Evidence for the subsynaptic zone as a preferential site for CHRN recycling at neuromuscular junctions

ABSTRACT

Vertebrate skeletal muscle contraction is mediated by nicotinic acetylcholine receptors (CHRN). Endocytosis and recycling of CHRN regulate their proper abundance at nerve-muscle synapses, i.e. neuromuscular junctions. Recent work showed that RAB5 is essential for CHRN endocytosis. Here, using in vivo-imaging of endocytosed CHRN and RAB-GFP fusion proteins, we deliver evidence for differential effects of RAB5-GFP, RAB4-GFP, and RAB11-GFP on CHRN endocytosis. Furthermore, while newly endocytosed CHRN colocalized with RAB5-GFP over large stretches of muscle fibers, RAB4-GFP and RAB11-GFP colocalized with endocytosed CHRN almost exclusively at neuromuscular junctions. In agreement with previous findings, this data suggests the existence of a specialized subsynaptic zone that is particularly relevant for CHRN recycling.

Neuromuscular junctions (NMJ) are synapses connecting motor neurons and skeletal muscle fibers in vertebrates. While a motor neuron can innervate several fibers, each fiber has one NMJ in adult muscle. NMJs, sized few tens of microns, comprise minute proportions of syncytial muscle fibers, which can be some centimeters long. To control voluntary movements, nicotinic acetylcholine receptors (CHRN) in NMJs are activated by acetylcholine, which is released from the motor neuron. CHRN are heteropentameric ligand-gated ion channels that accumulate at the NMJ postsynaptic membrane in band-like arrays at densities of > 10,000 µm⁻².¹ Specialized subsynaptic or fundamental nuclei express mRNAs for these junctional CHRN,² but some extrajunctional CHRN are found throughout the fiber as well.^1,3,4 To maintain constant postsynaptic CHRN densities, biosynthesis and degradation of receptors need to be balanced. Consequently, besides an activity-dependent tuning of CHRN gene expression,^5-7 the amount of CHRN at the NMJ is modulated by protein trafficking processes similar to those of other ion channels, e.g. AMPA receptors.^8,9 Upon endocytosis, CHRN can follow 2 different routes: recycling or degradation. While the degradation of CHRN uses macroautophagy and is regulated by the E3-ligase TRIM63,^10,11 recruitment of recycling CHRN to the synapse uses myosin V motor proteins,^12-14 and regulatory pathways, involving PKA,^13,15,16 PKC,¹⁶ and CaMK II.^17,18 CHRN trafficking can be studied using α-bungarotoxin (BGT), which binds CHRN in a highly selective and irreversible manner.²⁰ Injected into live animals, BGT labels surface-exposed postsynaptic CHRN and allows to study endo/lysosomal and recycling processes of CHRN.^19,21,22 Imaging of CHRN labeled with BGT-AlexaFluor647 (BGT-AF647) in live mouse tibialis anterior (TA) muscles, showed a steady-state amount of ∼9 endocytic CHRN vesicles per NMJ.²² Overexpression of a GTPase-deficient hyperactive Q79L mutant of RAB5, a major regulator of early endosomal steps,²³ amplified the number of CHRN puncta to ∼520.²² Together with the findings that RAB5^Q79L-GFP colocalized to 80% with CHRN puncta, that RAB5-GFP colocalized to CHRN puncta and endogenous RAB5 co-precipitated with CHRN,²² this strongly suggested a role of RAB5 in the endocytosis of CHRN.

