Myotonic Dystrophy type 1 (DM1, OMIM# 160900) is caused by an expanded CTG tract in the 3′ untranslated region (3′ UTR) of the dystrophia myotonica protein kinase (DMPK) gene that mainly affects muscular and neuronal lineages (Brook et al., 1992; Mahadevan et al., 1992; Udd and Krahe, 2012). The unstable CTG repeat continues to expand over the patients' lifetime resulting in somatic mosaicism (Martorell et al., 1997). The longest CTG expansions have been observed in the most severely affected tissues, including muscle, brain and heart (López Castel et al., 2011). Individuals born with very large CTG expansions can manifest congenital DM1 (CDM), which is the most severe form of the disease (De Antonio et al., 2016; Johnson et al., 2016; Nakamori et al., 2017).

DM1 patients experience muscular symptoms such as muscle weakness, myotonia or loss of muscle strength during disease progression (Bouchard et al., 2015). CDM exhibits features that are not seen in adult or classic DM1 patients, including severe muscle fibre immaturity (Farkas-Bargeton et al., 1988; Nakamori et al., 2017; Sarnat, 2011).

At the molecular level, DMPK transcripts containing CUG expansions form toxic RNA foci which sequester splicing factors such as muscleblind-like (MBNL) proteins, while also increasing levels of CUG-binding protein 1 (CELF1), leading to altered RNA splicing events causing DM1-related symptoms (Masuda et al., 2012; Wang et al., 2012). In addition, the hypermethylation of the CpG island upstream of the CTG repeat is seen exclusively in CDM patients and has been suggested to be a biomarker for CDM, arguing in favour of its significant contribution to disease severity (Barbé et al., 2017).

The current knowledge of muscle-specific DM1 mechanisms has been obtained from either mouse models, post-mortem human tissues, patient biopsies or tissue-derived myoblasts. Human embryonic stem cells (hESC) have previously been shown to be a suitable model for DM1 (Seriola et al., 2011; Yanovsky-Dagan et al., 2015; Barbé et al., 2017), although the effect of the DM1 expansion on in vitro myogenesis from hESC that carry DM1 has not been reported before.

In this study, we modelled early myoblast and myotube development, starting from hESCs, to investigate differences between healthy and DM1 cells. We detected disease-specific mechanisms at early developmental stages and revealed DM1-specific cellular and molecular pathway deregulation over the time course of early myogenic differentiation.

Six hESC lines were subjected to myogenic differentiation: three non-DM1 cell lines (VUB01, VUB02, VUB06) and three DM1 cell lines (VUB03-DM1, VUB19-DM1, VUB24-DM1), carrying each a differently sized CTG repeat expansion in the DMPK locus (Seriola et al., 2011; De Temmerman et al., 2008) (Table 1).

Table 1.

hESC lines used in this work

All six hESC lines were differentiated into myoblasts, selected, and from this stock further differentiated three times to multinucleated myotubes (van der Wal et al., 2017). We attempted to differentiate VUB24-DM1 three times also; however, only one differentiation experiment yielded enough myotube material for the RNAseq experiments. VUB03-DM1 was differentiated three times independently (subline 1, 2, 3) in order to control for differentiation variability. VUB19-DM1 had the smallest repeat size that increased with only 22 repeats from hESC to myotubes (Jonckheere-Terpstra test, P=0.009), while the repeat in VUB03-DM1 increased with 151 repeats (Jonckheere-Terpstra, test P=0.001). VUB24-DM1 had the largest CTG repeat size and showed the highest size variability (Fig. S1, Table S1).

Transcriptome analysis was carried out by RNA sequencing for all lines at the hESC, myoblast and myotube stage on one sample each. Principle component analysis (PCA) of all samples showed that the myoblast samples clustered apart from myotubes and hESCs, confirming the presence of three different cell identities (Fig. 1A; Fig. S2). In addition, no significant differences based on FDR<0.05 were found when comparing DM1 with non-DM1 samples of the same cell type, including core myogenic regulatory genes MYOD and MYOG, and loss of markers of undifferentiated state (Fig. 1B).

Fig. 1.

