Regeneration of neurons and fibers in the mammalian spinal cord has not …

• Abstract
• Background
• Methods
• Results
• Discussion
• References
• Figures

Overview of nerve regeneration

To assess the morphological changes that the RNC underwent during the regeneration period, longitudinal sections of body wall were stained with Toluidine blue and studied under light microscopy (Fig. 1). The first stage (2 dpi) was characterized by an injured RNC with fan-shaped stumps (Fig. 1B). Remarkably, the pre-existing CT band in the stumps were practically unaffected; if anything, it lost a bit of cohesiveness at its most proximal end, adjacent to the injury. By 6 dpi, small bundles of fibers had extended from both nerve stumps, giving the structure an overall needle-like shape laying on the substratum provided by the body wall and pointing towards the opposing stump (Fig. 1C). At this stage, the origin of nerve bundles extending beyond the stump seemed to be, mostly, the result of an outgrowth in the nerve EN component, which lies closer to the body wall. Still, the lack of a substratum in the injury gap appeared to be the main limitation for scouting bundles. Support for this comes from observations of fiber bundles, which in the absence of a direct intervening matrix between the stumps, were found close to the trough of the injury gap. Nerve re-contact occurred by 12 dpi, concurrent with the appearance of a provisional matrix that filled the injury gap (Fig. 1D). The regenerated RNC spanned the injury gap but remained disorganized with empty spaces that could be seen. This was clearly observed by using the differential interference contrast (DIC) filter in the microscope. Although the nerve had reconnected by 12 dpi, the CT band that normally separates the HN and EN was not present. The RNC structure was much improved by 20 dpi (Fig. 1E). At this stage the number and localization of nuclei, both in the HN and EN, was similar to that of an uninjured nerve. The CT band separating the HN and EN was already perceptible at this stage in some, but not all animals, as traces of thin filaments along the length of the nerve. Approximately one month after the injury (28 dpi), the regenerated nerve had a normal architecture (Fig. 1F). The nuclear arrangement and density of fibers was now similar to that of a normal uninjured nerve and the CT band was already present and compact, although in some areas it still seemed diffuse.

Regeneration of fiber subpopulations

To obtain an in-depth view of the events that occur in the regenerating RNC, we studied the responses of different fiber subpopulations to the injury. Using immunohistochemistry we were able to correlate the gross appearance of the RNC, studied in light microscopy, with the growth of particular fiber populations across the injury site of the regenerated RNC.

RN1

This is a monoclonal antibody raised in our lab against H. glaberrima RNC tissue [35] that provides very specific labeling of what appears to be a large fiber population of the holothurian nervous system (Fig. 2A). RN1 does not label the CT band that separates the HN and EN component, thus, this band could be detected as a non-labeled distinct strip of tissue embedded within the high density of RN1 immunoreactive fibers. As a result, this antibody served as a two-edged marker, to measure progress during the process of nerve regeneration, namely the tracking of fibers as they elongate, and the appearance of the CT band as evidence of tissue organization into HN and EN components.

