Defective Regulation of MicroRNA Target Genes in Myoblasts from Facioscapulohumeral Dystrophy Patients

Background: FSHD is characterized by the overexpression of double homeobox genes DUX4 and DUX4c . Results: We found 29 miRNAs differentially expressed between FSHD and normal myoblasts. Twelve of these miRNAs were up-regulated in myoblasts ectopically expressing DUX4c. Conclusion: DUX4c is linked to the abnormal miRNA expression profile observed in FSHD. Significance: We observe a defective gene regulation by miRNAs in FSHD.

Facioscapulohumeral muscular dystrophy (FSHD), 5 a dominant neuromuscular disease, is one of the most frequent muscular dystrophies in the Western world with a prevalence of 1 in 20,000. This dystrophy is characterized by weakness and atrophy of specific groups of muscles of the face, shoulder girdle, and lower extremities (1). The FSHD locus has been mapped to the subtelomeric region on chromosome 4q35, which contains an array of 3.3-kb-long macrosatellite repeats (D4Z4) (2). The length of this array varies from 35 to 300 kb in healthy subjects but is consistently smaller than 35 kb in FSHD patients (3).
Each D4Z4 repeat contains a functional promoter, an open reading frame for DUX4 (double homeobox protein 4) (4,5), and a number of regulatory elements (reviewed in Ref. 6). In healthy subjects, DUX4 is expressed during embryogenesis and down-regulated in the course of development (7,8). In general, its expression level is higher in the skeletal muscles of FSHD patients than in the muscles of healthy controls; however, it is not unusual to find FSHD samples with a normal level of DUX4 expression (7)(8)(9). Therefore, although DUX4 is generally considered as an FSHD inducer gene, other genes might contribute to the pathological phenotype: FRG1, FRG2 (FSHD region genes 1 and 2), ANT1 (adenine nucleotide translocator), and DUX4c (double homeobox 4, centromeric). Most of these genes have also been shown to be up-regulated in skeletal muscle tissue of some FSHD patients and primary myoblasts derived from them (for a review, see Ref. 6).
Transcriptional profiling performed in FSHD cells has demonstrated a defect in the myogenic differentiation program (9 -13), deregulation of genes related to oxidative stress (9,14,15), and deregulation of vascular smooth muscle and endothelial cell-specific genes (16) as well as cell cycle-related genes (17).
Ectopic overexpression of several 4q35 genes in mouse tissues or immortalized myoblasts cultured in vitro recapitulated some features of FSHD, suggesting that these genes could indeed contribute to the FSHD phenotype. Specifically, DUX4c and DUX4 have been shown to inhibit myogenic differentiation; DUX4 induced oxidative stress (18,19) and atrophy of myoblasts cultured in vitro (20); FRG1 overexpression in mouse muscles induced muscle atrophy (21).
The mechanism of the overexpression of these functionally important 4q35 genes in FSHD may be attributed, at least partially, to a modification of the nuclear matrix attachment prox-imally to the D4Z4 array (22) and subsequent perturbation of the three-dimensional structure of the FHSD locus (23,24). Furthermore, we have previously shown that the D4Z4 repeats contain a potent transcriptional enhancer (25,26), which interacts with the Krüppel-like transcription factor KLF15 and activates the expression of DUX4c and FRG2 genes (27).
The complexity of gene regulation in FSHD has been further enhanced by the recent finding that non-coding RNAs are implicated in epigenetic regulation of FSHD-related genes (28) BIO, muscle biopsies; MB, myoblasts; MT, myotubes; ND, non-determined number of D4Z4 repeats.

Sample
Type Patient  Tissue  Age  Sex  D4Z4 repeats  Source   N1  MB, MT  NO42  Quadriceps  24  Female  ND  This study  N2  MB, MT  NO44  Quadriceps  29  Male  ND  This study  N3  MB, MT  NO46  Quadriceps  31  Male  ND  This study  N4  MB, MT  NO47  Quadriceps  43  Male  ND  This study  F1  MB, MT  MO44  Pyramidal  54  Female  5/7  This study  F2  MB, MT  MO47  Quadriceps  38  Female  7  This study  F3  MB, MT  MO54  Quadriceps  25  Male  4  This study  F4  MB, MT  FSHD10  Trapezius  31  Male  5  This study   TABLE 2 Oligonucleotides used in the study The coordinates are given relative to the TSSs of the corresponding primary microRNAs (see Fig. 8).
