Mouse strains

C57BL/10ScSn-Dmd^mdx/J mice (mdx, RRID:IMSR_JAX:001801) were purchased from Jackson Laboratory and mated with transgenic mice expressing both a green fluorescent reporter expressed in Schwann cells, (B6:D2-Tg(S100B-EGFP)1Wjt/J, RRID:IMSR_JAX:005621) and a cyan fluorescent reporter expressed in motor axons (B6.Cg-Tg(Thy1-CFP)23Jrs/J, RRID:IMSR_JAX:003710). Use of these mice has been previously reported and all mice are available from Jackson Laboratories. mdx mice were crossed into fluorescent mice, and progeny back crossed into mdx for 8 generations to ensure a C57BL/10 isogenetic background in fluorescent mdx mice. 4–7 male mice per genotype and age were used for each experiment, for a total of 59 mice. Experimental procedures in the Formation and maintenance of neuromuscular synapses (2016–0158) protocol were approved by Texas A&M University IACUC.

In vivo two-color α-bungarotoxin

Mice were anesthetized via isoflurane inhalation and placed in a supine position. Neck hair was removed using a depilatory and the surgical site sterilized with povidone-iodine. An incision was made from chin to sternum and the skin retracted. The salivary glands were moved to the side to expose the STM. Surrounding connective tissue was gently removed and a non-saturating concentration (diluted 2 μg/mL in sterile lactated Ringers) of α-bungarotoxin conjugated to Alexa Fluor^™ 555 (ThermoFisher Scientific, Cat# 35451) applied to the surface of the muscles for 5 minutes (BTX-1). The neck cavity was then washed copiously with lactated Ringer’s ten times. For muscle injury experiments, superficial muscle fibers at areas close to the site of entry of the accessory nerve into the muscle were cut with a sterile #15 surgical scalpel. The contralateral muscle was also labeled with BTX-1, but not injured and used as an internal control (IC). For denervation experiments both STM muscles were labeled with BTX-1 and the accessory nerve of one muscle was severed near its insertion site into the muscle. The other muscle was spared and used as an IC. The salivary glands were moved back into place following labeling, washing, and injury. The skin was sutured with 8–0 silk and treated with 2% lidocaine hydrochloride jelly. Animals were allowed to convalesce for ten days post-surgery. Endpoints for the studies are labeled P38, P66, P160, and P450. P38 animals were 38 ± 3 days old, P66 animals were 66 ± 3 days old, P160 animals were P160 ± 3 days old, and P450 animals over 450 days old at the time of death. Mice were killed via I.P. injection of 0.15 mL of Euthasol. The STM was dissected, pinned at resting length to a sylgard dish and fixed using 4% phosphate buffered paraformaldehyde (PFA) for 20 minutes. After fixation the muscles were washed three times with phosphate buffered saline (PBS) and labeled with α-bungarotoxin conjugated to Alexa Fluor^™ 647 (ThermoFisher Scientific, Cat# 35450) for 5 minutes (BTX-2, diluted 2 μg/mL in PBS). DAPI (Thermo Fisher Scientific Cat# D3571, RRID:AB_2307445) was used to label nuclei (diluted 0.5 μg/mL in PBS). After three PBS washes, a longitudinal “filet” from across the surface of the muscle was dissected from distal tendon to proximal tendon and mounted on a microscope slide in anti-fade fluorescence mounting medium.

Image acquisition, and NMJ characterization and analysis

Junctions were counted and categorized using a Leica DMRX epifluorescence microscope and a 40X Oil objective (NA 1.4). Junctions were considered fragmented if they had 5 or more non-connected clusters of AChRs in close proximity to each other on the same myofiber. Junctions were categorized based of the brightness of BTX-1 and BTX-2 labels, as well as if they were continuous (< 5 AChR clusters), or fragmented (> 5 AChR clusters). They were classified as stable if BTX-1 and BTX-2 labels were both bright, or as dynamic if BTX-1 label was totally or mostly absent and BTX-2 label bright. In many cases dynamic junctions did not completely lose BTX-1, and small puncta of fluorescence was seen. It is likely that these puncta are BTX-1 labeled receptors tethered to the basal lamina and demarcate the initial synaptic site before myofiber degeneration. If both the BTX-1 and BTX-2 labels were faint, junctions were classified as lost, but this was rarely observed. In order to verify that dynamic junctions indeed lose their BTX-1 label, a subset of STMs were removed immediately following the initial exposure, at Day 0 without any convalescence. After dissection the Day 0 muscles were labeled with BTX-2. In all junctions BTX-1 and BTX-2 labels were bright and junctions were all categorized as stable. Characterization and counts were performed by multiple researchers on the same muscle preparation, and similar results obtained. 50 to 100 NMJs were characterized for each muscle preparation.

