TNF-α-Induced microRNAs Control Dystrophin Expression in Becker Muscular Dystrophy

Abstract

The amount and distribution of dystrophin protein in myofibers and muscle is highly variable in Becker muscular dystrophy and in exon-skipping trials for Duchenne muscular dystrophy. Here, we investigate a molecular basis for this variability. In muscle from Becker patients sharing the same exon 45-47 in-frame deletion, dystrophin levels negatively correlate with microRNAs predicted to target dystrophin. Seven microRNAs inhibit dystrophin expression in vitro, and three are validated in vivo (miR-146b/miR-374a/miR-31). microRNAs are expressed in dystrophic myofibers and increase with age and disease severity. In exon-skipping-treated mdx mice, microRNAs are significantly higher in muscles with low dystrophin rescue. TNF-α increases microRNA levels in vitro whereas NFκB inhibition blocks this in vitro and in vivo. Collectively, these data show that microRNAs contribute to variable dystrophin levels in muscular dystrophy. Our findings suggest a model where chronic inflammation in distinct microenvironments induces pathological microRNAs, initiating a self-sustaining feedback loop that exacerbates disease progression.

Figure 1. BMD Δ45–47 muscle shows variable dystrophin protein levels

(A) Western blot of BMD Δ45–47 muscle demonstrates variable dystrophin. Desmin and Coomassie-stain for myosin heavy chain (MYHC) used as loading controls. (B) Dystrophin transcript levels (RT-qPCR) do not correlate with dystrophin protein (Spearman’s correlation; r_s=−0.0332; p= 0.412). (C–E) DTMs increase in BMD ‘low’ dystrophin muscle. TaqMan TLDA miRNA arrays performed with ‘Normal’ (n=6) and BMD Δ45–47 (n=10) muscle. (C) Elevated DTMs in BMD muscle with low dystrophin. Table lists DTMs elevated in BMD ‘High’ and ‘Low’ muscle; fold changes and p-values shown (‘High’ >20% dystrophin (n=6) and ‘Low’ <20% dystrophin (n=4) vs. Normal (ANOVA with post-hoc contrast groups: Normal vs. ’Low’; Normal vs. ‘High’). (D) Inverse correlation between dystrophin and DTMs (defined in each sample using ≥1.5-fold increase vs. Normal). Plot shows miRNAs detected in individual muscle vs. % dystrophin (Spearman’s correlation; r_s=−0.726; ***p<0.0006). (E) DTMs inversely correlate with dystrophin protein. Plot shows miRNAs vs. % dystrophin protein in BMD and normal muscle (Spearman’s correlation r_s=−0.691;**p=0.0013; r_s =−0.610, **p=0.0048; r_s=−0.732, **p=0.0011 for miR-146b-5p, miR-374a, and miR-382, respectively). Refer also to Figure S1; Table S1.

Figure 2. miRNAs inhibit dystrophin protein translation in vitro

(A) Schematic of dystrophin 3′ untranslated region (UTR) reporter. The human dystrophin 3′UTR was cloned into the 3′ end of a Renilla reporter gene (psi-CHECK2 vector). The psi-CHECK2 vector co-expresses Firefly luciferase, and thus provided an internal transfection control. (B) miRNAs inhibit dystrophin 3′ UTR reporter activity. Individually, 14 dystrophin mRNA-targeting miRNAs were co-transfected with reporter into cells; % inhibition is provided in graph (n=4 replicates; ANOVA**p<0.01; ***p<0.001; ****p<0.0001 vs. negative (−) control). (C) Western blot of healthy human myotubes transfected with 50nM of indicated miRNAs. Tubulin (loading control) and densitometry values (% CTRL) are provided. (D) DTMs show synergistic inhibition. The 3 most potent DTMs (1nM, miR-146b, miR-374a, and miR-31) were transfected into cells individually, or in combination (referred to as ‘Biomix’); results reported as % inhibition (n=4 replicates; ANOVA, **p<0.01; ***p<0.001; ****p<0.0001 versus negative (−) control). (E) Schematic shows base-pairing of miRNAs with dystrophin 3′ UTR; called miRNA recognition elements or MREs. MRE mutants were constructed as shown; 4–5 nucleotide substitutions were made to reporter (mutated nucleotides in red). For miR-146a/b sequence x=c, y=a for miR-146b and x=t, y=g for miR-146a (blue). Mutagenesis was performed on 1 of 3 miR-374a MREs, however this mutant was anticipated to have little effect on reporter expression due to 2 non-mutated miR-347a MREs remaining (gray). (F) MRE mutagenesis reduces dystrophin inhibition. 50nM indicated miRNAs were co-transfected into cells along with dystrophin wild-type (white bars) or a MRE-mutant 3′UTR reporter (black bars). Mutated MRE construct matches transfected miRNA for each condition as indicated (n=4 replicates; Student’s t-test for wild-type versus mutant; #p<0.1; *p<0.05; **p<0.01). Refer also to Table S2.