(for a review, see Ref. 29). D4Z4 was found to be transcribed in both directions, and the resulting non-coding RNAs were shown to modulate DUX4 transcription (30). A high throughput analysis of microRNA expression in FSHD cells revealed several miRNAs differentially expressed in FSHD cells (17,31). Here, we have analyzed miRNA expression in primary myoblasts from healthy subjects and FSHD patients and found 29 miRNAs differentially expressed in FSHD samples. Twelve of these miRNAs, including myogenic microRNAs miR-1, miR-206, miR-133a, and miR-133b were induced by DUX4c overexpression, suggesting that DUX4c is a novel regulator of miRNA expression. Despite overexpression of several myogenic microRNAs, we did not observe a premature myogenic differentiation of FSHD myoblasts. The expression analysis of target genes of these miRNAs revealed that some of them are not repressed by myogenic miRNAs in FSHD cells.

EXPERIMENTAL PROCEDURES
Cell Culture Conditions and siRNA Transfection-The rhabdomyosarcoma cell line TE-671 (a kind gift of Dr. S. Leibowitz) was grown as described (26). Primary human myoblasts were isolated from skeletal muscles of healthy subjects as described (32) (for details, see Table 1), purified with an immunomagnetic sorting system (MiltenyiBiotec) using an anti-CD56/NCAM antibody according to the manufacturer's specifications. CD56positive myoblasts were seeded in collagen-coated Petri dishes (P1) and cultured in DMEM, 10% FCS, 1% Ultroser G at 37°C with 5% CO 2 . All experiments were carried out between P1 and P5 to avoid cell senescence. Myoblast purity was determined by staining for desmin. Proliferating primary human myoblasts were transfected as described (20), and RNA was prepared 24 h after transfection. The growth conditions of human immortalized myoblasts (a kind gift of Dr. V. Mouly) have been described previously (33). siRNA transfection has been performed using siPORTNeoFX reagent (reverse transfection protocol) as described (20). pcDNA3-MYOD plasmid was a kind gift of Anna Polesskaya.
Biopsies-The biopsies have been obtained in accordance with French national regulations. The origins of biopsies are listed in Table 1.
Reporter Gene Assays-5 ϫ 10 3 TE-671 cells were plated onto 24-well plates and after 24 h were cotransfected with 0.5 g of pGL3-based miR-1/206 or miR-133a/b reporter plasmids containing the luciferase gene fused to the 3Ј-UTR of the Hand2 (34) or Whsc2 gene (35), respectively, and 0.5 g of pCIneo-DUX4c, pCIneo-DUX4, or a GFP-coding plasmid (Stratagene) using JetPEI (Polyplus). Luciferase activity was determined 48 h after transfection using the Luciferase Assay System (Promega) and normalized to protein concentration (determined by BCA assay; Sigma) or the Dual-Luciferase Assay System and normalized to the activity of the reporter phRL-TK (Promega).
Chromatin Immunoprecipitation-Chromatin was prepared from 10 ϫ 10 6 primary myoblasts and myotubes from FSHD patients and normal controls and sonication-sheared, and 5 g was used for immunoprecipitation with 1 g of H3K9me3 antibodies (catalog no. ab8898, Abcam) or preimmune rabbit or mouse IgGs (Abcam) using the ChiP-IT Express Kit (Active Motif). Immunoprecipitated DNA was then amplified using specific primers (see Table 2) and TaqMan 2ϫ Gene Expression Master Mix (catalog no. 4369016, Applied Biosystems) on a custom TLDA array (Applied Biosystems). PCR amplification and data acquisition were performed using AB7900HT realtime PCR machine (Applied Biosystems). To study DUX4c binding to myomiR promoters, TE-671 cells were transfected with DUX4c-coding plasmid, and chromatin was immunoprecipitated using DUX4c antibodies (36).  Table 1 for sample description). Square intensity indicates the miRNA expression level normalized to the control (RNU44). Gray gradations were scaled from minimal and maximal values independently for each miRNA.
