Regulation of the Intermediate Filament Protein Nestin at Rodent Neuromuscular Junctions by Innervation and Activity

At neuromuscular junctions, nestin is confined to the postsynaptic apparatus

We investigated the localization of nestin immunoreactivity in nerves and muscles of the rat and mouse using four different anti-nestin antibodies. One of these, the monoclonal antibody 4E2, was typically used for labeling rat muscle; this antibody appears to recognize a truncated nestin protein (see Materials and Methods). As reported previously for 4E2 (Astrow et al., 1992, 1994) and in agreement with work of others (Carlsson et al., 1999; Vaittinen et al., 1999), we find that immunoreactivity for nestin is concentrated in the postsynaptic apparatus at the nmj and at the myotendinous junction.

To map more precisely the location of nestin immunoreactivity at the nmj, we imaged junctions both en face (Fig. 1A) and from the side (Fig. 1E) using confocal microscopy and high numerical aperture objectives. Similar results were obtained with the antibodies in mouse and in rat. Viewed en face, nestin immunoreactivity presented a negative image of the pretzel-like pattern of AChRs concentrated in synaptic gutters underneath the branches of the nerve terminal. The negative image results from the prominent nestin staining in the muscle cytoplasm located between these synaptic gutters (Fig. 1A, arrowhead) as well as at the edges of these gutters (Fig. 1D, arrowhead). A gap of ∼1 μm with a paucity of nestin label was present between the label for AChR and an intense band of nestin at the edges of each gutter (Fig. 1A–D; arrowhead in D). Nestin immunoreactivity is also located beneath the synaptic gutters (Fig. 1C). Here the staining has the appearance of bands and more intense puncta. The nestin bands generally run parallel to the bands of AChR (Fig. 1B). These AChR bands represent a diffraction-limited view of the openings of the secondary (junctional) folds that are present in the sacrolemma underneath the nerve terminal (Marques et al., 2000). The puncta of nestin staining are commonly situated between the bands of AChR labeling (Fig. 1, compare B, C), i.e., in the position of the protrusions of muscle cytoplasm between the secondary folds, the so-called “interfolds” (Couteaux and Spacek, 1988). Last, nestin extends away from the synaptic structures to muscle nuclei located at the synapses (Fig. 1A, arrow) and to the Z-lines located at the muscle sacrolemma, structures known as costameres (cf. Ervasti, 2003). This latter labeling vanishes within a few micrometers of the junction, either along the long axis of muscle fiber or into the depth of the fiber.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

A–H, Confocal images of nmjs viewed en face (A–D) and from the side (E–H) showing the location of nestin immunoreactivity. A, nmj from the extensor digitorum longus muscle of an adult rat (maximum projection of z-stack) labeled with fluorescent bungarotoxin (red), the anti-nestin antibody mAb 4E2, and a fluorescein-secondary antibody (green). Nestin immunoreactivity surrounds AChRs and extends along sarcomeric, Z-lines confined to the vicinity of the synapse and the surface of the muscle fiber. Nestin also surrounds junctional nuclei (one indicated with arrow in A). Nestin localizes prominently to the edges of the gutters containing AChR and fills many of the spaces between adjacent gutters (arrowhead in A). B–D, Higher-magnification, maximum projections of a portion of the synaptic gutter in an adult rat soleus muscle. Rhodamine bungarotoxin (white in B, red in D) shows the bands of receptors indicating the position of the mouths of the secondary folds of the junctional membrane. The fluorescein label for nestin (white in C, green in D) labels structures beneath the synaptic gutter (as determined by examining sequential images in the z-stacks). At the edge of the synaptic gutter, nestin immunoreactivity is displaced from the receptors by ∼1 μm (arrowhead in D), suggesting that a space exists between the receptors present near the tops of the folds and most of the nestin network. E, Image (maximum projection of 2 slices from a z-stack) of an nmj on the side of a muscle fiber labeled with Cy5 bungarotoxin (colored blue in image), the mAb 4E2 anti-nestin antibody (green), and a polyclonal antibody to ankyrin (red), a protein known to be concentrated near the bottoms of the synaptic folds. The blue, red, and green appear in primarily nonoverlapping bands at the synapse. F–H, Separate views of the three labels in the left portion of E. Fiduciary marks (+ symbols) were placed in the same positions in E–H; examination of these marks shows the ankyrin label (G) is interposed between the receptors (F) and the band of nestin label (H). Depressions in the side views of the AChR in E indicate cross-sections of the primary synaptic gutters (arrows). The nestin immunoreactivity surrounds a nucleus (arrowhead). Note that, although the most intense labeling for nestin is at the bottoms of the secondary folds, weaker nestin labeling does extend as periodic protrusions from this more intense labeling to near the AChR. This is most easily seen in the inset in E, showing the nestin label in a single, fortuitous z-section of this junction whose brightness and contrast have been adjusted so that the stronger nestin labeling beneath the gutters is saturated. Small arrows identify these protrusions into the layer of AChR. Scale bar: (in E) A, 9.4 μm; B–D, 5.3 μm; E, 2.5 μm.

