Direct Reprogramming of Mouse Fibroblasts into Functional Skeletal Muscle Progenitors

MyoD and Small Molecules Induce Progenitor-like Program in Fibroblasts

To induce conversion of fibroblasts into cells of the skeletal muscle lineage, we chose to overexpress MyoD, which has previously been shown to trigger transdifferentiation of different somatic cell types into post-mitotic myotubes (Davis et al., 1987) (Figure 1A, top row). To this end, we engineered a doxycycline (dox)-inducible lentiviral system (tetOP-MyoD), allowing for inducible and reversible activation of MyoD in target cells. MEFs transduced with lentiviral vectors co-expressing tetOP-MyoD and M2rtTA showed extensive nuclear staining for the MyoD protein after 24 hr of dox administration (Figure S1A) and gave rise to elongated myotubes after 24–96 hr of dox exposure, consistent with previous findings (Bergstrom et al., 2002) (Figure 1B, red arrowheads). We confirmed the myogenic identity of converted cells by immunofluorescence staining for the differentiation marker myosin heavy chain (MyHC), which was absent in uninfected MEFs (Figure 1B).

We next exposed MEFs undergoing MyoD-induced transdifferentiation to various small molecules and cytokines in an attempt to induce reprogramming into a proliferative, myogenic progenitor-like cell state in addition to generating mature myotubes (Figure 1A, bottom row). Specifically, we used compounds that were previously shown to enhance reprogramming into iPSCs, including the GSK3β inhibitor CHIR99021 (Bar-Nur et al., 2014, Vidal et al., 2014) (abbreviated as “G”) and the transforming growth factor β1 (TGF-β1) receptor (ALK5) inhibitor RepSox (Ichida et al., 2009, Maherali and Hochedlinger, 2009) (abbreviated as “R”). In addition, we tested the effect of the cyclic AMP agonist forskolin (abbreviated as “F”), which reportedly facilitates the transient expansion of satellite cells in vitro (Xu et al., 2013). These compounds were added individually or combinatorially to MEFs expressing MyoD and cultured in KnockOut DMEM media containing 10% KnockOut Serum Replacement (KOSR) and 10% fetal bovine serum (FBS), and supplemented with 10 ng/mL basic fibroblast growth factor (bFGF), which promotes satellite cell and myoblast growth (Xu et al., 2013). We found that individual treatment of MyoD-expressing MEFs with F, G, or R as well as combined treatment with R/G or F/G did not noticeably enhance the formation of MyHC⁺, multinucleated myotubes (Figure S1B). By contrast, F/R treatment led to a marked increase in the number of proliferative cells, as confirmed by increased 5-ethynyl-2′-deoxyuridine (EdU) incorporation, and this effect was further enhanced by addition of G (i.e., F/R/G condition) (Figure 1C). Cultures treated with F/R or F/R/G typically assembled into three-dimensional colonies (Figure 1B, green arrowheads) with emanating myotubes and interspersed small, proliferative cells that we considered might be muscle progenitors. Strikingly, we also detected spontaneous and robust contractions within cultures treated with these small molecules, which we rarely detected in MEFs expressing MyoD alone (Video S1. MEF Control, Related to Figure 1, Video S2. MEFs Expressing MyoD, Related to Figure 1, Video S3. MEFs Expressing MyoD and Exposed to F/R, Related to Figure 1).

To assess whether concomitant expression of MyoD and exposure to our small molecules could activate a skeletal muscle-specific stem or progenitor cell program in fibroblasts, we performed microarray analysis of (1) untreated MEFs, or MEFs expressing (2) MyoD alone (MyoD condition) or (3) MyoD in combination with F/R (MyoD + F/R condition) for 14 days, or (4) the immortalized myoblast cell line C2C12. We found that the expression of genes associated with satellite cells and uncommitted myoblasts (e.g., Pax7, Myf5, Fgfr4) were significantly increased in the MyoD + F/R but not in the MyoD condition (Figure 1D, top left), as compared with MEFs alone. We further observed exclusive induction of genes associated with mature muscle tissue including Myh1, Myh4, and Myf6 in the MyoD + F/R condition, whereas markers associated with alternative lineages such as cardiomyocytes (e.g., Tbx5, Gata4, Nkx2-5) were not expressed at appreciable levels (Figure 1D, top right and bottom). We confirmed induction of Pax7 by qPCR and immunofluorescence in MEFs and tail-tip fibroblasts (TTFs) expressing MyoD and exposed to F/R or F/R/G (Figures 1E and S1B). In agreement with these results, we found that MEFs exposed to the MyoD + F/R condition but not to the MyoD condition share expression of other stem and progenitor cell-associated genes when compared with a previously published dataset of quiescent satellite cells (QSCs) and activated satellite cells (ASCs) (Figures 1F and S1C). A comparison of the most highly differentially expressed genes and associated gene ontology categories between MEFs expressing MyoD and those expressing MyoD and exposed to F/R corroborated the conclusion that the induced cell populations represent skeletal muscle and not cardiac muscle (Figures 1G and 1H). Together, these results demonstrate that the combined expression of exogenous MyoD and treatment with small molecules induces several characteristics of myogenic progenitors in MEFs including proliferation and transcriptional activation of key satellite cell and myoblast markers, as well as differentiation and contraction.

