Alternative polyadenylation of Pax3 controls muscle stem cell fate and muscle function

See allHide authors and affiliations

Science  08 Nov 2019:
Vol. 366, Issue 6466, pp. 734-738
DOI: 10.1126/science.aax1694

Skeletal muscle during homeostasis

Muscle stem cells function in the regeneration of skeletal muscle after injury, but their role in homeostasis is unclear. De Morree et al. show that, in the absence of injury, stem cells in different muscles have different rates of spontaneous activation and fusion, which depend on the level of Pax3 protein (see the Perspective by Xi and Pyle). Regulation of Pax3 protein occurs posttranscriptionally through the small nucleolar RNA U1 and microRNA miR206. This work explains how muscle stem cells are maintained under normal conditions and shows that homeostatic muscle stem cell activation varies in different muscle groups.

Science, this issue p. 734; see also p. 684


Adult stem cells are essential for tissue homeostasis. In skeletal muscle, muscle stem cells (MuSCs) reside in a quiescent state, but little is known about the mechanisms that control homeostatic turnover. Here we show that, in mice, the variation in MuSC activation rate among different muscles (for example, limb versus diaphragm muscles) is determined by the levels of the transcription factor Pax3. We further show that Pax3 levels are controlled by alternative polyadenylation of its transcript, which is regulated by the small nucleolar RNA U1. Isoforms of the Pax3 messenger RNA that differ in their 3′ untranslated regions are differentially susceptible to regulation by microRNA miR206, which results in varying levels of the Pax3 protein in vivo. These findings highlight a previously unrecognized mechanism of the homeostatic regulation of stem cell fate by multiple RNA species.

Tissue homeostasis and regeneration depend on tissue-specific populations of stem cells (1), some of which maintain a quiescent state for prolonged periods of time (2, 3). In vertebrates, muscle stem cells (MuSCs) are required for skeletal muscle regeneration (46). Recent studies have revealed that MuSCs in sedentary mice contribute to the maintenance of adult myofibers, with high levels of contribution in diaphragm muscles and low levels in lower hindlimb muscles (7, 8). To understand the mechanisms that determine the extent of MuSC contribution to adult myofibers, we measured the extent to which MuSCs activate and enter the cell cycle in different uninjured muscles. When we pulsed mice in vivo with the nucleotide analog 5-ethynyl-2′-deoxyuridine (EdU) to label cells undergoing DNA replication, we observed a wide range in EdU incorporation. Diaphragm, gracilis, and triceps muscles exhibited the highest number of MuSCs that spontaneously broke quiescence and entered the cell cycle under homeostatic conditions, whereas hindlimb muscles showed the lowest number of such MuSCs (fig. S1, A and B).

MuSCs in the diaphragm, gracilis, and triceps muscles express high levels of the transcription factor Pax3, whereas MuSCs in most limb muscles do not (9). Given the established roles of Pax3 in driving cell proliferation during embryogenesis (10) and in response to stress (11, 12), we investigated whether Pax3 regulates the process of MuSC activation during homeostasis. We observed that Pax3 staining is positively correlated with EdU incorporation in MuSCs from different muscles (Fig. 1A). To directly test whether Pax3 plays a role in the balance between MuSC quiescence and activation, we conditionally deleted Pax3 from adult MuSCs using a Pax7-CreERT2 driver (Pax3cKO mice) and obtained a median reduction in Pax3 mRNA of 96 and 88% in diaphragm and hindlimb MuSCs, respectively. Deletion of Pax3 led to an increase in quiescence markers, a decrease in MyoD, and a decrease in EdU incorporation in the diaphragm, gracilis, and triceps MuSCs, both in vivo and in vitro (Fig. 1, B and C, and fig. S1, C to E).

Fig. 1 Pax3 controls MuSC activation under homeostatic conditions.

(A) MuSCs were isolated from mice treated with EdU for 3 days and stained for EdU and Pax3, and Pax3 levels were correlated with EdU levels. Data points represent individual mice, with muscles color coded as noted (low Pax3 levels are blue, high Pax3 levels are red). A linear regression was fitted through the data points (n = 5 mice). Dia, diaphragm; Gra, gracilis; Tri, triceps; Sol, soleus; EDL, extensor digitorum longus; TA, tibialis anterior; AU, arbitrary units. (B and C) MuSCs were isolated from WT, Pax3cHET, and Pax3cKO mice treated with EdU for 3 days, stained for EdU [(B), n = 11] or Pax3 [(C), n = 5], and counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Left panels: representative images, white arrows indicate DAPI-positive cells (B) or Pax3-positive cells (C); right panels: graph of quantification. ns, not significant. (D) Different strains of mice (WT and miR206−/−) were treated with EdU for 3 days. MuSCs were isolated and stained for Pax3 protein (left, n = 7) or EdU (right, n = 5). (E) Different strains of mice (WT, miR206−/−, and Pax3cKO) were treated with control or Pax3 knockdown AMOs and with EdU for 3 days. MuSCs were isolated and stained for Pax3 protein (left, n = 5) or EdU (right, n = 5, control is represented by the horizontal black line, normalized to 1). Mean + SEM; two-tailed Student’s t test, *P < 0.05, **P < 0.01, ***P < 0.001.

