A crush injury of the peripheral nerve provides a useful process for the investigation of reversible axotomy in which reinnervating axons successfully regrow to the neuromuscular junction (NMJ) (Gorio et al., 1983; Pestronk and Drachman, 1985; Hopkins et al., 1986; Rich and Lichtman, 1989; Reynolds and Woolf, 1992; Ann et al., 1994; Son and Thompson, 1995a; Kawabuchi et al., 1998). After a crushing of the peripheral nerve, axon terminals degenerate within a few days and are completely engulfed by Schwann cells (SCs). Regenerating motor nerves first arrive at the NMJs within 1 week after an operation and subsequent reinnervation to the NMJ is in progress after the third week (Pestronk and Drachman, 1985; Kawabuchi et al., 1998). Since the perineural connective tissue sheath remains intact, and the basal lamina tubes which guide the direction of axonal growth are not interrupted after nerve crushing, regenerating axons are guided directly back to their muscle (Ide et al., 1983; Brown and Hardman, 1987; Crews and Wigston, 1990). Accordingly, regenerating axons grow through the original pathways of SCs, and NMJs are reconstructed at the muscle's original NMJ sites. Shortly after contact between a neurite (growing motor axon) and a muscle fiber is established, redifferentiation of the muscle receptor sites and synapse reformation occur concomitantly during endplate regeneration (Rich and Lichtman, 1989; Jacob and Robbins, 1990a,b).
Recent work has advanced the understanding of SC–axon and axon–extracellular matrix interactions in sequences of neurite extensions and subsequent axon terminal maturation during development and regeneration in NMJs (see reviews by Sanes, 1989; Fawcett and Keynes, 1990; Reynolds and Woolf, 1993; Martini, 1994). A reciprocating interaction between neurons and SCs may be the key to successful peripheral nerve regeneration. Neurons have a profound action on SCs, affecting proliferation, differentiation, migration and growth of the SCs. SCs thus recruited to the injury site would begin to proliferate, and then act as local sources of trophic support, including neurotrophins and neurokines (Martini et al., 1990). SCs promote neurite growth, and they have been shown to regulate local intercellular actions involved in extension and axon guidance during muscle reinnervation.
Recent publications give substance to the role of terminal SCs (TSCs) which respond rapidly to changes in junctional innervation and thus appear to be involved in regenerative changes in the structure of adult NMJs. Son and Thompson (1995a,b) suggest that after complete or partial denervation TSCs extend processes and guide the growth of axons during reinnervation and sprouting. With labeling of mitosis for the TSCs, Love and Thompson (1998) have shown that NMJs gain SCs as they mature and a considerable amount of this mitosis occurs at the developing and regenerating junctions. Knowledge of how TSCs respond to nerve injury is therefore important for understanding the formation and maintenance of NMJs.
Previous studies have shown a less robust response to axotomy in aged animals, showing that age affects the regenerative and reinnervating capabilities of peripheral nerve fibers. Regeneration in the axotomized peripheral nerves is retarded (Black and Lasek, 1979; McMartin and O'Connor, 1979; McQuarrie et al., 1989) and there are fewer regenerated axons (Moyer et al., 1960; Tanaka and Webster, 1991) or fewer sprouts (Moyer et al., 1960; Vaughan, 1992) than in young adults. Reinnervation to the NMJs is also delayed (Drahota and Gutmann, 1961; Jacob and Robbins, 1990a,b; Kawabuchi et al., 1998). To date, multiple controlling systems have been seen as accounting for the causes of retarded regeneration during aging (Gutmann and Hanzlíková, 1966; Gutmann et al., 1968; Pestronk et al., 1980; McQuarrie et al., 1989; Tanaka and Webster, 1991). Four fundamental factors underlie age-related deficits. (1) Besides the fallout of motoneurons (Ishihara et al., 1987; Hashizume and Kanda, 1990), aged α-motoneurons suffer damage mainly in the distal parts of the motor axon, and this then initiates a disconnective/retractive process (Gutmann, 1970; Ansved and Larsson, 1990; Johnson et al., 1991; Johnson et al., 1995; Kawabuchi et al., 1995). This theory presumes a lack of supply of trophic substances from the muscle or a decreased sensitivity of motoneurons to trophic substances (Hefti et al., 1989; Kuno, 1990; Johnson et al., 1991; Johnson et al., 1995). (2) There is evidence that following nerve injury, aged motor neurons respond more slowly, do not expand the field of innervation and set an upper limit to the supply of regenerating nerves to NMJs (Pestronk et al., 1980; Jacob and Robbins, 1990a,b). (3) After injury in peripheral nerves Wallerian degeneration is delayed in aged animals, possibly due to changes in the responses of SCs, neurons, and macrophages (Tanaka and Webster, 1991; Verdú et al., 2000). (4) Reduction in the capacity of regenerating neuromuscular synapses in aged animals can be deduced. First, reduced axonal transport of the components necessary for axonal growth (slow transport) and functional nerve terminal maintenance (fast transport) have been documented (Black and Lasek, 1979; McMartin and O'Connor, 1979; Komiya, 1980; McQuarrie et al., 1989). A decrease in axon growth rate with increasing age (Black and Lasek, 1979; Komiya, 1980; Pestronk et al., 1980; reviewed by Bowe et al., 1989; Verdú and Navarro, 1995) is a notable sign of aging and is responsible for the genesis of axonal dystrophy (Johnson et al., 1995) and deficits in myelination (Tanaka and Webster, 1991). Second, motor axon terminals in old age suffer synaptic ineffectiveness (Grinnell and Herrera, 1981) and intrinsic disturbances in neurotransmission (Gutmann et al., 1971; Frolkis et al., 1976; Smith, 1979).
