Engineered FSHD mutations results in D4Z4 heterochromatin disruption and feedforward DUX4 network activation

Abstract

Facioscapulohumeral dystrophy (FSHD) is linked to contraction of D4Z4 repeats on chromosome 4q with SMCHD1 mutations acting as a disease modifier. D4Z4 heterochromatin disruption and abnormal upregulation of the transcription factor DUX4, encoded in the D4Z4 repeat, are the hallmarks of FSHD. However, defining the precise effect of D4Z4 contraction has been difficult because D4Z4 repeats are primate-specific and DUX4 expression is very rare in highly heterogeneous patient myocytes. We generated isogenic mutant cell lines harboring D4Z4 and/or SMCHD1 mutations in a healthy human skeletal myoblast line. We found that the mutations affect D4Z4 heterochromatin differently, and that SMCHD1 mutation or disruption of DNA methylation stabilizes otherwise variegated DUX4 target activation in D4Z4 contraction mutant cells, demonstrating the critical role of modifiers. Our study revealed amplification of the DUX4 signal through downstream targets, H3.X/Y and LEUTX. Our results provide important insights into how rare DUX4 expression leads to FSHD pathogenesis.

Keywords: Cell biology; Genomics; Molecular biology; Molecular mechanism of gene regulation.

Graphical abstract

Figure 1

Generation of SMCHD1 and/or D4Z4 mutant cells from health permissive skeletal myoblast (A) Western blot analysis the SMCHD1 protein expression in the cell lines used in the study. Lysates of immortalized control and FSHD1 and FSHD2 patient myoblasts, SMCHD1 mutants (SM), D4Z4 deletion mutants (DEL) and double mutants (DEL_SM) were subjected to western blot analysis using antibody specific for SMCHD1. β-Actin serves as a loading control. (B) Determination of the 4q and 10q D4Z4 repeat number. Examples of control and DEL1 mutant clones are shown. Genomic DNA was digested with EcoRI/HindIII (E/H) or EcoRI/BlnI (E/B) and subjected to PFGE. They were then blot-hybridized with the 4q/10q specific "1-kb" D4Z4 probe. E/H digestion leaves intact two 4q and two 10q D4Z4 arrays, while BlnI in an E/B only cleaves 10qD4Z4 repeat units. Size markers (in kb) are shown on the left. Arrowheads and stars indicate 4q and 10q D4Z4, respectively. The arrow indicates a band around 6.6 kb, which should be 2 D4Z4 repeat units caused by incomplete digestion. The two 4q D4Z4 repeat arrays are contracted, while the 10q D4Z4 bands size show no change. (C) D4Z4 gRNA targeting resulted in repeat contraction and recombination, leaving the last repeat with the DUX4 gene intact at the 4qA allele in DEL mutant cells. Top: schematic diagram of D4Z4 array in 4qA allele of parental cell with gRNA target sites for D4Z4 deletion (purple bars) as well as crRNA target sites designed for nanopore sequencing were shown at the top panel. D4Z4 cluster in 4qA, 4qB and 10q alleles of DEL3 were shown below. 10q D4Z4 sequences were confirmed by SNP analysis. The large triangle represented a 3.3 kb D4Z4 unit and its orientation. The small and partial triangle represented partial D4Z4 units and their orientation. The endonucleases (EcoRI/HindIII) cut sites, which generated the fragments detected in PFGE, are indicated. (D) Control, SMCHD1 mutant SM1 and D4Z4 contraction mutant DEL3 myoblast clones were differentiated and analyzed for DUX4 target (TRIM43, LEUTX and MBD3L2) RNA expression levels. Data are expressed as relative expression (mean ± standard deviation (SD)). Experiments are repeated (6 times for control, 5 times for SM1, and 8 times for DEL3). The gene expression over GAPDH was normalized to the corresponding gene expression value of the Control. ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001 vs. control by unpaired Student’s t test. (E) Corresponding data from double mutant (DEL4_SM_A) and FSHD2 were added to (D) for comparison. Experiments were repeated 10 times for DEL4_SM_A and 3 times for FSHD2 cells. Error bars indicate the standard deviation from the mean values. ∗∗∗p < 0.001. Multiple unpaired Student’s t-tests were performed, with each test comparing two groups of data. (F) DUX4 depletion by lentiviral shRNA abolished activation of DUX4 target genes (TRIM43, LEUTX and MBD3L2) compared to control shRNA (shCTRL). Cells were harvested at 4 days of myotube differentiation. DUX4 target expression levels were determined by RT-qPCR. Y axis is relative expression (mean ± SD) with the expression in shCTRL-transduced samples as one. ∗∗p < 0.01 and ∗∗∗p < 0.001 by unpaired Student’s t test, n = 3. (G) Stochasticity of target gene expression in the DEL3 cell line. Two aliquots of DEL3 (aliquots A and B) at different passage numbers and 3 replicates of a re-cloned DEL3 (total 23 samples) were differentiated for 4 days and analyzed for MBD3L2 expression by RT-qPCR. (H) Similar variegation was also observed as in (G) in 3 other DEL mutant clones (DEL1 (n = 8), DEL2 (n = 10) and DEL9 (n = 11). Control (n = 11). ∗p < 0.05 and ∗∗p < 0.01 by unpaired Student’s t test. (I) Comparison of MBD3L2 expression level in early myotube of Control (n = 11), single mutant DEL3 (n = 23), and double mutant DEL4_SM_A (n = 10). The dots on each boxplot represent the individual data, which was normalized to the mean value of Control, in each repeat. ∗∗∗p < 0.001 by unpaired Student’s t test.

