https://orcid.org/0000-0002-6965-7372
Abstract
Myotonic dystrophy type 1 (DM1) is a debilitating multisystemic disorder caused by a triplet repeat expansion in the 3’ untranslated region of dystrophia myotonica protein kinase mRNAs. Mutant mRNAs accumulate in the nucleus of affected cells and misregulate RNA-binding proteins, thereby promoting characteristic missplicing events. However, little is known about the signaling pathways that may be affected in DM1. Here, we investigated the status of activated protein kinase (AMPK) signaling in DM1 skeletal muscle and found that the AMPK pathway is markedly repressed in a DM1 mouse model (human skeletal actin-long repeat, HSALR) and patient-derived DM1 myoblasts. Chronic pharmacological activation of AMPK signaling in DM1 mice with 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR) has multiple beneficial effects on the DM1 phenotype. Indeed, a 6-week AICAR treatment of DM1 mice promoted expression of a slower, more oxidative phenotype, improved muscle histology and corrected several events associated with RNA toxicity. Importantly, AICAR also had a dose-dependent positive effect on the spliceopathy in patient-derived DM1 myoblasts. In separate experiments, we also show that chronic treatment of DM1 mice with resveratrol as well as voluntary wheel running also rescued missplicing events in muscle. Collectively, our findings demonstrate the therapeutic potential of chronic AMPK stimulation both physiologically and pharmacologically for DM1 patients.
Introduction
Myotonic dystrophy type 1 (DM1) is a genetic disorder characterized by a multisystemic phenotype that includes myotonia, progressive muscle wasting and weakness, cardiac defects, insulin resistance and cognitive impairments (1–3). DM1 is caused by an expansion of CUG triplet repeats located in the 3’ untranslated region (3’UTR) of the dystrophia myotonica protein kinase (DMPK) mRNA. The size of the trinucleotide expansion directly correlates with the severity of the disease, with healthy individuals carrying 5–37 repeats and affected DM1 patients having 50–>3000 repeats (4–6). This CUG expansion (CUGexp) forms a large secondary structure, leading to the aggregation of mutant mRNAs in the nucleus of DM1 cells (7).
Nuclear accumulation of CUGexp mRNAs triggers an imbalance in the level and localization of multiple RNA-binding proteins, thereby altering their normal cellular functions. In particular, muscleblind-like splicing regulator 1 (MBNL1) interacts with CUGexp repeats and is thus sequestered in DM1 nuclei with CUGexp mutant mRNA aggregates, which causes a decrease in the abundance of functional MBNL1 in the rest of DM1 cells (8). In contrast, CUGexp mRNA expression also leads to an increase in the level of CUGBP Elav-like family member 1 (CELF1 or CUGBP1) (9,10). Our laboratory further established that Staufen1, a multifunctional RNA-binding protein, is also increased in DM1 muscles (11,12). Other groups have identified additional RNA-binding proteins misregulated in DM1, including heterogeneous nuclear ribonucleoproteins (hnRNP) H and F, DEAD-box protein 6 (DDX6) and RNA helicase p68/DDX5 (13–18). It has been shown that these RNA-binding proteins regulate alternative pre-mRNA splicing of key transcripts in DM1, thereby participating in the etiology of DM1 symptoms. In particular, alternative splicing of the insulin receptor (INSR), voltage-gated chloride channel 1 (CLCN1 or ClC-1) and cardiac voltage-gated sodium channel (SCN5A) have been associated with symptoms including insulin resistance, myotonia and cardiac defects, respectively. In addition, missplicing of voltage-dependent calcium channel, sarco/endoplasmic reticulum calcium-ATPase (SERCA1) and ryanodine receptor (RyR1) are linked to alterations in excitation-contraction (EC) coupling and calcium homeostasis (9,19–23). Accordingly, DM1 is seen as a spliceopathy. In addition to alternative splicing, these RNA-binding proteins also assume other functions impacting DM1, including micro-RNA biogenesis for MBNL1 (24), modulation of translation and RNA stability for CUGBP1 (25,26) and stress response for Staufen1 and CUGBP1 (27,28).
For several years, we have been interested in examining the impact of phenotypic modifiers as a therapeutic approach for Duchenne Muscular Dystrophy (DMD) (29). As part of this line of work, we examined in detail the calcineurin (CnA) pathway and its therapeutic relevance for DMD. Collectively, our work showed that activation of CnA is highly beneficial for muscles of mdx mice, a DMD mouse model (30–34). In addition, we investigated activated protein kinase (AMPK) signaling which is known to promote expression of the slow, oxidative program in muscle fibers (35). Initial work from our laboratory and reproduced by others established that activation of AMPK signaling is highly beneficial for mdx mice. More specifically, we demonstrated that chronic AMPK activation with 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR) or metformin promotes the slow, oxidative myogenic program and improves the dystrophic phenotype in mdx mice through utrophin upregulation (36–39). Additionally, we and others showed that resveratrol (RSV), a natural polyphenol compound which enhances the activity of SIRT1, also leads to similar beneficial effects in mdx mice (40–42). Finally, stimulation of PPARβ/δ (known to cooperate with AMPK signaling) using the ligand GW501516 also promotes the transition towards slower, more oxidative fibers in mdx mice, leading to physiological improvements (43). Collectively, these data demonstrate the impact and potential of stimulating, PPARβ/δ and AMPK signaling to improve muscle functions in dystrophic muscle. In parallel to this work on DMD, we also examined events occurring in DM1 muscle, and recently discovered that CnA signaling is hyperactivated in DM1 muscle and that it represents a beneficial adaptation to the disease (44).
Given our long-term interest in the role of phenotypic modifiers in neuromuscular diseases, we wondered whether AMPK signaling is affected in DM1 and whether modulation of this pathway could be beneficial for the disease. Here, we investigated these questions and found that the AMPK signaling pathway is markedly repressed in muscle from a DM1 mouse model and patient-derived DM1 myoblasts. Our results also show that chronic activation of this pathway in DM1 with exercise or pharmacological AMPK or SIRT1 activators improves the DM1 spliceopathy in both DM1 mouse muscles and patient-derived human DM1 myoblasts. Such findings have important implications for the development of novel therapeutic strategies using pharmacological agents and exercise, either individually or even in combination.