Confocal micrographs were collected on either a Leica TPS II SP5 or a Zeiss LSM780 microscope using a 40X Oil objective (NA 1.4). Step size was set between 300–500 nm. Laser lines 405 nm, 458 nm, 488 nm, 563 nm, and 633 nm were used as necessary. Images were analyzed in ImageJ (ImageJ, RRID:SCR_003070) by creating maximum intensity projection images from collected stacks. Junctional Area was calculated by thresholding the BTX-2 labeled channel using the Auto Threshold tool provided by ImageJ. A polygon shaped region of interest was then circumscribed around the thresholded signal and the region of interest’s area measured. Receptor Area was measured as the thresholded area of the BTX label within the circumscribed region of the AChR. Dispersion Index [11] was calculated as Receptor Area/Junctional Area. The number of AChR fragments (Fragmentation) was measured using the objects counter in ImageJ by setting the minimum area for counts as 5 μm². While this measurement does not consider fragments in separate z-planes, as they may be superimposed in our maximum intensity projections, it is a conservative estimate of fragmentation. The actual number of fragments for both WT and mdx is likely underestimated by this measurement. tSC counts were made by counting only fluorescently labeled Schwann cells in proximity to the endplate having coincident nuclear label via DAPI. Axon branch points were measured by creating 3D rotations of the confocal images of fluorescently labeled motor neurons, and then counting the number of times the axon splits. Small varicosities of the motor terminal were not counted as branches.

Canine use

Dogs were used and cared for in accordance with the National Research Council’s Guide for the Care and Use of Laboratory Animals and housed at Texas A&M University. Procedures used were approved by university IACUC protocol, Standard Operating Procedures-Canine X-Linked Muscular Dystrophy (2012–052 or 2015–0110). Ages of dogs ranged from 1 to 6 years. 3 male wild-type (WT) controls and 7 male GRMD dogs were used, for a total of 10 animals. The animals used were undergoing necropsy for other purposes and the tissues shared. These dogs were part of a colony owned and maintained by Texas A&M University used for research purposes.

Canine muscle preparations

Following euthanasia for non-related procedures via I.V. injection of Euthasol, the middle portion of cranial tibial muscles was dissected and placed in 4% PFA for 20 minutes. After fixation, the muscles were washed three times with PBS. Muscle fiber fascicles were dissected from the section and labeled with BTX conjugated to Alexa Fluor^™ 555 overnight in PBS (diluted 2 μg/mL). Labeled fascicles were washed in PBS three times and then examined under a fluorescence dissecting microscope to identify the endplate band. A small portion of the fascicle containing the endplate band was cut from the fascicle and embedded in a 3% agarose gel. 40 μm longitudinal sections (parallel to the long axis of muscle fibers) were prepared using a vibratome. These sections were blocked using standard blocking solution (0.2% bovine serum albumin, 0.3% Triton 100X, 0.1% sodium azide in PBS) and labeled via mouse monocolonal antibodies to neurofilament (SMI-31, Covance, RRID:AB_531793, diluted 4 μg/mL in standard blocking solution), synaptic vesicles (DSHB Cat# SV2, RRID:AB_2315387, diluted 4 μg/mL in standard blocking solution) and rabbit polyclonal antibodies to Schwann cells (S100, Dako Cat# Ga504, diluted 4 μg/mL in standard blocking solution) in PBS overnight. After three PBS washes, goat anti-rabbit secondary antibodies conjugated to AlexaFluor^™ 647 (Thermo Fisher Scientific Cat# A-21245, RRID:AB_2535813), and goat anti-mouse secondary antibodies conjugated to AlexaFluor^™ 488 (Thermo Fisher Scientific Cat# A-11001, RRID:AB_2534069) were applied for two hours in PBS (diluted 2.5 μg/mL in standard blocking solution), along with BTX conjugated to AlexaFluor^™ 555 (diluted 2 μg/mL in standard blocking solution). DAPI was used to label nuclei (diluted 0.5 μg/mL in standard blocking solution). Following three additional PBS washes, the sections were mounted on microscope slides using an anti-fade mounting medium.

Statistics

mdx mice undergo progressive fragmentation of their AChR rich endplate

Using fluorescence confocal microscopy, we found changes to NMJ morphology in mdx mice similar to those reported previously [9–11]. We imaged mdx muscles near the onset of necrosis (P38) and onward and these changes were present in the youngest animals examined, and older adult mdx animals show morphological differences to the younger mdx mice indicating disease progression. When labeled with fluorescently tagged α-bungarotoxin (BTX), a highly specific ligand for the AChRs, the receptors at most individual mdx NMJs were fragmented into discontinuous aggregates. This differs from the pretzel-like receptor aggregates of WT animals, which are arranged into continuous branches (Fig 1A). Junctions with this abnormal clustering of AChRs have been classified as fragmented into “islands” [9, 11, 18]. Maximum intensity projections of confocal stacks were evaluated for the number of AChR fragments they contained. mdx NMJs had more fragments than WT at all ages (Fig 1B. Fragmentation: 336% increase at P38 t(32) = 7.974, P < 0.0001, 77% increase at P66 t(75) = 4.756, P < 0.0001, and 445% increase at P450, t(19) = 7.186, P < 0.0001, t-tests). Notably, at P38 in mdx there was a population of both fragmented (56.7%) and continuous junctions. We think it likely, as others have postulated [10], that dystrophic junctions remodel from continuous to fragmented as the animals age, that this remodeling is stochastic in its occurrence, and that this process is irreversible. The irreversibility is suggested by two observations. First, the proportion of junctions that are fragmented increases as mdx mice age and does not fall. At early time points (P14), WT and mdx animals have roughly the same proportion of adult-like, continuous junctions [10] but from P66 on nearly all the AChR aggregates are fragmented in appearance in mdx. Additionally, the degree of fragmentation becomes more severe as dystrophic animals age, with the number of fragments increasing from 7.3 in P38 mdx animals to 13.0 in P450 mdx animals (Fig 1. t(90) = 5.460, P < 0.001, 2-Way ANOVA, Bonferonni post-hoc test). In WT animals no such increase from P38 to P450 was evident, and the average number of fragments at all ages was less than five.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. WT and mdx NMJs.

