The H19 locus controls fetal growth by regulating expression of several genes from the imprinted gene network (IGN). H19 is fully repressed after birth, except in skeletal muscle. Using loss-of-function H19Δ3 mice, we investigated the function of H19 in adult muscle. Mutant muscles display hypertrophy and hyperplasia, with increased Igf2 and decreased myostatin (Mstn) expression. Many imprinted genes are expressed in muscle stem cells or satellite cells. Unexpectedly, the number of satellite cells was reduced by 50% in H19Δ3 muscle fibers. This reduction occurred after postnatal day 21, suggesting a link with their entry into quiescence. We investigated the biological function of these mutant satellite cells in vivo using a regeneration assay induced by multiple injections of cardiotoxin. Surprisingly, despite their reduced number, the self-renewal capacity of these cells is fully retained in the absence of H19. In addition, we observed a better regeneration potential of the mutant muscles, with enhanced expression of several IGN genes and genes from the IGF pathway.
INTRODUCTION
The H19 gene produces an imprinted maternally expressed non-coding RNA (Brannan et al., 1990). This transcript is very abundant during murine embryonic development and is as highly expressed as β-actin (Poirier et al., 1991). A highly conserved stem-loop structure in the first exon of the gene was discovered to produce a microRNA, called miR-675 (Pfeifer and Tilghman, 1994; Smits et al., 2008). Distinct roles during embryogenesis were identified for the long non-coding RNA and for miR-675 (Gabory et al., 2009,, 2010; Keniry et al., 2012). The production of mice carrying a targeted deletion of the gene, the H19Δ3 strain, allowed us to identify an overgrowth phenotype in the absence of the H19 gene. This phenotype is first detected in E15.5 embryos and continues in adult mice (Ripoche et al., 1997). Using both loss-of-function H19Δ3 and gain-of-function H19 transgenic (H19Tg) animals, we identified nine other imprinted genes whose expression was increased in the absence of H19 and was rescued by the transgene (Gabory et al., 2009). These genes belong to an imprinted gene network described by several authors (Varrault et al., 2006; Lui et al., 2008; Berg et al., 2011; Al Adhami et al., 2015). Our results suggest that H19 controls growth of the embryo through a balanced expression of imprinted genes.
We showed that this control of the imprinted gene network (IGN) in the embryo is driven by an interaction of the H19 non-coding RNA with the MBD1 methyl-CpG binding protein. This complex directly affects the expression of Igf2, Mest (Peg1) and Slc38a4, by modulating the H3K9me3 histone marks on the differentially methylated regions (DMRs) of these genes (Monnier et al., 2013). This suggests that imprinted genes are subject to epigenetic mechanisms on their DMRs that convey a strict control of their expression to produce a fine-tuned regulation of embryonic growth.
miR-675 has been shown to negatively regulate placental growth at the late stages of gestation, by affecting the expression of Igf1r. Its excision from exon 1 and its processing by Drosha into a pre-miRNA is inhibited by the HuR protein in early gestation placenta as well as in the embryo (Keniry et al., 2012).
H19 is strongly repressed after birth in all murine tissues, but it remains expressed in skeletal muscle and heart in the adult, suggesting an important function in these tissues (Poirier et al., 1991). Interestingly, H19 was also originally identified under the name of MyoH in the same selective screen that pinpointed MyoD as a key factor in muscular differentiation (Davis et al., 1987). Recently, it has also been observed that miR-675 promotes differentiation of myoblast cell lines (Dey et al., 2014).
Our aim was to investigate the role of H19 in vivo during adult myogenesis in our H19Δ3 model. Myogenesis is determined by several myogenic regulatory transcription factors, the MRFs, which include Pax7, Pax3, Myf5, Myod, myogenin and Mrf4 (Sabourin and Rudnicki, 2000; Buckingham, 2007; Buckingham and Relaix, 2007). Muscles are composed of multinucleated myofibers, resulting from the differentiation of myoblasts under the control of these different MRFs. During embryogenesis, primary and secondary myogenesis result in the production of the myoblasts and myotubes composing the final adult muscle fibers. Satellite cells, characterized by the expression of Pax7, are located at the surface of the myofibers and were identified as muscle stem cells. The satellite cells appear and proliferate in the embryo at the end of gestation (E16.5) and their number is considered definitive around 21 days after birth. These satellite cells then enter into quiescence and remain in this state all through the adult life. However, they are activated upon injury of a muscle and contribute to the regeneration capacity of muscle tissue, by proliferation and differentiation steps in order to produce newly formed myofibers. This is accompanied by an important self-renewal step to help maintain this stem cell population (Sambasivan and Tajbakhsh, 2007; White et al., 2010). Interestingly, the IGN appears to play a role in the satellite cells as shown by the expression of many imprinted genes in these cells (Yan et al., 2003; Berg et al., 2011; Al Adhami et al., 2015).
We show here that the overgrowth phenotype of the H19Δ3 mice is associated with muscle hypertrophy and hyperplasia. This phenotype is linked with higher proliferation of H19Δ3 myoblasts compared with wild-type (wt) myoblasts and with higher expression of Igf2 and reduced expression of myostatin (Mstn) in adult muscle. H19 is highly expressed in satellite cells and this led us to investigate the status of these cells in the mutant muscles. Unexpectedly, a 50% reduction in the number of satellite cells was observed in adult H19Δ3 mice compared with wt controls. To evaluate the biological function of these satellite cells in vivo, we investigated their regeneration capacity and their self-renewal status.