Results
AMPK signaling is decreased in DM1 mouse muscle
AMPK is a key regulator of muscle plasticity (35). To assess whether this pathway is affected in DM1, we first measured the level of AMPK signaling proteins in DM1 muscle. To this end, we used the DM1 transgenic mouse model containing 250 CTG repeats in the 3’UTR of a healthy gene (human skeletal actin-long repeat, HSALR), which is known to recapitulate the main features of the disease (45). We performed western blots using protein extracts obtained from extensor digitorum longus (EDL) muscles of 9-month-old control (FVB/N) and DM1 mice. Our results show that the total level of AMPK remained unchanged in DM1 mouse muscle compared to controls (P-value > 0.05) (Fig. 1A and B). However, we observed a marked ∼70% decrease in the level of phosphorylated AMPK (Thr172) over total AMPK in DM1 mice compared to WT mice (P-value = 0.02) (Fig. 1A and B). Activated AMPK promotes phosphorylation and activation of the transcriptional coactivator PGC-1α. Accordingly, total levels of PGC-1α were also decreased by ∼43% in DM1 mice compared to control (P-value = 0.007) (Fig. 1A and B).
AMPK signaling is decreased in DM1 mouse muscles and human myoblasts. (A) Representative western blots of AMPK, phospho-AMPK (Thr172) and PGC-1α in WT (FVB/N) and DM1 (HSALR) EDL muscles. (B) Quantifications of expression levels normalized to their respective β-actin controls. Phospho-AMPK was normalized to total AMPK level. N = 4–5 animals per group (9-month-old). (C–D) Representative western blots and quantifications of AMPK and phospho-AMPK (Thr172) over total AMPK in WT and DM1 myoblasts. (N = 2). T-tests were used, and asterisks indicate significance (**P-value ≤ 0.01).
We complemented these results in DM1 patient-derived cells. Human fibroblasts were converted into myoblasts by lentiviral-mediated transduction as previously described (11,27). We then performed western blots on control and DM1 myoblasts. While total AMPK levels were slighty increased in DM1 myoblasts, we observed a ∼35% decrease in the level of phospho-AMPK over total AMPK in DM1 myoblasts compared to controls (Fig. 1C and D). Together, these results show that the AMPK-PGC-1α axis is markedly repressed in DM1 muscle.
AICAR promotes expression of the slow, oxidative muscle phenotype in DM1 mouse muscle
AICAR is a pharmacological agonist of AMPK, which mimics the activating effect of AMP on AMPK (46). To determine the importance of the AMPK signaling for DM1, we treated 2-month-old DM1 mice for 6 weeks with daily subcutaneous injections of AICAR (500 mg/kg per day) (36,37). In parallel, daily saline injections were performed on DM1 mice as a control. AMPK activation by AICAR is known to promote the slower, more oxidative muscle phenotype in both wild-type mice and mdx mice (36,37). Here, we thus determined whether AMPK activation with AICAR is accompanied by changes in the metabolic properties of DM1 muscles.
First, we performed western blots on tibialis anterior (TA) muscles from DM1 mice. As expected, our results show a ∼1.8-fold increase in the level of PGC-1α in AICAR-treated DM1 mice compared to saline-treated DM1 mice (P-value = 0.008) demonstrating that AICAR indeed activated AMPK-PGC-1α signaling in treated DM1 mice (Fig. 2A and B). Next, we characterized the fiber-type composition of EDL muscles from DM1 mice. We performed immunofluorescence on cryostat cross-sections with myosin heavy chain (MyHC) type I-, IIa- and IIb-specific antibodies. While MyHC I and IIb remain unchanged (P-value > 0.05), we observed an increase in the percentage of oxidative MyHC IIa positive fibers from ∼11% in saline-treated DM1 mice to ∼19% in AICAR-treated DM1 mice (P-value = 0.0005) (Fig. 2C and D). Such findings are coherent with a switch from MyHC IIx towards IIa fibers in AICAR-treated mice. We also performed succinate dehydrogenase (SDH) staining on cross-sections as a measure of oxidative capacity of muscle fibers. Our results show an increase in the percentage of SDH-positive fibers in EDL muscles from AICAR-treated DM1 mice compared to saline-treated DM1 mice (P-value = 0.0008) (Fig. 2E and F).
AICAR promotes expression of the slow, oxidative muscle phenotype in DM1 mouse muscle. (A–B) Representative western blots and quantifications of PGC-1α in TA muscles. β-actin was used as a control. N = 3. (C) Representative immunofluorescence of MyHC IIa (red) in EDL muscles. Laminin (green) was used to delimit muscle fibers. Scale bars, 50 μm. (D) Percentage of MyHC I, IIa and IIb positive fibers. N = 6. (E) Representative SDH staining in EDL muscles. Scale bars, 200 μm. (F) Percentage of SDH-positive fibers. N = 6. (G–H) Representative western blots and quantifications of OXPHOS proteins (CI-NDUFB8, CIII-UQCRC2 and CV-ATP5A) in TA muscles. β-actin was used as a control. N = 6. T-tests were used, and asterisks indicate significance (*P-value ≤ 0.05, **P-value ≤ 0.01 and ***P-value ≤ 0.001).
To complement these experiments, we also performed western blots on protein extracts obtained from TA muscles using anti-oxidative phosphorylation (OXPHOS) antibodies (Fig. 2G). We observed a marked ∼5.6-fold increase in the level of complexes CI-NDUFB8, and ∼1.4- and ∼1.8-fold in CIII-UQCRC2 and CV-ATP5A in response to AICAR treatment in DM1 mouse muscles (P-value = 0.02, 0.001 and 0.0006, respectively) (Fig. 2G and H). Altogether, our results demonstrate both the efficiency of the chronic AICAR treatment and the role of this AMPK agonist in the induction of a shift towards slower, more oxidative muscle fibers in DM1 mice.
AICAR improves DM1 mouse muscle histology
In DM1, the CUGexp mRNA promotes an increase in the variability of fiber diameter and fiber hypertrophy (44,45). We thus performed a histological analysis on EDL muscle cross-sections in control, saline-treated and AICAR-treated DM1 mice. Muscle sections were stained with H&E and we analyzed the cross-sectional area (CSA). As expected, our data showed an increase in the mean fiber size from 1073 ± 53 to 1609 ± 101 μm2 (P-value = 0.0008, mean ± SEM) in control versus DM1 mice (Fig. 3A), showing fiber hypertrophy. This was paralleled by an increase in the coefficient of variation (CV) from 0.469 to 0.637 (Fig. 3A). In addition, we analyzed the distribution of CSA. As previously reported (44,45), we observed an increase in the percentage of large muscle fibers (>2000 μm2, P-value ≤ 0.01) in DM1 mice (Fig 3A and B). AICAR treatment of DM1 mice induced a robust decrease in the mean fiber size area and variation coefficient to control levels (P-value > 0.05 compared to WT) (Fig 3A and B). In addition, we also observed a marked decrease in the percentage of large muscle fiber (>2000 μm2, P-value ≤ 0.05 compared to DM1) to control levels (Fig. 3A and B). These results essentially show a complete reversion in the fiber variability to wild-type profiles.