Figure 2

Double mutants closely recapitulate patient cells (A) PCA analysis of myoblasts and early myotubes across all the cell types and clones. Top genes for each component are included in the Table S2. Differentiation days are indicated by shapes and cell types are indicated by colors according to the label legend. (B) Expression comparison of selected genes from PC1 and PC2 from (A). Top: boxplots of the selected top high genes expression of PC1 in control, FSHD patients and 3 types of mutant early myotubes. Bottom: boxplots of 7 DUX4 targets expression from the top 500 high genes of PC2. Expression values are in log₂ (normalized TPM +1). Significant values were calculated by Wilcoxon t-test (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05). (C) Volcano plots of significance (log₁₀ p-value) and log₂ fold change of double mutants (DEL_SM) compared to control at myoblast, early and late myotube stages. Significantly upregulated (blue) and downregulated (pink) genes and DUX4 targets (orange) are shown. DUX4 target genes in PC2 (B) are indicated in myoblasts and early myotubes. (D) Hierarchical heatmap of DUX4 target gene expression. A total of 63 target genes were selected based on previous studies.^, Expression values are in normalized TPM and log transformed. Gray shades indicate differentiation and colors indicate cell types.

Figure 3

Ontology analyses of common and distinct gene expression in patient and double mutant cells (A) The bubble plot shows gene ontology enrichment analysis of upregulated genes in the double mutants in myoblasts (pink), early (yellow) and late (blue) myotubes. The plot shows the selected top terms for each differentiation stage. X axis displays log₁₀ p-value and bubble size indicates number of genes in each term as indicated. (B) As in (A), bubble plot for downregulated genes. (C) Heatmap of log₂ fold change expression for the selective genes in FSHD patients and mutants and their related pathway. Asterisks indicate p value <0.01 and log₂FC > +1.5 or < −1.5. (D) The expression level of selected genes from (C) in double mutant myoblast (DEL4_SM_A) transduced with lentivirus carrying shControl or shDUX4. Real-time RT-qPCRs were performed for three biological replicates for each sample. Data are presented as mean ± SD; ∗∗p < 0.01, ∗∗∗p < 0.001, by one-tailed Student’s t test. Results presented as fold difference compared to shControl sample.

Figure 4

Heterochromatin changes in mutant cells (A) H3K9me3 ChIP-qPCR analysis of the DUX4 promoter region in FSHD1, FSHD2 and mutant myoblasts. Reduction of H3K9me3 is enhanced in double mutant cells. For both (A) and (B), signals were normalized to input of the corresponding samples. Error bars indicate the standard deviation from the mean values. Significant comparisons to the control are shown with the asterisks calculated by Student’s t test. The mean value of DEL mutant group is also compared to that of DEL_SM group using Student’s t test. (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05). (B) DNA methylation from MeDIP analysis. No reduction of DNA methylation was observed in SMCHD1 only and no additional effect in double mutant myoblasts. Error bars indicate the standard deviation from the mean values. Significant values were calculated by Student’s t-test (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001, ∗∗p < 0.01). (C) Comparison of DNA methylation levels at 4qA, 4qB, and 10q D4Z4 regions among the control, FSHD, and mutant cells. The MeDIP and input samples from (B) were amplified by using 4q/10q-D4Z4 specific PCR primers. The PCR products were sequenced and the 4qA, 4qB, and 10q D4Z4 specific sequence reads were analyzed. The 4qA(orange)/10q and 4qB(gray)/10q ratios of MeDIP were normalized with that of input. The data indicated relatively lower methylation at 4qA (but not 4qB) D4Z4 regions of FSHD1, D4Z4 deletion mutants and double mutants. p-values for significant differences versus the control sample are shown. Significant values were calculated by Student’s t-test (∗∗∗p < 0.001, ∗∗p < 0.01). All values are mean ± SD. (D) The effect of 5AzaC treatment on SMCHD1 mutants on DNA methylation. The control or SM1 cells were treated with 5AzaC for 24 h and allowed for 48 h of recovery before harvesting for MeDIP-qPCR. Error bars indicate the standard deviation from the mean values. Significant values were calculated by Student’s t-test (∗∗p < 0.01, ∗p < 0.05). (E) H3K9me3 ChIP-qPCR and MeDIP analysis at the DUX4 promoter region were performed to compare between Day 0 and Day 14 of DEL3 or Day 12 of DEL5 mutants. Error bars indicate the standard deviation from the mean values. Significant values were calculated by Student’s t-test (∗∗∗∗p < 0.0001, ∗∗∗p < 0.001).