Transcriptome Profiling-Human primary myoblasts were sacrificed directly on plates at 30% confluence using TRIzol. RNA was prepared using organic extraction and ethanol precipitation as described (38), followed by silica column cleanup on silica columns (Nucleospin RNA extraction kit, Macherey Nagel). RNA extracted from individual myoblast lines was Cy3labeled, mixed with a pool of RNA samples labeled with Cy5, and hybridized to Gene Expression microarrays (4 ϫ 44,000, catalog no. G4112F, Agilent) and scanned as instructed by the manufacturer. Scanned images were then analyzed using the Feature Extraction software (Agilent), and the treatment of the gene expression data was performed using R and Bioconductor. Spots with intensity lower than 50 or lower than background in more than 50% of biological replicates were removed from further analysis. The background correction and intensity normalization procedures were applied for the remaining ϳ30,000 probes using the Bioconductor package vsn (39). A background offset and a scaling factor for each array and dye channel were calculated using the least squares regression procedure, and then the generalized log transformation was applied. The ordinary least squares regression approach is based on the assumption that "most genes are not differentially expressed." However, in the case of myogenic differentiation, where many genes are differentially expressed, this assumption does not hold. Therefore, to apply the above mentioned approach to myogenic differentiation, the vsn least squares model was first applied to a subset of features and then extended for the whole set of features. To select the subset of features, a pool of samples prepared from five different proliferating myoblast lines before and after myogenic differentiation was hybridized to two additional microarrays. Then 14,358 features that did not exceed the cut-off value of 1.23-fold change between the pools were selected. To determine the differentially expressed genes, a t test analysis was conducted using the limma package from Bioconductor (39). Using this package, a linear model was fitted to the expression data for each gene. An empirical Bayes moderation of the S.E. values was performed. This method borrows information across genes in order to arrive at more stable estimates of each individual gene's variance, even for experiments with a small number of arrays.

Identification of miRNA Differentially Expressed in FSHD-
miRNA expression profiles in total RNA extracted from primary myoblast populations originating from four FSHD patients and four normal individuals were compared using a high throughput TaqMan qRT-PCR approach. To avoid contamination with connective tissue and inflammatory cells, we used an affinity purification procedure to isolate CD56 ϩ myoblasts. Of 365 microRNAs tested, 29 (hereafter called FSHDrelated microRNAs (FR-miRs)) were differentially expressed in FSHD myoblasts as compared with the controls (Fig. 1, A and B, and supplemental Table S1). These include 20 miRNAs up-regulated and 6 down-regulated in FSHD myoblasts (Fig. 1A). Differential expression of some of these FR-miRs, including miR-133a, miR-206, miR-21, and miR-100, was also discovered in FSHD cells in a pilot analysis of miRNA transcriptome using an Ambion microarray in FSHD myoblasts performed prior to this study (data not shown). miRNA expression has been also profiled in the same myoblasts following the induction of myogenic differentiation. Three microRNAs, miR-7, miR-21, and miR-594, were found differentially expressed in the resulting FSHD myotubes (Fig.  1B). In total, of 365 miRNAs tested, the expression of 186 miRNAs, including miR-517*, which was shown previously to be differentially expressed in FSHD muscles (31), could not be detected in our qRT-PCR assay.
Previously, miR-186, miR-15a, miR-23b, and miR-411 have been found to be differentially expressed in FSHD myoblasts (17,40). Of these, we could only confirm up-regulation of miR-15a (supplemental Table S1). Other miRNAs, including miR-30e and miR-331, shown previously to be overexpressed in Duchenne muscular dystrophy (31), were not found to be differentially expressed in the present study (supplemental Table S1). Therefore, all but one FSHD-related miRNAs identified in our study were not previously associated with FSHD.