To get deeper insights into the involvement of RAB5 on CHRN endocytosis, we here first compared the effects of RAB5 overexpression on CHRN endocytic carriers after different time periods, i.e., 1 h and 24 h of CHRN labeling. Since BGT needs several minutes to diffuse in the muscle and to enrich at NMJs, the 1 h injection time point served as acute labeling of CHRN. Conversely, at the 24 h time point, CHRN were marked for several hours at the moment of imaging. In the absence of exogenous RAB5, the number of BGT-positive vesicles after 24 h of BGT labeling was previously shown to be 9 ± 1.²² Here, we quantified BGT-positive vesicles after 1 h of BGT labeling using the same approach, i.e., in vivo-imaging. This yielded several 5 ± 1 (mean ± SEM, n = 6 muscles). To test the effect of RAB5 overexpression on CHRN endocytosis, TA muscles were transfected with RAB5-GFP and imaged 10 d later in vivo. CHRN were labeled with BGT-AF647 either 1 h or 24 h before imaging. Notably, in the presence of RAB5-GFP, the number of BGT-positive puncta augmented to 121 ± 15 and 31 ± 5 (both mean ± SEM, n = 4 muscles) for the 1 h and 24 h time points, respectively (Fig. 1A-C). Colocalization of CHRN puncta with RAB5-GFP was similar at both time points (Fig. 1D). To test, if the increase in BGT-positive puncta in the presence of RAB5-GFP was due to unspecific activation of endocytosis, we transfected muscles with RAB5-GFP and then, after 10 days, double-labeled muscles with 2 different fluorescent BGT-conjugates. A first injection with red fluorescent BGT-AF555 was used to saturate all BGT binding sites. One hour later, infrared fluorescent BGT-AF647 was injected and another hour later in vivo-imaging was performed. This showed a high amount of BGT-AF555 positive vesicles (82 ± 4; mean ± SEM, n = 4 muscles; Fig. S1), comparable to that observed before in the single labeling experiment. Conversely, the number of “unspecific” BGT-AF647 puncta was low with 5 ± 2 (mean ± SEM, n = 4 muscles), suggesting that RAB5 overexpression did not induce unspecific endocytosis.

Figure 1.

Overexpression of RAB5 enhances endocytosis of newly labeled CHRN. TA muscles were transfected with RAB5-GFP. In vivo-imaging of transfected muscles was performed 10 d later. CHRN were labeled with BGT-AF647 either 1 h (A) or 24 h (B) before microscopy. (A-B) Upper panels, maximum-z projections of confocal images from representative RAB5-GFP positive fibers. Scale bars, 10 µm. Lower panels, details of single confocal sections from boxed regions in upper images. Filled and empty arrowheads, CHRN-containing vesicles positive and negative for RAB5-GFP, respectively. Upper CHRN panels are shown without contrast enhancement. Overlays and lower panels were contrast enhanced to better visualize CHRN-positive vesicles. In overlay and overlay zoom pictures, BGT-AF647 and GFP signals are displayed in red and green, respectively. (C-D) Quantitative analysis of CHRN puncta per NMJ (C) and the fraction of these vesicles that were also GFP-positive (D). Mean ± SEM (n = 4 muscles each; *** P < 0.001; 3,054 vesicles from 41 NMJs were analyzed; n.s., not significantly different).

Next, we investigated the effects on CHRN endocytosis of RABs involved in receptor recycling, specifically RAB4 and RAB11.^25-27 Therefore, RAB4-GFP or RAB11-GFP were transfected into TA muscles, and CHRN were labeled with BGT-AF647 either 1 h or 24 h before in vivo-microscopy. Similar to RAB5-GFP (Fig. 1A and B), also RAB4-GFP and RAB11-GFP were enriched in punctate clusters at the NMJ region (Fig. 2A and B). Yet, numbers of CHRN puncta per NMJ were unchanged, both, between RAB4-GFP and RAB11-GFP and between the different time points of CHRN labeling. For simplicity, values of the 1 h and 24 h time points were pooled. In the presence of RAB4-GFP and RAB11-GFP, 26 ± 4 (mean ± SEM, n = 13 muscles) and 19 ± 3 (mean ± SEM, n = 11 muscles) CHRN puncta per NMJ were found (Fig. 2E). Colocalization rates of CHRN puncta with RAB4-GFP and RAB11-GFP were 45% ± 4% and 58% ± 4% (both mean ± SEM, n as for CHRN numbers), respectively (Fig. 2F). To verify that the observed distributions of RAB-GFPs and their effects on vesicle numbers and colocalization were not artifacts of GFP overexpression, we performed experiments in the presence of GDP-locked dominant-negative versions of RAB4 (RAB4^S22N)²⁸ or RAB11 (RAB11^S25N).²⁹ These showed completely different distributions compared with their wildtype counterparts: GFP fluorescence was evenly distributed throughout the cytoplasm (Fig. 2C and D). Additionally, RAB11^S25N-GFP accumulated in myonuclei (Fig. 2D). Colocalization of the dominant-negative RAB-GFP mutants with CHRN puncta was almost zero (Fig. 2F). Though, CHRN vesicle numbers were not significantly altered in the presence of the mutants as compared with corresponding wildtype RAB-GFP versions (Fig. 2E).

Figure 2.