AICAR improves DM1 mouse muscle histology. (A) Representative H&E staining of control, saline- and AICAR-treated DM1 EDL muscles. Inserts show mean CSA and CV. Scale bars, 50 μm. (B) CSA frequency. N = 5. (C) Percentage of central nuclei. N = 5–6. T-tests were used, and asterisks indicate significance (ns P-value> 0.05, *P-value ≤ 0.05, **P-value ≤ 0.01 and ***P-value ≤ 0.001).
It is well established that DM1 muscle fibers display an abundance of abnormal, centrally located nuclei. In agreement with this, our results show that while central nuclei are relatively rare in wild-type animals (<3%), saline-injected DM1 mice have ∼12% of central nuclei (P-value = 1x10−6) (Fig. 3A and C). These percentages are coherent with earlier findings (44,45). Strikingly, chronic AICAR treatment reverted the pattern of central nucleation in DM1 mice towards wild-type levels (P-value > 0.05 compared to WT) (Fig. 3A and C). Altogether, these results demonstrate that chronic AMPK activation with AICAR significantly improves the muscle histology of DM1 mice.
AICAR reverses CUGexp aggregation and MBNL1 sequestration
In DM1, CUGexp mRNAs accumulate within nuclei. These RNA aggregates can be visualized by fluorescence in situ hybridization (FISH). EDL muscle cross-sections were hybridized with a Cy3-(CAG)10 probe and nuclei stained with 4′,6-diamidino-2-phenylindole (DAPI). While no RNA foci were observed, as expected, in wild-type mice (not shown), approximately ∼38% of nuclei contained foci in saline-injected DM1 mice (Fig. 4A and B). In AICAR-treated DM1, we still observed nuclei containing RNA foci; however, the number of nuclei positively stained was strongly decreased by ∼40% (P-value = 7 × 10−9 compared to saline-treated DM1) (Fig. 4A and B), indicating that mutant CUGexp mRNA aggregates less following chronic AMPK activation.
AICAR reverses CUGexp mRNA aggregation, MBNL1 sequestration and reduces CUGBP1, Staufen1 and RBM3 levels. (A) Representative FISH showing nuclear RNA foci in EDL muscles. Scale bars, 10 μm. (B) Percentage of nuclei containing RNA foci. N = 5–6. (C) Representative western blots showing CUGBP1, Staufen1, MBNL1 and RBM3 levels in saline- and AICAR-treated DM1 TA muscles. (D) Quantifications of western blots. N = 6. (E) Representative immunofluorescence showing MBNL1 localization. Scale bars, 10 μm. (F) Percentage of nuclei showing MBNL1 aggregates in EDL muscles. N = 6. T-tests were used, and asterisks indicate significance (*P-value ≤ 0.05, **P-value ≤ 0.01 and ***P-value ≤ 0.001).
CUGexp expression is known to create an imbalance in the level and localization of multiple RNA-binding proteins (8,11,47). We thus analyzed by western blot the levels of CUGBP1, Staufen1 and MBNL1 using TA muscle extracts. While MBNL1 level was not significantly affected (P-value > 0.05), AMPK stimulation with AICAR decreased Staufen1 levels (P-value = 0.049) (Fig. 4C and D). CUGBP1 showed a trend towards a decrease in AICAR-treated DM1 mice, but it did not reach the significance level (P-value = 0.13) due to inter-individual variability. Recently, it was reported that RNA binding motif protein 3 (RBM3) is another RNA-binding protein misregulated by CUGexp (48). In our experiments, we observed a significant decrease of RBM3 in AICAR-treated DM1 mice (P-value = 0.004).
It is well established that MBNL1 is sequestered by CUGexp mRNA aggregates in the nucleus of DM1 cells (8). We therefore analyzed the localization of MBNL1 by immunofluorescence on EDL muscle cross-sections. As expected, MBNL1 is sequestered in the nucleus of saline-injected DM1 mice (Fig. 4E and F). AMPK activation by AICAR reduced the number of nuclei containing MBNL1 aggregates from 41% to 20% (P-value = 2 × 10−7) (Fig. 4E and F). The decrease in MBNL1 sequestration therefore matches the reduction observed in CUGexp mRNA aggregation (see above). Altogether, our results show that chronic AMPK stimulation decreases aggregation of toxic CUGexp mRNAs, resulting in a decreased sequestration of MBNL1 and reduced expression of Staufen1, RBM3 and CUGBP1.
AICAR corrects the DM1 spliceopathy
RNA-binding proteins misregulated in DM1 are known to act as important splicing factors. Accordingly, their imbalance affects normal alternative splicing patterns thereby causing the many symptoms associated with DM1. We therefore analyzed the pattern of key alternative splicing events known to be misspliced in DM1 by radioactive reverse transcription - polymerase chain reaction (RT-PCR) using EDL muscle extracts. First, we analyzed the splicing pattern of two sarcoplasmic reticulum proteins, namely the SERCA1 pump and the RyR1 channel. In this context, it has been proposed that skipping of exon 22 and 70 of SERCA1 and RyR1, respectively, affects the properties of both calcium pump and channel thereby affecting calcium homeostasis and EC coupling in DM1 cells (22,23). As previously described (44), we observed a marked increase in exon 22 and 70 exclusion in SERCA1 and RyR1 (P-value = 6 × 10−6 and 6 × 10−7, respectively) (Fig. 5A and B) in DM1 mouse muscle. AICAR treatment induced an almost complete reversion of alternative splicing patterns to wild-type levels, from ∼61% to ∼12% for SERCA1 and from ∼69% to ∼57% for RyR1 (P-value = 0.06 and 0.0003 compared to saline, respectively) (Fig. 5A and B). A similar result was obtained with TA muscle extracts (Fig. S1A and B).