Images of junctions lying on the edge of a muscle fiber (Fig. 1E–H) and thus viewed from the side confirmed the localization of staining obtained from en face views. The gap between AChR and nestin seen at the edge of the synaptic gutters was also present here. Thus, the gap outlines the entire gutter. To investigate whether the gap might be occupied by the secondary, junctional folds, double immunolabeling was performed with an antibody to sodium channels or to the ankyrin protein known to anchor these channels to the cytoskeleton. Both antibodies are known to label the bottoms and sides of junctional folds (Flucher and Daniels, 1989). The ankyrin label occupies the gap between the AChR and the intense nestin label (Fig. 1E–H). Similar results were obtained with an antibody to sodium channels (data not shown). Thus, an intense band of nestin labeling exists immediately beneath the junctional folds. Fingers of labeling are seen to penetrate toward the synaptic cleft (Fig. 1E, arrows in inset), suggesting that nestin immunoreactivity penetrates into the cytoplasm between the junctional folds. These findings suggest that most of the nestin immunoreactivity is not directly associated with the postsynaptic receptors, although, as suggested previously on the basis of electron microscopy, IFs may make attachments to the postsynaptic thickening at the top of the junctional folds (Ellisman et al., 1976). Together, these observations show that nestin is intimately associated with structural elements that define the synaptic contact. Nestin filaments are positioned so that they could support a number of features of the synapse, including junctional folding, the anchoring of synaptic nuclei, and the anchoring of the synaptic site to the membrane cytoskeleton and its attachments to the basal lamina.

Nestin is present in myelinating but not terminal Schwann cells in innervated muscle

As reported previously (Friedman et al., 1990), nestin immunoreactivity was found in myelinating Schwann cells (mSCs) (Fig. 2B,C). Images of small bundles of axons teased from the sciatic nerve or of intramuscular nerve branches showed that this immunolabeling was strong near the cell bodies of mSCs and at paranodes. Nestin immunolabeling was also present in ribbons, known as Cajal bands, located along internodes (Fig. 2B, arrowheads). Labeling with antibodies against both nestin and a general Schwann cell cytoplasmic marker, the protein S100B, showed that these nestin ribbons were contained within the thicker cytoplasmic processes of SCs located on the abaxonal surface of these internodes (Fig. 2B,C).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Nestin immunoreactivity is present in myelinating Schwann cells of the nerve but is not detectable in terminal Schwann cells that cover the neuromuscular junction. A–C, Image from a teased rat sciatic nerve. D, E, Image of a neuromuscular junction from an adult rat soleus muscle. A, D, Labeling with antibody to the SC marker S100B. B, E, Labeling with antibody 4E2. C, F, Color images merged from A,B and D,E, respectively, and including, in addition in blue, the nuclear marker 4′,6′-diamidino-2-phenylindole (C) and α-bungarotoxin-labeled AChRs (F). Asterisks indicate cell nuclei, the arrow in C indicates a node of Ranvier, and arrowheads in B indicate bundles of nestin filaments located within S100-labeled thickenings of SC cytoplasm on the abaxonal surface of the internode. Scale bar: A–C, 5 μm; D–F, 7.5 μm.