This present study is part of a series of studies analyzing the effects of advancing age on endplate reinnervation. We found in our earlier works, using an axonal marker, that age-related deficits during muscle reinnervation in Wistar rats certainly coincided with impaired SC-axon interactions in which fasciculation and terminal differentiation of reinnervating axons were involved (Kawabuchi et al., 1998). The three cellular components lying in close apposition in NMJs, i.e., the TSCs, motor nerve terminals and the sites of clustered acetylcholine receptors (AChR sites) exchange signals anterogradely and retrogradely affecting synapse formation, maintenance and repair (Hall and Sanes, 1993; Burden, 1998). It is then crucial to ascertain how advancing age affects the integrity of cellular interactions that would induce a delay in regenerative events in NMJs. With the same experimental paradigm of nerve crushing, the spatiotemporal sequence of morphological alterations in each of the TSCs, terminal axons, and AChR sites during regeneration and reinnervation of skeletal muscle was determined. The pattern of the changes was compared between adult and aged adult animals. Motor endplates were visualized by labeling the AChR sites with fluorescein isothiocyanate (FITC)-conjugated α-bungarotoxin (α-BT). Cryosections of the rat skeletal muscle were double-labeled with SC marker (S100) and α-BT, or with an axonal marker (PGP9.5-LI) and α-BT, and then observed using confocal laser scanning microscopy. PGP9.5, an ubiquitin protein (Wilkinson et al., 1989), is a suitable marker for detection of regenerating axons reinnervating different target organs, and is helpful for locating the major nerve pathways through target tissue (Navarro et al., 1997; Verdú and Navarro, 1997). The three-dimensionally reconstructed morphologies of each of the endplate components have demonstrated some aging-related abnormalities particularly affecting TSCs and AChR sites.
Size Changes of TSCs, Terminal Axons, and AChR Sites During Muscle Reinnervation
The mean area of S100-LI TSCs from adult unoperated controls was greater than that of the PGP9.5-LI terminal axons or α-BT stained postsynaptic AChR sites (Fig. 1). Because of the precise alignment between the terminal axon and the AChR site, there was a gross size matching between these two endplate components in adults, while in aged adults the areas of the AChR sites were greater than those of the terminal axons (Figs. 1B, C).
Compared with contralateral controls, the measures of area from each of the three endplate components of adults showed no significant changes in the early reinnervation period (1, 4 wpc), except for reduction of the 4-week-value for the AChR sites (Fig. 1). There was a decline in the areas of TSCs and AChR sites at 1 and 4 wpc in aged adults (Fig. 1), and age difference in the magnitude of loss in the size of TSCs and AChR sites at 4 wpc was pronounced [4 weeks/control for TSCs (μm2), −2% (510/520) in adult versus −32% (370/546) in aged adult, P < 0.005, chi-square test; for AChR sites, −30% (188/268) in adult versus −48% (186/355) in aged adult, P < 0.01)] (cf. Fig. 2). Although the dimensions of terminal axons and TSCs at 4 wpc from both age cohorts were similar to the contralateral controls (4W/control for terminal axons, +3% (287/278) in adult versus −10% (261/291) in aged adult, P > 0.05), the presence of multiple innervation contributed to this while the single-innervation profile was basically smaller-sized.
As compared with the values obtained at 4 wpc, regeneration in the motor endplate in 8- or 12-week-muscle in adults was in progression on the basis of restoration of the AChR site to the normal size. As well, there was a significant increase in the areas of TSCs and terminal axons, to the extent that these increases obviously exceeded the contralateral control values (Fig. 1). Apart from multiple innervation, this increment was mainly due to simple junctional enlargement and/or elaborate terminal arborization. Accordingly, the size of terminal axons at this stage was not proportional to that of the AChR site, indeed it was greater. In aged adults, in contrast to the terminal axons, restoration in TSCs and at AChR sites was obviously delayed [12 weeks/control for TSCs (μm2), +23% (650/530) in adult versus −9% (508/557) in aged adult; P < 0.005; for AChR sites, +15% (345/300) in adult versus −34% (243/370) in aged adult, P < 0.01] (cf. Fig. 2). Throughout the study, the measures of areas from each of the endplate components in the contralateral side were similar to, or a little greater than, those of the unoperated or sham controls. This did not, however, significantly change the time courses of muscle reinnervation as plotted by the measured data.
Immunolocalization for S100 and PGP9.5
The profiles of the SCs or SC processes, well-organized or not, have been identified by the SC marker S-100 in adult (Kawabuchi et al., 1998) and postnatal (Hirata et al., 1997) NMJs. Identical SC profiles could be followed to the NMJs throughout the study. In 1-week-muscle where Wallerian degeneration proceeds in the muscular nerve, and where the terminal axons in the NMJ are generally sparse, axonal debris possibly remains within the SCs. The localization of PGP9.5 immunostaining in these muscles might be confusing. Hirata et al. (1997) have shown that the profiles of PGP9.5-LI terminal axons are defined by reference to those of the S100-LI TSCs, even in earlier stages of postnatal development. Such reference was possible and the entire outline of terminal axons could be expressed in virtually all NMJs by PGP9.5 immunohistochemistry at 3 (adult) and 4 (aged) wpc. After this stage, the morphology in the regenerating NMJs was compatible with the terminal axons at different maturation stages. The feature of these terminal axons during reinnervation basically coincided with that seen on neurofilament plus synaptophysin immunohistochemistry (Kawabuchi et al., unpublished observations). Consequently, sequential reinnervation changes in our time course study, delineated by S100 and PGP9.5 in NMJs from both age cohorts, were useful for locating the major nerve pathways after regenerating axons arrived in NMJs, as well as for collecting data for morphometric analysis.
