Time course of changes in the three cellular components of the NMJ after chronic denervation
We analyzed the long-term morphological alterations associated with irreversible denervation of the cranial LAL muscle. Besides its usefulness to study the specific clinical consequences associated with the denervation of head/neck muscles [24], the LAL muscle offers several experimental advantages, as it is a superficially exposed, flat, and thin cranial muscle, allowing repeated in vivo manipulation and microscopic observation of NMJs in whole-mount preparations [23, 25, 26]. We have recently refined a muscle denervation procedure to specifically target muscles from the cranial region, by resecting a 5-mm segment of the right posterior auricular nerve branch (Fig. 1a), while isolateral muscles from non-injured animals were used as control [23]. Denervated LAL muscles were dissected at different times after nerve resection and stained to reveal the three cellular components of the NMJ. Control NMJs exhibited full innervation of pretzel-like AChR aggregates and were completely covered by tSCs (Fig. 1b). In turn, at different times after nerve resection, motor axons degenerate (evidenced by decreased staining in motor terminals and motor axons), postsynaptic domains went through morphological modifications from a pretzel-like to a fragmented shape (Fig. 1b, arrows), and tSCs migrated towards the vicinity of denervated NMJs (Fig. 1b, arrowheads) and projected elongated cellular processes (Fig. 1b, empty arrowheads).
We first studied the formation of ectopic AChR clusters, a postsynaptic hallmark of NMJ denervation [27]. As expected, we found that the percentage of ectopic AChR aggregates increased since 30 days after denervation, representing around 40% of total AChR aggregates, significantly higher than 1.42 ± 0.87 % obtained in controls (Fig. 1c, d). The increasingly high proportion of ectopic AChR aggregates correlated with drastic alterations in the gross morphology of the postsynaptic domains (Fig. 1c), making it unfeasible to recognize previously innervated NMJs. Therefore, the next analyses in the chronic denervation model were performed at 7 and 30 days after nerve resection. Both time points were characterized by a marked decrease in the measurement of presynaptic area, which likely corresponded to small remnants of axon terminal debris (Fig. 1e) and subsequent endplate denervation. The postsynaptic domain was also subjected to morphological changes, as the quantification of the area of AChR aggregates (excluding ectopic aggregates) showed a progressive decrease, already detectable 7 days after damage (Fig. 1f). Consequently, the apposition between pre and postsynaptic domains showed a nearly null overlap after nerve damage at both time points (7 and 30 days) (Fig. 1g).
We next analyzed the behavior of tSCs after NMJ chronic denervation. tSCs were identified by their positive S100B staining, their distribution on or in the vicinity of NMJs, and by their characteristic shape [7, 28, 29]. Indeed, soon after facial nerve resection (7 days), tSCs projected cell processes and their somas migrate out from the synaptic region (Fig. 1h, arrowheads), which led to a significant decrease in the average number of tSC somas at the NMJ (1.60 ± 0.38) compared to controls (2.30 ± 0.15; **p < 0.01, one-way ANOVA) (Fig. 1i). At longer denervation times (30 days), the average number of tSCs was less than 1 per NMJ (Fig. 1i); consequently, the number of tSCs located within a 50 μm radius of NMJs showed a significant increase shortly after denervation (7 days); however, the number of tSC in the periphery of the NMJ exhibits a trend to decrease 30 days after injury (Fig. 1j).
Collectively, our findings reveal that the cranial LAL muscle exhibits morphological responses in the three cellular components of the neuromuscular synapse after chronic denervation, comparable to those described in hind-limb muscles.
Transient NMJ denervation alters the behavior of its cellular components
To study the behavior of the three NMJ cellular components upon reversible axonal damage, the right posterior auricular branch of the facial nerve was crushed for 30 s (Fig. 2a). Five days after nerve crush, most endplates were denervated (Fig. 2b, inset). Endplate re-innervation was detected 15 days after injury where reinnervated NMJs exhibited a similar shape that control muscles (Fig. 2b, inset). Quantification of presynaptic parameters showed that both nerve terminal perimeter and area were decreased to almost undetectable levels at the time previously described for Wallerian degeneration (5 days post-injury) (Fig. 2c, d) [5]. Interestingly, even though morphological parameters of the nerve terminal, such as perimeter (Fig. 2c) and area (Fig. 2d) recovered values since 30 days after injury, they remained significantly lower than control after reinnervation. As comparable long-lasting alterations were observed in the area of postsynaptic AChR aggregates (Fig. 2e), the simultaneous decrease in the size of both pre and postsynaptic domains gave rise to smaller NMJs than controls, allowing full NMJ reinnervation 30 days after nerve injury (Fig. 2f). Finally, to analyze if tSC behavior was altered as a consequence of short-term NMJ reinnervation, we quantified the number of tSCs per NMJ (Fig. 2g, h), the proportion of NMJs bearing tSC projections (Fig. 2g, i), and the proportion of NMJs having tSCs in the periphery (Fig. 2g, j). No significant alterations were found at the different times analyzed after reversible nerve injury compared to controls.