AICAR corrects the DM1 spliceopathy and expression of CLCN1. (A) Radioactive RT-PCR analysis showing alternative splicing patterns of SERCA1, RyR1, TNNT3 and CLCN1 in control, saline- and AICAR-treated DM1 EDL muscles. (B) Percentages of SERCA1 E22 exclusion, RyR1 E70 exclusion, TNNT3 F inclusion and CLCN1 Ex7a inclusion, with mean ± SEM. N = 3–5. (C) Representative immunofluorescence showing CLCN1 expression and localization in EDL muscles. Scale bars, 50 μm. T-tests were used, and asterisks indicate significance (ns P-value > 0.05 and ***P-value ≤ 0.001).
The fast skeletal muscle troponin T3 (TNNT3) is another mRNA with aberrant alternative splicing in DM1. This protein plays a crucial role for muscle contraction. Several exons are alternatively spliced in TNNT3 (exons 4, 6, 7, 8, Fetal, 16 and 17) (49). Inclusion of Fetal exon (exon F) in TNNT3 affects the biological function of TNNT3 protein, in particular its calcium-regulated ATPase activity (50). While completely excluded in normal adults (∼2%), the exon F remains included in combination with other exons, in DM1. We reproduced these patterns in saline-injected DM1 mice (∼37%, P-value = 8 × 10−6) (Fig. 5A and B) and observed that AICAR treatment significantly reduced the presence of TNNT3 isoforms containing exon F (8%, P-value > 0.05 compared to WT) (Fig. 5A and B). These data show that AMPK activation with AICAR has a positive effect on alternative splicing in DM1 muscle.
AICAR corrects alternative splicing and expression of the chloride channel
The muscle-specific CLCN1 is responsible for chloride ion exchange at the membrane. In adult muscle, the main CLCN1 mRNA isoform excludes exon 7a. In DM1, exon 7a inclusion is increased leading to limited expression or expression of a non-functional chloride channel (19). Indeed, exon 7a contains an in-frame termination codon leading to the following: (i) non-sense mediated decay causing a reduction in CLCN1 mRNA and (ii) the production of a truncated and non-functional CLCN1 proteins. Reduction of functional CLCN1 protein expression has been associated with myotonia seen in DM1 skeletal muscle (51). In agreement with the literature, we observed an increase in exon 7a inclusion in saline-injected DM1 mice compared to wild-types (from ∼13% to ∼28%, P-value = 1 × 10−6). Our results show that AICAR treatment reduced exon 7a inclusion to wild-type levels (∼15%, P-value > 0.05 compared to WT control) (Fig. 5A and B) thus reverting alternative splicing patterns towards wild-type mice.
To complement these results, we performed immunofluorescence using anti-CLCN1-specific antibodies on EDL muscle cross-sections. While our no primary antibody control showed no signal (not shown), CLCN1 antibodies preferentially labelled the sarcolemma of muscle fibers in wild-type mice (Fig. 5C). As expected, saline-injected DM1 mice showed a reduction in CLCN1 signal compared to wild-types (Fig. 5C). Coherent with our splicing data, AICAR treatment resulted in a striking rescue in the expression of CLCN1 at the sarcolemma of muscle fibers (Fig. 5C). Altogether, these results show that a key alternative splicing event linked to CLCN1 expression and myotonia is rescued by AMPK stimulation by AICAR.
AICAR treatment prevents CUGexp aggregation in human DM1 myoblasts and shifts alternative splicing towards WT conditions
As shown above, AMPK activation has beneficial effects in DM1 mouse muscles. To investigate whether these benefits could be recapitulated in human cells, DM1 patient myoblasts were treated with an increasing dose of AICAR (0, 0.5, 1 and 2 mm) for 24 h. Healthy human myoblasts were used as controls.
First, alternative splicing profiles were analyzed by RT-PCR. In wild-type myoblasts, we observed ∼100% inclusion of SERCA1 exon 22 and ∼0% in DM1 myoblasts (Fig. 6A and B). At 0.5 mm, AICAR treatment induced an increase in exon 22 inclusion (∼79% exon 22 exclusion). Higher doses of AICAR showed a more pronounced, dose-dependent increase in exon 22 inclusion (∼66 and 56% exon 22 exclusion at 1 and 2 mm, respectively, P-value < 0.001) (Fig. 6A and B). These results, therefore, recapitulate in human cells our results obtained in DM1 mice.
AICAR prevents CUGexp aggregation and shifts alternative splicing towards WT conditions in human DM1 myoblasts. (A) Human DM1 myoblasts were treated with an increasing dose of AICAR (0, 0.5, 1 and 2 mm). Untreated WT myoblasts were used as controls. Representative RT-PCR of SERCA1, ZFAND1, MDM4 and GPCPD1 mRNAs. (B) Quantifications of RT-PCRs. N = 3. (C) Representative FISH showing nuclear RNA foci in DM1 myoblasts. (D) Number of RNA foci per nuclei. N = 4. (E–F) Representative western blots and quantifications of AMPK and phospho-AMPK over total AMPK showing activation of AMPK signaling in human DM1 fibroblasts treated with 1 mm AICAR. (N = 2). T-tests were used, and asterisks indicate significance (ns P-value > 0.05, *P-value ≤ 0.05, **P-value ≤ 0.01 and ***P-value ≤ 0.001).
We then analyzed other alternative splicing events previously described as aberrant in mesenchymal precursor cells derived from DM1 human embryonic stem cells (48). Thus, splicing of Zinc Finger AN1-Type Containing 1 (ZFAND1) exon 3, Mouse Double Minute 4 P53 Regulator (MDM4) exon 7 and Glycerophosphocholine Phosphodiesterase 1 exon 5 were analyzed. All three show an increase in exon inclusion in DM1 cells (Fig. 6A and B). AICAR treatment induced a striking dose-dependent decrease in exon inclusion for these three splicing events. In fact, the higher dose of AICAR (2 mm) completely rescued missplicing to control wild-type levels (Fig. 6A and B).
To further examine the impact of AICAR on endogenous CUGexp DMPK mRNA aggregation, we performed FISH on DM1 myoblasts treated with increasing doses of AICAR. As previously described (27), untreated DM1 myoblasts contain ∼4.7 RNA foci per nuclei (Fig. 6C and D). AICAR treatment induced a marked, dose-dependent decrease in the number of nuclear RNA foci to ∼2.8 and ∼1.4 foci per cell (P-value = 0.0003 and 4 × 10−7, respectively) (Fig. 6C and D). Altogether, these results demonstrate that AMPK stimulation decreases mutant DMPK mRNA aggregation and corrects aberrant alternative splicing in human DM1 myoblasts.