(A) Maximum intensity projection confocal images of P66 WT, P66 mdx and P450 mdx NMJs. Images from left to right: AChR, motor axon terminal arbor, tSCs, and composite images. In the composite image, AChRs are pseudocolored red, motor axons cyan, and tSCs green. Scale bar = 20 μm. “#” indicates examples of AChR fragments; arrowheads, examples of axon terminal branch points; asterisks cell bodies of tSCs. In P450 mdx, not all tSCs are marked. Some non-specific labeling via the S100 promoter of cells that cannot be positively identified as Schwann cells is seen. These cells are not labeled as tSCs in the panels nor included in measurements. (B) Junctional measurements in WT and mdx NMJs from P38 to P450 as described in “Image acquisition and analysis”. Asterisks above solid bars denote significance between genotypes (t-test, * P < 0.05, ** P < 0.001, *** P < 0.0001). 15 to 90 NMJs were compared for each measurement. Exact n can be found in text as t(df) = t-value. n = df + 2. n is pooled number of individual NMJs compared. Dashed lines denote significance between ages (2-Way ANOVA, Bonferroni post-hoc test, P < 0.05).

https://doi.org/10.1371/journal.pone.0205926.g001

mdx mice undergo progressive NMJ expansion and a dispersion of AChRs within the synaptic area

Fragmentation is one of the most striking postsynaptic changes in mdx NMJs, but other measurable differences are also evident. In order to approximate the size of the entire synaptic structure (i.e. the AChR aggregate, the motor axon terminal, and the tSCs) a polygon was drawn around the labeled receptor aggregate and its area was measured. This polygon included the totality of the receptors but also included areas between the receptor aggregates that were not labeled with BTX. Thus, the polygon included areas containing tSC somata and processes, and the motor axon terminals ramifying above and between the synaptic contacts, even if these entities were not fluorescently labeled or imaged. We designate this measurement the Junctional Area. At all ages, the Junctional Area was larger in mdx than in WT (Fig 1B. P38 t(36) = 6.581, P < 0.0001, P66 t(84) = 2.319, P = 0.0228, P450 t(19) = 2.537 P = 0.0201, t-tests). WT Junctional Area ranged from ~35% to ~80% of the mdx Junctional Area.

There was also an increase in Junctional Area in mdx from P38 to P450 (Fig 1B, t(90) = 5.089, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test), and P66 to P450 (Fig 1B, t(90) = 5.380, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test) that was not noted in WT animals. P450 mdx Junctional Area was about 2.5x that of P38 mdx.

An effort was made to measure just the area that AChRs occupied on the sarcolemma, not including the motor axon and tSCs. The Receptor Area was defined as the area within the Junctional Area that fluoresced intensely when labeled with BTX conjugated to an AlexaFluor^™ marker. mdx displayed increased Receptor Area from WT only at P38 and P450 (Fig 1B. P38, t(41) = 3.284, P = 0.0021 and P450, t(24) = 2.195 P = 0.0381, t-tests). At P38 mdx Receptor Area was 1.3x as large as P38 WT, while at P450 mdx Receptor Area was 1.5x as large as P450 WT. In WT endplates, there were no significant changes in Receptor Area throughout life. Interestingly the Receptor Area in mdx increased from ages P38 to P450 (Fig 1B. t(90) = 4.277, P < 0.001, 2-way ANOVA, Bonferroni post-hoc test). There was no increase in Receptor Area from P38 to P66 in mdx. It should be noted that Receptor Area is not a receptor density measurement. While the space the receptors occupy on the myofiber increases, the density may or may not change at the same time. We chose not to rely on fluorescence intensity measurements to estimate receptor density as this measurement varies with junctional position in the muscle preparation thickness.

We also investigated the spread of receptors within the synaptic area. To quantify this spread, the Receptor Area was divided by the Junctional Area and termed the Dispersion Index (DI), a measure previously used by Pratt and colleagues [11]. The lower this ratio the more spread there is of the receptor clusters within the junction. Junctions in the mdx animals were consistently more disperse, i.e. had a lower DI than age matched WT (Fig 1B. P38 t(34) = 5.926 P < 0.0001, P66 t(88) = 3.117, P = 0.0025, P450 t(38) = 5.034, P < 0.0001, t-tests). The DIs of mdx animals were ~70%-90% of WT. Interestingly the spread of the receptors within the junction at P450 mdx is 1.2x that of P38 mdx animals (t(90) = 2.264, P < 0.05, 2-Way ANOVA, Bonferroni post-hoc test). As with Fragmentation, this suggests that as mdx animals age an increasing number of NMJs remodel from the healthy continuous state to the large, fragmented dystrophic state. The results also suggest that individual NMJs can become increasingly remodeled with time. By P66 nearly all the mdx NMJs are remodeled, but significant differences from P66 to P450 can be seen in all the postsynaptic characteristics measured in mdx. It is likely that ongoing degeneration and regeneration of the STM myofibers is driving these progressive alterations.