Despite the presence of nestin in mSCs, no immunoreactivity was detected in the nonmyelinating, tSCs that cover the nerve terminals at the nmj (Fig. 2D–F).

Nestin immunoreactivity is restricted postsynaptically to neuromuscular junctions during early postnatal development, and denervation arrests this restriction

To examine how nestin immunoreactivity appears at junctions during development, immunolabeling was performed using the 4E2 antibody on rat muscles during early postnatal development. Immunoreactivity was present in the center of each muscle fiber at postnatal day 0 (P0) as well as at the tendons, but, unlike the distribution in adult muscles, labeling extended longitudinally along the fibers for a considerable distance from the plaques of AChRs (supplemental Fig. 2A, available at www.jneurosci.org as supplemental material). In these fibers, there was no obvious preferential localization to the junction. Indeed, in contrast to adult muscles, the locations of junctions could not be reliably identified by the nestin label, and, very often, the brightest label was at sites near but in no obvious relationship to the junctions. In addition labeling was obvious in thin, secondary myotubes that were growing longitudinally in each direction toward the tendons; labeling was particularly strong at the tapered ends of these developing fibers as well as sites in which other fibers had come in contact with the tendons. Later in postnatal development (P10), the labeling in the center of the muscle became confined to regions closer to the AChR plaque (supplemental Fig. 2B, available at www.jneurosci.org as supplemental material). By P21, the labeling resembled that reported above in adult muscles (data not shown). If muscles were denervated at the time of birth and examined 2 weeks later, the distribution of nestin immunoreactivity vis-à-vis the AChR plaques was like that present at the time of denervation (supplemental Fig. 2C, available at www.jneurosci.org as supplemental material), i.e., all along the muscle fibers, showing that the localization of nestin to junctions during development requires innervation.

In adult rodents, denervation causes a rapid reduction of nestin immunoreactivity in the postsynaptic muscle fiber, its appearance in terminal Schwann cells, and its upregulation in Schwann cells of the nerve

In confirmation of previous results using the 4E2 antibody (Astrow et al., 1994), we found that denervation caused loss of nestin immunoreactivity from the postsynaptic muscle fiber and its appearance in tSCs. Postsynaptic immunoreactivity like that seen in Figure 1 rapidly diminishes in denervated rat muscles and is almost undetectable in most muscle fibers by day 3 (Fig. 3). Similar observations were made in mouse muscles in which the decline in postsynaptic immunoreactivity was obvious but less pronounced than in the rat. We confirmed these results using the three other nestin antibodies (data not shown). This contrasts to tSCs, which change from immunonegative (Fig. 2D–F) to strikingly immunopositive over the same period of time. We previously used this immunoreactivity with the antibody 4E2 to visualize the processes extended by SCs after denervation (Son and Thompson, 1995a,b). The mSCs along the intramuscular nerves that were weakly labeled before denervation become more strongly immunopositive after denervation (data not shown). As the junctions become reinnervated, the normal distribution of nestin immunoreactivity is restored (Astrow et al., 1994) (Fig. 4). An absence of nestin expression in tSCs was seen at neonatal junctions, and these neonatal cells also became immunoreactive after denervation (supplemental Fig. 4, available at www.jneurosci.org as supplemental material). Thus, unlike the muscle expression, nestin expression in tSCs mirrors the adult pattern from the time of birth.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