RESULTS
H19Δ3 muscles display hypertrophy and hyperplasia
H19Δ3 mice display a general increase in mass of 8-10% compared with wt littermates (Ripoche et al., 1997; Gabory et al., 2009; Fig. S1). We linked this phenotype to muscle overgrowth because the tibialis anterior mass in H19Δ3 mutants showed an increase of 25% compared with wt animals (6-week-old mice; Fig. 1A). To identify the cause of this overweight phenotype, tibialis muscles were collected from 6-week-old male adults and sections were labeled with laminin (Fig. 1B). The cross-section area (CSA) was determined and showed both hypertrophy (2-fold increase, Fig. 1C,D) and hyperplasia (1.3-fold increase, Fig. 1E) of H19Δ3 muscles compared with wt muscles.
H19 deletion induces muscle hyperplasia and hypertrophy. (A) Mass of individual tibialis muscles from wt and H19Δ3 6-week-old males (n=12 and n=15, respectively). (B) Histological section of tibialis muscle from wt and H19Δ3 mice immunostained with anti-laminin antibody. (C) Histogram of fiber CSA (µm2) from tibialis of adult males (n=3/genotype), mean of fiber CSA is also represented as a bar graph (D). (E) Number of fibers per tibialis section from adult males (n=3 mice/genotype). (F) Transcript levels of genes from the IGF pathway and of myostatin (Mstn) were quantified by RT-qPCR in adult tibialis muscle (n=4 mice/genotype). (G,I) Histograms of fiber CSA (µm2) from tibialis of P6 males (n=4/genotype) (G), and of P21 males (n=3/genotype) (I). (H,J) Number of fibers per tibialis section from P6 males (n=4/genotype) (H), and from P21 males (n=3/genotype) (J). Data are shown as means±s.e.m. Asterisks indicate that data are statistically significant (*P<0.05, **P<0.01, ***P<0.001; Student's t-test). CSA, cross-section area.
H19 deletion induces muscle hyperplasia and hypertrophy. (A) Mass of individual tibialis muscles from wt and H19Δ3 6-week-old males (n=12 and n=15, respectively). (B) Histological section of tibialis muscle from wt and H19Δ3 mice immunostained with anti-laminin antibody. (C) Histogram of fiber CSA (µm2) from tibialis of adult males (n=3/genotype), mean of fiber CSA is also represented as a bar graph (D). (E) Number of fibers per tibialis section from adult males (n=3 mice/genotype). (F) Transcript levels of genes from the IGF pathway and of myostatin (Mstn) were quantified by RT-qPCR in adult tibialis muscle (n=4 mice/genotype). (G,I) Histograms of fiber CSA (µm2) from tibialis of P6 males (n=4/genotype) (G), and of P21 males (n=3/genotype) (I). (H,J) Number of fibers per tibialis section from P6 males (n=4/genotype) (H), and from P21 males (n=3/genotype) (J). Data are shown as means±s.e.m. Asterisks indicate that data are statistically significant (*P<0.05, **P<0.01, ***P<0.001; Student's t-test). CSA, cross-section area.
To determine whether these phenotypes are established during the postnatal stage, tibialis muscles were collected from H19Δ3 and wt mice at postnatal day (P) 6 and P21. No difference in the CSA between the mutant and the wt muscles was found at either of these stages; therefore, hypertrophy is only established after P21 (Fig. 1G,I). By contrast, hyperplasia begins earlier than hypertrophy: the number of fibers was slightly higher at P6, although the difference was not statistically significant (12% increase, P=0.24), but by P21, H19Δ3 tibialis muscles showed significant hyperplasia (39% increase; Fig. 1H,J).
Two major signaling pathways control skeletal muscle growth: the insulin-like growth factor (IGF) pathway acts as a positive regulator of muscle growth, and the myostatin pathway acts as a negative regulator. To investigate these pathways, we performed expression analysis by RT-qPCR on wt and H19Δ3 adult muscles. We identified an increase in the level of Igf2 expression (50% increase) and a reduction of Mstn expression level (−50%) in the absence of H19 (Fig. 1F). Expression of other members of the IGF pathway such as Igf2r, Igf1 and Igf1r showed no significant difference between H19Δ3 and wt samples.
We concluded from these observations that hyperplasia is present at the postnatal stage. The number of muscle fibers is normally set between E18 and birth (White et al., 2010). Our results suggest that an increased proliferation leading to hyperplasia occurs in the H19Δ3 muscles, probably during the late fetal stage. By contrast, hypertrophy sets in only at the adult stage in the mutant muscles. The expression profiles of Igf2 and Mstn strongly suggest that these genes are responsible for the hypertrophic phenotype of the mutant mice.
H19 deletion induces depletion of satellite cells
Because many imprinted genes, including H19, are expressed in satellite cells, we investigated whether the absence of H19 would have an effect on these cells. We first evaluated the expression of Pax7 – a specific marker of satellite cells – in whole tibialis muscles by RT-qPCR (Fig. 2A). Interestingly, there was a significant decrease (−35%) in the expression of this gene in H19Δ3 muscles compared with wt muscles.