Age Difference of Terminal Schwann Cells and Terminal Axons in 1-Week-Muscles
Labeling for S100 at 1 wpc in adult NMJs clearly defined chain-like cellular strands as SC strands, and the junctional clusters of elongated or ellipsoid cells with profuse process extensions as TSCs (Fig. 3A). Most SCs and TSCs in aged adults exhibited irregular contours and disorganization (Fig. 3B). In both age cohorts, the distribution of PGP9.5 immunostaining in nerve bundles was generally sparse and there were few of the terminal basket-like patterns of terminal axons in NMJs.
Delay in Close Correspondence Between TSCs and AChR Sites in Aged Animals During Muscle Reinnervation
In the adult cohort, at 3 wpc the S100-LI nerve bundles, made up of loosely fasciculated masses of a chain of cell bodies and processes, crossed motor endplates, and connected several motor endplates, consistent with so called “bridges” (Son and Thompson, 1995a, b) (Fig. 4A). Greater portions of the α-BT stained AChR plaques displayed green-colored FITC fluorescence, indicating that S100 labeling only partially covered the areas of AChR plaques in the NMJs. After 4 wpc, there was a precise covering of the AChR geometry by round- or spindle-shaped S100 labeling identical to that of TSCs (Fig. 4B). Progressively, a closer correspondence between the TSCs and AChR plaques was verified by the gradual increase in the incidence of NMJs where more than 50% of the AChR geometry apposed the S100 labeling (Figs. 4B, 5A). In the aged cohort, at 3 wpc the S100-LI cell strands were generally sparse, and only poorly formed thin processes extending from the cell body contacted the underlying AChR plaque in most of the NMJs (Fig. 4C). Besides a delay in arrival in the NMJs, as shown by the slowing rate of close correspondence (> 50%) between TSC and the underlying AChR site (Fig. 5B), the feature of TSCs at 8 wpc consisted of persistent poor formation (Fig. 4D).
Delay in Close Correspondence Between Terminal Axons and AChR Sites in Aged Animals During Muscle Reinnervation
In adult NMJs, multiple PGP9.5-LI extensions from nerve trunks entered the NMJs at 3 wpc. After 4 wpc, reoccupation by the extensions over greater portions (more than 50%) of the AChR geometry was in progress (90% of the junctions at 8 wpc)(Figs. 4E, 5C). Progressively, much of the postsynaptic receptor region became covered by well-defined elaborate arborizations (Fig. 4F). In the aged adult cohort, the pattern of endplate reoccupation coincided with that of the adults, in that the geometry of PGP9.5 immunostaining over the area of AChR plaques gradually increased (Figs. 4G, 4H, 5D). However, the rate of increase in the area of correspondence was slowed compared to that of the adult. At 4 wpc, the distal extensions, with generally low immunostaining and ill-defined contours, were mostly traced to the NMJs, but terminal expansions were poorly formed (Fig. 4G). Although 50% of the NMJs at week 8 had greater portions (> 50%) of the AChR geometry apposed by the PGP9.5-LI terminal arborizations (Fig. 5D), the anatomical relationship between these two endplate components was clearly incomplete (Fig. 4H). In both age cohorts the pattern of close correspondence of the TSCs or terminal axons at 12 wpc was similar to that seen in the unoperated controls (Fig. 5). The same was true in the sham or contralateral controls at 12 wpc.
Degenerated Schwann Cells in Aged Adults
Early in reinnervation (3 and 4 wpc), S100-LI profiles identical to TSCs exactly located in the NMJs, and there was a profuse, elongated extension of processes from these cells in adults (Fig. 4A). The S100 immunostaining in aged adults displayed degeneration profiles of the SC strands and TSCs up to 12 wpc, whereas these were infrequently found in adults. The SC strands took a meandering and undetermined course, without any uniformity in size and shape (Figs. 6A, B). Dispersed remnants or debris, affecting both the preterminal and terminal regions, were thus formed and pitiful extensions of processes from the TSCs were evident (Fig. 6B). Late in reinnervation the SC strands displayed either remarkable tangles of poorly fasciculated, unaligned, unordered and distorted bundles (Fig. 6C), or poor development (Fig. 6D). The postsynaptic AChR plaques apposing these TSCs manifested thin flat plaques or poorly defined contours (Figs. 6A–D). Throughout the study, closer views of the TSCs revealed bizarre profiles such as abnormal arrangements and ill-defined contours, as well as diverse sizes and shapes (Figs. 7A, B). Besides the large portions of the AChR geometry which TSCs failed to occupy (Fig. 7C), sets of collapsed SC strands and AChR associations, disrupting the usual chain-like arrangement, alignment, and organization, and instead taking a loop-like figures (Fig. 7D), were the characteristics of the late reinnervation period.
Regenerative Growth of Terminal Schwann Cell Processes
Compared with contralateral controls, the measures of length of TSC processes (mean ± standard deviation) at 3 wpc in adults showed a significant increase (controls versus 3 weeks, 14.2 ± 6.0 versus 36.4 ± 17.6 μm, P < 0.001, t-test). Their subsequent decline at 8 wpc (3 versus 8 weeks, 36.4 ± 17.6 versus 22.4 ± 15.6 μm, P < 0.001) indicated eventual retraction and withdrawal in regenerative growth. Age changes consisted of shorter TSC processes at 3 wpc (adult versus aged adult, 36.4 ± 17.6 versus 28.2 ± 12.5 μm, P < 0.01), and conversely, longer ones at 8 wpc (adult versus aged adult, 22.2 ± 16.5 versus 31.4 ± 15.6 μm, P < 0.01), indicating delay in retraction and withdrawal. NMJs interconnected by the TSC processes were most frequent at week 3 in adults (40% of the NMJs), whereas in the aged adults it was at week 4 (35% of the NMJs). The incidence of these changes gradually decreased thereafter.