After denervation, nestin immunoreactivity disappears from the postsynaptic apparatus in rat muscle but appears in terminal Schwann cells. Maximum projection of confocal images of the nmj of an adult rat soleus muscle denervated by resection of the sciatic nerve 3 d earlier. The immunoreactivity for nestin (4E2) is present in the processes of tSCs, as determined by colabeling of the SC marker S100B (data not shown); the cell bodies of the tSCs are only weakly labeled. Immunoreactivity in the postsynaptic muscle fiber is very weak or absent, in contrast to innervated fibers (Fig. 1). Scale bar, 10 μm.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Denervation causes expression of a GFP reporter controlled by the nestin neural enhancer in terminal Schwann cells that persists until the nerve regenerates. A–F, Wide-field fluorescence images were captured vitally from the same, small region of the surface of the sternomastoid muscle before denervation (Day 0, A, B) and 5 d (Day 5, C, D) and 33 d (Day 33, E, F) after crush of the muscle nerve. Other experiments show the nerve regenerates to this muscle within 10 d after crush. At each time point, an image was collected of the AChRs labeled with a nonblocking concentration of rhodamine bungarotoxin (A, C, E) and an image of the GFP fluorescence in the same field (B, D, F). Arrow marks an endplate shown at higher magnification in the insets in both the AChR and nestin–GFP panels. Before denervation, there is little GFP fluorescence, even postsynaptically in muscle fibers. Within 5 d of denervation, GFP is expressed in SCs above the junctional sites as well as in the intramuscular nerves (out of focus in background of D). In contrast to the nestin seen by immunostaining (Fig. 3), the soluble GFP fills the cell bodies and nuclei of the SCs. The expression of GFP is greatly diminished by day 33, ∼3 weeks after this region of the muscle is reinnervated. Scale bars: 200 μm; insets, 20 μm.

Expression of GFP under the control of the nestin promoter and neural enhancer can be used to vitally observe the reaction of terminal Schwann cells to denervation

To examine the possibility that a mouse line bearing a transgene containing the nestin promoter and intronic neural enhancer coupled to the coding sequence for soluble GFP (Mignone et al., 2004) might label muscle fibers and/or their SCs, we conducted repeated, vital imaging of junctions in the sternomastoid muscle of these mice using techniques developed by Lichtman et al. (1987). We identified nmjs by labeling with a low, nonblocking concentration of rhodamine–bungarotoxin (cf. Rich and Lichtman, 1989; Akaaboune et al., 1999). Before denervation, we could detect no GFP in mSCs, tSCs, or muscle fibers (Fig. 4B). Only rare, unidentified cells in the connective tissue in the muscle were GFP positive (data not shown). However, if we reimaged muscles within 2 d of denervation, bright GFP fluorescence was present in both mSCs and tSCs and the processes they extended over and beyond the denervated junction (Fig. 4D). The number of junctions with GFP-positive cells and the intensity of their label increased over the next day (data not shown). The GFP label persisted throughout the period of denervation but then generally disappeared as nerve fibers regenerated and reinnervated the old synaptic sites (Fig. 4F). High-magnification examination of such junctions in vivo showed that the processes extended by tSCs during denervation could be easily identified (Fig. 5B,D). When we compared the immunostaining of these processes with anti-nestin antibody and the fluorescence from GFP, we saw that the same processes were labeled but that the GFP label extended slightly (∼5 μm) beyond the labeling with the nestin antibody, suggesting that cytoplasmic GFP fills the tips of processes beyond the placement of nestin filaments (supplemental Fig. 3, available at www.jneurosci.org as supplemental material). Moreover, the vital imaging revealed that tSC processes are quite dynamic. Within an interval as short as 24 h, some processes disappeared, others had grown distances of up to 75 μm, and new processes had appeared (Fig. 5D). After denervation, bright GFP fluorescence also appeared in the mSCs contained within the endoneurial tubes leading to each endplate as well as in the intramuscular nerves (Figs. 4D, 5B,D, asterisks). These results suggest that the so-called neural enhancer present in the nestin gene (Zimmerman et al., 1994) drives expression in SCs after the removal of axons. Because the GFP expression reports the activity of the nestin enhancer elements, these observations also suggest that the regulation of nestin expression in tSCs occurs at the transcriptional level.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

The dynamics of processes formed by terminal Schwann cells after denervation can be vitally imaged in mice that have a nestin enhancer–GFP transgene. A–D, Vital images acquired using a rhodamine filter set to image AChR (A, C) and a GFP filter set (B, D). The same denervated junction was imaged 2 d (A, B) and 3 d (C, D) after section of the muscle nerve. The axon innervating this particular junction arrived over the pathway of an endoneurial tube whose mSCs express GFP (asterisk in B, D). An unidentified cell is present in each GFP image (arrowhead). Although the tSCs remain associated with the AChRs at each time, they also grow processes away from the endplate. A process extended before day 2 (arrow in B) has grown an additional 75 μm by day 3 (arrow in D); a process that was absent at day 2 has grown from the synaptic site by day 3, extending ∼75 μm (double arrowhead in D); a branch of a process at day 2 (+ symbol in B) is absent at day 3. Scale bar, 12.5 μm.