H19 controls the number of satellite cells. (A) Transcript level of the Pax7 gene was quantified by RT-qPCR in adult tibialis muscle (n=4 mice/genotype). (B) Immunostaining of satellite cells. Arrowheads indicate Pax7+ cells (green) on isolated myofibers. Nuclei are counterstained with DAPI (blue). (C) Mean number of satellite cells per EDL myofiber of wt and H19Δ3 adult mice (n=mean of 130 fibers from four mice/genotype). (D) Mean number of satellite cells per EDL myofiber of wt, H19−/+ and H19−/+;Tg adult mice (n=4 mice/genotype). (E) Mean number of satellite cells per EDL myofiber of wt and H19Δ3 P21 mice (n=3 mice/genotype). (F) EDL cryosection from P6 males immunostained with Pax7 (green) and laminin (red) antibodies. Nuclei are counterstained with DAPI (blue). Arrowheads indicate Pax7+ cells. (G) The mean number of Pax7+ cells per 100 fibers/section represented as a bar graph (n=4/genotype). (H) EDL cryosection from P21 males immunostained with Pax7 (green) and Ki67 (red) antibodies. Nuclei are counterstained with DAPI. Laminin staining is shown in white in the merged image. Arrowheads indicate only Pax7+ cells. Stars indicate Pax7+/Ki67+ positive cells. (I) The ratio of Pax7+/Ki67+ versus Pax7+/Ki67− cells from either wt or H19Δ3 sections is represented as a bar graph (n=4/genotype). Data are shown as means±s.e.m. Asterisks indicate that data are statistically significant [*P<0.05, ***P<0.001; Student's t-test (A,C,E,G,I) and Kruskal–Wallis test with Dunn's post test (D)].
H19 controls the number of satellite cells. (A) Transcript level of the Pax7 gene was quantified by RT-qPCR in adult tibialis muscle (n=4 mice/genotype). (B) Immunostaining of satellite cells. Arrowheads indicate Pax7+ cells (green) on isolated myofibers. Nuclei are counterstained with DAPI (blue). (C) Mean number of satellite cells per EDL myofiber of wt and H19Δ3 adult mice (n=mean of 130 fibers from four mice/genotype). (D) Mean number of satellite cells per EDL myofiber of wt, H19−/+ and H19−/+;Tg adult mice (n=4 mice/genotype). (E) Mean number of satellite cells per EDL myofiber of wt and H19Δ3 P21 mice (n=3 mice/genotype). (F) EDL cryosection from P6 males immunostained with Pax7 (green) and laminin (red) antibodies. Nuclei are counterstained with DAPI (blue). Arrowheads indicate Pax7+ cells. (G) The mean number of Pax7+ cells per 100 fibers/section represented as a bar graph (n=4/genotype). (H) EDL cryosection from P21 males immunostained with Pax7 (green) and Ki67 (red) antibodies. Nuclei are counterstained with DAPI. Laminin staining is shown in white in the merged image. Arrowheads indicate only Pax7+ cells. Stars indicate Pax7+/Ki67+ positive cells. (I) The ratio of Pax7+/Ki67+ versus Pax7+/Ki67− cells from either wt or H19Δ3 sections is represented as a bar graph (n=4/genotype). Data are shown as means±s.e.m. Asterisks indicate that data are statistically significant [*P<0.05, ***P<0.001; Student's t-test (A,C,E,G,I) and Kruskal–Wallis test with Dunn's post test (D)].
We then compared the number of satellite cells present on muscle fibers from H19Δ3 and wt mice. For these experiments, the extensor digitorum longus (EDL) muscle was used because of easier dissection of the muscle held by the tendon. Individual fibers were immunostained with anti-Pax7 antibodies and DAPI counterstaining (Fig. 2B). The number of satellite cells present on fibers showed a striking 50% reduction in adult H19Δ3 fibers, compared with wt fibers (Fig. 2C). In addition to this experiment, the number of satellite cells was compared in muscle fibers from H19−/+, wt and H19−/+;Tg littermates. The results showed a similar, although less pronounced, decrease in the number of satellite cells (−25%) in H19−/+ compared with wt muscle (Fig. 2D). The difference in the number of satellite cells in the two experiments is probably due to the different backgrounds on which these mice were maintained, 129/SV for the first and C57BL/6/CBA/129 for the experiment including the transgene. Interestingly, in H19−/+;Tg mice, the presence of the H19 transgene was able to rescue the reduced number of satellite cells and restored values similar to counts in wt muscles.
Because satellite cells develop during late gestation and continue to proliferate during the postnatal stage up to P21, we performed anti-Pax7 labeling on P6 and P21 muscles. For the P21 stage, individual fibers were isolated from the EDL muscle and immunostained with anti-Pax7 antibodies and DAPI as above (Fig. 2E). For the P6 stage, because of the small size of the muscle samples, Pax7 detection was performed on frozen sections and included a co-labeling with anti-laminin antibodies and DAPI counterstaining (Fig. 2F,G). A comparison of H19Δ3 and wt muscles showed no difference in the number of satellite cells at these early stages. The reduction in the number of satellite cells therefore occurs during adult life. This could suggest that the establishment of quiescence is delayed in the mutant mice, with H19Δ3 satellite cells continuing to be incorporated and fused into muscle fibers later than the P21 stage and thus reducing their number in the adult muscle.
To investigate this, we performed simultaneous Ki67 and Pax7 immunostaining at the P21 stage, when the satellite cells enter into quiescence (Fig. 2H,I). The ratio of Pax7+/Ki67+ versus Pax7+/Ki67− at this stage shows the ratio of active satellite cells to quiescent cells. The number of active satellite cells was higher in the H19Δ3 P21 sections. This suggests that the H19Δ3 satellite cells have not yet reached full quiescence at this stage, reinforcing the idea that entry into quiescence is delayed in the mutants compared with wt muscles.