Prior work from our laboratory has shown that the microRNA miR206 inhibits Pax3 expression in MuSCs isolated from hindlimb muscles (9). We analyzed miR206−/− mice and found that they have increased Pax3 protein levels and corresponding increases in EdU incorporation in limb MuSCs. These changes were comparable to the levels seen in the diaphragm MuSCs (Fig. 1D and fig. S1, F and G), which were reverted by conditional deletion of Pax3 (fig. S1, H to J). Pax3 mRNA levels did not change, which suggests that miR206 regulates Pax3 protein translationally (fig. S1K). Finally, we injected wild-type (WT) and miR206−/− mice with antisense vivo-morpholino oligonucleotides (AMOs), complementary to the translation initiation site in Pax3, to block Pax3 translation. This led to reduced Pax3 protein levels and reduced EdU incorporation in diaphragm MuSCs in both WT and miR206−/− mice (Fig. 1E and fig. S1L) but not in Pax3cKO mice (Fig. 1E and fig. S1L). Conversely, miR206-blocking AMOs led to increased Pax3 protein levels and EdU incorporation in WT limb MuSCs but not in MuSCs of miR206−/− or Pax3cKO mice (fig. S1, M to O). We conclude that Pax3 increases the propensity of quiescent MuSCs to exit the quiescent state.

Previous results from our laboratory have shown that MuSCs in the diaphragm express Pax3 transcript isoforms with shorter 3′ untranslated regions (3′UTRs) that lack binding sites for miR206 (9). Using single-molecule fluorescence in situ hybridization (smFISH), we found that most diaphragm MuSCs express mainly short isoforms, whereas most limb MuSCs express mainly long isoforms (Fig. 2, A and B, and fig. S2, A to C). We observed similar isoform distribution patterns in miR206−/− MuSCs, which suggests that miR206 does not affect Pax3 isoform expression patterns (fig. S2, D and E). This indicates a mechanism by which MuSCs tune Pax3 protein levels. In MuSC cDNA libraries, we observed increased levels of U1 small nucleolar RNA (snRNA) in limb MuSCs compared with diaphragm MuSCs, but we observed comparable levels of other known alternative polyadenylation factors (Fig. 2C and fig. S2F). Moreover, using multiple sequence alignments of vertebrate Pax3 paralogs, we identified two conserved motifs upstream of the most proximal polyadenylation site (PAS) that match the consensus sequence for U1 snRNA (fig. S2G). This suggests that U1 snRNA could directly interact with Pax3 mRNA. U1 snRNA is the RNA scaffold of the small nucleolar riboprotein U1, the macromolecular complex that recognizes the 5′ splice site, acting outside of the spliceosome to protect native transcripts from premature transcription termination and polyadenylation (13). Knockdown of U1 snRNA has resulted in the expression of shorter transcript isoforms (14), consistent with limb MuSCs expressing longer Pax3 isoforms and higher U1 snRNA levels.

Fig. 2 U1 snRNA controls Pax3 length and sensitivity to miR206.

(A and B) MuSCs from WT and Pax3cKO were isolated from the indicated muscles and stained with probe libraries complementary to the open reading frame (ORF) or the 3′UTR of Pax3. (A) Representative smFISH images with assay schematic shown above; (B) graph of quantification (n = 4). (C) MuSCs were analyzed for snRNA levels by reverse transcription polymerase chain reaction. Values are relative to U2 snRNA (n = 9). (D) MuSCs from mice treated with control AMOs, U1–site blocking AMOs (hereafter, U1 site AMOs), or PAS1-blocking AMOs (hereafter, PAS1 AMOs) were stained by smFISH, and Pax3 isoforms per cell were quantified (n = 4). Mean + SEM, one-tailed Student’s t test, *P < 0.05, ***P < 0.001.

To evaluate the role of U1 snRNA in alternative polyadenylation in MuSCs, we created reporter genes that express the 3′UTR of Pax3 downstream of green fluorescent protein (fig. S2H). Knockdown or overexpression of U1 snRNA led to a reduction or increase of distal amplicons, respectively (fig. S2, I and J). To evaluate the role of U1 snRNA in MuSCs in vivo, we treated mice systemically with AMOs to knock down U1 snRNA. We observed a decrease in long Pax3 isoforms in limb MuSCs, with no significant change in diaphragm MuSCs (fig. S2, K to M).