The increase in the amounts of mature miRNAs in FSHD myoblasts could be explained by an increased transcription of their genes, resulting in elevated expression of both pri-miRNA precursors and mature miRNAs; alternatively, an elevated rate of miRNA precursor processing could result in the increase in mature forms with or without a decrease in the expression of pri-miRNAs. Classical myogenic microRNAs miR-1, -133a/b, and -206 (41) were among the miRNAs most strongly overexpressed in proliferating FSHD myoblasts (Figs. 1A and 3B and supplemental Table S1). We have studied the expression of their precursors in the FSHD and control cells. Human myogenic miRNA precursors pri-miR-133a-2 (chr. 20) (see Fig. 8), pri-133a-1, and pri-miR-1-2 (chr. 18) were significantly upregulated in FSHD myoblasts, whereas pri-miR-1-1 (chr. 20), pri-miR-206, and pri-miR-133b (chr. 6) demonstrated a trend for up-regulation in FSHD cells as compared with the normal   Fig. 2A). As expected, the expression of human myogenic miRNA precursors and their mature forms was strongly up-regulated following myogenic differentiation in vitro of normal and FSHD myoblasts (Figs. 2A and 3B). In FSHD myotubes, we have observed a decrease in the expression level of all myogenic miRNA precursors ( Fig. 2A) but not in the expression level of mature forms of these miRNAs as compared with controls (Fig. 3B).
Next, we have used ChIP followed by qPCR to examine the chromatin structure of promoters of miR-1, miR-133a/b, and miR-206 before and after myogenic differentiation and in FSHD. As expected, we have observed a significant reduction of the heterochromatin-specific histone 3 modification H3K9me3 on these promoters. A similar reduction of H3K9me3 was observed in proliferating FSHD myoblasts as compared with normal controls, supporting our observation that the expression of myogenic miRNA genes is higher in FSHD cells (Fig. 4). We conclude that the up-regulation of the mature forms of myogenic microRNAs can be explained by the up-regulation of their precursors in myoblasts and myotubes from the FSHD patients.
FR-miRs Are Differentially Expressed during Myogenic Differentiation-A defect of myogenic differentiation in FSHD has been documented previously by several groups (9, 13, 32).
Here, we have found that classical myogenic microRNAs, miR-1, -133a, -133b, and miR-206 were up-regulated in FSHD myoblasts. This prompted us to test whether the expression of other FR-miRs is also myogenesis-dependent. Intriguingly, nine FR-miRs up-regulated in FSHD myoblasts were also up-regulated during normal myogenic differentiation (Fig. 3A).
The up-regulation of myogenesis-dependent miRNAs in proliferating FSHD myoblasts could be explained by a premature myogenic differentiation of FSHD cells. We thus measured the expression of myogenic markers, including desmin, myosin heavy chain 1, and others, and did not find evidence of their up-regulation in FSHD myoblasts (Fig. 3C). Therefore, the elevated expression of myogenesis-related FR-miRs in FSHD myoblasts was not due to their premature myogenic differentiation.
DUX4c Activates FR-miRs-Recent studies have indicated that the DUX4 and DUX4c transcription factors are overexpressed in FSHD and could play an important role in the onset of this disease (18 -20). To test whether DUX4 and DUX4c could be responsible for the observed up-regulation of FR-miRs in FSHD, we overexpressed these transcription factors in rhabdomyosarcoma TE-671 cells. Twelve of FR-miRs up-regulated in FSHD were found to be also up-regulated by DUX4c overexpression (Fig. 5A). Interestingly, six of these FR-miRs, including  Table 2; their positions relative to the transcription start site of the corresponding pri-miRNA are indicated below). Relative enrichment was calculated as target DNA enrichment compared with input (percentage of input) normalized to percentage of input of the Alu repeat. B, DUX4c is bound to the promoter region of miR-1-1/miR-133a-2 pri-miRNA on chromosome 20 and miR-1-2/miR-133a-1 pri-miRNA on chromosome 18. Chromatin was immunoprecipitated using antibodies against DUX4c or preimmunization IgGs and quantified using qPCR. The RNU48 promoter region was used as a control, and data are expressed as percentage of input; *, p Ͻ 0.5; **, p Ͻ 0.01 (Student's t test). Error bars, S.E.