Figure 5

Inhibition of DNA methylation increases DUX4fl RNA and protein and robustly upregulates target genes in mutant cells (A) Hierarchical heatmap of DUX4 target gene expression for control, DEL3, and DEL3_SM_A mutants with or without 5AzaC at myoblast and early myotube. A total of 63 target genes were selected based on previous studies.^, Expression values are in normalized TPM and log transformed. (B) DUX4 target genes were greatly affected by 5AzaC treatment in mutant cells. Control and SM1 myoblasts were treated with or without 5AzaC for 48 h. Then 5AzaC was removed from the media, and differentiation was induced. 4 days later, RT-qPCR of DUX4 target genes were performed. The gene expression data were normalized to GAPDH level in each sample, and then normalized to the LEUTX value of 5AzaC treated SM1. Data are presented as mean ± SD; ∗p < 0.05, by one-tailed Student’s t-test. (C) Comparison of DUX4 target genes level between early myotubes of 5AzaC treated SM and DEL mutants. SM and DEL myoblasts were treated with 5AzaC for 48 h right before differentiation. At day 5 of differentiation, the mRNA expression level of DUX4 target genes was assessed by real-time RT-PCR, relative to SM1. Data are presented as mean ± SD; ∗p < 0.05, by one-tailed Student’s t-test. Representative images of in situ detection of LEUTX RNA (red) with or without 5AzaC treatment are shown on the right (blue: DAPI). Scale bar 10 μm. (D) DUX4 depletion inhibited DUX4 target gene upregulation induced by 5AzaC treatment in mutant cells. SM1 and DEL3 cells were treated with 5AzaC and induced differentiation same as (B). During 5AzaC treatment, cells were infected with lentivirus containing shCTRL or shDUX4. For each cell line, LEUTX expression level after DUX4 depletion was shown as fold difference compared to the control. Data are presented as mean ± SD; ∗∗p < 0.01, by one-tailed Student’s t-test. (E) 5AzaC facilitated DUX4fl expression in DEL3 early myotubes. Left: the schematic diagrams of mRNA transcripts for DUX4fl, the DUX4s isoform and DUX4 homologs (DUX4c and DBET), the black regions, which represent >99% homology to DUX4fl, could be detected by corresponding DUX4 4ZZ probes or 2ZZ probes, but not by both. Therefore, the overlapping signals from 4ZZ and 2ZZ probes represent the DUX4fl transcripts. Middle panel, example images of the RNAScope results of the 4ZZ probes (green), 2ZZ probes (red) and the overlapping foci (yellow). DAPI is in blue. Scale bar = 10 μm. Right, 5AzaC treatment increased the percentage of myotubes with overlapping foci of 4ZZ and 2ZZ probes. Data from 3 independent experiments are presented as mean ± SD. ∗p < 0.05, by one-tailed Student’s t-test. (F) Examples of DUX4 protein expression in double mutant myoblasts and myotubes. Immunofluorescence for DUX4 on DEL4_SM_A cells after 5 days of differentiation. Nuclei were counterstained with DAPI (blue). Scale bar 50 μm. (G) Quantification of DUX4 protein expression with and without 5AzaC in double mutant myoblasts (left) and early myotubes (right). Mutant late myotubes and FSHD1 and FSHD2 patient early and late myotubes are shown for comparison. The DUX4 integrated density values in myoblasts/myotubes were measured using ImageJ software. Top 3% values in each group were used for graph and data analysis. All the data were normalized to the corresponding mean value of the 5AzaC-treated DEL_SM samples. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, by unpaired Student’s t test. Totally 600 myotubes or 1200 myoblasts were observed in each group. (H) Replotting the data in (G, right panel) for the frequency of DUX4 IF staining positive myotubes. All values are mean ± SD. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, by unpaired Student’s t test.