Age Differences of Regenerating Preterminal and Terminal Axons
At 3 wpc the less-developed forms of PGP9.5-LI terminal axons in the adult NMJs were identified on the basis of less delineated and uneven outlines, and poorly defined terminal branching. After 4 wpc well-organized PGP9.5-LI terminal axons in adult NMJs gradually became common (Figs. 4E, F). Some notable features of PGP9.5-LI extensions in aged adults, which were infrequent in the adult, were as follows: Early in reinnervation, (1) conglomerates of thin and thick extensions connecting to undefined terminal structures (Fig. 4G). (2) accumulation on NMJs of disorganized and unaligned extensions, as well as profuse but disorderly developed terminal expansions (Fig. 8A). Mid-to-late in reinnervation, (1) extensive tangles of thin, elongated, distorted extensions with irregular contours (Fig. 8B), (2) aggregates of abundant thick extensions around NMJs (Fig. 8C), and (3) long meandering extensions arising from nerve bundles (Fig. 8D).
Early in reinnervation, representative features of endplate-to-endplate connections consisted of multiple (Fig. 9A) or single (Fig. 9B) extensions connecting from terminal to terminal over several NMJs. With progression of reinnervation, multiple extensions usually converged into a single NMJ. Additional substrates for endplate-to-endplate connections were noted, i.e., single, slender projections arising from the node of Ranvier in the nerve bundle (Fig. 9C). In aged adults, the multiple axon-multiple NMJ relationship persisted until late in reinnervation. Mid-to-late in reinnervation, endplate-to-endplate connections manifested as single extensions giving rise to terminal (Fig. 9D) or preterminal (Fig. 9E) sprouting or supplying several NMJs close to each other (Fig. 9F). Poorly defined terminal structures of these NMJs were a common feature (Figs. 9D, E).
Degeneration and Regeneration in Postsynaptic AChR Sites
The AChR plaque was circular or doughnut-shaped, consisting of multiple AChR-rich branches in adults (Fig. 10A). AChRs were fairly evenly distributed within each branch. In addition, AChRs appeared enriched along the borders of most of the branches, forming a relatively smooth outline. Early in reinnervation (weeks 1, 3 and 4) a majority of the AChR plaques exhibited intact profiles, while some of them were less delineated, small, flat and slender, displaying shrinkage and hypersegmentation (Figs. 10B, C). Disorganization, reduced density, and fragmentation of the AChR plaque into abundant patches was also detected (Fig. 10C). At 8 wpc some AChR plaques had well-defined contours of FITC fluorescence, similar to those seen in the unoperated controls, while the remainder displayed thickening of the fringes and a few small round cupulae with distinct contours (Figs. 10D, E). In aged adults, besides the changes described above, a strikingly abnormal morphology of AChR sites was noted throughout the study. We observed that AChR plaques broke up into series of discrete boutons with portions of irregular contours (Figs. 10F, G), or were made up of dispersed faint, fine granular FITC fluorescence (Figs. 10G, H). Branch borders were often ragged or frayed rather than smooth and sharply delineated (Figs. 10G, H). At 8 wpc staining of the AChR plaques recovered but their outlines were still uneven, irregular and less-delineated, and unorganized (Figs. 10I, J).
Muscle Fiber Atrophy
In both age cohorts nerve crush induced long-lasting atrophy of muscle fibers in which their cross-sectional area (Fig. 11), perimeter and diameter, versus those of contralateral controls, significantly reduced from intact levels. In aged adults the degree of reduction in the area of the muscle fiber was pronounced [adult versus aged adult, −21% (1484/1882 μm2) versus −51% (1421/2909 μm2 at week 4, P < 0.005, chi-square test; −32% (1366/2010 μm2) versus −42% (1618/2810 μm2) at week 8, P < 0.01)], and restoration to the original size was delayed.
Persistent Abnormalities at 12 weeks Post-crush
Sixty to eighty NMJs from 3 rats at 12 wpc from each of the age cohorts were examined. The two age cohorts displayed persistent changes in which both the SCs and axons were involved, such as terminal sprouting, endplate-to-endplate connections, and multiple innervation. Sixty-five per cent of aged NMJs showed these changes in TSCs, including additional abnormalities such as poor development or randomly aligned process extensions. Forty per cent of aged NMJs displayed persistent changes in terminal axons, including additional abnormalities as shown in Figs. 8C, 8D, and 9F. Sixty percent of aged postsynaptic AChR sites, compared to 30% of adult ones, were aberrant, showing persistent degeneration and regeneration changes, though the degree of abnormality varied among junctions.
Unusual Figures of Terminal SCs and Terminal Axons in Unoperated Controls
In some NMJs from adults, areas of the AChR site were devoid of an overlying terminal axon, consistent with signs of continued reorganization with focal deafferentiation and reinnervation (Cardasis, 1983; Kawabuchi et al. 1995)(Figs. 12A, B). The TSCs normally show a cluster of oval or round cells with a few minute processes. Unusual morphology in aged adult controls consisted of poorly developed TSCs with unordered spatial alignment, variability in size and shape, and abundant longer processes (Fig. 12C), or meandering preterminal axons (Fig. 12E). The underlying AChR sites showed complexity in branching pattern, and areas of faint or irregular activity, all within the same NMJ (Figs. 12D, F). These aberrant figures are consistent with the coincidence of degenerative and regenerative events in aged motor endplates (Cardasis, 1983; Cardasis and LaFontaine, 1987; Kawabuchi et al. 1995).
Variability of Data Between the Two Independent Researchers
Three sections for each of the morphometric analyses of the area of correspondence and the sizes of the three endplate components, at 4 and 8 wpc from both age cohorts, analyzed by one investigator (M.K.), were examined independently by another investigator (Z.C.). There was no discernible difference in the measured data, and the minor discrepancy (1–5%) between these two independent observers was due to the misidentification of the territory of the terminals, or the overlooking of faintly stained fine neural profiles arising from terminal axons or TSCs, which should have been included as a positive count.