Most of the data collected with this antibody confirm the observations made using the Toluidene blue dye; however, some interesting differences were found. The fan-shaped morphology of the nerve stumps at the injury site observed at 2 dpi was mostly the result of a swelling (enlargement) in the EN component, while the HN component was not severely affected (Fig. 2B). At this stage, the HN component proximal to the injury appeared to retract in relation to the EN. At 6 dpi, the stumps acquired a club-shaped appearance and RN1-labeled fiber bundles could be observed protruding toward the opposite stump (Fig. 2C). Unfortunately, because of the high density of labeling, individual nerve fibers inside these pioneering bundles were hard to identify. However, the RN1 fiber bundles were always observed to be in direct contact with the substratum of the body wall and continuous with the injured stump. At this stage we observed the first difference in the rate of regeneration between the EN and HN components; the EN component provided the first scouting bundles that traverse the substratum towards the injury site, while RN1-labeled fibers of the HN appeared to be in a latent state. At 12 dpi, RN1 fibers from opposing nerve stumps came together, forming the newly regenerated RNC (Fig. 2D). At this stage, although RN1 fibers filled the area between the stumps, the separation of the RNC into its respective components had not occurred yet. During this stage RN1 labeled side branches could be observed originating from the same nerve bundles that form the connection between the stumps. These branches might be some that had gone astray into other injured tissues or they might be precursors to peripheral nerves. By 20 dpi the new RNC showed the first hint of connective tissue separation between the HN and EN components, albeit, rudimentary and irregular at times (Fig. 2E). The EN component was now structurally comparable to that of an uninjured RNC, as its wandering bundles of fibers were no longer present or at least apparent. Additionally, the regenerated RNC had a more cylindrical shape in accordance with the rest of the nerve; nonetheless, the staining intensity was now less than in uninjured areas. In spite of all these improvements in the EN component, the HN in most animals was still underdeveloped compared to uninjured counterparts in the same section. The new RNC was restored by 28 dpi with an HN and an EN component clearly separated from each other by the CT band between them (Fig. 2F).

α-GFS

α-GFS is a polyclonal antibody made in our lab against the neuropeptide GFSKLYF-amide that recognizes a subpopulation of neuronal fibers and somas in both HN and EN components of the RNC of H. glaberrima [33].

At 2 dpi, immunolabeling with GFS clearly showed the fan-shaped morphology of the cut RNC and the fraying of the nerve fibers close to the injury (Fig. 3A). The reduction in fiber density close to the injury was accompanied by an increase in labeling intensity close to the nerve stumps, which could be attributed to the appearance of α-GFS-labeled varicosities along the fibers. At 6 dpi, GFS-immunoreactive fibers had sprouted from the nerve stumps and the overall structure was clearly pointing towards the opposite stump (Fig. 1B). It is important to note that that the previously observed varicosities were also present in these migrating fibers. Similar to what was observed for RN1, fibers expressing the GFS neuropeptide also reconnected by 12 dpi (Fig. 3C). The labeling with GFS showed that immunoreactive fibers in the HN followed those of the EN; instead of using the body wall as a substrate. Even when the path taken by some fibers might be twisted and indirect, the overall axis of the fibers inside the regenerated EN was parallel to the long axis of the RNC. By 20 dpi both components were present in the regenerated RNC. No major differences were observed a month after transection (28 dpi) between regenerated RNC segment and its uninjured counterpart (Fig. 3D).

α-Galanin

In the RNC this antibody labels a relatively small population of fibers [34]. At 2 dpi, galanin immunoreactive fibers underwent a change in their morphology forming round varicosities with very intense labeling. This change was more conspicuous closer to the injury site, where the nerve fibers have became frayed; they were present in the HN and EN component of the RNC. At 6 dpi, varicosities were also found in the needle-shaped bundle of fibers exiting the nerve stump (Fig. 4A). Once the RNC reconnected at 12 dpi, the area of the injury could be easily identified by the presence of these large varicosities. At 20 dpi, galanin-positive fibers were present in both components of the RNC, although with a higher density in the EN (Fig. 4B). Galanin-positive fibers passing between these two components were also apparent at multiple points along the regenerated RNC, as is usual in an uninjured nerve. At 28 dpi, the large varicosities had disappeared and α-galanin produced a labeling that was virtually undistinguishable from that of a normal RNC.

In summary, all neuronal markers used showed that the injury gap that separates the two RNC stumps is slowly filled by a growth of fibers that originate from the stump ends in the first two weeks of regeneration (Fig. 4C). This process is complete by 20 dpi and a RNC with high similarity to the uninjured RNC is in place by 28 dpi.