TABLE 3 MicroRNAs differentially expressed in FSHD myoblasts (MB) and myotubes (MT) and during myogenic differentiation induced in normal human myoblasts
*, this study. Arrows pointing upward indicate up-regulation, and those pointing downward indicate down-regulation. -, no change. ND, microRNA is undetectable.

FSHD MT (this study)
Normal myogenesis

DUX4 (this study)
the classical myogenic microRNAs miR-1, miR-133a, miR-133b, and miR-206, were also up-regulated during normal myogenic differentiation (Table 3). DUX4 overexpression mostly inhibited the miRNA expression with the exception of two microRNAs, miR-32 and miR-206, that were up-regulated in FSHD and induced by DUX4 overexpression (Fig. 5A). We have next tested the effect of DUX4 and DUX4c overexpression in TE-671 cells on the expression of myogenic microRNA precursors (pri-miRs). We have found that DUX4c overexpression led to the up-regulation of pri-miR-133b and pri-miR-206 precursors from chromosome 6 and pri-miR-1 and pri-miR-133a precursors from chromosome 20 but not from chromosome 18 (Fig. 2B). DUX4 overexpression did not significantly alter the expression of myogenic microRNA precursors (Fig. 2B). Forced expression of MYOD can induce myogenic differentiation of rhabdomyosarcoma cells, although inefficiently Primary myoblasts from one normal subject and one FSHD patient were transfected via siRNA against DUX4c, DUX4, or control siRNA. A, the expression level of miRNAs was quantified via qRT-PCR using the ⌬⌬Ct method using RNU44 as control gene and control transfected myoblasts as a reference sample (expression level was arbitrarily set to 1). A representative experiment is shown. B, the expression level of DUX4 and DUX4c or GAPDH genes was measured via RT-PCR. Insets show close-ups of miRNA expression data in proliferating myoblasts. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001; #, statistically significant difference (p Ͻ 0.05) as compared with the myoblasts transfected with the same siRNAs. MB, myoblasts; MT, myotubes. Error bars, S.E. (42,43). We overexpressed MYOD in TE-671 rhabdomyosarcoma cells and observed moderate up-regulation of pri-miRNA precursors of myogenic microRNAs. Interestingly, DUX4c knockdown prevented MYOD-dependent up-regulation of pri-miR-1 and -133a on chromosome 20 and pri-miR-133b and pri-miR-206 on chromosome 6 (Fig. 2C). Furthermore, we have found that DUX4c was bound to the promoter regions of pri-miR-1 and -133a on chromosomes 18 and 20 (Fig. 4B).
To confirm that the myogenic miRNAs induced by DUX4 and DUX4c, were functional, we next transfected immortalized myoblasts with DUX4-and DUX4c-expressing plasmids along with miR sensor plasmids containing a luciferase reporter gene fused to the 3Ј-UTRs of a gene containing either miR-133a/b (35) or miR-1/206 (34) recognition sites. DUX4c overexpression resulted in a reduced expression of both sensors but not of their mutant versions, indicating that DUX4c-induced myogenic miRNAs were functionally active, whereas the effect of the DUX4 overexpression on the miR reporter activity was not statistically significant (Fig. 5B).
Furthermore, to test whether DUX4c is required for the upregulation of the expression of myogenic microRNAs during myogenic differentiation, we have knocked down DUX4 and DUX4c expression in myoblasts and myotubes originating from one healthy donor and one FSHD patient and tested the expression of mature myogenic microRNAs in these cells. We have observed that DUX4c knockdown resulted in a significant decrease of the expression of all four myogenic microRNAs (miR-1, miR-133a, miR-133b, and miR-206) in normal and FSHD myoblasts (Fig. 6, A and B). We conclude that DUX4c might be required for the up-regulation of the myogenic microRNAs during myogenic differentiation in both normal and FSHD cells. Taken together, these results indicate that DUX4c contributes to the up-regulation of myogenic microRNAs during myogenic differentiation of both normal and FSHD cells.