mdx NMJs undergo a remodeling of the presynaptic apparatus

It has been established that mdx mice undergo bouts of myofiber necrosis and regeneration due to their high susceptibility to myofiber damage. Correspondingly, muscle fiber damage has been shown to rearrange the NMJ. Based on observations in vivo of deliberate muscle fiber damage, this is thought to occur through the motor axon terminal, which is normally directly apposed to the receptor islands that appear after regeneration of each damaged fiber. The motor terminal was found to extend processes along the regenerating muscle fiber and it was proposed that the motor neuron directs the insertion of AChRs into new contact sites in the sarcolemma, resulting in a larger and fragmented endplate with a more complex motor axon terminal arbor [19]. The motor terminal was found to extend processes along the regenerating muscle fiber and it was proposed that the motor neuron directs the insertion of AChRs into new contact sites in the sarcolemma, resulting in a larger and fragmented endplate with a more complex motor axon terminal arbor. It is probable this phenomenon is occurring in dystrophic mice as well. To investigate this possibility, measurements on the presynaptic apparatus of NMJs, namely the tSCs and motor axon terminal, were performed. mdx animals were crossed to transgenic mice that express eGFP driven by the S100 promoter in Schwann cells [20], and CFP in motor axons driven by the Thy1 promoter [21]. The number of motor axon branch points in proximity to the endplate was considered a measure of motor neuron terminal rearrangement. Branching pattern differences between mdx and WT animals were not significant except at P450 (Fig 1B. Branch Points, t(21) = 4.816 P < 0.0001, t-test). It was noted, however, that the gross morphology of dystrophic terminal arbors was visually distinct from WT. As the motor neuron contacts the muscle fiber at the endplate of healthy WT animals, branches of the terminal tend to run smoothly along the tracks formed by the AChR gutters. mdx axonal arbors were seen to have small, bulbous varicosities that made connections to the fragmented islands of AChRs (Fig 1A). These varicosities were not counted as branches because they are very short. The number of axonal branches did increase from 7.5 to 11.1 (Fig 1B. t(52) = 4.010, P < 0.001, 2-way ANOVA, Bonferroni post-hoc test) in mdx mice from P38 to P450. This trend was not apparent in WT mice, consistent with previous reports that normal junctions can grow and shrink without changing their branching pattern [22].

The number of tSCs was counted for both mdx and WT junctions based on GFP expressed in these cells and coincident nuclear DAPI label (S1 Fig). Dystrophic endplates consistently had more tSCs than age matched WT endplates (Fig 1A and 1B. Terminal Schwann Cells, P38, t(8) = 2.880 P = 0.0205, P66 t(33) = 3.655, P = 0.009, P450 t(21) = 6.711, P < 0.0001, t-tests). In mdx the number of tSCs increased with age, from P38 to P450 (Fig 1B. t(45) = 3.441, P < 0.01, 2-Way ANOVA, Bonferroni post-hoc test). The phenomenon was not apparent in WT mice. On average WT P38 animals had 2.8 tSCs per NMJ, while P450 had 3.1. In contrast, at the same ages, mdx NMJs had on average 5.0 tSCs at P38 and 7.9 at P450. These findings fit well with observations that tSCs are added to the NMJ during the process of junction growth during development [23]. During periods of little to no growth in healthy adult mice no tSCs are added. However, during the expansion of the endplate from P38 to P450 in mdx mice, as suggested by increased Junctional Area, there is also an expected addition of tSCs.

GRMD dogs display similar changes in NMJ morphology to those in mdx mice

mdx mice have a mild phenotype and near-normal lifespan compared to humans afflicted with DMD. Therapies that are effective in treating dystrophic mice have not translated well to humans in clinical settings [24, 25]. We, therefore, sought to determine if the morphological changes seen in mdx mice also occur in GRMD, a canine model more clinically relevant to DMD [26]. We harvested the midsection of cranial tibial muscles in the pelvic limb of adult GRMD and WT dogs at necropsy performed for other non-related experiments. AChRs, Schwann cells, and motor neurons were fluorescently labeled via immunofluorescence and BTX. When imaged using confocal microscopy, changes to the NMJ like those in mdx mice were found. The postsynaptic AChR aggregate, labeled via fluorescent BTX, was clearly abnormal (Fig 2A). The same NMJ measurements made on murine synapses were also made for canine preparations. The NMJs of GRMD dogs were 1.8x as large as their WT counterparts (Fig 2B. Junctional Area, t(383) = 6.627, P < 0.0001, t-test). The spread of receptors in GRMD NMJs is 1.6x that of WT (Fig 2B. Dispersion Index, t(383) = 13.72, P < 0.0001, t-test). Finally, the number of fragments of AChRs was higher in dystrophic dogs than in WT (Fig 2B. Fragmentation, t(383) = 10.59, P < 0.0001, t-test). WT animals had an average of 1.6 fragments per NMJ, and GRMD 5.8. No significant difference was seen in the Receptor Area of the two groups. This result is similar to P66 mice and may represent a homeostatic event in affected animals that compensates for remodeling of their NMJs at least at the ages examined.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. GRMD NMJ morphology.