H19 controls proliferation of myoblasts
Since the fate of satellite cells is to proliferate, differentiate and self-renew, we first evaluated the proliferation capacity of H19Δ3 myoblasts compared with wt myoblasts. Primary myoblasts were isolated from limb muscles of H19Δ3 and wt adult mice. The myoblasts were maintained in culture in growth medium. Their growth rate was evaluated for 7 days and by day 7, the cell number had doubled in cells lacking H19 compared with wt myoblasts (Fig. 3A). We then used the gain-of-function transgenic mice and isolated myoblasts from wt, H19−/+ and H19−/+;Tg littermates. We observed a rescue of the enhanced growth rate of the H19−/+ myoblasts in the presence of the H19 transgene (Fig. 3B). This suggests that the H19 RNA itself is controlling myoblast number.
H19 controls myoblast growth. Growth curves for (A) wt and H19Δ3 primary myoblast cultures (from two mice per genotype and triplicate time points) and (B) wt, H19−/+ and H19−/+;Tg primary myoblast cultures (from three mice per genotype and triplicate time points). Data are shown as means±s.d. Asterisks indicate that data are statistically significant (*P<0.05, ***P<0.001; two-way ANOVA, Bonferroni post test).
H19 controls myoblast growth. Growth curves for (A) wt and H19Δ3 primary myoblast cultures (from two mice per genotype and triplicate time points) and (B) wt, H19−/+ and H19−/+;Tg primary myoblast cultures (from three mice per genotype and triplicate time points). Data are shown as means±s.d. Asterisks indicate that data are statistically significant (*P<0.05, ***P<0.001; two-way ANOVA, Bonferroni post test).
H19Δ3 satellite cells show normal regeneration capacity
To evaluate the biological function of the satellite cells in vivo, we investigated their capacity to perform muscle regeneration. Quiescent satellite cells are activated after muscle injury and participate in the production of new myofibers. Injection of snake venom (cardiotoxin, Ctx) into the tibialis muscle induces injury of the muscle and allows to evaluate the regeneration capacity of satellite cells in vivo. We performed a first Ctx injection in tibialis muscles from H19Δ3 and wt male mice and collected injured muscles 4 weeks later. Sections of the injured muscles were stained with Hematoxylin and Eosin. The sections displayed normal muscle fibers with centrally located nuclei and there were no differences in regeneration capacity at this stage between H19Δ3 and wt samples (Fig. 4B, Fig. S2).
Regeneration assay by cardiotoxin (Ctx) injection. (A) Representation of the multiple Ctx injection assay. Arrowheads indicate time of collection of the muscles. (B) Sections of adult tibialis muscle from wt and H19Δ3 mice after one Ctx injection stained with Hematoxylin and Eosin. (C) wt and H19Δ3 muscles after four Ctx injections, with control contralateral uninjected muscle. (D) Mass of injured tibialis versus contralateral uninjected tibialis from wt and H19Δ3 males (n=11). (E) Histogram of fiber CSA (µm2) after four Ctx injections in wt and H19Δ3 muscles. Arrow points to larger fibers in the mutant muscle. Transcript levels of (F) Pax7 and Ki67 genes and (G) the IGF pathway genes quantified by RT-qPCR in adult wt and H19Δ3 injured tibialis muscles (n=4 mice/genotype). Data are shown as means±s.e.m. Asterisks indicate that data are statistically significant (*P<0.05, **P<0.01, ***P<0.001; Student's t-test).
Regeneration assay by cardiotoxin (Ctx) injection. (A) Representation of the multiple Ctx injection assay. Arrowheads indicate time of collection of the muscles. (B) Sections of adult tibialis muscle from wt and H19Δ3 mice after one Ctx injection stained with Hematoxylin and Eosin. (C) wt and H19Δ3 muscles after four Ctx injections, with control contralateral uninjected muscle. (D) Mass of injured tibialis versus contralateral uninjected tibialis from wt and H19Δ3 males (n=11). (E) Histogram of fiber CSA (µm2) after four Ctx injections in wt and H19Δ3 muscles. Arrow points to larger fibers in the mutant muscle. Transcript levels of (F) Pax7 and Ki67 genes and (G) the IGF pathway genes quantified by RT-qPCR in adult wt and H19Δ3 injured tibialis muscles (n=4 mice/genotype). Data are shown as means±s.e.m. Asterisks indicate that data are statistically significant (*P<0.05, **P<0.01, ***P<0.001; Student's t-test).
Repeated injury is a suitable model to assess the self-renewal ability of satellite cells and the long-term regeneration potential of skeletal muscle. We reasoned that if H19Δ3 satellite cells had a self-renewal defect, multiple injections of Ctx would lead to poor regeneration after several rounds of injury. To investigate this question, we performed four series of injections of Ctx into the tibialis muscle of 6-week-old adult male mice, each separated by a period of 28 days (Fig. 4A). The experiment was repeated three times on 5-6 mice of each genotype in each experiment. After the third injection, muscles were collected after 28 days and sections were stained with Hematoxylin and Eosin. Regeneration was complete and very similar to what was observed after only one injection, suggesting that self-renewal of the satellite cells occurred normally in the mutant muscles.
It was described previously that the strongest change in gene expression is detected 7 days after injury (Andersen et al., 2013; Al Adhami et al., 2015). Therefore, muscles were also collected at 7 days after the fourth injection. Interestingly, the injured H19Δ3 muscles were larger than the wt injured muscles (Fig. 4C). A 12% increase in mass was observed for the mutant injured muscle compared with the wt injured muscle (Fig. 4D; n=11, P=0.0128).
Sections of these injured muscles showed similar regeneration efficiency, with centrally located nuclei. We performed anti-laminin immunolabeling on these muscle sections and evaluated the number of fibers and their CSA. No difference in the ratio of Ctx/control number of fibers between wt and H19Δ3 muscles was detected. By contrast, a small percentage of fibers were larger in H19Δ3 compared with the wt muscles (Fig. 4E).