To test whether U1 snRNA regulates Pax3 isoform expression through the conserved binding motifs in the Pax3 3′UTR, we designed AMOs complementary to these motifs to compete with U1 snRNA at the Pax3 3′UTR. We observed a switch to shorter Pax3 isoforms in mice treated with these AMOs (Fig. 2D and fig. S2N). Next, we designed AMOs complementary to two key conserved sequences in the most proximal PAS (PAS1) to compete with key polyadenylation factors and select for a more distal PAS (fig. S2O). We observed an increase in longer isoforms in diaphragm MuSCs in mice treated with these AMOs (Fig. 2D and fig. S2N). We conclude that U1 snRNA induces the expression of longer Pax3 transcript isoforms in MuSCs in vivo and that it works through conserved motifs within the Pax3 transcript.

We next explored whether changes in Pax3 isoform expression lead to differences in Pax3 protein levels. Because only the long Pax3 isoforms contain miR206 binding sites, only these isoforms are expected to be sensitive to changes in miR206 levels. We tested this expectation by treating miR206−/− mice with U1 site or PAS1 AMOs and then transfecting the purified MuSCs with a miR206 mimic. After 24 hours in vitro, we measured Pax3 protein levels. In limb MuSCs (which express mainly long isoforms) from control AMO-treated miR206–/– mice, the miR206 mimic reduced Pax3 protein levels compared to control (miR1 mimic) (Fig. 3A). In contrast, limb MuSCs from miR206−/− mice treated with U1 site AMOs did not show any reduction in Pax3 protein levels, whereas limb MuSCs from miR206−/− mice treated with PAS1 AMOs did show a strong reduction in Pax3 protein levels in response to miR206 (Fig. 3A). We observed comparable expression patterns in the diaphragm MuSCs, which at baseline express mainly short isoforms (Fig. 3A). Knockdown or overexpression of U1 snRNA led to increased or decreased Pax3 protein levels, respectively, in WT MuSCs but not in miR206−/− MuSCs (fig. S3, A and B).

Fig. 3 U1 snRNA controls MuSC activation by controlling Pax3 levels.

(A) MuSCs from miR206−/− mice treated with control, U1 site AMOs, or PAS1 AMOs were transfected with miR206 or control (miR1) and stained for Pax3 (n = 6). (B) Different strains of mice (WT and miR206−/−) were treated with increasing amounts of control or U1 knockdown AMOs. MuSCs were isolated and stained for Pax3. Control-treated mice are represented by the horizontal black line (n = 5). (C) Different strains of mice (WT and miR206−/−) were treated with control AMOs, U1 site AMOs, or PAS1 AMOs, and MuSCs were isolated and stained for Pax3 protein (n = 5). (D) Different strains of mice (WT, miR206−/−, and Pax3cKO) were treated with control AMOs, U1 site AMOs, or PAS1 AMOs, and MuSCs were plated in the presence of EdU for 24 hours and stained for EdU (n = 5). Mean + SEM, one-tailed Student’s t test, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0005.

Having shown that longer isoforms are more susceptible to miR206 regulation, we next tested whether such regulation occurs in vivo. Knockdown of U1 snRNA led to dose-dependent increases in Pax3 protein levels in WT but not miR206−/− mice (Fig. 3B and fig. S3C). We then treated WT mice with U1 site and PAS1 AMOs. Blocking the U1 snRNA binding sites led to increased Pax3 protein levels in WT MuSCs but not miR206−/− MuSCs, whereas blocking PAS1 led to decreased Pax3 protein levels (Fig. 3C and fig. S3, D and E). We conclude that U1 snRNA induces a switch to longer Pax3 isoforms, which leads to inhibition of Pax3 protein expression by miR206.

We next considered whether U1 snRNA, by regulating Pax3 protein levels, also affects MuSC activation. We treated miR206−/− mice with U1 site or PAS1 AMOs, purified the MuSCs, and transfected them with miR206 in the presence of EdU. For mice treated with PAS1 AMOs, transfection of miR206 led to reduced EdU incorporation in both limb and diaphragm MuSCs. For the mice treated with U1 site AMOs, no change in EdU incorporation was observed by miR206 transfection (fig. S3F). Overexpression of U1 snRNA led to reduced EdU incorporation levels in WT MuSCs but not miR206−/− or Pax3cKO MuSCs (fig. S3G).

