Wnt/β-catenin signaling suppresses DUX4 expression and prevents apoptosis of FSHD muscle cells

Abstract

Facioscapulohumeral muscular dystrophy is a dominantly inherited myopathy associated with chromatin relaxation of the D4Z4 macrosatellite array on chromosome 4. DUX4 is encoded within each unit of the D4Z4 array where it is normally transcriptionally silenced and packaged as constitutive heterochromatin. Truncation of the array to less than 11 D4Z4 units (FSHD1) or mutations in SMCHD1 (FSHD2) results in chromatin relaxation and a small percentage of cultured myoblasts from these individuals exhibit infrequent bursts of DUX4 expression. There are no cellular or animal models to determine the trigger of the DUX4 producing transcriptional bursts and there has been a failure to date to detect the protein in significant numbers of cells from FSHD-affected individuals. Here, we demonstrate for the first time that myotubes generated from FSHD patients express sufficient amounts of DUX4 to undergo DUX4-dependent apoptosis. We show that activation of the Wnt/β-catenin signaling pathway suppresses DUX4 transcription in FSHD1 and FSHD2 myotubes and can rescue DUX4-mediated myotube apoptosis. In addition, reduction of mRNA transcripts from Wnt pathway genes β-catenin, Wnt3A and Wnt9B results in DUX4 activation. We propose that Wnt/β-catenin signaling is important for transcriptional repression of DUX4 and identify a novel group of therapeutic targets for the treatment of FSHD.

Figure 1.

Culture medium supplemented with KOSR improves myoblast differentiation and produces myotubes that express DUX4. (A) Immunofluorescent microscopy of primary human FSHD myoblasts (2082) differentiated using HS/ITS or KOSR and stained with an antibody specific for the C-terminus of DUX4 (red) and MHC (green). Nuclei are counterstained with DAPI (blue). Scale bar = 50 μm. (B) High magnification images of insets numbered in (A). Scale bar = 10 μm. (C) Scatter plot displaying the number of nuclei per cluster and demonstrating that myotubes differentiated in medium supplemented with KOSR contain large clusters that have more nuclei. *P < 0.05 by Mann–Whitney test for non-parametric distribution. (D) Fold change in the percentage of total nuclei positive for DUX4 immunofluorescence within MHC(+) myotubes for two non-FSHD and three FSHD-derived myoblasts when comparing HS/ITS or KOSR differentiation conditions. Results are pooled between the three cell lines (FSHD1(2082), FSHD2(2305) and FSHD2(1881). P < 0.05 based on Student's t-test. (E) Percentage of DUX4(+) nuclear clusters as a fraction of total number of nuclei in myotubes. (F) FSHD myoblasts were fused with mouse C2C12 myoblasts which do not carry the DUX4 gene. Mouse nuclei are distinguished by pronounced pericentric heterochromatin staining with DAPI and several examples are circled in yellow. DUX4(+) mouse nuclei (circled in yellow and immunostained red for DUX4) are indicated. (G) Quantitative RT-PCR of DUX4-activated genes, MBD3L2 and CCNA1 in three FSHD-derived cell lines [FSHD1(2082), FSHD2(2305), FSHD2(1881)]. Data is displayed as the change in expression level when myoblasts are differentiated in KOSR compared with HS/ITS. *P < 0.05 by Student's t-test.

Figure 2.

Myoblasts from FSHD-muscle biopsies die when efficiently differentiated to myotubes. (A) Drawing of photo arrangement shown in (B) showing a cytopathic lesion represented as a white circle. The microscope was programmed to take pictures (shown as blue squares) spanning the lesion. (B) Immunofluorescent microscopy of a 49 images spanning a 4 × 5 mm lesion from an FSHD myotube culture differentiated in medium supplemented with KOSR for 48 h. Green = MHC. (C) Numbered fields from (B) shown at increased magnification to demonstrate that nuclei adjacent to the lesion (Inset #2) are DUX4(+), whereas nuclei outside of the lesion are DUX4(−) (Inset #1). Red = antibody to C-terminus of DUX4. Green = antibody to MHC. Scale Bar = 50 μm.

Figure 3.

Optimization of DUX4 knockdown. (A) FSHD myoblasts were transfected with two siRNAs targeting DUX4 (DUX4-1 and DUX4-4). Twenty-four hours later, the medium was changed to KOSR differentiation medium and 48 h later harvested and assayed for DUX4 transcripts by RT-PCR. (B) Cells prepared as in (A), were assayed by qRT-PCR for expression of the two DUX4-activated genes, MBD3L2 and CCNA1. (C) Cells prepared as in (A) were fixed and stained for DUX4 (red) and MHC (green) and counterstained with DAPI (blue). Scale Bar = 50 μm.