Expression levels of Pax7 and Ki67 (also known as Mki67) were analyzed by RT-qPCR on samples collected 7 days after the fourth injection (Fig. 4F). Higher expression levels of these two genes (1.5-fold and 2-fold increase, respectively) were detected in the mutant samples compared with the wt samples. This suggests that although there are fewer satellite cells in the mutant muscle, H19Δ3 satellite cells are correctly activated after injury and even show increased proliferation compared with wt cells during the regeneration process. This is in agreement with the increased growth rate of mutant myoblasts compared with wt myoblasts, observed in vitro in the primary cultures (Fig. 3).
To investigate the reason for the increased mass and size of H19Δ3 regenerated muscles, expression levels of genes involved in growth control were analyzed. Whereas in uninjured muscle, only Igf2 expression was increased in mutants, others genes of the IGF pathway – Igf1, Igf1r and Igf2r – showed higher expression levels (1.5-fold increase) in the mutant samples after injury (Fig. 4G). This upregulation of IGF pathway genes certainly accounts for the increased muscle mass and the hypertrophic fibers observed after the multiple Ctx injuries.
We concluded from these results that the H19Δ3 satellite cells have a self-renewal capacity that is similar to that of wt satellite cells. Regeneration occurs normally in mutant mice, if not better than in wt mice. The mass increase observed for the regenerated mutant muscles could be associated with the increased proliferation of the H19Δ3 myoblasts (suggested by the increase in Ki67 expression) and with hypertrophic fibers with high levels of expression of genes from the growth-controlling IGF pathway.
Imprinted gene network expression is induced upon muscle regeneration
Imprinted genes are co-expressed and co-regulated in fetal tissues and belong to an IGN. We had shown previously that H19 controlled nine genes of the IGN during fetal development (Gabory et al., 2009; Monnier et al., 2013). To identify potential genes involved in the H19 mutant muscle phenotype described above, we performed RT-qPCR experiments on muscles collected at different stages from H19Δ3 and wt mice. Several genes from the IGN were tested, such as Igf2, Dlk1, Cdkn1c, Dcn, Peg3, Mest, Igf2r and Slc38a4.
No major significant alteration of expression was observed in muscles collected from P6 mice, with the exception of Igf2 and Dlk1 (15-20% increase; Fig. 5A). We also tested P21 proliferating myoblast cultures and found no change in expression of any of these genes (Fig. S3). This suggests that the effect of the H19 gene on the modulation of the IGN occurs essentially in the embryo as we had shown previously (Gabory et al., 2009) and not at postnatal stages. In adult muscles, Igf2 expression remained upregulated (+50%) and was accompanied by a decrease in expression of Dcn and Slc38a4 (−20% and −50%, respectively; Fig. 5B). The reduction in Dcn expression can be correlated with the observed decrease in Mstn expression levels (Fig. 1F), because both genes are involved in the TGF-β pathway (Kishioka et al., 2008).
IGN expression is induced upon muscle regeneration. (A) Transcript levels of the IGN genes in wt and H19Δ3 tibialis muscles from P6 mice, (B) in uninjured wt and H19Δ3 adult tibialis muscles, (C) in uninjured wt compared with injured wt Ctx adult tibialis muscles and (D) in injured wt Ctx compared with injured H19Δ3 Ctx reported to uninjured muscle transcript levels. Data are shown as means±s.e.m. Asterisks indicate that data are statistically significant (*P<0.05, **P<0.01, ***P<0.001; Student's t-test).
IGN expression is induced upon muscle regeneration. (A) Transcript levels of the IGN genes in wt and H19Δ3 tibialis muscles from P6 mice, (B) in uninjured wt and H19Δ3 adult tibialis muscles, (C) in uninjured wt compared with injured wt Ctx adult tibialis muscles and (D) in injured wt Ctx compared with injured H19Δ3 Ctx reported to uninjured muscle transcript levels. Data are shown as means±s.e.m. Asterisks indicate that data are statistically significant (*P<0.05, **P<0.01, ***P<0.001; Student's t-test).
We then performed the same analysis of these imprinted genes on uninjured and injured tibialis muscles, collected at 7 days after four rounds of Ctx injury. We first evaluated a possible effect of Ctx treatment on the reference genes and found no significant difference in their expression between control and Ctx samples (Fig. S4). A comparison of the expression profiles of the imprinted genes in wt mice showed a striking increase in levels of expression of H19 (6-fold), Igf2 (46-fold), Dlk1 (4.5-fold), Cdkn1c (2-fold), Peg3 (8-fold), Mest/Peg1 (10-fold) and Igf2r (1.5-fold) (Fig. 5C, Fig. S5). In the H19Δ3 injured muscles, a further enhancement of expression was detected for Dlk1 (3.4-fold), Cdkn1c (2-fold), Dcn (1.5-fold), Peg3 (2-fold) and Igf2r (1.6-fold; Fig. 5D and Fig. S5). For the evaluation of Dlk1 expression, primers for both the membrane-bound and soluble forms of Dlk1 were used and produced very similar results. We concluded from these results that H19 and a set of genes from the IGN are strongly reactivated during regeneration and formation of new myofibers in adult muscle.