To provide further evidence for the role of U1 snRNA in the control of homeostatic MuSC activation, we treated WT mice with U1 site AMOs or PAS1 AMOs. Blocking the U1 binding sites led to increased EdU incorporation in WT limb and diaphragm MuSCs, whereas blocking the proximal PAS led to decreased EdU incorporation (Fig. 3D). However, the AMOs had no effect in miR206−/− or Pax3cKO mice (Fig. 3D). These data suggest that U1 snRNA controls MuSC activation under homeostatic conditions via alternative polyadenylation of Pax3 (fig. S3H).

To explore the functional relevance of Pax3 expression in MuSCs in vivo, we conditionally deleted Pax3 in MuSCs in 3-month-old mice and saw no difference in MuSC numbers 2 weeks or 9 months later (fig. S4, A to C). Using enhanced yellow fluorescent protein (eYFP) as a lineage tracer for MuSCs (fig. S4D), we observed distinct eYFP expression in myofibers after 9 months. Approximately 50% of myofibers were positive in the diaphragm muscles and 10% were positive in the tibialis anterior (TA) muscles of WT mice (Fig. 4, A and B), consistent with previous reports (7, 8). In those mice in which Pax3 had been conditionally deleted, fewer diaphragm myofibers exhibited evidence of fusion of MuSCs (Fig. 4, A and B, and fig. S4E). In these mice, we observed a significant decrease in myofiber size in diaphragm and triceps muscles, but not in TA muscles (Fig. 4C and fig. S4F). Consistently, Pax3cKO mice displayed a decrease in grip strength (fig. S4G) and reduced performance in a treadmill assay (Fig. 4D). We conclude that, in muscles in which MuSCs express high levels of Pax3, Pax3 expression is critical for structural and functional homeostasis of the tissue.

Fig. 4 Pax3 determines MuSC contribution to mature myofibers.

(A to C) TA and diaphragm muscles of WTPax7Cre-ERT2;eYFP/eYFP and Pax3cKO;eYFP/eYFP mice were sectioned 9 months after recombination and stained for eYFP (green) and wheat germ agglutinin (WGA, red) and counterstained with DAPI (blue). (A) Example images with eYFP-positive myofibers marked with asterisks; (B) graph of quantifications of eYFP-positive myofibers (n = 6); (C) graph of mean fiber diameter (n = 6). (D) WT and Pax3cKO mice were tested on a treadmill with an endurance running protocol. Total distance run was plotted for individual mice (n = 15). Mean + SEM, two-tailed Student’s t test, *P < 0.05, **P < 0.01, ***P < 0.001.

Previous work has revealed that the frequency of fusion of MuSCs with mature myofibers differ in various muscles (7, 8). Here, we provide evidence of a molecular basis for this heterogeneity, demonstrating that Pax3 expression regulates the propensity of MuSCs to contribute to muscle homeostasis. Furthermore, we describe the molecular mechanisms by which Pax3 is differentially expressed in MuSCs in different muscles. We find that U1 snRNA interacts with Pax3 mRNA to determine transcript isoform expression patterns and, thus, sensitivity to inhibition by miR206. Together, U1 snRNA, miR206, and Pax3 mRNA create a rheostat of Pax3 protein expression that controls spontaneous activation and cell cycle entry of MuSCs under homeostatic conditions.

Supplementary Materials

Materials and Methods

Figs. S1 to S4

References (1525)

References and Notes

Acknowledgments: We thank S. Conway for providing Pax3fl/fl mice, E. Olsen for providing miR206−/− mice, and C. Keller for providing Pax7CreERT2/CreERT2 mice. We thank C. Cain and L. Rott and the Palo Alto VA Flow Cytometry Core for assistance with flow cytometry experiments. Funding: This work was supported by funding from the Glenn Foundation for Medical Research, the Muscular Dystrophy Association (MDA313960 to A.d.M.), the FSH Society (to A.d.M. and T.A.R.), the Lundbeck Foundation (R232-2016-2459 to J.F.), The Danish Council for Independent Research (5053-00195 to J.F.), the NIH (R01 AR062185, R37 AG023806, and P01 AG036695 to T.A.R.), and the Departments of Veterans Affairs (BLR&D Merit Review to T.A.R.). Author contributions: A.d.M. and T.A.R. initiated the project. T.A.R. provided guidance throughout. A.d.M., J.D.D.K., Q.G., A.K., B.B., and C.T.J.v.V. performed molecular and cell biology experiments and analyzed data. A.d.M., J.F., A.U., and M.Q. performed and analyzed muscle function and histology studies. A.d.M. and T.A.R. wrote the manuscript. Competing interests: The authors declare no competing interests. Data and materials availability: All data are available in the manuscript or the supplementary materials.

Stay Connected to Science

Navigate This Article