy Data Analysis.

Sections used for electron microscopy labeling of β-tubulin 3 (immunogold-silver) were incubated in a primary solution containing mouse anti–β-tubulin 3 (1:1,000 dilution) for 24 h at room temperature. After incubation, the sections were rinsed in PBS (0.1 M, pH 7.4) and incubated in a 1:50 dilution of donkey anti-mouse IgG bound to 0.4 nm nanogold (Nanoprobes) for 2 h for detection of new neurons. The gold particles were fixed to the tissue by incubating the sections in 1% glutaraldehyde in 0.01 M PBS for 10 min. The particles were enlarged for microscopic examination by reaction in a silver solution (HQ Silver; Nanoprobes) for 4 min at room temperature. The sections were then postfixed in 2% osmium tetroxide in 0.1 M PB, dehydrated and flat-embedded in epon (19% EM Bed-812; 36% DDSA, 44% NMA, and 1% BDMA; Electron Microscopy Sciences) between two pieces of Aclar plastic (Allied Signal).

To assess the distribution of β-tubulin 3 immunogold particles, data analysis was performed on ultrathin sections obtained exclusively from the outer (0.1–2 μm) surface of the flat-embedded tissue, where there was optimal penetration of immunoreagents. The β-tubulin 3 immunogold particles were distributed in kytoplasm and processes.

We identified mature synapses on serial sections and defined them by the presence of the following features in at least one section: a postsynaptic density, more than four presynaptic vesicles within 100 nm of the presynaptic membrane, and a clearly defined synaptic cleft (31).

Neural Circuit Test with the Multielectrode Dish System.

The multielectrode dish system (Alpha MED Scientific) was prepared basically as described previously (32). The device had an array of 64 planar microelectrodes, each 50 × 50 μm, arranged in an 8 × 8 pattern (interelectrode distance, 150 μm). Preparation of acute spinal cord slices as well as the electrophysiological recordings protocol were almost the same as described previously (32) with slight modifications; i.e., the spinal cord tissue including the lesion area was sliced longitudinally.

Statistical Analysis.

Unless stated otherwise, data are presented as mean ± SD. The Shapiro–Wilk test was used for data normality analysis. Levene’s test was used to test for homogeneity of variance. One-way ANOVA and Bonferroni analysis (multiple comparison for three groups), or the Student t test or Mann–Whitney U test was used to determine statistical differences between two groups. P < 0.05 was considered to indicate a statistically significant difference.