Figure 3. DTMs reduce dystrophin in vivo

(A–B). miRNA pool (called “Biomix” with 70% or 1.05μg miR-146b; 25% or 0.375 μg miR-374a; and 5% or 0.075 μg miR-31) was injected into TA of 6-week old C57BL10/J mice (group termed miRNA, n=6/group). Equivalent amount of control was injected into left TA (group termed CTRL). Muscles were harvested 7 days later. (A) miRNA injection of wild-type mice to observe effects on steady-state dystrophin. Left; Representative immunofluorescence images overlaid with tattoo dye from brightfield to delineate site of injection; red=dystrophin; green=tattoo dye; white arrows denote where dystrophin levels are decreased in miRNA, but not in CTRL injected muscles; scale bars=100μM. Right; Average pixel count (dystrophin levels) around injection site (Student’s t-test; *p<0.05). (B–C) miRNA injection after injury in wild-type mice to observe effects of de novo dystrophin expression. Muscle injury was inflicted using notexin. Three days post-injection miRNAs were injected into the right TA; CTRL was injected into the left. Mice were sacrificed 7 days post-injury (n=3/group). (B) miRNAs reduce dystrophin expression post-injury. Western blot of CTRL or miRNA-injected muscle to show dystrophin. Loading controls are provided. (C) Left; Dystrophin immunostaining in CTRL and miRNA-injected mice. Central nucleation demarcates regenerated fibers (red=dystrophin, blue=DAPI). Right; average pixel count (dystrophin levels) per field (Student’s t-test; **p<0.01). Refer also to Figure S2, Table S2.

Figure 4. DTMs are elevated in dystrophic muscle and increase with age

(A) miRNAs are elevated in dystrophic dogs. Levels of miR-146b, miR-146a, miR-223 in the vastus lateralis (VL) muscle of 6-month old GRMD (n=9) compared to aged matched wild-type dogs (n=3) (Student’s one tailed t-test; #p<0.1, p<0.05, **p<0.01, ***p<0.001). (B) DTMs increase with disease progression. Levels of miR-146a, miR-146b, miR-223 in VL muscle biopsies of 1 and 6 month-old GRMD dogs (n=6/group). (C) DTMs increase with age in mdx mice. Left; DTMs in TA of mdx mice (12 days, n=3; 8 weeks, n=4). miR-223 and miR-31 levels are shown (Student’s t-test; ***p<0.001; *p≤0.05). Right; Representative H&E of cross-sections from TAs of 12-day and 8-week old mdx mice where scale bar = 100μM. (D–E) DTMs are elevated in whole extensor digitorum longus muscle (EDL) and in purified myofibers from mdx EDL. (C) DTMs in whole EDL of mdx or age-matched wild-type mice (Student’s t-test, n=4 per group; **p<0.01; *p<0.05; #p<0.1) (D) DTMs in purified myofibers from the contralateral EDL of the same mdx and wild-type mice from panel C. Note in C and D, miRNA upregulation is maintained in purified (n=4/group; Student’s t-test with Mann-Whitney correction for non-Gaussian distribution; **p<0.01; *p<0.05; #p<0.1). Refer also to Figure S3, Table S3.

Figure 5. DTMs are inversely correlated with exon skipping success in vivo

(A–B) 4-week-old mdx mice were given PMO (single high intravenous dose, 800 mg/kg) driving exon 23 skipping. 4 weeks post-treatment, muscles were analyzed for miRNA expression via RT-qPCR and for dystrophin via SILAM Mass Spectrometry (n=3 muscles). (A) miRNAs influence intra-variability in dystrophin rescue. Dystrophin and miRNA levels are shown for tibialis anterior (TA), gastrocnemius (gastroc) and diaphragm muscle from a single PMO-treated mouse (ANOVA, *p<0.05; **p<0.01; ***p<0.001). (B) Inter-subject variability in dystrophin rescue influenced by miRNAs. Plot of dystrophin protein as % wild-type (y axis) and a combinatorial score of 7 miRNAs (miR-146b, miR-374a, miR-31, miR-223, miR-146a, miR-382 and miR-320a scored as low, moderate or high, x axis; 63 total measures). Measurements determined using triceps, TA and gastroc of treated mdx (ANOVA, **p<0.01).

Figure 6. DTMs are induced by NFκB-mediated inflammation

(A) mdx H-2K myotubes treated with indicated drug, were induced with TNFα; DTMs assayed by RT-qPCR. miR-146a and miR-223 increase with TNFα; VBP15 or prednisolone (Pred) pre-treatment inhibits induction; miR-146b and miR-382 decreased with VBP15, but not Pred (n=5/group; ANOVA, ****p<0.0001, ***p<0.001, **p<0.01; *p<0.05; #p<0.1). (B) Muscles from 6-month-old mdx mice treated with Pred (5 mg/kg/day) or VBP15 (45 mg/kg/day) as described (Heier et al., 2013). miR-146a, miR-146b and miR-223 decreased with both drugs (n=8/group; ANOVA; ****p<0.0001, **p<0.01, *p<0.05). (C) Muscles from 8-week-old mdx mice treated with Pred (5 mg/kg/day) or VBP15 (15 mg/kg/day) as described (Heier et al., 2013). miR-146a and miR-223 are reduced by both drugs, whereas miR-382 increases with Pred, but not VBP15 (n=5/group; ANOVA; **p<0.01, *p<0.05). (D) DTMs associated with the NFκB pathway are preferentially elevated in “Severe” BMD muscle. Left; Representative H&E staining showing “Mild” and “Severe” BMD pathology. Right; miR-146a, miR-146b and miR-223 levels in BMD “Mild and “Severe” muscle (Student’s t-test **p<0.01; *p<0.05; #p<0.1). Scale bars=200μM. Refer also to Table S4, Figure S4.