AICAR is known to have potential toxic effects on cells which vary according to cell type (52). We therefore analyzed the viability of DM1 fibroblasts exposed to AICAR using a 3-[4,5-Dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) assay. Control and DM1 fibroblasts were cultured and treated with an increasing dose of AICAR (0, 0.5, 1 and 2 mm) for 24 h. Our results show a 91–94% of viability at 0.5 mm, 80–83% at 1 mm and 73–74% at 2 mm for control and DM1 fibroblasts, respectively (Fig. S2). Although AICAR showed a relative decrease in cell survival at tested concentrations, a sharp reduction in foci number was observed in viable cells at the same doses (Fig. 6C and D). Moreover, measurents of RNA foci were conducted on those cells that remained viable. Together, this indicates that our findings are toxicity-independent.
Finally, activation of AMPK signaling was evaluated by western blots after treating DM1 fibroblasts with 1 mm of AICAR. Our results show a ∼2.6-fold increase in the level of phospho-AMPK over total AMPK in AICAR-treated DM1 fibroblasts compared to untreated cells demonstrating that AICAR indeed activated AMPK-PGC-1α signaling in human cells in culture (Fig. 6E and F).
RSV partially corrects aberrant alternative splicing in DM1 mouse muscle
RSV is a natural polyphenol compound known to activate SIRT1 and AMPK signaling (53–56). To confirm the importance of AMPK signaling with another pharmacological drug, we treated 2-month-old DM1 mice for 6 weeks with RSV-supplemented diet (∼100 mg/kg per day) (40). In parallel, an age-matched control group of DM1 mice was fed with standard diet for the duration of the treatment.
To assess the effect of RSV on RNA toxicity, we analyzed the alternative splicing profile of SERCA1, RyR1, TNNT3 and CLCN1 by radioactive RT-PCR with TA muscle extracts, as performed following AICAR injections in Figure 5. While SERCA1 did not show any change in alternative splicing profiles upon RSV treatment (P-value = 0.4), RyR1, TNNT3 and CLCN1 showed a clear trend towards restoration to wild-type levels (P-value = 0.06, 0.18 and 0.07, respectively) (Fig. 7A and B). However, given the high variability observed in RSV-treated DM1 mice, these results did not reach the significance threshold (P-value > 0.05). These data show nonetheless that a 6-week RSV administration recapitulates, albeit more modestly, the effects of AICAR in DM1 mice.
Chronic RSV administration decreases DM1 spliceopathy. 2-month-old DM1 mice were fed with control diet or chow supplemented with RSV (100 mg/kg per day) for 6 weeks. (A) Radioactive RT-PCR analysis showing alternative splicing patterns of SERCA1, RyR1, TNNT3 and CLCN1 in control, DM1 and RSV-treated DM1 TA muscles. (B) Percentages of SERCA1 E22 exclusion, RyR1 E70 exclusion, TNNT3 F inclusion and CLCN1 Ex7a inclusion, with mean ± SEM. N = 3–5. T-tests were used, and asterisks indicate significance (ns P-value > 0.05, *P-value ≤ 0.05, **P-value ≤ 0.01 and ***P-value ≤ 0.001).
Physical activity has a beneficial impact on alternative splicing in DM1 mouse muscle
Exercise is known to elicit muscle plasticity through the activation of AMPK signaling (35). We thus wondered whether a physiological stimulus such as chronic exercise in the form of voluntary wheel running could trigger the same beneficial effects as drug-induced AMPK activation in DM1 mice. To test this hypothesis, DM1 mice aged 4–8 months were provided with free access to an angled rotating wheel for 8 weeks. Age-matched sedentary DM1 littermates and controls (mix of FVB/N and HSA-short repeat, HSASR mice) were housed in cages without running wheels. At the end of the 8-week period, exercise wheels were removed 24 h prior to the collection of tissue samples to avoid any acute effects of physical activity.
We then analyzed by RT-PCR the effect of voluntary running on patterns of alternative splicing of key genes known to be misspliced in DM1 by RT-PCR with EDL muscles, as performed above with AICAR and RSV. Remarkably, all splicing events analyzed, i.e. SERCA1, RyR1, TNNT3 and CLCN1, showed a clear and significant improvement in alternative splicing towards control wild-type levels. More specifically, we observed a marked ∼28% decrease in exon 22 exclusion and ∼6% in exon 70 exclusion in SERCA1 pump and RyR1 channel (P-value = 0.003 and 0.02, respectively). We also observed a ∼14% reduction of TNNT3 exon F inclusion towards wild-type (P-value = 0.01). Finally, we obtained a ∼10% reduction in the myotonia-associated exon 7a inclusion for CLCN1 (P-value = 0.003) (Fig. 8A and B). These data clearly show that 8 weeks of voluntary running improve the DM1 spliceopathy thereby recapitulating physiologically in DM1 mice, the effects of AMPK activation by pharmacological activators.
Physical activity has a beneficial impact on alternative splicing in DM1 mouse muscle. 4- to 8-month-old DM1 mice were provided with free access to an exercise wheel for 8 weeks. (A) RT-PCR analysis showing alternative splicing patterns of SERCA1, RyR1, TNNT3 and CLCN1 in ‘sedentary’ controls and DM1, and ‘exercised’-DM1 EDL muscles. (B) Percentages of SERCA1 E22 exclusion, RyR1 E70 exclusion, TNNT3 F inclusion and CLCN1 Ex7a inclusion, with mean ± SEM. N = 10–15. T-tests were used, and asterisks indicate significance (*P-value ≤ 0.05, **P-value ≤ 0.01 and ***P-value ≤ 0.001).
Regulation of AMPK signaling in DM1 muscle. AMPK signaling is repressed in DM1 muscle. Pharmacological (AICAR or RSV) or physiological (exercise) activation of the AMPK pathway has a beneficial impact on DM1. It promotes the slower, more oxidative muscle phenotype, and improves DM1 muscle histology by decreasing muscle fiber hypertrophy and central nucleation. It also triggers a decrease in nuclear CUGexp RNA aggregation, a rescue in RNA-binding misregulation and a correction of alternative splicing towards wild-type conditions.