Overexpression of RAB4-GFP and RAB11-GFP and of their dominant-negative mutants do not modulate CHRN vesicle numbers. TA muscles were transfected with GFP constructs fused to RAB4 (A), RAB11 (B), RAB4^S22N (C), or RAB11^S25N (D). In vivo-imaging of transfected muscles was performed 10 d later. CHRN were labeled with BGT-AF647 either 1 h or 24 h before microscopy. (A-D) Upper panels, maximum-z projections of confocal images from representative GFP-positive fibers. Scale bars, 10 µm. Lower panels, details of single confocal sections from boxed regions in upper images. Filled and empty arrowheads, CHRN-containing vesicles positive and negative for GFP, respectively. Upper CHRN panels are shown without contrast enhancement. Overlays and lower panels were contrast enhanced to better visualize CHRN-positive vesicles. In overlay and overlay zoom pictures, BGT-AF647 and GFP signals are displayed in red and green, respectively. (E-F) Quantitative analysis of CHRN puncta per NMJ (E) and the fraction of these vesicles that were also GFP-positive (F). Mean ± SEM (n = 13, 11, 6, and 6 muscles RAB4, RAB11, RAB4^S22N, and RAB11^S25N, respectively; * P < 0.05, *** P < 0.001; n.s., not significantly different; 4,784 vesicles from 219 NMJs were analyzed).

Previous reports demonstrated that RAB5, RAB4, and RAB11 cooperate at different levels of the endocytic/recycling pathways,^30,31 and occupy distinct domains on endocytic/recycling carriers.³² However, those observations were mostly made in cultured cells and rarely in live animal tissue. To understand how vesicles containing CHRN colocalize with RABs in their in vivo-environment, we here looked in more detail on the relationship between BGT-AF647 and RAB-GFP fluorescence signals. All analyzed RAB-GFPs either showed point-to-point colocalization with BGT-signals or cap-like arrangements. In the latter, RAB-GFP signals were just in partial coincidence with BGT puncta (see lower panels in Fig. 3A and B). Additionally, RAB11-GFP often completely surrounded groups of BGT puncta, mostly close to (< 5 µm) the NMJ (see lower 2 panels in Fig. 3C).

Figure 3.

GFP fusion proteins of RAB4, RAB5, and RAB11 show either point-to-point colocalization with CHRN puncta or they mark domains on these carriers. TA muscles were transfected with GFP constructs fused to RAB4 (A), RAB5 (B), or RAB11 (C). In vivo-imaging of transfected muscles was performed 10 d later. CHRN were labeled with BGT-AF647 1 h before microscopy. Panels depict details of single confocal sections from representative GFP-positive fibers. Filled and empty arrowheads, CHRN-containing vesicles positive and negative for GFP, respectively. All panels were contrast enhanced to better visualize CHRN-positive vesicles. Overlays, BGT-AF647 and GFP signals in red and green, respectively. Large saturated regions in CHRN channel and overlay pictures show parts of adjacent NMJs. Scale bars, 2 µm.

As mentioned, RAB4-GFP (Fig. 2A), RAB5-GFP (Fig. 1A and B), and RAB11-GFP (Fig. 2B) accumulated in the immediate vicinity of the NMJ (see additional Fig. 4A for RAB4-GFP). To further address their specific functions in CHRN endocytosis and recycling, we performed a detailed quantitative analysis of the correlation between RAB-GFP and CHRN puncta distribution. This revealed a differential enrichment of CHRN carriers throughout the fiber, depending on the overexpressed RAB protein. While RAB4-GFP and RAB11-GFP induced a concentration of BGT-positive vesicles in the vicinity of the NMJ, RAB5-GFP led to a more even distribution of CHRN vesicles along the fiber (Fig. 4B). Further, while for RAB4-GFP and RAB11-GFP > 50% of CHRN carriers colocalizing with the respective RAB-GFP were found in < 5 µm distance from the NMJ (left panels in Fig. 4C), this value was reached for RAB5-GFP only in a perimeter of > 20 µm (upper right panel in Fig. 4C). A gradient of RAB5-GFP positive CHRN-containing vesicles was absent when BGT was injected 1 h before imaging (lower right panel in Fig. 4C).

Figure 4.