Target Genes Escape the Repression by Myogenic miRNAs in FSHD-Our observation that myogenic microRNAs are up-regulated in FSHD myoblasts could be explained by a premature myogenic differentiation of these cells. However, myogenic differentiation markers were not up-regulated in FSHD (Fig.  3C). We therefore hypothesized that the overexpression of myogenic microRNAs observed in FSHD cells could be insufficient for the induction of the myogenic differentiation; alternatively, the functional activity of these miRNAs could be perturbed. miRNAs regulate gene expression either at the level of transcript stability by inducing deadenylation and degradation of mRNA or by inhibiting the translation of their target genes (44). If a gene is controlled at the level of transcript stability by a given miRNA, its expression level will be low when the expression level of the corresponding miRNA is high; in other words, the expression levels of an miRNA and its target gene should be inversely correlated. To test whether the differential expression of miRNAs between FSHD and normal myoblasts had a functional significance, we thus decided to investigate the expres-

Functional classification of genes targeted by miR-1, -133a, -133b and -206 in normal and FSHD myoblasts
The number of target genes down-regulated during myogenic differentiation of normal or FSHD myoblasts is indicated. *, less than 50% of target genes of a given functional class down-regulated during normal myogenesis are down-regulated during myogenic differentiation of FSHD myoblasts. The complete list of target genes of each microRNA analyzed can be found in supplemental Table S2. sion level of their predicted target genes. The majority of validated microRNA target genes that are currently available from the public databases miRTar Base and MiRWalk are regulated at the level of translation, but these are outnumbered by genes regulated by miRNAs at the level of transcription (45,46).
We thus chose to focus on target genes that are potentially regulated by myogenic miRNAs at the level of transcript stability. For this purpose, the following strategy was adopted. First, the RNA22 algorithm was used to predict target genes for each FR-miR (47). Then the expression level of RNA22-predicted target genes was tested in samples with high and low expression levels of FR-miRs (normal myotubes and myoblasts, respectively). Only those target genes with expression levels inversely correlated with FR-miR expression levels were considered as "supported by microarray data" (supplemental Table S2). We then analyzed the expression of supported myogenic miRNA target genes in FSHD as compared with normal myoblasts. In FSHD myoblasts with myogenic microRNAs up-regulated, a significant number of microarray-supported target genes were not down-regulated (Table 4). In the case of genes of miR-1 and miR-133b, only 57.6 and 68.4% of microarray-supported target genes were found down-regulated in FSHD myoblasts, respectively, indicating that myogenic microRNAs fail to down-regulate the totality of their microarray-supported target genes in FSHD myoblasts. Functional classification of the microarray-supported targets of the myogenic microRNAs has shown that in agreement with previous reports, they might be involved in the control cell cycle and myogenesis-related gene expression (41). In addition, these microRNAs could be implicated in the regulation of kinase activity, DNA damage response, ubiquitination, and others (Fig. 7, Table 4, and supplemental Table S3).
Interestingly, although some functions of myogenic microRNAs were well preserved in FSHD myoblasts, others seemed to be severely compromised with less than 50% of genes still down-regulated. These include cell cycle regulation, DNA damage response, ubiquitination/proteolysis, chromatin organization and modification, and cytoskeleton organization (Fig. 7,  green). Taken together, our data provide evidence of an abnormal expression of myogenic miRNA target genes in FSHD.

DISCUSSION
Significant progress in the understanding of the pathogenesis of FSHD has been made possible by a combination of genomic (8,48,49), transcriptomic (9 -11, 13, 17), and proteomic approaches (11,14). Recently, non-coding RNAs have emerged as major players in FSHD (for a review, see Ref. 29). miRNA profiles of FSHD cells have been reported in several studies (17,31), but so far, no model explaining the miRNA deregulation observed in FSHD has been put forward.