(A) Confocal maximum intensity projection of adult WT and GRMD NMJs harvested from the cranial tibial muscle. Composite images pseudocolored red-AChR, cyan-motor axon, green-tSC. Scale bar = 20 μm. Note the difference in magnification between WT and GRMD. (B) Junctional measurements in WT and GRMD NMJs as described in “Image acquisition and analysis”. Asterisks denote significant differences (t-test, P < 0.0001). Exact n can be found in text as t(df) = t-value. n = df + 2. n is number of pooled individual NMJs compared.

https://doi.org/10.1371/journal.pone.0205926.g002

When the presynaptic apparatus of the NMJ was investigated, no differences in the number of motor axon branches were noted (Fig 2A and 2B, t-test), similar to the case of younger mdx reported above. In some preparations, it was evident that GRMD axon terminals also synapsed on the myofiber in varicosities (S2 Fig) similar to that of mdx. However, this was not apparent in all images, and is likely due to incomplete labeling using the antibodies to synaptic vesicles and neurofilament. Consistently weaker labeling was observed in the canine preparations than those obtained in mice. It is possible that older GRMD dogs would show increased changes like those in the P450 mdx mice, but this was not investigated. Finally, there was no difference in the number of tSCs between GRMD and WT dogs, but again immunofluorescent labeling was the method used in dog preparations. In the mouse investigations, the use of transgenic fluorescent proteins in the Schwann cells and the motor neurons allowed for more faithful labeling. Regardless, the similarities between mdx and GRMD NMJs suggest that remodeling occurs in all mammals that lack dystrophin. This includes boys with DMD. What exactly is causing these changes has not been satisfactorily investigated, however. Using the mdx mouse, we were able to investigate if myofiber degeneration and regeneration were indeed causative in remodeling.

AChR replacement is elevated in mdx mice throughout life

We investigated whether the dynamics of the AChRs could provide evidence of how the NMJ becomes remodeled. The rationale behind the approach is that loss of receptors from the myofiber membrane should indicate recent degeneration of the myofiber [27]. As the muscle fiber undergoes necrosis, the AChRs on its surface would also be destroyed during breakdown of the sarcolemma. Upon subsequent regeneration of the myofiber, new AChRs would have to be inserted into the membrane to allow the resumption of neuromuscular transmission. If a loss of receptors is coincident with abnormal endplate morphology, this would suggest that necrosis and regeneration at the endplate spurs morphological changes to the NMJ. To investigate this possibility, the AChRs of STMs were labeled with a sub-saturating dose of fluorescently tagged BTX (BTX-1) in live mdx and WT mice (Fig 3B). The concentration and exposure time have been shown previously to not affect neuromuscular transmission or NMJ morphology [28]. Following a 10-day recovery period (endpoints of P38, P66, P160, and P450) that allows for the occurrence of any ongoing myofiber degeneration and regeneration, the labeled muscle was dissected from the animal and labeled with a second spectral variant of fluorescently tagged BTX (BTX-2). This method is termed the in vivo two-color BTX method. Using epifluorescence microscopy, the junctions were then categorized based on both their morphology (fragmented or continuous) and the presence of each of the two spectral labels. If BTX-1 was lost and BTX-2 was bright, junctions were categorized as dynamic. If the brightness of the BTX-1 label was similar to BTX-2, junctions were categorized as stable (Fig 3A and 3B). If both BTX-1 and BTX-2 were weak, the junction was categorized as lost. It has been noted that there is rarely a complete removal of the BTX-1 label even after a bout of myofiber degeneration that follows deliberate damage to fibers. This residual label appears punctate and probably marks the initial synaptic site where some receptors remain tethered to the basal lamina after myofiber degradation [19, 29]. S3 and S4 Figs show a number of examples of categorization, indicate the reproducibility of this technique, and show the stochastic nature of BTX-1 loss. In WT mice the frequency of dynamic junctions is consistently low from P38 to P450, indicating no dramatic loss of receptors over the 10-day interval. However, when mdx animals were investigated, the proportion of dynamic junctions was 3.3x to 10x as large as WT controls at P38 to P450 (Fig 3A. P38 t(9) = 6.468, P < 0.001, P66 t(6) = 5.433, P < 0.001, P160 t(6) = 5.712, P < 0.001, P450 (6) = 3.719, P < 0.01, 2-Way ANOVA, Bonferroni post-hoc test). Lost junctions were considered dynamic in these measurements, as they also indicate degeneration of the myofiber. Surprisingly, there was no significant effect of age on the frequency of dynamic junctions in mdx muscles, indicating that receptor replacement in the STM is not confined to an early period during the lifetime of mdx mice but occurs constantly from P38 to P450 (Fig 3A). While dynamic junctions were used in this report to measure myofiber degeneration and regeneration, there was some evidence of central chains of nuclei directly under dynamic, fragmented junctions (S1 Movie).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Receptor replacement and NMJ categorization following two-color BTX.