Myogenic differentiation from hESC to the myoblast stage is equally successful in DM1 and control cell lines. (A) PCA plot of RNA sequencing results of all lines in the undifferentiated state, the myoblast and the myotube stage. The PCA is based on the results for coding genes with a count per million greater than one in at least two samples. (B) Expression of the undifferentiated state markers NANOG and POU5F1, and myogenic regulatory factors MYOD and MYOG over the course of myogenic differentiation, as measured by RNA sequencing. (C) Results of the FACS purification of the myogenic differentiation from hESCs towards myoblasts, using a C-MET+ and HNK1− selection for three non-DM1 lines and three DM1 cell lines (one line in triplicate). Data are shown as means±s.d., t-test P=0.1935. (D) GO terms for muscle-related gene sets that are significantly enriched in the differential gene expression between hESC and the myoblast stage, using Gene Set Enrichment Analysis. The plot only shows those GO terms that are exclusively enriched for DM1 (dark-grey bars) and non-DM1 (light-grey bars) samples. The full list of 32 commonly enriched gene sets can be found in the Table S2. (E) The heatmap shows the differential gene expression of the genes of the hallmark gene set ‘myogenesis’ of the molecular signatures database, for DM1 and non-DM1 hESC to myoblast differentiation. Genes in grey have an FDR>0.05. The grey bars indicate genes for which the expression level did not change between two developmental stages. The full list of included genes can be found in the Table S4. N/A, not applicable; ND, not determined; NS, not significant; TMM, trimmed mean of M values; SC, human embryonic stem cells; MB, myoblasts; MT, myotubes; FC, fold change; NES, normalized enrichment score.

Fig. 1.

Myogenic differentiation from hESC to the myoblast stage is equally successful in DM1 and control cell lines. (A) PCA plot of RNA sequencing results of all lines in the undifferentiated state, the myoblast and the myotube stage. The PCA is based on the results for coding genes with a count per million greater than one in at least two samples. (B) Expression of the undifferentiated state markers NANOG and POU5F1, and myogenic regulatory factors MYOD and MYOG over the course of myogenic differentiation, as measured by RNA sequencing. (C) Results of the FACS purification of the myogenic differentiation from hESCs towards myoblasts, using a C-MET+ and HNK1− selection for three non-DM1 lines and three DM1 cell lines (one line in triplicate). Data are shown as means±s.d., t-test P=0.1935. (D) GO terms for muscle-related gene sets that are significantly enriched in the differential gene expression between hESC and the myoblast stage, using Gene Set Enrichment Analysis. The plot only shows those GO terms that are exclusively enriched for DM1 (dark-grey bars) and non-DM1 (light-grey bars) samples. The full list of 32 commonly enriched gene sets can be found in the Table S2. (E) The heatmap shows the differential gene expression of the genes of the hallmark gene set ‘myogenesis’ of the molecular signatures database, for DM1 and non-DM1 hESC to myoblast differentiation. Genes in grey have an FDR>0.05. The grey bars indicate genes for which the expression level did not change between two developmental stages. The full list of included genes can be found in the Table S4. N/A, not applicable; ND, not determined; NS, not significant; TMM, trimmed mean of M values; SC, human embryonic stem cells; MB, myoblasts; MT, myotubes; FC, fold change; NES, normalized enrichment score.

After the first differentiation step, we found no significant differences in the percentage of C-MET+/HNK− cells in cell cultures by flow cytometry, suggesting that all lines differentiated with similar efficiency (Fig. 1C). We investigated the activation of muscle-related gene sets using gene-set enrichment analysis during the course of differentiation between DM1 and non-DM1 samples. In the differentiation from hESCs to myoblasts, we found 36 muscle-related GO gene sets that were enriched in non-DM1 samples, and 39 that were enriched in DM1 samples, 32 of which were in common (Fig. 1D; full list of pathways in Table S2). The heatmap in Fig. 1E shows the differentially expressed genes of the hallmark gene set ‘myogenesis’ (of the molecular signatures database). We only included those genes that were significantly differentially expressed in the non-DM1 lines (FDR<0.05 in non-DM1 hESC to myoblast differential gene expression). Table S4 shows all the included genes and their log2FC and FDR. Overall, both groups show a comparable expression pattern, with both groups similarly inducing muscle-related genes, confirming that DM1 and non-DM1 cell lines undergo the first part of myogenic differentiation with equal efficiency.

We then investigated the second stage of the differentiation from myoblasts towards myotubes and found that DM1 lines showed a statistically lower number of MF20+ cells (t-test, P=0.0003). VUB24-DM1 showed the lowest numbers of positive cells (Fig. 2A,B) and, moreover, VUB24-DM1 myotubes cluster together with the myoblasts in the PCA analysis (Fig. 1B). A gene set enrichment analysis of the differentiation from myoblasts to myotubes, showed 21 GO gene sets that were enriched in DM1 samples, versus 38 that were enriched in non-DM1 samples, of which 20 muscle related GO gene sets were commonly enriched (Fig. 2C; full list in Table S3). Taking into account the genes from the hallmark gene-set ‘myogenesis’, there is a considerable number of genes that are not significantly induced in the DM1 group, which are highly induced in the non-DM1 group (Fig. 2D; Table S5).