DISCUSSION
H19 plays an important role during development in limiting growth of the embryo via a regulatory control of genes belonging to the IGN (Gabory et al., 2009,, 2010). Since H19 remains expressed in the adult muscle, the H19Δ3 mutant mice provide an interesting model to study the role of this gene in the adult. Our main findings show a role for the H19 gene in the control of the number of satellite cells and of their proliferation in the adult muscle. We also show an effect of the H19 gene in the process of muscle regeneration, in a functional test of multiple rounds of injury. Interestingly, the analysis of the expression of imprinted genes from the IGN revealed a remarkable upregulation of these genes during regeneration accompanied by increased muscle size in the mutant mice. Our study of the H19Δ3 mice provides a model for the regulation of the IGN in the context of muscle injury. This suggests that the embryonic regulatory function of H19 on the IGN is reactivated during the regeneration process occurring in the adult muscle.
Our first observation was an increase in the mass of the H19Δ3 muscles with hypertrophy and hyperplasia at the adult stage, compared with wt mice. Muscle fiber number is determined during development at the secondary myogenesis step, which occurs during late gestation and the total number of fibers is unchanged between birth and adult stage (White et al., 2010). We suggest that the increased proliferation observed in H19Δ3 myoblasts compared with wt myoblasts must occur during the late fetal stage. This leads to the hyperplasia observed in adult muscle. Muscle fiber size, by contrast, is under the control of growth-controlling genes (Musarò et al., 2001). Hypertrophy observed in the adult mutant muscles could therefore be correlated to the increase in the expression levels of Igf2 and Dlk1, combined with the reduced expression of Mstn. Our initial experiments in the H19Δ3 embryos had revealed an upregulation of Igf2 in the absence of H19 (Ripoche et al., 1997). Igf2 belongs to the IGF pathway, which confers positive growth regulation. Interestingly, Igf2 expression is normally downregulated after birth and is replaced by expression of Igf1. In the case of the H19Δ3 mutants, the absence of H19 seems to help maintain Igf2 expression in adult muscle. Dlk1 is another imprinted gene whose expression is enhanced in H19Δ3 muscle, at least at the P6 stage. Dlk1 is highly induced during normal muscle development and during regeneration (Andersen et al., 2009; Waddell et al., 2010). Dlk1 is also one of the genes responsible for the callipyge phenotype in cattle (Charlier et al., 2001) and has been shown to induce hypertrophy (Davis et al., 2004; Waddell et al., 2010). One of the other main regulatory pathways of muscle growth is the myostatin pathway. Mstn-null mice show hypertrophy and hyperplasia of their muscles (Lee and McPherron, 2001). Exceptional muscle development of the Belgian Blue cattle has also been associated with a mutation in Mstn (Grobet et al., 1997). Myostatin is considered to be a powerful inhibitor of muscle growth. Its reduced expression in the H19Δ3 muscles suggests that it could be a good candidate for the overgrowth phenotype of these mutants. A tight correlation between Mstn and Igf2 expression has been previously established (Kalista et al., 2012; Clark et al., 2015). Taken together, although other genes might also participate in this phenotype, we suggest that the upregulation of the growth-promoting Igf2 and Dlk1 genes at the early postnatal P6 stage coupled with the reduced expression of Mstn in the adult muscle are good candidates for the hypertrophic phenotype observed in the H19Δ3 mutant muscles.
Primary myoblasts in culture show an enhanced growth rate in the absence of H19, which can be rescued by the H19 transgene. Although we have not pursued this study to identify the possible underlying mechanisms, this can be correlated with our previous identification of H19 acting as a tumor suppressor (Yoshimizu et al., 2008). We surprisingly observed a 50% reduction in the number of satellite cells present on the mutant muscle fibers. Satellite cells are produced during late gestation and proliferate during the early postnatal period. They cease proliferating and enter into quiescence around 21 days after birth, with their number remaining constant after this stage (Schultz et al., 1978; White et al., 2010). To understand the reason for this reduction, we compared the number of satellite cells in wt and mutant muscle fibers at different postnatal stages, P6 and P21, and found no difference. The reduction in the number of satellite cells was only observed in the adult muscle. The rescue of the number of satellite cells in the H19−/+;Tg strongly suggests that the H19 gene is involved in the control of the number of satellite cells at some stage of their establishment.
H19 has previously been shown to affect quiescence of another category of stem cells – the hematopoietic stem cells (Venkatraman et al., 2013). A possible explanation for the observed reduction is that the entry into quiescence of satellite cells is delayed in H19Δ3 mutants. The number of satellite cells is normally fixed at a certain stage, but the mutant cells might continue to be incorporated into adult muscles after P21 and only enter into quiescence at a later stage. This would lead to a reduced number of quiescent satellite cells in the adult mutant muscle. Our experiment showing a higher level of Pax7+/Ki67+ satellite cells in the mutant P21 muscles confirms this delay in the entry into quiescence. Quiescence perturbation could be a common feature of stem cells deficient for the H19 gene.
At the molecular level, satellite cells might show a delay in their entry into quiescence because a certain number of genes involved in proliferation are not properly downregulated in the satellite cells. A comparison of expression profiles of quiescent and activated satellite cells reveals a strong induction of several imprinted genes upon activation (Pallafacchina et al., 2010). These genes are Mest, Ndn, Cdkn1c, Gnas, Kvlqtot1 (all approximately 3- to 4-fold induction), Slc38a4, Peg12, Igf2 (6-fold induction) and especially Dlk1 (30- to 50-fold induction). If genes such as Igf2 and Dlk1 are not correctly repressed in the active H19Δ3 satellite cells at the time of their entry into quiescence, this could lead to continued proliferation and would result in a delay in the time to quiescence in the mutant muscles.