Domains for colocalization of CHRN carriers with RAB5-GFP on the one hand and RAB4-GFP and RAB11-GFP on the other hand do not perfectly overlap. TA muscles were transfected with GFP constructs fused to RAB4, RAB5, or RAB11. In vivo-imaging of transfected muscles was performed 10 d later. CHRN were labeled with BGT-AF647 either 1 h or 24 h before microscopy. (A) Maximum-z projections of confocal images from a representative RAB4-GFP positive fiber. Scale bar, 50 µm. Filled and empty arrowheads, CHRN-containing vesicles positive and negative for GFP, respectively. All panels were contrast enhanced to better visualize CHRN-positive vesicles. Overlay, BGT-AF647 and GFP signals in red and green, respectively. (B-C) Quantitative analysis of the spatial distribution of CHRN puncta per NMJ (B) and the average numbers of CHRN carriers colocalizing with either RAB4, RAB5, or RAB11 (as indicated) as a function of distance from the NMJ (C). In B, all data taken with 1 h and 24 h labeled CHRN were pooled for the different RAB members. In C, data for 1 h and 24 h with RAB5 were separated. Mean ± SEM (in B: n = 13, 8, and 11 muscles for RAB4, RAB5, and RAB11, respectively; in C: n = 13, 4, 4, and 11 muscles for RAB4, RAB5–1h, RAB5–24h, and RAB11, respectively. * P < 0.05, *** P < 0.001; n.s., not significantly different; 5,837 vesicles from 260 NMJs were analyzed).

Finally, since endosome maturation typically involves a switch from RAB5 to RAB7,²⁴ we studied if CHRN puncta colocalize with RAB7-GFP. Therefore, we performed in vivo-imaging of BGT-labeled vesicles in the presence of RAB7-GFP. TA muscles were transfected with RAB7-GFP, injected with BGT-AF647 9 d later and then imaged on day 10. This showed a partial colocalization between BGT-positive puncta and RAB7-GFP, mostly in form of capping or surrounding structures (Fig. S2).

Taken together, this study addressed the involvement of RAB5, RAB7, RAB4, and RAB11 in CHRN endocytic/lysosomal processing and recycling. Interestingly, the effect of RAB5-GFP on the number of CHRN endocytic puncta was much stronger when BGT-labeling occurred 1 h before imaging as compared with the 24 h time point (Fig. 1). This was not due to different RAB5-GFP expression levels at these time points, since GFP-fluorescence intensity was similar in both cases. Given that BGT labeling 1 h before imaging is likely to mark both newly arrived and older receptors, while visualization after 24 h of BGT injection will only reveal the fate of older CHRN, these results suggest that RAB5-GFP exhibited a differential activity for distinct CHRN receptor populations. Indeed, earlier in vivo-radioiodine studies showed that, depending on muscle activity status, CHRN can assume at least 3 different half-life times, i.e., 1 d, 8 d, or 13 d.^5,33-35 In normal NMJs, most CHRN exhibit the 13 d half-life, and ∼15–20% of CHRN show the 1-d kinetics.^5,35 The 8-d kinetics is typical for receptors that were synthesized and reached the postsynapse in the innervated state, but then immediately after BGT-pulse labeling the muscle was denervated (termed “acute denervation”). This demonstrates that distinct CHRN populations with different recycling and lifetime behaviors exist on the postsynapse, being consistent with the possibility of a selective activity of RAB5 for such populations. The fact, that at the 24 h time point only 31 CHRN puncta per NMJ were found while there were 121 puncta at the 1 h time point, suggests that many CHRN in these vesicles were either recycled to the postsynapse or degraded. Bruneau et al. found that within a 24 h period there is a loss of roughly 40% of pre-existing CHRN. In the same period, 20% of CHRN recycle and 20% of CHRN are newly made to replenish the NMJ.¹⁹ Accordingly, one can assume that 20% of CHRN are degraded within a 24 h period, also fitting to data from BGT-radioiodine approaches.^5,35 Additionally, RAB5 overexpression was found to enhance CHRN degradation rate,²² and homotypic fusion of endocytic vesicles²⁴ likely contributed to a further reduction in the amount of BGT-positive puncta after 24 h.