Anatomical Relationships Between TSCs, Terminal Axons and AChR Sites During Muscle Reinnervation Following Nerve Crushing
Results reported here demonstrate the spatiotemporal patterns of endplate reinnervation and regeneration following nerve crushing, as three-dimensionally superimposed features of TSCs, terminal axons and postsynaptic AChR sites, as shown by confocal laser scanning microscopy. We note that confocal reconstructions of immunolabeling using SC or axon markers are useful and allow complete three-dimensional structural information to be precisely correlated with the distribution of individual immunostained counterparts in NMJs.
Changes in 1-week-muscle in adults mainly consist of an injured phase of the motor axons and a reparative phase of the SCs. By postoperative week three the well-organized SC strands extend into a majority of the NMJs and a wealth of multiple thin PGP-LI extensions already supply the NMJs, indicating that most preterminal and terminal SCs are associated with regenerating axons. At this stage a robust process extension from TSCs is underway and the terminal axon branches approached adjacent NMJs along the TSC processes, forming endplate-to-endplate connections. We conclude from this that 3 weeks post-nerve crush is a critical time point when major morphological nerve changes convert from a degenerative to a regenerative phase, i.e., regrowing axons return to the target muscle fiber and motor reinnervation and reformation of NMJs commences. Regeneration was well in progress after 4 wpc and we found that during synaptic reconnection there was a synchronous growth among the TSCs, terminal axons and AChR plaques, each of which rapidly restores the original figures. Regenerating TSCs and axon terminals covered a large fraction of the subsynaptic membrane over the following weeks, usually precisely reestablishing the endplate shape.
Aging Causes Delay in Reinnervation
The complete regeneration process rate declines with age, and in 24-month-old-groups regeneration was delayed and recovery clearly lowered with respect to younger groups. TSCs and axon profiles were observed in a majority of the aged NMJs as early as 4 wpc, a delay of at least by 1 week compared to adults. Concomitant with this, we found some specific aging-related abnormalities in the preterminal and terminal regions, such as the poor formation and damage to SC strands and TSCs, poorly fasciculated axon bundles and poorly defined terminals, as well as fine granular degeneration with frayed borders of postsynaptic AChR sites. It is likely that these abnormalities are responsible for the delay in reinnervation and regeneration in NMJs and similarly in the process of endplate reoccupation. On the basis of occasional degeneration changes of SC sheaths and TSCs, it may be that some motoneurons have lost portions of their connections with the target muscles. The pattern of reinnervation, which sets limit to the supply into the NMJs due to a fallout in portions of the regenerating nerve fibers, then substantiated the notion that aging produces a partial or a complete loss of skeletomotor connections from the target (Johnson et al., 1995), or a failure to develop and maintain an enlarged terminal field in response to nerve injury (Jacob and Robbins, 1990a,b; Johnson et al., 1995).
Impaired Outgrowth of TSC Processes and Terminal Sprouting
Our results from both age cohorts demonstrate the extensive processes elaborated by TSCs upon denervation and subsequent reinnervation. In adult rat muscle, TSCs rapidly extend profuse networks of processes that are maintained for several weeks after a crushing injury in the absence of reinnervation (Reynolds and Woolf, 1992; Woolf et al., 1992). Son and Thompson (1995a,b) suggest that the regenerating axons, on growing back to transiently denervated NMJs, use TSC processes as bridges to interconnect the motor endplates. Profuse extension of SC processes and bridge formation after peripheral nerve damage, which accompanied nerve terminal outgrowth, has been implicated in guidance of regenerating axons to denervated endplates. In adults, we observed the ability of TSCs to give rise to exuberant networks of processes early to mid in reinnervation and regeneration of NMJ, but small and short TSC processes became common progressively with time, indicating an eventual retraction and withdrawal. The poor process extension from TSCs in the early reinnervation period in aged adults is thought to result from malformation of the TSCs in the NMJ.
Our results effectively coincide with the accurate reoccupation of former synaptic sites following peripheral nerve injury, as documented for amphibians (Letzinsky et al., 1976) and mammals (Lüllmann-Rauch, 1971; Rich and Lichtman, 1989). On the basis of consistently high levels of close correspondence of the AChR site to TSCs or terminal axons after 4 wpc, it is evident that the mechanism of close apposing between AChR sites and presynaptic elements in adults operates from early in the reinnervation period.
With aging the postsynaptic AChR sites within the endplate were delayed in being covered by regenerating TSCs and terminal axons. We found an initial drastic decline of close correspondence between AChR sites and TSCs, and between AChR sites and terminal axons at 4 wpc when there was the lowest magnitude of endplate occupation. This represents a pronounced spatial misalignment, such as areas of regenerating TSCs or terminal axons that have not reoccupied the AChR site. Electron microscopy of unoperated aged NMJs on occasion showed a small portion which remained exposed despite ongoing endplate reoccupation, and thus motor axons appear to delay or fail to provide terminal branches to some gutters (Fahim and Robbins, 1982; Cardasis, 1983; Cardasis and LaFontaine, 1987). It is likely that once reaching a gutter, the regenerating axons sometimes fail to precisely reestablish the original endplate shape and to extend over its entire length due to spatial mismatch, under persistent damage of TSC and AChR sites.