Satellite cells display stem cell properties of self-renewal, proliferation and differentiation. To further investigate the biological capacities of H19-deficient satellite cells, we focused on the regeneration capacity of these cells by injuring tibialis muscles. A first injection of Ctx resulted in efficient regeneration after 28 days both in wt and H19Δ3 injured muscles. A delay in the regeneration efficiency at early times (14 days post injury) has been observed (Dey et al., 2014). We also observed this delay in the H19Δ3 injured muscles as early as day 7, but regeneration appeared to be rescued with time. This delay could be an initial reflection of the reduced number of satellite cells. As muscle repair proceeds, more efficient proliferation of the H19Δ3 myoblasts compensates the reduced number of satellite cells and results in efficient regeneration after 28 days. Subsequent multiple rounds of injury show that the mutant satellite cells continue to perform their function and maintain a high capacity for muscle repair. A significant increase in the mass of tibialis muscles after four rounds of Ctx injury was observed in H19Δ3 mice. This improvement could be related not only to the higher growth rate of mutant myoblasts but also to the enhanced expression of genes from the IGF pathway, because Igf1 has been shown to play a role in improving regeneration (Pelosi et al., 2007). Overall, the self-renewal capacity of the H19-deficient satellite cells does not seem to be impaired, and once activated, their efficient proliferation capacity leads to improved regeneration.
We followed the expression levels of genes from the IGN in whole muscle at the P6 and adult stage, in P21 myoblast cultures and during regeneration experiments. Very few genes were affected during postnatal growth and at the adult stage in H19Δ3 whole muscle. By contrast, in wt Ctx-injured muscles, seven imprinted genes showed increased expression (H19, Igf2, Dlk1, Cdkn1c, Peg3, Mest and Igf2r). The involvement of genes from the IGN in regeneration has been previously described (Yan et al., 2003; Al Adhami et al., 2015). Our data provide a detailed quantification of the increased expression of IGN genes during wt regeneration. The strong increase in expression of H19 and Igf2 after injury can be related to the observed twofold increase in the expression of Myod (Fig. S6). This myogenic factor binds the mesodermal enhancers controlling expression of the H19-Igf2 locus (Cao et al., 2010; Borensztein et al., 2013). Therefore, an increase in Myod expression is expected to lead to a correlated increase in the expression of these two imprinted genes. In addition, we observe an enhanced upregulation of most imprinted genes in the mutant injured mice compared with the wt injured mice. In the embryo, the absence of H19 also induces upregulation of these genes, suggesting that the IGN is subject to a fine-tuned control to achieve correct development of the embryo (Gabory et al., 2009). The regenerating muscle is recapitulating an embryonic program, as evidenced by the expression of embryonic forms of myosin heavy and light chains in new myofibers (Whalen et al., 1990; Jerkovic et al., 1997). There is strong evidence for maternally expressed genes to limit growth, whereas paternally expressed genes enhance growth (Haig, 2004). A coordinated upregulation of these genes during regeneration therefore provides a simultaneous balanced correlation between growth enhancement and growth reduction. This results in adequately controlled muscle production. Enhanced expression of these genes during regeneration in H19Δ3 muscles could explain the improved regeneration effect in the mutant mice after multiple rounds of injury, although an important role for the IGF pathway must also be taken into account.
The H19 locus produces a long non-coding full-length RNA and a microRNA, miR-675. In placenta, miR-675 is shown to control growth by repressing Igf1r, a direct target of the miR (Keniry et al., 2012). It was also previously shown that the miR-675 plays a role in inducing differentiation of myoblasts (Dey et al., 2014). During enhanced regeneration of H19Δ3 muscles, Igf1r expression is clearly upregulated, suggesting that the miR-675 might be responsible for part of this effect. Interestingly, a study performed on purified satellite cells has shown that miR-675 is not expressed in quiescent satellite cells but is expressed in activated satellite cells (D. Castel and S. Tajbakhsh, personal communication). Alternatively, since the H19 RNA interacts with MBD1 to bring H3K9me3 repressive marks on some of its targets (Igf2, Mest and Slc34a8), H19 could act by repressing imprinted genes involved in the myogenic pathways through epigenetic modifications (Monnier et al., 2013). Another long non-coding RNA, linc-MD1, plays an important role in myogenesis, by targeting a myomiR called miR-133 (Legnini et al., 2014). Because the H19 RNA has also been shown to act as a sponge for let7 (Kallen et al., 2013), it could also interfere with the levels of other myomiRs. Whether the different effects of the H19 locus on the control of the number and growth rate of satellite cells and on regeneration efficiency can be attributed to the full-length RNA or to miR-675 remains to be elucidated.
Finally, the H19-Igf2 locus is associated in humans with the Beckwith-Wiedemann syndrome, in which affected children show an increased predisposition to tumors such as Wilms’ tumors and rhabdomyosarcomas. This occurs particularly in cases of paternal unidisomies, and therefore is correlated with a lack of maternally expressed genes (such as H19) of the region (Cooper et al., 2005). H19 has been shown to be a tumor suppressor (Hao et al., 1993; Yoshimizu et al., 2008). The enhanced muscle regeneration observed in our mutant mouse model could be considered as a first step of increased proliferation of myofibers lacking H19, possibly leading to tumorigenesis. However, we have never observed tumors in our H19Δ3 mouse colony over the years. Provided the mouse model can be applied to a human situation, this suggests that occurrence of rhabdomyosarcomas requires the absence of more than one tumor suppressor gene in children with Beckwith-Wiedemann syndrome.