Based on previous findings showing that the postsynaptic organization is not significantly affected upon muscle denervation [5, 30], we next conducted detailed analyzes of the postsynaptic morphology at different times after reinnervation. The perimeter of AChR aggregates decreased immediately after nerve injury and recovered values similar to control after 2 weeks; however, a decrease was observed 3 months after injury (Fig. 3a). The endplate area decreased 5 days after damage to reach control values as soon as re-innervation was accomplished; similarly, the endplate perimeter transiently decreased but exhibited control values since 15 days post nerve injury (Fig. 3b, c). Other postsynaptic parameters, such as endplate diameter (Fig. 3d) and postsynaptic compactness (Fig. 3e), defined as the endplate area occupied by AChRs, were transiently and slightly decreased but exhibited control values 90 days post nerve injury. In sharp contrast with the degenerative model (Fig. 1d), even though few scatter ectopic AChR clusters were observed during early denervation, they were no longer detectable upon NMJ reinnervation, as described [27]. Altogether, our findings thus far reveal that some changes occurring upon NMJ denervation in the LAL muscle are rescued with short-term reinnervation, while others remain altered long after reinnervation.
The stability of the NMJ postsynaptic domain is negatively affected by nerve damage
We next sought to analyze the stability of AChRs within the endplate, as its rapid removal from the postsynaptic membrane is a hallmark of NMJ denervation [10]. With this aim, we followed an in vivo two-color BTX method [31] by which postsynaptic AChRs were labeled in vivo with a non-saturating dose of a fluorescently tagged BTX (BTX-1) and after dissection, the LAL muscles were labeled with a different fluorescently tagged BTX (BTX-2). Using confocal microscopy, AChR aggregates were categorized as “stable” if BTX-1 and BTX-2 labels were of similar intensity or as “unstable” if BTX-1 labeling was mainly absent and BTX-2 intensity was comparatively higher. Considering that after chronic denervation most postsynaptic domains lose their pretzel-like morphology and exhibit a fragmented morphology (Fig. 4a), and ectopic AChRs reach a significant proportion a month later (Fig. 1d), we performed these experiments in a time frame covering 15 days post nerve resection. In control muscles, as AChR removal from the postsynaptic membrane was around 50% of the total AChRs (Fig. 4b, Ctrl), most postsynaptic structures were classified as stable (Fig. 4c). As expected, the proportion of stable postsynaptic structures decreased after denervation, concomitantly with an increase of unstable structures (Fig. 4b, c). To perform similar analyses during NMJ regeneration, we first characterized the time point of NMJ reinnervation in the LAL muscle (Fig. 4d). We found that regenerated axons reached the LAL muscle by 7 days after nerve crush, while different degrees of partial NMJ reinnervation were observed in the period between 8 and 10 days after injury. At 11 days, most postsynaptic domains were fully reinnervated (Fig. 4d). Interestingly, our two-color BTX analyses showed a significantly higher proportion of unstable postsynaptic domains between 10 and 21 days after nerve injury (Fig. 4e, f). While AChR stability tended to recover 37 days after nerve crush injury, a slight but statistically significant biphasic behavior was observed after 60 days. Finally, at 90 days, the percentage of unstable and stable structures was comparable to uninjured control muscles (Fig. 4f). Our data reveal that postsynaptic stability is highly impaired upon chronic denervation of the LAL muscle. Interestingly, postsynaptic stability is not immediately recovered after NMJ reinnervation, as this parameter is rather delayed by a time period of several weeks and only recovers control values three months after damage.
Short-term denervation leads to a long-term adaption of the NMJ structure
As the organization of AChR aggregates within the endplate is altered in conditions affecting NMJ integrity and function, we next sought to analyze postsynaptic morphology after nerve damage. AChR clusters were categorized into mature pretzels (complex and highly branched shapes), collapsed (shrunk, non-branched structures), and fragmented pretzels (i.e., those having more than six distinctive AChRs aggregates or “fragments”) (Fig. 5a). Soon after facial nerve resection, collapsed structures (Fig. 5b; red bars) begin to appear, concomitant to a marked decrease of pretzel-like shapes (Fig. 5b; green bars), resulting in their absence from 45 days after nerve resection onwards. After 30 days, most postsynaptic domains become fragmented (Fig. 5b; gray bars). A detailed observation allowed us to identify and quantify two different organization patterns of fragmented postsynaptic structures that, according to previous evidence [32], were classified as (i) fragmented smooth, exhibiting discrete fragments with defined edges, and (ii) fragmented blurred, as those having multiple small fragments displaying diffuse edges (Fig. 5a). In chronically denervated NMJs, we found an evident prevalence of fragmented blurred morphologies from as early as 7 days (around 50%) to 30 days onwards (> 90%) (Fig. 5c). To complement the idea that different forms of NMJ fragmentation arise from NMJ degeneration, we analyzed AChR stability in both types of fragmented structures (Fig. 5d). We found that fragmented blurred structures are unstable, evidenced by a strong reduction of BTX-1 detection compensated by higher BTX-2 staining (Fig. 5e).