Here we have identified a specific miRNA profile of FSHD myoblasts and myotubes (Fig. 1). In total, we have found 29 Of these, only three microRNAs were found to be differentially expressed in FSHD myotubes versus controls. Intriguingly, miRNAs differentially expressed in myoblasts and myotubes are not the same. A loss of differential expression of the majority of FR-miRs in differentiated myotubes may be explained by a higher variability of miRNA expression that we observed in myotubes. Some FR-miRs described in this study were previously found to be overexpressed in other muscular dystrophies (e.g. miR-1 and miR-133 in Duchenne muscular dystrophy and miR-107 in limb-girdle muscular dystrophy) (31, 50). We have not been able to confirm the expression of some miRNAs found to be differentially expressed in FSHD by others (miR-186, -23, -411, and -517*) (17,31,40). The reason for this discrepancy might be the difference in cell culture procedures (e.g. using CD56 ϩ myoblasts versus non-purified myoblasts).
A significant proportion of the FR-miRs identified in this study are also up-regulated during normal myogenesis, among them the classical myogenic miRNAs miR-1, miR-133a, miR-133b, and miR-206 (Fig. 8) that are under control of the myogenic differentiation factors MyoD, Mef2, and SRF (51). However, the up-regulation of myogenesis-dependent miRNA in FSHD cannot be explained by a premature onset of myogenic differentiation in FSHD myoblasts, because myogenic markers were not up-regulated in FSHD cells. This suggests that one or FIGURE 8. Genes coding for human myogenic microRNAs miR-1, miR-133a, miR-133b, and miR-206. miRNA genes are transcribed into 5Ј-capped and 3Ј-polyadenylated primary microRNA precursors (pri-miRs) that are processed by Drosha/Pasha endonucleases into pre-miRs, which are exported to cytoplasm, where the final Dicer-dependent processing step and mature miRNA production take place (for a review, see Ref. 44). Positions of PCR primers used for pri-miR expression analysis via qRT-PCR and chromatin structure analysis via ChIP are indicated with black and white arrows, respectively. Human miR-1 and miR-133a are encoded within two common bicistronic genes on chromosomes 20 and 18. A, intron-exon structure and genome context (Ensembl) of the putative pri-miR precursor of miR-1-1 and miR-133a-2 on chromosome 20. The transcription start site (TSS) of this precursor is defined by the c20orf166 TSS according to CAGE tags, TSS seq, and H3K4me3 ChIP data (60). B, intron-exon structure and genome context (Ensembl) of the putative pri-microRNA precursor of miR-1-2 and miR-133a-1 on chromosome 18. The TSS of this precursor is given according to Ref. 60. The structure of the human miR-206 and -133b genes is currently unknown; however, it has been suggested that it might be similar to that of the corresponding mouse genes (61). In mice, miR-133b is expressed as a common precursor containing both miR-206 and miR-133b; miR-206, however, is expressed from its own internal promoter (61). C, putative intron-exon structure and genome context (Ensembl) of primary microRNA precursor of miR-206 and miR-133b on chromosome 6 deduced from its mouse homolog linc-MD1 (61); genomic coordinates are given according to genome assembly hg19; TSSa is according to Ref. 62; TSSb is according to Ref. 60. Genomic coordinates are given according to human genome assembly hg19. Pre-miRNA coordinates were extracted from Ensembl. Genomic coordinates of pri-miRNA precursors, including their putative TSSs, were extracted from the Refs. 60, 62, and 63. Conversion of genomic coordinates, if needed, was performed via the UCSC Genome Browser LiftOver utility. more factors distinct from classical myogenic differentiation regulators drive the overexpression of these miRNAs in FSHD myoblasts.