MATERIALS AND METHODS
Mouse strains
All experimental designs and procedures were in agreement with the guidelines of the animal ethics committee of the Ministère de l'Agriculture (France). H19Δ3 and H19Tg gain-of-function mice were previously described (Ripoche et al., 1997; Gabory et al., 2009). The H19Δ3 strain is maintained on the 129SV/Pas background. Mutant H19Δ3 females were mated with H19Tg males (C57BL/6/CBA background) to obtain H19−/+;Tg mice. H19−/+;Tg females were mated with wt males to obtain litters of H19−/+, H19−/+;Tg and wt mice on a similar genetic background. H19−/+(H19mat Δ3/pat wt) mice are identical to H19−/− mice because of the imprinted status of the gene.
Histology and immunostaining
Tibialis anterior and EDL muscles of P6, P21 or 6-week-old male mice from each genotype were dissected and fixed for 4 h at 4°C in 4% paraformaldehyde (PFA) and washed in PBS as described previously (Le Grand et al., 2012). After incubation in PBS and 10% sucrose overnight at 4°C, samples were incubated in PBS with 10% gelatin and 10% sucrose, then 10% gelatin and frozen in isopentane cooled by liquid nitrogen. Sections were obtained (7 µm) using a Leica CM1850 cryostat and stained with Hematoxylin and Eosin or immunostained using anti-Pax7 (Santa Cruz, sc-81648; 1:100), anti-Ki67 (Abcam, ab15580; 1:400) and anti-laminin (Santa Cruz, sc59854; 1:100) antibodies. Secondary antibodies were Alexa Fluor 488 anti-mouse IgG, Alexa Fluor 546 anti-rabbit IgG and Alexa Fluor 647 anti-rat IgG (Life Technologies). The sections were mounted and the images were captured with an Olympus BX63 microscope. Image analysis was performed with ImageJ (NIH) software.
Single myofiber isolation and staining
Single myofibers were isolated from the EDL muscles of P21 or 6-week-old male mice as previously described (Pasut et al., 2013). Freshly isolated fibers were fixed in 4% PFA, blocked in PBS containing 5% horse serum and 0.5% Triton X-100 for 1 h at room temperature and stained with anti-Pax7 antibodies (Santa Cruz, sc-81648; 1:100) at 4°C overnight. The samples were washed in PBS and incubated with secondary antibody Alexa Fluor 488 anti-mouse IgG (Life Technologies). Nuclei were counterstained with DAPI.
Primary myoblast isolation and cell proliferation assay
Primary myoblasts were isolated from limb muscles of P21 and 6-week-old male mice as described previously (Le Grand et al., 2012). For the adult 129SV/Pas wt and H19Δ3 cultures, two mice for each genotype were dissected. For the P21 mice, four mice for each genotype were dissected. For the C57BL/6/CBA background, three mice for each genotype were used. Muscles were digested in collagenase B (10 mg/ml)/dispase (3.2 mg/ml) (Roche) at 37°C for 30 min. Cells were filtered, centrifuged and plated on collagen-coated dishes and cultured in growth medium [F10 Ham's medium (Gibco) supplemented with 20% fetal bovine serum, 5 ng/ml bFGF and 1% penicillin-streptomycin].
To determine the cell proliferation rate, early passages of primary myoblasts were seeded (50,000 cells per well) on six-well plates coated with collagen and counted at different time points (day 0 to day 7). Experiments were performed in triplicate for each primary culture from each genotype and for each time point.
Muscle injury and regeneration assay
Muscle injury was induced by injecting 50 µl snake venom cardiotoxin I (Ctx, Latoxan) (12 µmol/l in PBS) with 26 G needles into the left tibialis muscle of 6-week-old anesthetized males as described previously (Yan et al., 2003; Castets et al., 2011). For multiple regeneration assays, four successive Ctx injections were performed with an interval of 4 weeks between injections. Muscles were collected 7 days after the fourth Ctx injury, to extract RNA or perform histological analysis. The multiple regeneration experiment was performed three times on 5-6 male mice for each experiment and for each genotype.
RNA isolation and quantitative PCR assay
Total RNA was extracted using the Qiagen miRNeasy mini kit (Qiagen) with DNase treatment to remove the genomic DNA. 500 ng of total RNA was reverse-transcribed using the PrimeScript RT reagent kit (Takara). Quantitative real-time PCR was performed on 10 ng cDNA using SYBR Premix Ex Taq (Takara) in a LightCycler 480 apparatus (Roche). Gene expression levels were normalized to the geometric mean of the expression level of Mrpl32, Sdha and EiF4a2 housekeeping genes, as determined by Genorm analysis (Vandesompele et al., 2002). Primer sequences used for this study are listed in Table S1.
Statistical analysis
Statistical analysis was performed using Prism software. Student's t-test, two-way ANOVA with Bonferroni post test and Kruskal–Wallis test with Dunn's post test were used, as indicated in the figure legends.
Acknowledgements
We thank Dominique Daegelen, Pascal Maire, Fabien Le Grand, Andrew Keniry and Shahragim Tajbakhsh for many helpful discussions. We are very grateful to Athanassia Sotiropoulos and Didier Montarras for critical reading of the manuscript. We thank the Animal House Core Facility for their constant help and the Sequencing and Genomic Core Facility of the Cochin Institute for their technical support and advice.
Author contributions
C.M., P.M., Y.L. and A.G. performed experiments and analyzed data. M.B. helped with Ctx injection of the mice. L.D. directed the project, analyzed data and wrote the manuscript. All authors contributed to critical reading and editing of the manuscript.
Funding
This work was supported by funding from the Association Française contre les Myopathies (AFM) and from the Agence Nationale de la Recherche [ANR-14-CE11-0022-02 ‘Twothyme’]. C.M. was awarded a fellowship from the Région Ile-de-France (DIM Biotherapy).
References
Competing interests
The authors declare no competing or financial interests.