(A) Maximum intensity projection images of STM whole mounts that have undergone the two-color BTX procedure. Examples are from P66 mice. BTX-1 is the first receptor label. BTX-2 is the second applied 10 days later. Imaging follows muscle dissection. In mdx there is a loss of BTX-1 and frequently the BTX-2 label shows fragmentation. Scale bar = 20 μm (B) Two-color BTX experimental design. The STM of mice was exposed and labeled with BTX-1 at Day 0. The animal was allowed to recover from the surgery for 10 days and then euthanized. The labeled muscle was dissected from the animal and labeled with BTX-2. The lower portion of (B) illustrates four predicted outcomes of the experiment. In red is a possible morphology of BTX-1 labeled AChR aggregates at Day 0 of the experiment, in green BTX-2 label at Day 10. Colocalization of BTX-1 and BTX-2 is indicated with yellow. Intact myofibers are shown as pink tubes, and a degenerating myofiber is shown as a gray tube. Scenario 1 shows a stable, continuous junction. 2 a dynamic, fragmented junction. 3 a stable fragmented junction, and 4 another dynamic, fragmented junction. (C) Receptor replacement in WT and mdx STM. Receptor replacement, calculated as percentage of dynamic junctions, in mdx mice is constantly elevated compared to WT. Asterisks denote significance (* P < 0.05, *** P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test). No effect of age was found on receptor replacement. (D) Categorization of P38 WT and mdx junctions following two-color BTX. Bars below labels denote significant difference within genotypes. Not all significant differences within genotypes are shown for sake of brevity. Dashed lines above graph denote significance between genotypes (2-Way ANOVA, Bonferroni post-hoc test, P < 0.05). A full account of statistical significance can be found in S1 Table. Exact n can be found in text as t(df) = t-value. n = df + 2. n is number of mice compared.

https://doi.org/10.1371/journal.pone.0205926.g003

When the morphology of junctions was assessed between mdx and WT animals, we noticed that the predicted decrease in the number of stable, continuous junctions in mdx (Fig 3D. t(9) = 12.10, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test) was accompanied by an increase in the number of dynamic, fragmented junctions (Fig 3D. t(9) = 5.763, P<0.001, 2-Way ANOVA, Bonferonni post-hoc test). This suggests that junctions shift from continuous to fragmented during muscle regeneration. P38 muscles were used in this investigation, as after this time point a great majority of mdx endplates are fragmented, and once fragmented NMJs have not been shown to revert back to a continuous morphology. In P38 mdx animals, the two NMJ morphologies present allowed us to assess whether or not receptor replacement has any bearing on NMJ morphology. When the morphologies of dynamic junctions were compared within P38 mdx animals dynamic, fragmented junctions were seen at a higher frequency than dynamic, continuous (Fig 3D. t(30) = 5.992, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test). Again, this suggests that in mdx animals, AChR replacement is associated with NMJ remodeling. This correlation was not apparent in P38 WT animals. In WT animals, we found a small fraction of dynamic junctions (both fragmented and continuous) with no significant difference between their morphologies (Fig 3D). While we predicted that in both WT and mdx mice dynamic junctions would preferentially be fragmented, the WT observation that receptor replacement had no effect on junction morphology may be a result of the overall low levels of fiber damage even in normal muscles. As such, any striking differences in the morphology of dynamic junctions might have been masked. While not indicated in Fig 3 for sake of brevity, a full account of NMJ categorization differences (i.e. dynamic, fragmented vs stable, fragmented) is presented in S1 Table. Overall, these results suggest that cycles of myofiber degeneration and regeneration lead to remodeling of their junctions in dystrophy. An alternative explanation is that more rapid receptor replacement is a characteristic of mdx junctions because dystrophin tethers receptors to the membrane as suggested from gamma counting of radio labeled AChRs in whole muscles [30]. This is unlikely given the presence of both dynamic and stable junctions in the same muscle. Elevated receptor replacement was encountered at only a subset of junctions. In fact, dynamic junctions were observed on fibers neighboring those with stable junctions. If dystrophin loss causes a decrease in AChR ½ life, it would be expected that individual endplates would all lose similar amounts of receptors. Also, this rapid replacement is not a constant characteristic of fragmented junctions. Many fragmented junctions still retain high levels of BTX-1 (i.e. were classified as stable), suggesting that they suffered damage before, rather than during, the 10-day experimental period. Thus, these experiments suggest that the increased receptor replacement is caused by myofiber necrosis resulting from mdx muscle’s increased susceptibility to damage. We suggest the receptor replacement is not a primary effect of dystrophin loss, but a secondary effect resulting from dystrophic muscles having increased degeneration and regeneration. We performed further experiments to address whether myofiber damage itself was indeed causing the observed changes in NMJs.

Dynamic junctions show morphological differences from stable mdx junctions

When morphological measurements of the NMJ were compared between dynamic and stable endplates of mdx mice, slight but significant differences were observed. At P38 the DI of dynamic endplates note more spread of the AChR at dynamic junctions (stable = 0.3725 ± 0.02, dynamic = 0.3069 ± 0.016, t(25) = 2.261, P = 0.0327, t-test). The same trend was evident at P66 (stable = 0.3738 ± 0.013, dynamic = 0.3098 ± 0.028, t(44) = 2.169, P = 0.0216, t-test). Additionally at P66, dynamic junctions had a lower receptor area than stable junctions (stable = 508.4 ± 34.21, dynamic = 304.4 ± 36.33, t(44) = 3.246, P = 0.0009, t-test). n is the number of pooled individual NMJs compared. No other attributes measured were significantly different between dynamic and stable junctions. At P66 almost all endplates were fragmented. These results suggest that not only can myofiber degeneration and regeneration cause a morphological shift from continuous to fragmented, but also that an already fragmented endplate can rearrange further following additional rounds of myofiber degeneration and regeneration. Continued remodeling may explain the increase in receptor dispersion in older mdx animals. While not measured, there were no obvious receptor clusters in mdx that were not covered by axon terminals. This corresponds well with our hypothesis that nerve terminal growth following myofiber regeneration is responsible for endplate remodeling. In other words, the nerve likely grows and branches along the myofiber, and directs new sites of receptor accumulation, in non-continuous islands. The regenerating muscle does not first deposit AChRs, which the nerve then innervates.