Although GFP was absent from the tSCs at junctions of neonatal nestin–GFP mice, it was induced after denervation, showing a change in transcriptional activity in SCs at this earlier developmental stage like that in the adult (supplemental Fig. 4, asterisks, available at www.jneurosci.org as supplemental material). This GFP induction occurred before the death of these cells that follows neonatal denervation (Trachtenberg and Thompson, 1996).

Absence of GFP expression in innervated muscle in nestin transgenic mice shows that control elements other than the neural enhancer dictate postsynaptic expression

There was no GFP evident in the postsynaptic apparatus of the nestin–GFP transgenic mice either before or after denervation. However, because GFP is a soluble protein, any postsynaptic expression might be diluted by the large cytoplasmic volume of the muscle fiber. To test this possibility, we conducted experiments using a second transgenic mouse with an identical promoter and intronic sequences but whose reporter was CFP with a nuclear localization signal. When we imaged normally innervated muscles from this mouse, we saw no nuclear labeling except for rare elongated nuclei that appeared to be apposed to the surface of muscle fibers (data not shown). No labeling was seen at nmjs or in intramuscular nerves. However, after denervation, CFP-labeled nuclei were found in these two locations (supplemental Fig. 5, asterisks, available at www.jneurosci.org as supplemental material). These nuclei belong to SCs because they are located within cells labeled by an antibody to an SC marker (the protein S100B). Thus, just as in the case of the cytoplasmic GFP reporter mice, there is an upregulation of expression of a neural enhancer-driven, nuclear-localized CFP reporter in SCs after denervation. The absence of expression of both the cytoplasmic and nuclear reporter postsynaptically in muscle before and after denervation suggest that the neural enhancer present in intron 2 of the nestin gene is responsible for nestin expression in denervated tSCs but not for its expression postsynaptically in muscle fibers.

Nestin mRNA is localized to the postsynaptic apparatus in muscle fibers

The motor nerve terminal promotes the transcription of a number of genes important for the construction of the synapse from the subsynaptic myonuclei. These “synapse-specific” genes are transcribed far less, if at all, in nuclei elsewhere in the muscle (Sanes and Lichtman, 2001). Such synapse-specific gene expression is thought to be controlled by local signals arising either directly or indirectly from the nerve terminal. To determine whether nestin transcription in each muscle fiber is controlled in a synapse-specific manner, we performed in situ hybridization on rat muscles, using probes for nestin mRNA. A strong signal was obtained from muscle fibers at their nmjs (Fig. 6B). Little or no hybridization was detected outside the synaptic zone, and the hybridization resembles that present at nmjs when probes for the α subunit of the AChR are used (Fig. 6C). This result provides direct evidence (in addition to that provided by expression arrays) that nestin mRNA is confined to the nmj (Kishi et al., 2005; McGeachie et al., 2005; Nazarian et al., 2005) and that nestin is transcribed in a synapse-specific manner.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Nestin mRNA is localized to the neuromuscular junction in innervated muscles and to Schwann cells after denervation. Cross-sections of the tibialis anterior muscles from an adult rat. A, Sections stained for cholinesterase (AChE) to mark synaptic sites. B, An adjacent section to that in A processed for in situ hybridization using a digoxigenin-labeled antisense riboprobe for nestin. Nestin mRNA (black arrowhead in B) is concentrated at a synaptic site marked by cholinesterase staining (white arrowhead in A). Sense probes yielded no signal (data not shown). C, Another section probed for AChRα. The location of the label in C, known to be in the muscle fiber, resembles that in B (compare arrowheads in each). D, E, Sections of an innervated (D) and an 8-d-denervated (E) muscle processed as above and probed for nestin. Intramuscular nerves, indicated by white arrows, have labeled cells in the denervated (E) but not the innervated (D) muscle. Scale bar, 50 μm.