Discussion
In this study, we assessed the status of AMPK signaling in skeletal muscle from DM1 mice and patient-derived myoblasts. We found that the AMPK pathway is markedly decreased in DM1 muscles. In order to assess the importance of AMPK signaling for DM1, we used pharmacological and physiological activators of AMPK signaling. AICAR-dependent AMPK activation promoted expression of the slower, more oxidative phenotype and, importantly, improved DM1 muscle histology by decreasing muscle fiber hypertrophy and central nucleation. In addition, AICAR reverted multiple events associated with RNA toxicity in DM1, including CUGexp mRNA nuclear aggregation, MBNL1 sequestration, RNA-binding protein imbalance and aberrant alternative splicing (Fig. 9). Improvements in CUGexp mRNA aggregation and alternative splicing were also recapitulated in DM1 patient-derived myoblasts. Furthermore, exercise and RSV, which also activate AMPK signaling, improved the DM1 spliceopathy in DM1 mice. Altogether, our data demonstrate the therapeutic potential of chronic AMPK stimulation both physiologically and pharmacologically for DM1 patients.
In skeletal muscle, AMPK acts as a master regulator of cellular energy homeostasis and muscle plasticity (35). Indeed, AMPK is activated in response to a decrease in energy state (e.g. nutrient deprivation) or an increase in energy expenditure (e.g. exercise) and is therefore a sensor of energy balance. Once activated, AMPK participates in several events including the activation of transcriptional activators and co-activators, including PGC-1α, which stimulates gene expression and mitochondrial biogenesis and function (57). Given these pivotal roles, AMPK signaling has been found to be altered in a variety of physiological and pathological conditions, including obesity, sedentary lifestyle and insulin resistance (58). Here, we report for the first time that AMPK signaling is repressed in DM1 mouse muscles and in DM1 myoblasts in culture. Given the fact that AMPK signaling is such an important regulator of muscle plasticity, it is expected that repression of AMPK signaling in DM1 will have a direct impact on muscle structure, function and metabolism. In agreement with this, we found that stimulation of AMPK signaling in DM1 muscle promotes muscle plasticity towards a slower, more oxidative phenotype and improved muscle histology.
The current widely accepted pathogenic model for DM1 stipulates that expression of CUGexp mRNAs causes RNA toxicity. In particular, CUGexp mRNAs aggregate in the nucleus of DM1 cells and alter the level and localization of several RNA-binding proteins. Their imbalance, in turn, affects alternative splicing profiles of many pre-mRNAs causing the characteristic spliceopathy. Our results show that chronic AICAR treatment ameliorates all aspects of RNA toxicity including RNA foci formation, MBNL1 sequestration, RNA-binding protein imbalances, aberrant alternative splicing and CLCN1 expression.
Of relevance to our findings, it is now known that variations in energy metabolism can directly affect alternative splicing. For example, amino acid deprivation affects alternative splicing in yeast (59). Moreover, starvation in mice modulates hnRNPs and serine/arginine-rich (SR) proteins thereby controlling G6PD alternative splicing in hepatocytes (60–62). More recently, RSV and metformin were shown to regulate INSR alternative splicing in several human cell lines in culture (63) and in peripheral blood lymphocytes of non-DM1 patients with type 2 diabetes (48), respectively. Taken together with our current data, these findings raise the possibility that AMPK directly regulates alternative splicing.
Despite these observations, it still remains unclear how AMPK signaling acts to induce pleitropic benefits on the DM1 phenotype. In this context, several mechanisms may be envisaged. For example, it appears that activation of AMPK signaling with different means in cultured cells and in muscle in vivo causes parallel reductions in the steady-state levels of mutant transcripts in nuclei. Given that the effects of AICAR seen in the current study are reproduced in both DM1 mouse muscles as well as in patient cells, such reductions in the steady-state levels of mutant mRNA foci likely take place post-transcriptionally. Thus, the benefical effects of AICAR under these conditions can occur through the following: (i) the targeting of CUGexp-transcripts by RNA-binding proteins involved in mRNA stability, which expression and/or subcellular localization are affected by AMPK signaling such as HuR (64–66) and AUF1 (67,68); (ii) the displacement of MBNL1 from RNA foci (Fig. 4E and F); or (iii) the AMPK-dependent regulation of RNA-binding proteins including CUGBP1, Staufen1 and RBM3 (Fig. 4C and D and (48)). Current work in our laboratory is focused on dissecting these various possibilities in order to define the exact cellular event(s) modulated by AMPK activation that result(s) in several benefical effects.
Our results further show that both AICAR and RSV, two pharmacological activators of AMPK signaling, improve the DM1 spliceopathy. However, the effect of the RSV treatment did not reach the significance threshold, likely due to the fact that RSV targets SIRT1 rather than AMPK directly. Note also that saline-treated DM1 mice used in the RSV experiments display a milder defect in TNNT3 and CLCN1 splicing compared to the one used for AICAR experiments contributing to the lower significance (Figs 5B and 7B). Despite the smaller effect, we nonetheless observed a clear trend towards an improvement of DM1 spliceopathy. These data are in fact in excellent agreement with earlier studies showing that RSV and metformin modulate pre-mRNA alternative splicing in culture, in patient-derived DM1 fibroblasts and embryonic stem cell lines, respectively (48,63). They are also in accord with a recent study reporting that a short-term (1 week) AICAR treatment improves RNA toxicity in DM1 mice (69). Interestingly, and by contrast to our current findings, no significant improvements were reported on muscle histology with the short-term treatment (69), indicating a duration-dependent effect of AICAR treatment. Collectively, these studies highlight that the chronic use of drugs targeting AMPK signaling represents an excellent therapeutic strategy to alleviate symptoms associated with the DM1 pathology including the spliceopathy.
Over the years, multiple signaling pathways have been reported to be affected in DM1 by CUGexp mRNA expression. These pathways include the following: (i) PKC and Glycogen synthase kinase-3β which participate in the phosphorylation and stability of CUGBP1 (70,71); (ii) the double-stranded RNA-dependent protein kinase PKR (72); (iii) TWEAK/Fn14 and NF-κB (73); (iv) mammalian target of rapamycin (mTOR) (74); and (v) CnA signaling (44). While the full extent of perturbations in signaling pathways in DM1 muscle is still not completely understood, altogether these studies show a consistent and global imbalance in major intracellular signaling pathways in DM1. Given the predominant roles of these pathways in several aspects of normal and diseased muscle biology, the concomitant misregulation of these pathways most likely contributes profoundly to the complexity of DM1 muscle phenotype. This misregulation of multiple key signaling cascades, coupled to the imbalance in several RNA-binding proteins, further highlights the complexity of the cellular and molecular events that lead to the multisystemic DM1 phenotype. The fact that several pharmacological approaches have successfully corrected individual signaling pathways and resulted in phenotypic improvements in various animal models is very encouraging for DM1 patients. However, it also highlights the importance of developing combinatorial therapies targeting multiple pathways and events to optimize correction of the DM1 phenotype.