Fiber damage at the endplate region of the muscle causes increased receptor replacement, in both mdx and WT animals

In order to test whether dynamic junctions are indeed a result of myofiber degeneration and regeneration, the same two-color BTX method was performed in conjunction with deliberate injury of the STM. Briefly, after the BTX-1 label was applied a superficial cut was made across the muscle with a scalpel near the endplate region to spur myofiber necrosis. The contralateral muscle was also labeled but not injured and used as an internal control (IC). When this was performed in P38 WT animals, the proportion of dynamic junctions in damaged muscles was 19.1x that of the IC (Fig 4A. P38 t(6) = 3.111, P = 0.0208. t-test). In P66 WT animals the proportion of dynamic junctions in damaged muscles was 31.6x as large as IC. (Fig 4A. P66, t(6) = 3.963, P = 0.0074, t-test). When deliberate injury was performed at the endplate area in P38 mdx the proportion of dynamic junctions in damaged muscle was 1.7x as large as that of the IC, and at P66 the proportion of dynamic junctions in damaged muscle was 2.1x as large as the IC (Fig 4A. P38 t(6) = 3.013, P = 0.0236; P66 t(6) = 3.772, P = 0.0093, t-test). This shows that myofiber damage near the endplate area is sufficient to induce increases in AChR replacement in both WT and mdx mice.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. NMJ response to myofiber injury near the endplate in WT and mdx mice.

(A) Receptor replacement following myofiber injury in WT and mdx STM. Both WT and mdx mice show increases in receptor replacement following damage as assessed by percentage of dynamic junctions. Asterisks denote significant differences (t-test, * P < 0.05, ** P < 0.001). (B) NMJ categorization following myofiber injury in P38 WT and mdx. Following muscle injury the proportion of dynamic, fragmented junctions is significantly increased and the proportion of stable, continuous junctions is decreased compared to ICs in both WT and mdx animals (2-Way ANOVA, Bonferroni post-hoc test, P < 0.001). Bars below the graphs indicate significant differences in the proportion of junctions within IC or Damaged groups (2-Way AVOVA, Bonferroni post-hoc test, P < 0.05), while dashed lines above the graphs indicate significant differences between the two groups (2-Way ANOVA, Bonferroni post-hoc test, P < 0.0001). Not all significant differences within genotypes are shown for sake of brevity. This information can be found S2 and S3 Tables. (C) Increases in receptor replacement following myofiber damage in WT and mdx. After damaging the STM, WT and mdx animals had similar increases in receptor replacement. This was true for both P38 and P66 animals. (t-test, n.s.). This result was interpreted to mean that WT and mdx animals have similar degenerative and regenerative responses to cut damage near the endplate, and that the magnitude of our injury was constant. Exact n can be found in text as t(df) = t-value. n = df + 2. n is the number of mice compared.

https://doi.org/10.1371/journal.pone.0205926.g004

In order to see if this muscle damage-induced receptor replacement had an effect on the morphology of junctions, dynamic fragmented and dynamic, continuous junctions were compared within injured muscles. In other words, we investigated whether dynamic junctions were preferentially fragmented in the muscle injury paradigm. P38 animals were again used because at this time in mdx there is a population of both fragmented and continuous junctions, while after P66 most mdx junctions are fragmented. Therefore, in P38 animals any differences in dynamic junction morphology caused by myofiber damage could be noted. Within damaged P38 mdx STMs the proportion of dynamic, fragmented junctions was 9.6x that of the proportion of dynamic, continuous junctions (Fig 4B. t(15) = 8.279, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test). Within damaged WT muscles the proportion of dynamic, fragmented junctions was 4.5x that of the proportion of dynamic, continuous junctions (Fig 4B. t(15) = 2.725, P < 0.05, 2-Way ANOVA, Bonferroni post-hoc test). The prevalence of dynamic, fragmented junctions over continuous within injured muscles suggests myofiber necrosis and regeneration is correlated with NMJ remodeling. A full account of NMJ type within the groups (i.e dynamic, continuous vs stable, fragmented in WT damaged muscles, etc) is given in S2 and S3 Tables.

The hypothesis that myofiber degeneration and regeneration cycles cause increases in receptor replacement and is correlated to NMJ remodeling is further substantiated when the ICs are investigated. Within mdx ICs, the proportion of dynamic, fragmented junctions was 6.4x that of the proportion of continuous junctions that were also dynamic (Fig 4B. t(15) = 4.179, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test). This trend was not noted for WT IC muscles. In non-injured WT muscles, it is unlikely that the muscles have been damaged naturally, whereas mdx muscles begin to show signs of myofiber degeneration around 3 weeks of age [31–33]. Again, because the receptor replacement in WT ICs is so small, it is possible any NMJ remodeling correlated to replacement may have not been appreciated. These results suggest that muscle damage does cause increases in receptor replacement, and that it is correlated to NMJ remodeling.