We found damage of regenerating SC strands, as well as poor development of TSCs and their processes, persisting up to a late phase of reinnervation. The molecular mechanisms that regulate SC survival and maintenance during peripheral nerve regeneration are not known. It has been shown that migration and cell division participate in the addition of TSCs, thereby giving an adequate length to the motor endplate during early postnatal development and regeneration (Reynolds and Woolf, 1992; Love and Thompson, 1998). The mechanism for addition of TSCs by division at NMJs and/or from SC migration would exist to increase the number of TSCs and cover the regenerating motor endplate. In contrast, Komiyama and Suzuki (1991, 1992) demonstrated a progressive reduction of proliferation of SCs in aged nerves due to a decrease in mitogenic factors produced during degeneration. Here, we have not defined loss and addition of TSCs, nor did we observe mitotic TSCs after nerve crush. However, degeneration figures of SCs and TSCs in old age lead us to suggest that the sequences of formation and/or maintenance during SC growth in the preterminal and terminal regions are impaired. It is interesting to note that one of the neurotrophic factors, neuregulin, has mitogenic and survival effects on SCs, as well as effects on SC motility (Marchionni et al., 1993; Carraway and Burden, 1995; Mahanthappa et al., 1996; Trachtenberg and Thompson, 1997).
Degeneration and Regeneration of AChR Sites
AChR plaques at original junctional sites after denervation have been reported to survive for long periods in frogs, mice and rats (Frank et al., 1975; reviewed by Fambrough, 1979). We confirmed this finding, by our own observations on the process of reversible axotomy, that the original junctional AChRs remained stained throughout the observation period. Denervation accelerates the degradation of AChRs (Brett et al., 1982; Bevan and Steinbach, 1983; Csillick et al., 1999), and shrinkage of the subneural apparatus after denervation has been noted in acetylcholinesterase histochemistry (Csillick, 1967), and by in vivo staining (Rich and Lichtman, 1989). Denervation per se is primarily responsible for muscle atrophy. Shrinkage of the postsynaptic AChR site at the endplate following denervation may then account for the reduction in the muscle fiber size. Degeneration figures of AChR sites such as in smaller sizes, irregular contours and lowered staining is a result of successive, denervation-related changes in the postjunctional membrane (Krause and Wernig, 1985; Csillick et al., 1999). Using knockout mice lacking the components of the dystrophin-glycoprotein complex, Grady et al. (2000) provided evidence for its involvement in the formation and maintenance of AChR sites. In aged adults, besides signs of shrinkage, long-lasting morphological changes were observed, including fragmentation of the AChR clusters into fine granular aggregates with indistinct borders of the branches. Many of these changes were very similar to those observed in mutant mice. These features may then involve shrinkage-associated derangement in the reconstruction and/or maintenance of AChR sites.
As with the features of simplicity with well-defined contours, the small discrete structures comprising portions of AChR plaques, occurring in the later phase of reinnervating steps, are consistent with those reported for secondary consequences of degeneration and regeneration or transitions to the mature form of the subneural apparatus (Csillik, 1967; Krause and Wernig, 1985). The distribution of AChR sites over the entire interval studied could then be the result of receptor disappearance at some areas and reconstruction of receptors in other regions, while the proportion of reconstructed AChR plaques increases with time. Like the other nerve regeneration models (Womble, 1986; Ishikawa et al., 1988), the development of regeneration in postsynaptic AChR sites was associated with contact between regenerating motor axons and muscle fibers. The structure of AChR sites gradually regenerates with the progression of motor endplate reinnervation, but aging causes a delay in reinnervation.
Persistent Anomalous Changes in Aged Adults
After regenerating nerve fibers arrive at their targets, long-lasting morphological and physiological changes involving the internodal distance and diameter were observed for more than one year (Bowe et al., 1989; Hildebrand et al., 1985, 1986). This appears to account for the delayed recovery of motor function after peripheral nerve lesions (Drahota and Gutmann, 1961; Vaughan, 1992; Verdú and Navarro, 1995). Late in reinnervation (8, 12 wpc), under persistent muscle fiber atrophy, the presynaptic elements in NMJs (TSCs and terminal axons) were shown to increase in area relative to contralateral values, and terminal complexity either remains constant or increases by sprout extensions from TSCs or terminal axons. During this time period, it was noted that the size of the endplate surface occupied by AChR-rich membrane and muscle fibers gradually increased. The gross morphology of regenerating NMJs may be continually remodeled over periods of months, and such anatomical plasticity in the target region is most likely an adaptational and repair process (Johnson et al., 1995).
In adults, these aberrant changes persist but gradually decline, and in a preliminary study of ours it has been shown that even on the 20th week the residual abnormalities in aged adults do not seem to regress and are still more frequent compared to adults (Kawabuchi and Zhou, unpublished observations). Specific aging-related abnormalities consisted of varying features of endplate-to-endplate connections, disorganized, disordered axonal branching, poorly-organized TSCs, and poorly-formed AChR plaques. Some possible explanations for the long-lasting incomplete regeneration and reinnervation are as follows: (1) Endplate-to-endplate connections during muscle reinnervation were shown to arise from terminal axons or the node of Ranvier. The nodal sprouts, whether branching in the nerve bundle or from single extensions, may in part coincide with the well-investigated collateral nodal sprouts, the hallmarks of partial denervation (Brown et al., 1980), i.e., a response of motoneurons to develop an enlarged terminal field in response to nerve injury. The relationship of nodal sprouting in aged motor endplates with the age-related motoneuronal dysfunction has been stressed (Caccia et al., 1979; Larsson and Edström, 1986; Kawabuchi et al., 1995). We have shown that the multiple axon-multiple NMJ relationship usually converts to the multiple axon-single NMJ type, suggesting elimination of redundant innervation. Mid to late in reinnervation, some aberrant changes of endplate-to-endplate connections in aged adults consisted of persistent multiple axon-multiple NMJ relationship, as well as single axonal projections with poorly regenerated terminals. This pattern of muscle reinnervation suggests a reduced motoneuronal capacity in compensatory collateral sprouting. (2) When