Figure 4.

Myotube death is DUX4 dependent. (A) FSHD myoblasts were transfected with non-targeting (SCR) or DUX4-targeting siRNAs and differentiated into myotubes. Ninety-six hours after the initiation of differentiation, dishes were stained with Coomassie blue, and imaged at low magnification (×10) to demonstrate the widespread loss of myotubes in the culture. Scale bar = 0.5 mm. (B) FSHD myoblasts were transfected with non-targeting or DUX4-targeting siRNAs and differentiated into myotubes. Ninety-six hours post-differentiation, dishes were stained with antibodies to MHC and the myotube fusion index was calculated. P < 0.05 by Student's t-test. (C) Average number of contracted MHC(+) myotubes containing clusters of pyknotic nuclei per well of a 384-well dish 96 h following transfection of the DUX4-targeting siRNA. P < 0.05 by Student's t-test. (D) FSHD1(2349) myoblasts were transfected with non-targeting, DUX4-targeting and p53-targeting siRNA and stained for MHC 96 h post-differentiation. Scale bar = 50 μm. (E) FSHD1(2349) myoblasts cultured in 384-well dishes were treated with non-targeting, DUX4-targeting or FRG1-targeting siRNAs and assayed for myotube viability by staining with an anti-MHC antibody (green) and counterstained with DAPI (blue). Scale bar = 50 μm.

Figure 5.

Activation of the Wnt/β-catenin signaling reduces DUX4 expression in FSHD myotubes. (A) RT-PCR of non-FSHD and FSHD-derived cells cultured in myoblast proliferation or differentiation medium in the absence or presence of GSK3β inhibitor (GSKi). Transcription of the Wnt target gene, AXIN2, was monitored as a control for activation of Wnt/β-catenin signaling. (B) Primary FSHD(2349) myoblasts were differentiated in medium containing GSKi or 250 ng/ml recombinant Wnt3A (rWnt3A) or relevant vehicle controls (DMSO for GSKi or 0.1% CHAPS for rhWnt3A). Myotubes were fixed at 48 h to quantify DUX4 protein expression. (C) Immunofluorescent microscopy images of cells prepared in (B). Scale Bar = 50 μm. Myotube index for the image displayed in each condition is displayed in white text in the lower left corner of the panel (MI). (D) Primary FSHD(2349) myoblasts were cultured as in (B) and fixed at 96 h to measure myotube loss. Photos from the 96 h timepoint stained with MHC (green) are shown. The percentage of DUX4(+)/MHC(+) nuclei from a 48 h timepoint of the matching experiment are shown in white text in each quadrant. Scale bar = 200 μm. (E) Myotube index of experiments shown in (D). P < 0.05 by Student's t-test. (F) FSHD2 myotubes were treated with GSKi to confirm that the effect of Wnts on DUX4 protein levels was independent of D4Z4 chromatin structure. DUX4(+) nuclei were quantified. P < 0.05 by Student's t-test. (G) C2C12 myoblasts infected with a lentivirus encoding D4Z4(FFL) were cultured in the presence of absence of GSKi, and Luciferase activity was measured. P < 0.05 by Student's t-test.

Figure 6.

Knockdown of endogenous Wnt/β-catenin signaling pathway components activates DUX4. (A) Quantification of DUX4(+) nuclei from myotubes transfected with siRNAs targeting CTNNB1 (the gene encoding β-catenin) (siCTNNB1) or Wnt3A (siWnt3A.) P < 0.05 by Student's t-test (B) FSHD1(2349) myoblasts were differentiated in medium supplemented with KOSR and containing 100 nmol IWP2 or vehicle control. The fold change of DUX4(+) nuclei is shown. (C) FSHD1(2349) myoblasts were differentiated in medium containing rWnt3A with or without IWP2. DUX4 was quantified by immunohistochemstriy. P < 0.05 by ANOVA. (D) Non-FSHD(2401) myoblasts were seeded onto wells of at 384-well dish and transfected with siRNAs to the genes shown on the X-axis (si[GENE]). Wells were stained for DUX4 and inspected for DUX4 expression. Two DUX4(+) nuclear clusters were observed when Wnt9B was knocked down. P < 0.05 by ANOVA. (E) Knockdown of Wnt9B (siWnt9B) results in a 10-fold increase in DUX4(+) nuclei than siSCR controls. P < 0.05 by Student's t-test. (F) Model depicted two states of a truncated D4Z4 array existing in equilibrium in a myotube culture. On the left of the equation, silenced heterochromatinized arrays are shown coated in white circles. When the array reaches a chromatin conformation amenable for DUX4 transcription (right of the equation), the canonical Wnt pathway serves to repress DUX4 transcription.