Several studies have demonstrated the beneficial impact of moderate exercise for neuromuscular disorders. In DMD, weaker muscles and more fragile muscle fibers are hallmarks of the disease. Accordingly, DMD patients are more susceptible to muscle damage caused by exercise. However, despite these physiological dysfunctions, a milder exercise regimen has proven to be beneficial in mdx mice as well as for DMD patients (75,76).
In comparison to DMD, little is known about the impact of exercise in DM1 (77,78). Here, we show for the first time that voluntary wheel running in DM1 mice improves the DM1 spliceopathy which, in the widely accepted RNA toxicity model, is thought to be responsible for the major symptoms of the disease. It is well established that AMPK is activated in muscle by endurance training and that it mediates its beneficial effects (35). Thus, our results obtained with the pharmacological AMPK activators AICAR and RSV suggest that the positive impact of exercise on the DM1 spliceopathy takes place through AMPK stimulation.
Symptoms associated with DM1 include muscle weakness, pain and fatigue, along with cognitive impairments. As a consequence, patients affected with DM1 often have a lower ability to carry out certain daily activities (79). Therefore, the commitment of patients to a regular physical activity regime could rapidly become a major challenge. AMPK activators such as AICAR, RSV and metformin are sometimes referred to as ‘exercise mimetics’. Indeed, AMPK agonists such as AICAR enhance training performance and endurance along with metabolic changes, thereby recapitulating some of the effects associated with exercise (80). Therefore, a therapeutic approach including the use of AMPK activators could simulate or potentiate the effects of moderate exercise for DM1 patients especially for those with limited ability and/or motivation to exercise. In addition, the use of pharmacological AMPK activators may have the advantage of having a pleiotropic effect, which is crucial given the multisystemic nature of DM1. Altogether, our results show that moderate exercise and pharmacological AMPK activation represent an achievable therapeutic strategy for treating DM1 patients.
Materials and Methods
Antibodies
The antibodies used were anti-AMPKα (2532s, Cell Signaling/New England Biolabs, Whitby, ON, Canada), anti-phospho-AMPKα (Thr172) (2535, Cell Signaling/New England Biolabs), anti-OXPHOS cocktail (ab110413, Abcam/Cedarlane, Burlington, ON, Canada), anti-CUGBP1 (sc-20003, Santa Cruz Biotechnology, Mississauga, ON, Canada), anti-Staufen1 (ab73478, Abcam/Cedarlane), anti-MBNL1 (H00004154, Abnova/Cedarlane), anti-MyHC Type I, IIa and IIb (BAF8, SC71, and BFF3, respectively; Developmental Studies Hybridoma Bank, Iowa City, IA) and anti-β-actin (sc-47778, Santa Cruz Biotechnology).
Animal models and treatments
All procedures were approved by the University of Ottawa Animal Care Committee and were in compliance with the guidelines of the Canadian Council on Animal Care and the Animals for Research Act. Wild-type FVB/N (Charles River, Sherbrooke, QC, Canada), control HSASR and DM1 HSALR line LR20b (45) were used in this study. DM1 mice were treated daily for 6 weeks with either AICAR (TRC Canada, Toronto, Canada; 500 mg/kg per day, subcutaneously) or saline (control), as previously described (36, 37). A separate cohort of animals was fed for 6 weeks with a standard chow diet (Teklad 2018; Harlan Research Models and Services, Madison, WI) or chow supplemented with RSV (100 mg/kg per day, Toronto Research Chemicals, Toronto, ON, Canada), as previously described (40). Finally, another group of DM1 mice was provided with free access to an angled rotating wheel for 8 weeks. Exercise wheels were removed the day prior dissections. At the end of each study, muscles were dissected and either (i) frozen and crushed in liquid nitrogen for protein and RNA extraction or (ii) embedded in Tissue-Tek OCT compound (VWR, Mississauga, Canada) and flash frozen in isopentane cooled with liquid nitrogen for cryostat sectioning. Left and right EDL muscles were used for histology and RNA extraction, whereas TA muscles were used for RNA and protein extraction. The larger size of the TA is advantageous for protein extraction and western blotting.
Cell culture conditions, Lentivirus infections and treatments
Control (GM03377) and DM1 (GM03132) human fibroblasts (Coriell Cell Repositories) were grown according to instructions in growth medium (DMEM 10% fetal bovine serum [HyClone, Thermo Fisher Scientific], 100 U/ml penicillin, 100 μg/ml streptomycin). Lentiviral particles were produced by transient transfection of 293 T cells (American Type Culture Collection/Cedarlane) with lentivector pCDH-MyoD (11,27), along with psPax2 and pMD2.G packaging plasmids. The conditioned medium containing viral particles was collected and used to transduce primary cells overnight in the presence of 6 μg/ml of Polybrene (Sigma-Aldrich, Oakville, ON, Canada). Infected cells were grown for 3–5 days before treatments and analyses. Patient-derived myoblasts were treated with AICAR for the indicated time and concentration.
Immunofluorescence and fiber typing
Cryostat cross-sections were fixed for 15 min (4% formaldehyde in phosphate-buffered saline [PBS]), permeabilized for 30 min (0.1% Triton in PBS) and blocked for 30 min (5% goat serum, 0.1% Triton in PBS). Sections were incubated with the primary antibody diluted in blocking buffer for 1 h at 37°C or overnight at 4°C. Then, cross-sections were thoroughly washed and incubated for 1 h with Alexa secondary antibodies (Invitrogen Life Technologies/ThermoFisher Scientific) diluted in blocking buffer. Following washes, sections were mounted with Vectashield mounting medium (Vector Labs/Cedarlane) containing DAPI for staining of nuclei. For fiber typing, cryostat cross-sections were processed using the M.O.M. immunodetection kit (Vector Labs/Cedarlane) with Texas Red streptaviding detection system (Vector Labs/Cedarlane) following manufacturer instructions.