Fig. 2.

Myoblasts obtained from DM1-hESC have a decreased ability to progress to the myotube stage. (A) Percentage of nuclei within a myosin heavy chain (MF20) positive myotube (n=3). *** P=0.0003, t-test. (B) Immunostaining for myosin heavy chain (MF20) for three control non-DM1 cell lines and three DM1 cell lines after myotube differentiation. (C) GO terms for muscle-related gene sets that are significantly enriched in the differential gene expression between the myoblast and myotube stage, using Gene Set Enrichment Analysis. The plot only shows those GO terms that are exclusively enriched for DM1 and non-DM1 samples. The full list of gene sets, including 20 common pathways, can be found in Table S3. (D) The heatmap shows the differential gene expression the hallmark gene set ‘myogenesis’ of the molecular signatures database, for DM1 and non-DM1 myoblast to myotube differentiation. Genes in grey have an FDR>0.05. The grey bars indicate genes for which the expression level did not change between two developmental stages. The full list of included genes can be found in the Table S5. NES, normalized enrichment score; MB, myoblasts; MT, myotubes; FC, fold change.

Fig. 2.

Myoblasts obtained from DM1-hESC have a decreased ability to progress to the myotube stage. (A) Percentage of nuclei within a myosin heavy chain (MF20) positive myotube (n=3). *** P=0.0003, t-test. (B) Immunostaining for myosin heavy chain (MF20) for three control non-DM1 cell lines and three DM1 cell lines after myotube differentiation. (C) GO terms for muscle-related gene sets that are significantly enriched in the differential gene expression between the myoblast and myotube stage, using Gene Set Enrichment Analysis. The plot only shows those GO terms that are exclusively enriched for DM1 and non-DM1 samples. The full list of gene sets, including 20 common pathways, can be found in Table S3. (D) The heatmap shows the differential gene expression the hallmark gene set ‘myogenesis’ of the molecular signatures database, for DM1 and non-DM1 myoblast to myotube differentiation. Genes in grey have an FDR>0.05. The grey bars indicate genes for which the expression level did not change between two developmental stages. The full list of included genes can be found in the Table S5. NES, normalized enrichment score; MB, myoblasts; MT, myotubes; FC, fold change.

In order to further explore transcriptional differences in the myogenic differentiation we compared the top positively and negatively enriched Hallmark pathways in differential gene expression from hESCs to myoblasts and from myoblasts towards myotubes (Fig. 3). Interestingly, during the differentiation to myoblasts, we found that IL6-JAK-STAT3 signalling and TNFA signalling via NFKB were positively enriched in both experimental groups (Fig. 3A). On the other hand, the interferon alpha response, belonging to interferon type I, was positively enriched in the differentiation of the DM1 hESC to myoblast only. In the second step of differentiation, the interferon type I pathway appears negatively enriched only in the DM1 cells (Fig. 3A). Remarkably, the canonical WNT pathway was only activated in the non-DM1 myoblast-to-myotube transition. The mTORC1 signalling, an important pathway in myogenesis, was negatively enriched in both groups in the transition from myoblast to myotube (Rion et al., 2019).

Fig. 3.

Pathway analysis in DM1 and non-DM1 samples during myogenic differentiation shows differences in inflammatory response and mTORC1 signalling. (A) Top ten up- and downregulated pathways. The top panel presents myoblasts compared to hESC, the lower panel represents myotubes compared to myoblasts. (B) Heatmaps representing the log fold change of genes belonging to the interferon alpha response, mTORC1 and canonical WNT signalling. The grey bars indicate genes for which the expression level did not change between two developmental stages. The full list of genes, their log2FC and FDR are listed in Table S6. Grey lines indicate genes with FDR>0.05. NES, normalized enrichment score; FC, fold change.

Fig. 3.

Pathway analysis in DM1 and non-DM1 samples during myogenic differentiation shows differences in inflammatory response and mTORC1 signalling. (A) Top ten up- and downregulated pathways. The top panel presents myoblasts compared to hESC, the lower panel represents myotubes compared to myoblasts. (B) Heatmaps representing the log fold change of genes belonging to the interferon alpha response, mTORC1 and canonical WNT signalling. The grey bars indicate genes for which the expression level did not change between two developmental stages. The full list of genes, their log2FC and FDR are listed in Table S6. Grey lines indicate genes with FDR>0.05. NES, normalized enrichment score; FC, fold change.