One of these factors could be DUX4c, an inducer of the MYF5 gene, involved in myogenic determination (36). Previously, we had shown that DUX4c is overexpressed in FSHD myoblasts and myotubes (27,36). DUX4c is overexpressed in most FSHD patients tested, although in some patients, DUX4c is missing from the chromosome 4q with the shortened D4Z4 repeat array (52,53), which argues against a causal role of this gene in the FSHD etiology. Unfortunately, in these patients, neither DUX4c expression levels nor miRNA profiles have been determined; therefore, we cannot exclude the possibility that DUX4c originating from a homologous chromosome contributes to the pathological phenotype in these patients.
Here, we have obtained multiple lines of evidence that DUX4c is an activator of the expression of the myogenic microRNAs. We have demonstrated that DUX4c overexpression induced expression of myogenic microRNAs and their precursors in TE-671 rhabdomyosarcoma cells. Conversely, DUX4c knockdown resulted in down-regulation of microRNAs in primary myotubes from normal subjects and FSHD patients. Interestingly, the effect of DUX4c knockdown on the expression of myogenic microRNAs was stronger in FSHD as compared with control myotubes. At the moment, we cannot conclude whether the stronger impact of DUX4c knockdown on the expression of myogenic miRNAs in FSHD cells is a general phenomenon, because only one cell line from a normal control and an FSHD patient was used in the study; however, our results clearly indicate that DUX4c might be a novel regulator of the expression of myogenic microRNAs.
Ectopic overexpression of any of the four classical myogenic microRNAs is sufficient for induction of the myogenic differentiation in skeletal myoblasts (54). However, premature differentiation is not observed in FSHD myoblasts expressing all four myogenic miRNAs. Failure to induce myogenic differentiation can be explained by a low level of overexpression of myogenic miRNAs in FSHD myoblasts; alternatively, one can suggest that the function of these miRNAs is impaired in FSHD myoblasts.
Indeed, we observed that several target genes of the myogenic miRNAs were not down-regulated in FSHD myoblasts, as would be expected if they were still under control of these miRNAs (Fig. 4). A similar effect has been observed previously in a mouse model with simultaneous up-regulation of a microRNA, namely miR-206, and its target gene urotropin (55). Functional classification of myogenic miRNA target genes indicated that whereas most functions remained virtually unchanged, genes specifically related to cell cycle control, DNA damage response, and ubiquitination escaped miRNA-dependent repression in FSHD myoblasts.
There are several possible explanations for this phenomenon. First, the miRNA-dependent repression pathway could be impaired in FSHD. Second, some other factors could override the miRNA-dependent regulation of these genes in FSHD. In favor of the former model, genes encoding the components of RNAi machinery, including all four Argonaute genes and the Dicer 1 gene have been reported to be down-regulated in FSHD myotubes (9), suggesting that miRNA-dependent transcript destabilization and pre-miRNA processing might be perturbed in FSHD cells. In addition, altered pre-miR processing might lead to generation of miRNA isoforms with an altered spectrum of target genes (for a review, see Ref. 56). In favor of the latter model, it has been demonstrated that several potent transcription factors, including DUX4 (7-9) and KLF15 (27), are up-regulated in FSHD. DUX4 up-regulates the expression of the cyclin A gene (57,58), which could, at least in part, explain why cell cycle-related genes escape the control of FR-miRs. Furthermore, up-regulation of one miRNA target gene may lead to saturation of RNAi machinery and a consequent reduction in the miRNA-dependent repression of other target genes (59).
The specific escape of cell cycle and DNA damage responserelated genes from FR-miR control, whatever the mechanism is, indicates that FSHD myoblasts simultaneously express two competing biological programs: (i) the myogenic differentiation program as indicated by the overexpression of myogenesisrelated microRNAs orchestrated by DUX4c and the concomitant repression of the majority of their target genes and (ii) the proliferation program as indicated by the overexpression of cell cycle and DNA damage-related genes. Because the successful completion of myogenic differentiation requires cell cycle arrest, the simultaneous expression of two incompatible programs may explain why FSHD myoblasts demonstrate a defect in myogenic differentiation, as observed previously (10,11,13,14), rather than prematurely engage into the myogenic differentiation program.