Histochemical staining
For H&E staining, cryostat cross-sections were stained with hematoxylin and eosin, dehydrated with successive 70, 95 and 100% ethanol washes, cleared with toluene and mounted with Permount (Fisher Scientific, Ottawa, ON, Canada). For SDH staining, cross-sections were air dried, incubated for 1 h at 37°C with SDH incubation medium (1.5 mm nitroblue tetrazolium chloride, 130 mm sodium succinate, in 0.1 M PBS, pH 7.4), rinsed and mounted with aqueous mounting medium.
RNA FISH
Cryostat cross-sections were covered with 40% formamide 2× saline-sodium citrate (SSC) for 10 min and incubated for 2 h with 10 ng of Cy3-labelled (CAG)10 oligonucleotide probe in hybridization buffer (40% formamide, 2× SSC, 0.2% BSA, 10% dextran sulfate, 2 mM vanadyl adenosine complex, 1 mg/ml tRNA and salmon sperm DNA). After washes, slides were mounted with Vectashield medium containing DAPI (Vector Labs/Cedarlane).
Image acquisition
Images were visualized on a Zeiss Axio Imager. M2 upright microscope, equipped with Zeiss 63× Plan-Apochromat 1.4 Oil, 40× Plan-Apochromat 1.4 Oil, 20× Plan-Apochromat 0.8, 10× Plan-Apochromat 0.45 and 5× EC Plan-NeoFluar 0.16 objective lenses. Images were acquired with Zeiss AxioCam MRm. Images were processed with the Zeiss AxioVision software, Northern Eclipse (Empix Imaging, Mississauga, Canada), Photoshop CS5 (Abobe Systems, San Jose, CA) or Image J (National Institutes of Health, Bethesda, MD). Muscle CSA, mean and CV (standard deviation of muscle fiber / mean muscle fiber diameter) were measured using Image J.
Western blotting
Dissected muscles were crushed in liquid nitrogen, and muscle powder resuspended in RIPA buffer (50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS and protease inhibitors [Complete; Roche/Sigma-Aldrich]) or Urea/Thiourea buffer (7 M urea, 2 M thiourea, 65 mm chaps, 100 mm DTT, 10 U DNase I, protease inhibitors [Complete; Roche/Sigma-Aldrich]). Protein concentration was determined using the BCA protein assay kit (Pierce/ThermoFisher Scientific) or CB-X Protein Assay kit (G-Bioscience, St. Louis, MO), respectively. A 20- to 30-μg amount of total proteins was separated by SDS-PAGE and transferred onto nitrocellulose membranes. Non-specific binding was first blocked with 1X PBS containing 5% skim milk, and membranes were then incubated with primary antibodies. After thorough washing with 1X PBS with 0.05% Tween, membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (Jackson Immunoresearch Laboratories/Cedarlane). After washes, signals were revealed using ECL reagents (Fisher Scientific) and autoradiographed with X-Ray films (Fisher Scientific). Quantifications were performed with the Image Lab software (Bio-Rad) or Image J (National Institutes of Health).
RNA extraction, reverse transcription, regular and real-time quantitative PCR
Total RNAs were extracted from samples using TriPure (Roche/Sigma-Aldrich). A 1-μg amount of RNA was DNase-treated (Ambion/ThermoFisher Scientific), and cDNAs were synthesized using MuLV Reverse Transcriptase (Applied Biosystems/ThermoFisher Scientific). PCRs were performed using GoTaq (Promega/Fisher Scientific), radioactive PCR as previously published (11) and real-time quantitative PCR (MX3005P, Stratagene, La Jolla, California, USA) using the QuantiTect SYBR Green PCR Kit (QIAGEN, Toronto, ON, Canada) according to the manufacturer’s instructions. The sequences of mouse primers for alternative splicing analyses were as follows: m-SERCA1 (fwd 5′-ATCTTCAAGCTCCGGGCCCT-3′, rev 5′-CAGCTTTGGCTGAAGATGCA-3′); m-RyR1 (fwd 5′-GACAATAAGAGCAAAATGGC-3′, rev 5′-CTTGGTGCGTTCCTGATCTG-3′) (23); m-TNNT3 (fwd 5′-TCTGACGAGGAAACTGAACAAG-3′, rev 5′-TGTCAATGAGGGCTTGGAG-3′), and m-CLCN1 (fwd 5′-GGAATACCTCACACTCAAGGCC-3′, rev 5′-CACGGAACACAAAGGCACTGAATGT-3′) (49). The sequences of human primers for alternative splicing were as follows: h-SERCA1 (fwd 5′-ATCTTCAAGCTCCGGGCCCT-3′, rev 5′-CAGCTCTGCCTGAAGATGTG-3′) (23); h-ZFAND1 (fwd 5′-GGCAGCGAGATTTTCTTCCA-3′, rev 5′-GCCACAAGTTCTCTCTCAGC-3′); h-MDM4 (fwd 5′-TCTCCGTGAAAGACCCAAGC-3′, rev 5′-GCTCTGAGGTAGGCAGTGTG-3′); and h-GPCPD1 (fwd 5′-TGGAAAGCAACCATTGTACTCAG-3′, rev 5′-TGTGGATTCCAAATTGTCCATCG-3′) (48).
MTT assay
Cells were seeded in 96-well plates with 100 μl of growth medium. Proliferative cells were incubated with an increasing dose of AICAR for 24 h. Then, cells were incubated with 0.5 mg/ml of MTT (Sigma-Aldrich) for 4 h. Cells were washed with PBS and formazan crystals solubilized with 100 μl of DMSO for 30 min. Absorbance was measured at 570 nm and no-cell background subtracted. Data were expressed as a percentage of viable cells compared to untreated control.
Statistical analysis
Student’s t-tests were used to determine whether differences between groups were significant. The level of significance was set at P-value ≤ 0.05 (*P-value ≤ 0.05, **P-value ≤ 0.01 and ***P-value ≤ 0.001). Means ± SEM are presented throughout.
Author Contributions
A.R.C. and B.J.J. designed the experiments. A.R.C., A.A.R. and G.B. performed the experiments. A.R.C., A.A.R., G.B. and B.J.J. analyzed the data. A.R.C. and B.J.J. wrote the manuscript.
Acknowledgements
The authors are grateful to J. Lunde for technical assistance with this study.
Conflict of Interest statement. None declared.
Funding
This work was supported by the Canadian Institutes of Health Research (CIHR); the Association Franse contre les Myopathies (AFM); and the Muscular Dystrophy Canada (MDC).
References