In situ hybridization with a probe for nestin mRNA was also performed at nmjs after denervation. Signal was present, but it was not possible to judge whether the signal in the postsynaptic muscle fiber was decreased (data not shown), because the results from the immunostaining and from the GFP transgenic mice predict that expression is increased in the adjoining tSCs. That SCs upregulate nestin mRNA expression is suggested by comparison of the hybridization signal within the intramuscular nerves of denervated but not innervated intramuscular nerves (Fig. 6D,E).

Blockade of transmitter release from the junction upregulates nestin expression in Schwann cells and downregulates this expression in muscle fibers

The localization of nestin mRNA to the postsynaptic region and the loss of immunoreactivity for nestin protein after denervation suggest a special role for the nerve in regulating nestin expression. To examine the possibility that nerve-evoked activity plays a role in this expression, botulinum toxin was applied topically to the soleus muscle in adult rats and to sternomastoid muscles in mice. Muscles were examined at the time of the terminal experiment by stimulation of the muscle nerve to verify a deficiency in the ability of the motor axons to evoke muscle twitches. When muscles were examined after 3–4 d of blockade, no substantial changes in immunoreactivity in the postsynaptic apparatus or in tSCs were seen. However, after 8–10 d of blockade of rat soleus, immunoreactivity in the postsynaptic apparatus was diminished (67% of junctions in three muscles; data not shown), and immunoreactivity appeared in tSCs and the processes these tSCs extended away from the synaptic sites (63% of the junctions in three muscles; data not shown). Similar findings were made in the sternomastoid muscles of adult mice probed with a polyclonal antibody to nestin. These findings suggest that transmitter release from the nerve terminal plays a role in maintaining the postsynaptic expression of nestin in muscle and in suppressing it in tSCs.

Stimulation of denervated muscle maintains postsynaptic nestin immunoreactivity but does not prevent the upregulation of nestin expression in terminal Schwann cells

Many of the effects of denervation on muscle gene expression are mediated via the electrical/contractile activity imparted to the fiber by its innervation (Lømo, 1980). To test whether the postsynaptic localization of nestin might be influenced by muscle activity, we denervated rat hindlimb muscles and, at the time of denervation, began extrinsic stimulation of the soleus muscle with currents administered via electrodes placed at the tendon at each end of the muscle. Stimulation of muscles was able to preserve immunoreactivity for nestin at the former synaptic sites (Fig. 7, compare C, D) for periods of denervation of up to 17 d, the longest periods tested. The same period of denervation resulted in marked diminution in nestin immunoreactivity in muscles that were either not stimulated or that received sham stimulation. We conclude that muscle activity maintains the postsynaptic nestin protein. It may do so by either promoting transcription at the synapse or altering the turnover of nestin protein. In contrast, imposing this activity on denervated muscles did not alter the immunoreactivity appearing in tSCs (Fig. 7D), nor did it apparently alter their extension of processes, as has been reported previously (Love et al., 2003).

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Stimulation of denervated muscle maintains postsynaptic immunoreactivity for nestin but does not affect its upregulation in terminal Schwann cells. A–D, Junctions in a rat soleus muscles 7 d after denervation by sciatic nerve resection were identified by labeling of AChR with rhodamine bungarotoxin (A, B) and immunostaining for nestin using the 4E2 antibody (C, D). Junction in A and C was from denervated muscle left unstimulated. Junction in B and D was from a muscle of a littermate that was similarly denervated but stimulated directly by implanted electrodes with 100 Hz trains lasting 1 s repeated every 100 s (average frequency, 1 Hz). Although both muscles have labeling in tSCs and the processes they extend from the denervated junction, the stimulated muscle retains postsynaptic labeling whereas the unstimulated muscle has lost it. Scale bar, 10 μm. E, Nestin labeling with mAb 4E2 in three rat soleus muscles from three different animals. Two of the muscles were denervated for 7 d, and one of these was stimulated for the same period of time. Seventy-five junctions in each muscle were scored for the presence of labeling only in tSCs (white bars), only in the postsynaptic apparatus underneath AChR (black bars), or in both (gray bars). Results shown here were typical of at least five cases in which stimulation was successfully applied for 7–14 d. In a few cases, stimulation was begun some days after loss of postsynaptic immunoreactivity in companion muscles. In these cases, immunoreactivity was restored to the postsynaptic area.