Cell division

Pulse chase

RNCs are ganglionated nerves where the neuronal soma and supporting cells can be mainly found in the periphery. To study the possibility that cell division is occurring in the RNC as a result of the injury, animals were injected with the thymidine analog, BrdU, and then checked for its incorporation into the nuclei. Initially, at 2 dpi few cells within the proximal stumps, both in the HN and EC bands, of the injured nerves were undergoing cell division; no cell division was seen distal to the injury site (Fig. 5). At 6 dpi, the highest numbers of dividing cells were seen in the proximal region, with about 6% of cells (6.81 ± 0.42 in HN and 5.36 ± 0.4 in EN) undergoing cell division. Cell division distal to the injury site never exceeded 2%. By 12 dpi, cell division in the proximal stumps remained comparatively similar (6.54 ± 0.23 in HN and 383 ± 0.21 in EN) to the previous stage. However, an impressive surge in cell division occurred in the regenerating RNC that is forming across the injury site, where close to 20% of the cells (22.2 ± 1.06 in HN and 14.9 ± 1.15 in EN) were undergoing cell division. Subsequently, at 20 dpi, a sharp drop in cell division was observed in the newly formed RNC. Proximal to the injured nerve, cell division did not exceed 2%. At the last stage studied (28 dpi), cell division was mainly restricted to the HN of the regenerated RNC, where cell division was 3.28 ± 0.7% in HN and 1.26 ± 0.3 in EN, everywhere else, cell division was less than 1%.

Cell regeneration and birth dating

The next step was to determine whether the neurons in the regenerated RNC were originating from dividing precursors. For this we used two neuronal markers that label independent neuronal populations in the RNC, anti-GFS [33] and anti-Nurr1 (unp. obs.). These markers labeled neuronal cell bodies in both the EN and HN components of the RNC (Fig. 6A &6B). Initially, experiments were made by injecting animals with BrdU at different time periods and sacrificing animals at 30 dpi. However, these animals showed few if any GFS or Nurr1 neurons in their regenerated RNC. Thus, we decided to extend the experiment until day 62 dpi. For this, animals were separated into 6 groups and each group received BrdU injections during a particular 6-day period. All animals were sacrificed at 62 dpi. The rationale for this experiment was that only those cells that stop proliferation close to the last BrdU injections would retain the BrdU label until 62 dpi. Cells that kept proliferating would dilute out the BrdU and not be labeled. Thus, we would expect that dividing neuroblasts would stop dividing and then differentiate, so that most of the labeled cells that had stopped dividing within the RNC would be neurons.

Sections of the regenerated RNC were obtained and the cells in an area of equivalent length of RNC were determined on the microscope field of view (40×) were counted at 62 dpi. Results showed that the total number of cells (as measured by counting Hoescht-labeled nuclei) within the regenerated RNC did not differ from that in the proximal and distal RNC, suggesting that cell regeneration had occurred after 2 months (Fig. 6C). At this time, neurons expressing GFS or Nurr1 were also found within the regenerated RNC, albeit, in smaller numbers when compared to those within the proximal and distal RNC (Fig. 6D &6E)

Birthdating studies showed a peak during the second and third week of regeneration (groups 2 & 3, 8–22 dpi) when close to 3% of the cells (group 2 – 3.16 ± 0.54%, group 3 – 2.81 ± 0.59%) stopped dividing, and were still BrdU labeled at 62 dpi (Fig. 7A &7B). This would suggest that it is during this time period that most neurons have stopped dividing and started differentiation. To further verify that some of the dividing cells were neurons, we used double labeling with the neuronal markers GFS and NURR1. The results were somewhat unexpected. None of the cells that expressed GFS were double labeled with BrdU in any of the injected groups, even when over 300 GFS neurons were observed. In contrast, some cells labeled with Nurr1 were also labeled with BrdU (Fig. 7C). These cells were present only in the HN component of animals injected with BrdU during 8–22 dpi (which corresponds to the peak in cells that stopped proliferation). However, the number of neurons is rather small as only 2.5% of all NURR1 immunoreactive cells present in these groups were BrdU labeled (Fig. 7D).