In-depth comparison of BGT-positive vesicle distributions in the presence of heterologous RAB4, RAB5, and RAB11 (Fig. 4) indicated that the subsynaptic region represents a domain of preferential interaction of RAB4 and RAB11 with CHRN, while RAB5 colocalizes with CHRN carriers throughout the fiber. In combination with the understanding of RAB5 as a general endocytosis orchestrator and RAB4 and RAB11 as organizers of vesicle recycling, this data are consistent with a specialization of the subsynaptic region for CHRN recycling. Conversely, endocytosis of CHRN might occur all along the fiber. These assumptions are in agreement with previous reports on the enrichment of other components of the CHRN recycling machinery at the NMJ,^{12,13,15-18,36,37} and consolidate the concept of the subsynaptic region as a specialized zone of muscle fibers, where synapse-relevant molecules, such as CHRN, experience a special treatment from biogenesis over function to recycling or degradation.

Material and methods

Animals and transfection

Use and care of animals were as approved by German authorities and according to national law (TierSchG7). Experiments used adult C57BL/10J mice kept at the local animal facility. Animals were anaesthetized using either intraperitoneal (i.p.) injection of Xylavet® 20 mg/ml (cp-pharma) and Zoletil® 100 (Laboratoires Virbac) or inhalation of Isoflurane (cp-pharma, AP/DRUGS/220/96). For transfection, cDNA encoding heterologous fusion proteins (10 µg) was injected into TA, followed by electroporation as described in detail previously.^38,39

CHRN labeling and expression plasmids

For in vivo-microscopy of transfected TA muscles, these constructs were used:

RAB4-GFP, RAB5-GFP, RAB11-GFP (all kindly provided by Dr. Walter Volknandt, Frankfurt, Germany), and EGFP-RAB7A (Addgene, 28047). RAB4^S22N and RAB11^S25N were cloned into pEGFP-C3 backbone. Using overlapping PCR mutagenesis, serine 22 (RAB4^S22N) or serine 25 (RAB11^S25N) residues were mutated to asparagine using the following primers: RAB4^S22N forward primer TTA TTG GAA ATG CAG GAA CTG GCA AAA ATT GCT TAC T, RAB4^S22N reverse primer TCA ATA AAC TGA TGA AGT AAG CAA TTT TTG CCA GTT C, RAB11^S25N forward primer GGA GAT TCT GGT GTT GGA AAG AAC AAT CTC TTG TCT CGA TTT ACT C, RAB11^S25N reverse primer GAG TAA ATC GAG ACA AGA GAT TGT TCT TTC CAA CAC CAG AAT CTC C. Labeling of endocytic CHRN was performed by injecting 31 pmoles of BGT-AF647 (Life Technologies, B35450) either 24 h or 1 h before confocal microscopy. In Fig. S1, the same amount of BGT-AF555 (Life Technologies, B35451) was injected additionally 1 h before BGT-AF647.

In vivo-microscopy, data analysis and statistics

In vivo-microscopy was performed 10 d after transfection of TA muscles with a DMRE TCS SP2 confocal microscope equipped with Leica Confocal Software 2.61, a KrAr laser (488 nm), a HeNe laser (633 nm), and a HCX PL APO 63x/1.20 W CORR CS water immersion objective (all Leica Microsystems, Mannheim, Germany). Images were taken at 8-bit and 1024 × 1024-pixel resolution with either 1 x or 2 x electronic zoom. Fluorescence was elicited at 488 nm and 633 nm wavelengths for GFP and BGT-AF647, respectively. Emission signals used the SP settings 500–570 nm and 650–750 nm for GFP and BGT-AF647, respectively. Image analysis used ImageJ software (NIH, Bethesda, MD). Quantitative analysis of vesicle numbers and colocalization was performed as described.²² Briefly, BGT-AF647 image stacks were crosstalk-subtracted and screened for CHRN signals, yielding the amount of CHRN-positive structures. Each CHRN-positive structure was marked by a circular region of interest (ROI) using the ImageJ “Oval” tool. ROIs were imported into ImageJ “ROI manager” and screened in the GFP-channel for colocalization. Distances of CHRN-positive puncta to the NMJ were measured by applying the shortest straight line from the signal to the outer edge of the NMJ fluorescence signal using ImageJ “Straight” tool. Numeric data were handled using Microsoft Excel 2010. Data shown in graphs represent mean ± SEM. Significance was tested using Student's t-test or Welch test when applicable. To this end, Kolmogorov-Smirnow-test for normal distribution and ANOVA F-test for homo/heteroscedasticity were performed. Significance levels are indicated in all respective figures. For data compilation, Adobe Illustrator (Adobe Systems Software) and ImageJ were used.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.