In the NMJ reinnervation paradigm, mature pretzels displayed a marked biphasic behavior, as their proportion decrease at 30 and 90 days after nerve crush injury, with partial recovery at 60 days (Fig. 5f; green bars). These changes mirrored an inverted biphasic behavior of fragmented structures (gray bars) (Fig. 5f). Remarkably, after a discrete period showing a similar proportion of both fragmented structures (15 days after nerve crush), we found that opposite to chronic denervation, the proportion of fragmented smooth structures become significantly higher upon NMJ reinnervation (Fig. 5g). As in the chronic denervation paradigm, fragmented blurred structures displayed higher instability than smooth fragmented ones (Fig. 5h, i). Together, these studies reveal the existence of different types of NMJ postsynaptic fragmentation, whose relative abundance likely correlates with their reinnervation potential.
Previous findings in hind-limb muscles showed transient NMJ poly-innervation during reinnervation [5]. To evaluate whether NMJ regeneration at the LAL muscle also exhibited this parameter, we used 3D projections of the z-stacks obtained from confocal microscopy to quantify mono- and poly-innervated NMJs, as 2D images often do not suffice to distinguish between these two types of innervation (an additional figure shows this in more detail (see Additional file 1) as well as additional movie files (see Additional files 2, 3, 4 and 5)). According to previous reports, we observed that early NMJ reinnervation was accompanied by poly-innervation, as 34.30 ± 13.01% (***p < 0.001, one-way ANOVA test) of NMJs were innervated by more than one motor axon at day 15 (Fig. 6a). Remarkably, a similar proportion of NMJ poly-innervation persists until 3 months after facial nerve crush (Fig. 6b). When we analyzed NMJ poly-innervation in the different postsynaptic morphologies, we found that the initial abundance of poly-innervated postsynaptic pretzel-like shapes (Fig. 6c; green bars) turned into an increased proportion of poly-innervated fragmented structures 30 and 90 days after nerve injury (Fig. 6c; gray bars). From these, poly-innervation was significantly higher in fragmented NMJs exhibiting smooth morphology 15 and 90 days after nerve injury (Fig. 6d; blue bars).
Altogether, our results indicate that some morphometric parameters of the NMJ, including AChR aggregates area, endplate diameter, and postsynaptic stability, are altered soon after denervation and remain altered long after NMJ reinnervation. Moreover, other morphometric parameters, including the fragmentation of postsynaptic structures and marked poly-innervation seem to remain irreversibly altered, suggesting that nerve injury leads to an adaptive reminiscence of the NMJ after damage.
Delayed functional recovery of the neuromuscular synaptic transmission after endplate reinnervation
To study how the sustained effects observed on both NMJ morphology and stability after denervation impact on NMJ functionality, we studied synaptic transmission at different time points after nerve crush injury in ex vivo nerve–muscle LAL preparations using intracellular electrophysiological recordings. We analyzed spontaneous miniature endplate potentials (mEPPs) and evoked endplate potentials (EPPs); we also obtained information on quantal content (QC, the number of quanta released per action potential) and possible changes in short-term plasticity (facilitation and depression) during repetitive stimuli. The average amplitude of mEPPs was increased 21 days after injury (Fig. 7a, b), while their frequency decreased at 21 days and increased at 90 days (Fig. 7a, c). The mean amplitude of EPPs was significantly reduced 15 days (34.64 ± 4.37 mV; ****p < 0.0001, Mann–Whitney test) and 21 days (51.23 ± 3.53 mV; ***p < 0.001, Mann–Whitney test) after nerve crush injury, compared to uninjured controls (85.21 ± 8.12 mV) (Fig. 7d, e). This evoked response recovered control values 2 months after injury (Fig. 7d, e), consistent with our observations that mice begin to recover ear movement one week after nerve crush. Similarly, the QC decreased 15 and 21 days after nerve damage but recovered 60 days after nerve injury (Fig. 7f).
No differences in paired-pulse facilitation (PPF) at frequencies ranging from 5 to 50 Hz were observed between control and muscles 15, 21, 60, and 90 days after denervation (Fig. 7g), indicating no differences in the probabilities of neurotransmitter release. In contrast, short-term depression of EPP amplitude was slightly but significantly increased shortly after NMJ reinnervation (Fig. 7h), suggesting a subjacent decline of the QC under continuous stimuli which, among other possibilities, could be explained by a decrease in the refilling of the readily releasable pool (RRP) of vesicles. This parameter was also slightly decreased 3 months after nerve injury (Fig. 7h).
In summary, our findings reveal that although degenerative and regenerative paradigms of nerve injury led to morphological and functional alterations of the NMJ, some of them are efficiently recovered after reinnervation (Fig. 8). Other parameters, including AChR aggregates area, nerve terminal area and perimeter, endplate diameter, and postsynaptic stability, were recovered only three months after nerve injury. Importantly, NMJ poly-innervation and fragmentation remained altered long after muscle reinnervation has been accomplished (Fig. 8). Despite the observed long-term consequences of short-term reinnervation, synaptic transmission at the NMJ is recovered to control levels 2 months after nerve crush injury, suggesting that the observed altered morphological features are part of adaptive mechanisms that take place during the regenerative process.