Apoptosis

To determine the cellular events that occurred during RNC regeneration following a transection-type injury, we decided to examine if programmed cell death was occurring (Fig. 8). As soon as 2 dpi, apoptotic cells were present in the HN component of the proximal stumps. The highest number of apoptotic cells in the proximal stumps occurred at 6 dpi, in both the HN and EC bands, although the magnitude of apoptotic cells was higher in the HN band. By 12 dpi and on to 28 dpi, few if any apoptotic cells were observed in the nerve proximal to the injury site. In contrast, at 12 dpi, a large wave of apoptosis occurred in the regenerated RNC that had just formed over the injury gap. The number of apoptotic cells in the regenerating RNC dropped in the two subsequent stages, until at 28 dpi, no apoptosis was observed in the EN and only in 1% of cells of the HN band.

Spherule-containing cells

Two types of spherule-containing cells have been described in our system: morulas, which are highly basophilic and stain dark purple with toluidine blue, and spherulocytes, which are recognized by a monoclonal antibody prepared against sea cucumber tissues [28,32]. Only about 5% of spherule-containing cells are recognized by both markers.

Few (less than1 per visual field), if any, morula cells were found within the normal RNCs (see Fig. 1). (As mentioned earlier, fields of view, FOV, were obtained with the 40× objective and encompassed an equivalent area in all animals). Changes in their number within the RNC were already seen at 2 dpi, when they invaded the nerve stumps. They achieved their highest numbers at this stage, when the average number of cells per field of view within the proximal stumps was 6.9 + 0.2 cells per FOV (Fig. 9A, and see Fig. 1). Most cells were closely apposed to the CT band that separates the HN and EN components; this is the same location where they are normally found in an uninjured nerve. However, as a response to the injury, morulas were also found interspersed within the neuropile, more commonly where the nerve stumps were engorged, or fan-shaped. By 6 dpi, the number of cells in the proximal stump remained the same (6.4 + 0.24 cells per FOV) as in the previous stage. However, some of these cells were also associated with the RNC cell bodies, in both the HN and EN, adjacent to the coelomic side and also with the regrowing fibers that originated from the nerve stumps. At 12 dpi the number of morulas in the proximal RNC decreased to levels just slightly higher (1.6 + 0.1 cells per FOV) than the basal level and remained at this level until 28 dpi. The changes in the morula population only occurred proximally to the injury site since distal to the injury site the cell numbers were not significantly different from uninjured nerves at any stage studied. The number of cells was much higher in the new RNC bridging the injury site at 12 dpi, on average there were 7.12 ± 0.35 cells per FOV. At 20 dpi, the average number of cells in the regenerated RNC declined to 2.61 ± 0.25 cells per FOV; similar to the number of cells found at 28 dpi (2.89 ± 0.23 cells per FOV)(see Fig. 1).

The number of spherulocytes within the RNCs also underwent changes during regeneration (Fig. 9D). These cells were not normally found in uninjured nerves, however, as soon as 2 dpi they were found within the proximal stumps (1.8 ± 0.27 cells per FOV); and to a lesser extent distal to the injury (1.1 ± 0. 07 cell per FOV). A consecutive increase was seen in the RNC stump by 6 dpi, when on average 3.23 ± 0.34 cells per FOV were observed in the RNC. By 12 dpi, the number of spherulocytes proximal to the regenerated RNC decreased, but still remained significantly higher than the basal level (see Fig. 9B). This was accompanied by a relatively high infiltration of these cells into the new RNC that had on average 4.34 ± 0.51 cells per FOV. The number of cells seen in the regenerated RNC by 20 dpi decreased to 2.24 ± 0.25 cells per FOV; the range of cells in proximal and distal parts relative to the injury site was 1.59 ± 0.17 to 1.34 ± 0.04 per FOV. Distally, the spherulocyte population remained unchanged from 3 dpi until 20 dpi at less than 1.4 cells per FOV. At 28 dpi, spherulocytes could only be detected in the regenerated portion of the RNC (see Fig. 9C), and although their levels were lower than those of previous stages, they were still significantly higher than in uninjured nerves.