We further investigated these pathways by analysing the expression of all genes in each pathway (as listed in the Hallmark gene sets). Fig. 3B and Fig. S3 show the heatmaps of the log2FC of all genes in each pathway with an FDR<0.05 in at least one of the groups. Table S6 lists all genes included in the heatmaps and their log2FC and FDR. Overall, the majority of genes in the three inflammatory pathways (IL6-JAK-STAT3 signalling, TNFA signalling via NFKB and interferon alpha response) are upregulated in the first part of the differentiation, while in the second part most genes are downregulated, and the only pathway showing differences between the two groups is the interferon alpha response.

In the interferon alpha response pathway, 77% of the genes are upregulated in the DM1 samples and 65% in the non-DM1. Of these upregulated genes, 71% were more strongly induced in the DM1 samples. During the second part of the differentiation, the majority of the interferon alpha response genes were downregulated in the DM1 samples while remaining mostly unchanged in non-DM1. These data support the observation that this pathway was specifically enriched in the DM1 samples during the first part of the differentiation.

With regards to the WNT signalling, the heatmap in Fig. 3B does not reveal pronounced differences in terms of significantly differentially expressed genes between the non-DM1 and DM1 samples, suggesting that the differential enrichment predicted by GSEA is likely not of strong biological relevance.

Finally, the mTORC1 shows striking differences between the two groups. The non-DM1 samples downregulate 87% (62/71) genes in the pathway, suggesting that the mTORC1 signalling is being repressed during the transition from myoblast to myotube. Conversely, the DM1 samples only downregulate 19% (14/71) of the genes, potentially unveiling abnormal activation of the mTORC1 signalling in DM1 myotubes.

In sum, the TNFα signalling via NFKB pathways is equally enriched in both groups. Conversely, the interferon alpha response pathway was only enriched in the DM1 samples, with a stronger up- and downregulation of its genes in the course of differentiation. Further, the mTORC1 signalling showed remarkable differences, with the DM1 cells failing to downregulate this signalling pathway during the progression from myoblast to myotube.

We analysed the methylation status of 23 CpG sites upstream of the CTG repeat, including the CTCF1 region, and 11 downstream CpG sites, spanning the CTCF2 region, in all lines included in this study, at the three stages of differentiation, hESCs, myoblasts and myotubes (Fig. 4).

Fig. 4.

The upstream CpG methylation in the DMPK locus increases over differentiation in DM1 samples but does not affect the expression of DMPK and its flanking genes. (A) Average methylation levels of the CpG sites upstream of the CTG repeat in each sample. The upstream methylation is shown for 23 CpG sites and all epi-alleles were analysed after massive parallel sequencing for three DM1 and three non-DM1 cell lines (one-way ANOVA: VUB03-DM1 1: P<0.0001 2: P<0.0001 3: P<0.0001, VUB24-DM1: P<0.0001, VUB19-DM1: P=0.2989, VUB02: P=0.710, VUB06: P=0.326, VUB01: P=0.977; *** indicates significant differences). (B) mRNA levels of DMPK, SIX5 and DMWD over the course of myogenic differentiation. Samples are grouped according to their CpG methylation status upstream of the CTG repeat. Data are presented as mean±s.e.m. TMM, trimmed mean of M values; SC, hESC; MB, myoblasts; MT, myotubes.

Fig. 4.

The upstream CpG methylation in the DMPK locus increases over differentiation in DM1 samples but does not affect the expression of DMPK and its flanking genes. (A) Average methylation levels of the CpG sites upstream of the CTG repeat in each sample. The upstream methylation is shown for 23 CpG sites and all epi-alleles were analysed after massive parallel sequencing for three DM1 and three non-DM1 cell lines (one-way ANOVA: VUB03-DM1 1: P<0.0001 2: P<0.0001 3: P<0.0001, VUB24-DM1: P<0.0001, VUB19-DM1: P=0.2989, VUB02: P=0.710, VUB06: P=0.326, VUB01: P=0.977; *** indicates significant differences). (B) mRNA levels of DMPK, SIX5 and DMWD over the course of myogenic differentiation. Samples are grouped according to their CpG methylation status upstream of the CTG repeat. Data are presented as mean±s.e.m. TMM, trimmed mean of M values; SC, hESC; MB, myoblasts; MT, myotubes.

In the undifferentiated state, two of the three DM1 lines showed methylated alleles for the upstream site, while the third line (VUB19-DM1) showed no methylation, as do the control lines. VUB03-DM1 and VUB24-DM1 carry expansions up to 1600 and 3600 repeats, respectively, while the largest expansions in VUB19-DM1 are only 600 repeats (Table 1;