Additional evidence that myofiber damage causes NMJ remodeling is found when damaged muscles are compared directly to ICs. Following deliberate myofiber injury in P38 WT muscles the proportion of dynamic, fragmented junctions was 15.4x that of IC dynamic, fragmented junctions (Fig 4B. WT, t(6) = 3.398, P < 0.01, 2-Way ANOVA, Bonferroni post-hoc test). In damaged P38 mdx muscles the proportion of dynamic, fragmented junctions was 1.9x that of IC dynamic, fragmented junction (Fig 4B. mdx t(6) = 4.289, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test). There was also a corresponding loss of stable, continuous junctions in damaged muscles compared to ICs for both genotypes (Fig 4B. WT, t(6) = 5.299, P < 0.001, mdx t(6) = 5.820, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test). This investigation suggests that myofiber necrosis and regeneration at the endplate region is sufficient to shift endplate morphology from continuous to fragmented.

To investigate whether WT and mdx muscles responded to endplate damage in a similar fashion, the increase in the proportion of dynamic junctions in injured muscles compared to ICs was calculated. WT and mdx muscles showed no difference in response to injury at P38 or P66 (Fig 4C. n.s. t-test). This indicates that baseline replacement due to exogenous injury is the same between WT and mdx mice, and that the magnitude of the injury was the same. As the receptor replacement response due to myofiber damage at the endplate area of WT mice mimics that seen in uninjured mdx mice, we believe it likely that degeneration and regeneration of the muscle is the cause of both receptor replacement and remodeling.

The motor axon is needed to restructure the postsynaptic endplate

Above, we have shown that muscular dystrophy in mdx mice causes increases in AChR replacement. The increase is due to myofiber degeneration and regeneration at the synaptic site and is correlated with abnormal endplate morphology. We next investigated the role that the innervating motor axon plays in receptor dynamics and endplate remodeling in mdx mice. It has previously been suggested in mdx mice that the nerve is necessary to restructure the endplate upon muscle cell regeneration [34]. However, in this experiment whether or not the endplate area of the muscle was injured could not be directly assessed. To investigate the role of innervation in remodeling specifically regenerated NMJs, the two-color BTX approach was again utilized, but the nerve that innervates a single STM (accessory nerve) was severed at the time of the first BTX exposure. After the 10-day convalescent period the nerve had not regrown to the endplate region and no NMJs were innervated. The contralateral muscle was left innervated as an IC. The procedure was initiated in P28 mice. Again, at the endpoint of this experiment (P38) there are still a large number of junctions that are not fragmented in mdx mice, so we are able to record any changes in the proportion of fragmented or continuous junctions with and without the nerve. Receptor replacement in denervated mdx muscles was elevated from ICs (Fig 5B. t(6) = 3.227, P = 0.018, t-test). This was not surprising as denervation and/or blocking muscle contractions is known to promote AChR replacement [28, 35]. However, in comparison to innervated muscles, denervated P38 mdx STMs had an increase in the number of dynamic, continuous junctions (Fig 5C. t(6) = 5.484, P < 0.001, 2-Way ANOVA, Bonferroni post-hoc test) and a corresponding decrease in dynamic, fragmented junctions (Fig 5C. t(6) = 3.684, P < 0.01, 2-Way ANOVA, Bonferroni post-hoc test).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Innervation is required for NMJ remodeling in P38 mdx mice.

(A) Confocal maximum intensity projection of P38 mdx innervated (IC) and denervated STMs. “#” denotes dynamic, fragmented junctions. Asterisks denotes dynamic, continuous junctions. In the composite image BTX-1 has been pseudocolored red, BTX-2 green, and motor axon cyan. Note that in the denervated muscle there is no motor axon labeling in the maximum intensity projections. This indicates motor axon loss and no regeneration during the 10-day recovery period. Additionally, both the IC and denervated collections have NMJs with little or no BTX-1 label. This indicates AChR replacement attributed to myofiber degeneration/regeneration in both NMJs with and without innervation Scale bar = 20 μm. (B) Receptor replacement in P38 mdx following denervation. Receptor replacement, as assessed by dynamic junctions, is elevated in P38 mdx denervated muscles as compared to IC (t-test, P < 0.05). (C) NMJ categorization in P38 mdx mice after denervation. Following denervation, the proportion of dynamic, fragmented junctions is significantly decreased, and the proportion of dynamic, continuous junctions is decreased compared to non-denervated IC (2-Way ANOVA, Bonferroni post-hoc test, P < 0.001). Bars below the graphs indicate significant differences in the proportion of junctions within IC or denervated groups (2-Way AVOVA, Bonferroni post-hoc test, P < 0.05), while bars above the graphs indicate significant differences between the two groups (2-Way ANOVA, Bonferroni post-hoc test, P < 0.0001). Not all significant differences within genotypes are shown for sake of brevity. This information can be found in S4 Table. Exact n can be found in text as t(df) = t-value. n = df + 2. n is number of mice compared.

https://doi.org/10.1371/journal.pone.0205926.g005