regenerating nerve fibers continue to grow toward their targets, some misrouting occurs (Bernstein and Guth, 1961; Brushart and Mesulam, 1980; Brown and Hardman, 1987; Thomas, 1989). Occasional fairly long, meandering paths and undetermined courses taken by the SC strands or axons, as well as irregularity in the caliber along the sprouted preterminal region, were characteristic features of aging. Derangement in the guidance mechanism of regenerating nerves might account for this continuing unordered formation processes involved in neurite extension and termination. It is generally accepted that a crush-type injury does not disrupt the perineurial connective tissue sheath and the SC basal lamina which direct regenerating axons back to their denervated site. It is possible that malformation and poor growth of the SC sheaths may affect the guidance mechanism mediated by the basal lamina tubes. However, the causes of aging-related deficits appear to be complex, and the relationship between the observed abnormalities and reduced rate of axonal elongation due to changes in the axonal transport system (Black and Lasek, 1979; Tanaka and Webster, 1991; Vaughan, 1992), as well as disturbances in local control mediated by the reciprocal interactions of regenerating axons and the SC surface (Hall and Sanes, 1993; Reynolds and Woolf, 1993; Langenfeld-Oster et al., 1994; Scherer and Salzer, 1996) deserves further study. (3) Derangement in the spatiotemporal relationship between nerve and muscle is a likely source of age-related differences observed in NMJs. First, with aging, the trophic support for neurons begins to decrease and thus a reduction of neuronal support with aging may account for the more pronounced atrophy of aged muscle fibers following denervation. Second, we found persistent shrinkage and damage of AChR sites in regenerating NMJs, concomitant with the long-lasting degeneration changes in TSCs and terminal axons. The rate of recovery in the AChR site was slowed compared to that of the TSCs. This data suggests that the long-lasting lessened spatial availability of the contact region delineated by the AChR site might affect coordination of successive steps in synaptic differentiation (Hall and Sanes, 1993; Burden, 1998) or target availability (Guth, 1956; Engh et al., 1971; reviewed by Bowe et al., 1989 and Fawcett and Keynes, 1990; Vaughan, 1990). This is consistent with some aspects of the involvement of age-related neurotrophic disorders on reinnervation and regeneration in NMJs, as postulated for the maintenance and survival of the regenerating nerves during aging (Hefti et al., 1989; Johnson et al., 1991; Johnson et al., 1995). It should be noted that, on the other hand, the neurons and TSCs at this stage of an animal's age are capable of some compensatory responses to peripheral nerve injury.
All procedures were approved by the Committee of Ethics on Animal Experiments in the Faculty of Medicine, Kyushu University. The criteria for the age of adult and aged adult, as well as the procedures for nerve crushing, were the same as previously reported (Kawabuchi et al., 1998). Briefly, thirty male adult (4 months old; body weight, 315 ± 19g) and forty-six aged adult (24 months old; body weight, 535 ± 55g) rats (Wistar/Ms) were used. We checked the differences in the rates of reinnervation and nerve regeneration between rats with and without aging-related symptoms. Eight aged rats, that were sick or morbid, were excluded from the study. The rest of the aged rats (38 rats) were classified into two groups, one devoid of any clinical symptoms (12 rats), and the other displaying two types of physical abnormalities (26 rats), i.e., (1) gait disturbance and an obvious atrophy of hindlimb muscle (4 rats), and (2) clumsiness of the hindlimb or stride at a slow pace without muscle atrophy (22 rats). No regenerating nerve fibers in the muscular nerve bundles of the group (1) cohort, even at 8 weeks postcrush (wpc), were detected in a preliminary examination. There was no discernible difference in the patterns of nerve regeneration, whether or not aged animals displayed clumsiness of the hindlimb or had a slow stride. It is considered that these symptoms mainly derive from inactivity or nutritional factors in laboratory animals that have been kept long-term in cages, as suggested by Cardasis and Padykula (1981). The present report refers to the symptom-free (12 rats) and the group (2) cohort (22 rats) to follow up the time courses of aging-related changes.
Surgery and Tissue Preparation
The distance that nerve fibers must regenerate between the site of the crush injury and that of the entry of the deep peroneal nerve into the tibialis anterior muscle was designed to be equivalent (20 mm) in the two age groups. The proximal measurements between the nerve-injured region and the lateral edge of the pelvic bone were 34.2 ± 3.2 (mean ± standard deviation) in adult (n = 5) and 35.5 ± 2.4 mm in aged adult (n = 7). After deeply anesthetizing the animals with ether followed by sodium pentobabiturate (50 mg/kg body weight, subcutaneous injections), the right sciatic nerve was exposed and crushed for 20 sec using microforceps. After nerve crushing was achieved at the maximum pressure, the wound was closed and animals were allowed to recover. At the time of perfusion the leg muscles operated on became smaller than the unoperated ones in all cases in the study. A 15–25% reduction in muscle weights at 4 wpc was noted.
The post-crushing survival periods in the two age cohorts were as follows: 1, 3, 4, 6, 8, and 12 weeks. The contralateral limb served as control. Six each of the 30 adult rats and the 34 aged adult rats were used as sham-operated (n = 3) controls, where the nerve was exposed but not damaged, as well as unoperated (n = 3) controls. The tibialis anterior muscles from 3–5 rats from each of the adult and aged adult groups were examined every day post crushing. Following anesthesia with ether and sodium pentobabiturate, the animals were perfused through the aorta with phosphate buffer saline (PBS), pH 7.6, followed by 300–500 ml of fixative composed of 4% paraformaldehyde in 0.1 M phosphate buffer at pH 7.4. The muscles were dissected out and were then immersed in graded sucrose solution of up to 20% in 0.1 M phosphate buffer for cryoprotection. All the samples were immediately frozen in isopentane cooled with solid CO2 and stored in a sealed plastic tube maintained at −80°C until use. Free floating sections were used in all experiments, because in preliminary tests the staining intensity was found to be greater in these than in dry mounted sections. Fifty μm longitudinal sections were cut on a freezing microtome (Frigocut, Germany) at −20°C and were collected in PBS. Usually a series of 5 sections per muscle were collected from 3 animals. The sections were taken exclusively from the middle core of the muscle, where the distribution of nerve bundles and nerve terminals across the long axis of the muscle, as well as the degree of reinnervation, were almost identical at any level of the sections.