Figure 6

Identification of early and late DUX4 target genes that are differentially sensitive to myotube differentiation (A) Double mutant cells were treated with or without 5AzaC as indicated. IF signals of H3.X/Y and LEUTX in myoblasts (left panel) and LEUTX in early myotubes (right panel) were quantified as integrated intensity in each myoblast/myotube using ImageJ software. FSHD2 patient early myotubes are shown for comparison. The DUX4 integrated density values in each myoblast/myotube were measured using ImageJ software. Based on the highest positive myoblasts/myotubes number of all, same number of values in each group were used for graph and data analysis. All the data were normalized to the corresponding mean value of the 5AzaC treated samples (release day1 for the myoblasts). Red dots represent mean values. (n = 300 myotubes or 1000 for myoblasts). ∗∗∗p < 0.001, by one-tailed Student’s t-test. (B) Incorporation of H3.X/Y into DUX4 targets in control and double mutant at Day 4 with or without 5AzaC. Cells were treated with 5AzaC for 48 h before differentiation. Significant incorporation of H3.X/Y is shown by the asterisks with the indicated comparisons. Significant values were calculated by Student’s t-test (∗∗p < 0.01, ∗p < 0.05). All values are mean ± SD. (C) Identification of early and late DUX4 target genes that are insensitive and sensitive to myotube differentiation, respectively. Boxplots of representative early and late DUX4 target gene expression in double mutant cells were shown to compare myoblast, early and late myotube stages as indicated. Expression values are in log₂ normalized TPM. (D) The effects of H3.X/Y or DUX4 shRNA depletion on LEUTX expression. Double mutant DEL4_SM_A myoblasts were infected with lentivirus expressing control, H3.X/Y or DUX4 shRNA and differentiated into myotubes. RNA was harvested at early myotubes stage (differentiation days 3–5) as described in the STAR methods section. Three biological replicates for each sample were performed. Data are presented as mean ± SD; ∗∗p < 0.01, ∗∗∗p < 0.001, by one-tailed Student’s t test. Results presented as fold difference compared to shControl differentiated sample. (E) Control and DEL4_SM_A myoblasts were transduced with a lentiviral empty vector or a lentiviral vector expressing H3. X. Differentiation was induced at 48 h after transduction. For myoblasts or early myotubes as indicated, the mRNA expression level of the downstream target genes was assessed by real-time RT-qPCR. Data are presented as mean ± SD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, by one-tailed Student’s t test. Results presented as fold difference compared to empty vector infected double mutant cells. (F) Similar experiments as in (E), but DEL4_SM_A myoblasts were transduced with a lentiviral vector expressing LEUTX. Data are presented as mean ± SD; ∗p < 0.05, ∗∗p < 0.01, by one-tailed Student’s t test. Results presented as fold difference compared to empty vector infected double mutant cells. (G) Overexpression of H3.X and LEUTX in MB or MT was assessed by western blot. Pan histone H3 antibody was used as control as indicated. Lanes 1 and 2: mock transfection. Lanes 3 and 4: H3.X OE. Lanes 5 and 6: LEUTX OE. The endogenous LEUTX is upregulated in H3.X OE myotubes (lane 4). (H) TF binding motifs at the promoter of H3.X/Y. Binding motifs for DUX4 and the putative LEUTX motif (OTX2) within 1 kb upstream and 0.5 kb downstream of the transcription start site for H3.X/Y were identified by using the MoLoTool provided in HOCOMOCO v11, with p-values less than or equal to 0.001. Visualization was done on the UCSC genome browser using GENCODE v36 for the H3.X/Y genes model. (I) The effects of control, H3.X/Y or LEUTX overexpression (as in E–G) on 63 DUX4 target genes in DEL_SM myotubes Day 5 are assessed by RNA-seq and displayed in boxplots. p-values are calculated using Wilcoxon t-test indicated at the top. (J) Similar analysis was performed with H3.X/Y or LEUTX shRNA depletion compared to the same control as in (I) on DUX4 target gene expression in DEL_SM myotubes Day 7.

Figure 7

Schematic models of the establishment of the FSHD gene expression phenotype For a Figure360 author presentation of this figure, see https://doi.org/10.1016/j.isci.2024.109357. (A) Positive feedback loop between H3K9me3 and SMCHD1. SMCHD1 interacts with D4Z4 chromatin in an H3K9me3-dependent manner and also maintain H3K9me3. (B) Coherent feedforward mechanism of DUX4 and target gene expression. While DUX4 is critical for the initial activation of its target genes, the early target H3.X/Y expression is essential for efficient expression of other downstream target genes, including the late target TF, LEUTX. LEUTX in turn promotes further expression of H3.X/Y. H3.X/Y as well as LEUTX (and possibly other DUX4 target TFs) contribute significantly to the expression of other DUX4 target genes. (C) Two key processes in FSHD pathogenesis. D4Z4 heterochromatin disruption induced synergistically by D4Z4 and SMCHD1 mutations (and/or other epigenetic modifiers) enables stabilization and enhancement of DUX4fl expression. Once activated by DUX4, DUX4 target genes undergo cross-regulation contributing to the establishment of the FSHD gene expression phenotype.