Sections were rinsed twice in PBS and in 0.4% Triton-X 100 in PBS, respectively, and incubated in 1% bovine serum albumin (BSA) in PBS supplemented with 0.05% NaN3 for 1 hr. The primary antibodies applied to the sections for 24∽48 hr were as follows: polyclonal anti-PGP9.5 (Ultraclone) diluted 1:1600∽2000, and polyclonal anti-S100 (Nichirei) diluted 1:40. For detection of bound antibodies, the following procedures were used: incubation with biotinylated secondary antisera (goat antirabbit-IgG), dilution 1:200 (Vector Labs, Burlingame, CA), followed by streptavidin labeled with Texas red, dilution 1:200 (Vector Labs), for 2 hr. Between each step, sections were washed three times for 15–30 min in PBS. All procedures were performed at 4°C. After washing in several changes of PBS, sections were placed on a slide pretreated with 0.1% Poly-L-Lysine (Sigma Chemical Company, St. Louis, MO) to prevent contracture of the muscle section. Following immunostaining, the geometry of the postsynaptic AChR site was defined by α-bungarotoxin (α-BT) staining. Labeling of this site was achieved by 4–6 hr incubation at room temperature with FITC-conjugated α-BT (dilution 1:50, Molecular Probes, Eugene, OR). Sections were washed for 1–2 hr in several changes of PBS and mounted with Vectashield (Vector).
Confocal Laser Scanning Microscopy
Counterstained materials were observed with a confocal laser scanning microscopy system (CLSM-GB 200 Olympus, Japan) equipped with an argon/ krypton ion laser allowing simultaneous scanning of two fluorescent dyes. For observations of single-stained materials of either FITC or Texas red, the wave length of the excitation laser light was restricted to 488 nm for FITC and to 568 nm for Texas red by switching the dicrhoic mirrors in order to avoid cross talk. For simultaneous observations of the materials double stained by FITC and Texas red, both 488 and 568 nm laser lights were used at the same time for excitation. In these cases we carefully verified in advance that no detectable cross talk signals from the other fluorescence dye could be recognized in each of the experiments, especially for observations of double positive sites. Optical sections at intervals of 0.5 to 1μm were projected on a single plane extending for 10 to 40 μ m in thickness. Using a 40X or 60X water immersion objective lens, the “three-dimensional ” organization of the motor endplate can be analyzed by reconstructing the series of images.
To combine the data from different animals, it was determined that at least 15 NMJs per section were required for statistical reliability to ensure that inter-animal variability was not significant. So for quantitative analysis of each of the three dimensionally reconstructed junctional components, one 50 μm section with the necessary number of NMJs was selected per animal. Three sections from 3–5 animals from both age cohorts were processed for morphometry using the NIH image program: (1) Age changes of endplate size at 1, 4, 8 and 12 wpc, as determined by the area of S100-like immunoreactive (S100-LI) TSCs, PGP9.5-like immunoreactive (PGP9.5-LI) terminal axons, or α-BT-stained AChR sites were quantified. The point of origin for a motor endplate delineated by S100 immunostaining was identified as a local cluster of several round or oval cells. The area of an axon terminal included the projection area of the terminal axon and its branches at the AChR site, and the point of origin was defined as the first branch point of the terminal axon. To avoid errors of statistical reliability, the motor endplates with undefined borders, due to distant neighbouring groups of TSCs or terminal axons, were not counted. Further, motor endplates lying en face in the major plane of the section were selected for measurement. The original size of the planar area was determined from an automated procedure on the basis of the length of the scale on the microphotographs. (2) The Texas red-stained S100 or PGP9.5 immunolabeling is normally red and becomes yellow when it overlies sites of green FITC fluorescence in AChR staining. Following measurement of the area in the yellow- and green-color images, the ratio of overlap at 1, 4, 8 and 12 wpc was estimated by the mean percent of the yellow image divided by the green one. The degrees of overlap at each time point were classified into three groups (0%, < 50%, > 50%). The pseudocolor images were converted to greyscale images, and an intensity threshold on the greyscale image, sufficient to allow precise measurement, was defined at the site of the overlap. The extent of the greyscale image was always referenced to that of the original color image (yellow or green). (3) Age-related differences in the extent of process extension at 3 and 8 wpc were quantitatively assessed by determining the length of the processes extending out from the TSC. Any extensions with more than 100 μm length were not counted. Measurements were expressed as mean μm ± standard deviations. (4) Examinations of the cross-sectional area, diameter and perimeter of muscle fibers were made for the contralateral controls and muscle preparations at 4, 8 and 12 wpc.
Morphometric analyses of the time courses of the degree of overlap and the sizes of individual endplate components were performed on original images independently by two investigators (M.K. and Z.C.) and their results were compared to evaluate accuracy.
For the areas of TSCs, axon terminals and AChR sites and the cross-sectional sizes of muscle fibers, mean and standard deviations were calculated. Two forms of Student's t-test were then utilized to compare the means of each parameter in both age cohorts. Age-related differences in the degree of the size reduction in TSCs, terminal axons, AChR sites and muscle fibers at 4, 8, and/or 12 wpc were compared with a chi-square test. For both t-tests and chi-square tests, differences between the two values were considered significant if the probability value (P) was found to be less than 0.05.
We thank Mr. Takaaki Kanemaru (Morphology Core, Faculty of Medicine, Kyushu University) and Yasuhiro Hirakawa for their help in preparing photomicrographs. The English usage in this manuscript was revised by Mr. Kerry Greer (Edanz Editing Co., Japan). The Ministry of Education, Culture and Science of Japan grants were given to M.K.