Transgene-Independent Progenitors Clonally Self-Renew and Express Key Satellite Cell and Myoblast Markers

We determined the temporal requirement for exogenous MyoD expression and small molecules during the establishment and maintenance of our progenitor-like cell population. To this end, we treated MyoD-transduced MEFs with dox for 0, 2, 4, 6, 8, or 12 days in the presence of F/R, followed by 7 days of dox withdrawal in the continuous presence of F/R, before scoring for contractile colonies containing Pax7⁺ cells (Figure S1D). We found that as little as 2 days of dox induction (i.e., exogenous MyoD expression) followed by 17 days of dox-independent growth was sufficient to generate contractile colonies with associated Pax7⁺ cells (Figures S1D–S1F). Individually picked colonies gave rise to stable cultures that could be propagated for at least 6 months or 20–24 passages in the absence of dox and containing both mononucleated cells and contracting myotubes (Figures S1D–S1F). However, continuous exposure of dox-withdrawn cultures to F/R was required for their maintenance as small-molecule removal led to their collapse (data not shown). We further derived transgene-independent myogenic progenitor cultures from subsets of MEFs or TTFs that were pre-sorted based on expression of the fibroblast-associated surface marker Thy1, confirming their origin from fibroblasts rather than a contaminating myogenic cell type (Figures 2A–2D). We also used an Oct4 lineage ablation system (Oct4-CreER;R26-LSL-DTA) (Bar-Nur et al., 2015) to rule out the possibility that our myogenic cells are generated indirectly via pluripotent intermediate cells, which subsequently differentiated (Bar-Nur et al., 2015, Maza et al., 2015) (Figures 2E–2H). Thus, we conclude that small molecules and growth factors collaborate with transient MyoD expression to endow MEFs and TTFs with a stable myogenic progenitor-like cell state. Once established, these cells no longer require exogenous MyoD expression and can be propagated long-term in the continuous presence of small molecules, demonstrating that they have acquired a transgene-independent self-renewing state. Based on these characterizations, we provisionally termed these cells “induced myogenic progenitor cells” (iMPCs).

We next used qRT-PCR, microarray, and immunofluorescence analyses to assess whether transgene-independent iMPC cultures express markers associated with different stages of myogenesis (see Figure S2A for a scheme). Indeed, we found that bulk iMPC cultures at both low (∼P5) and high (∼P15) passage (Figures 3A–3C and S2B–S2D; Video S4) as well as clonal iMPC lines that were derived from singly plated mononucleated cells (Figures 3D–3F and S3A; Video S5) expressed the satellite cell marker Pax7 and the differentiation markers MyHC and Myog. Importantly, both bulk and clonal iMPC cultures activated the endogenous MyoD locus, confirming our earlier observation that dox exposure is only required for the initiation and not maintenance of iMPCs (Figures 3B, 3C, 3E, 3F, and S3A). We noted that Pax7⁺ cells lacked MyHC expression, supporting the notion that these cells resemble undifferentiated myogenic stem or progenitor cells (Figure 3C). Quantification of cells expressing these marker combinations within bulk iMPC cultures further revealed the co-existence of iMPC subsets that expressed Pax7 and Myf5 or MyoD as well as subsets that expressed Pax7 but neither Myf5 nor MyoD (Figures 3G, 3H, S3B, and S3C). This observation suggests that the Pax7⁺ population contains more primitive cells resembling QSC-like cells as well as more committed progeny resembling ASCs. Notably, clonal iMPC cultures also contained Pax7⁺/MyoD⁺ double-positive cells but lacked Pax7⁺/MyoD⁻ cells, suggesting that the subcloning procedure may have selected for the outgrowth of more committed iMPC subsets (Figure S3D). This notion is consistent with the recent observation that differentiated myofibers and niche signals are critical for the prolonged maintenance of primitive subsets of Pax7⁺ satellite cells ex vivo (Quarta et al., 2016).

To further explore the similarity of transgene-independent Pax7⁺ iMPC subsets to QSCs or ASCs, we derived iMPCs from MEFs carrying a satellite cell-specific Pax7-nGFP transgene (Sambasivan et al., 2009). We detected up to 10% Pax7-nGFP⁺ cells in dox-independent iMPCs, confirming robust activation of the endogenous Pax7 locus (Figure 3I). Differences in the frequency of recorded Pax7⁺ iMPCs using immunofluorescence analysis (Figure 3H) and the reporter system are likely due to method-specific detection limits. Like satellite cells, sorted Pax7-nGFP⁺ iMPCs were small by forward/side scatter and occasionally gave rise to adjoining cells in culture that exhibited equivalent GFP expression, suggesting active cell division (Figures 3I and 3J). By contrast, sorted Pax7-nGFP⁻ cells were larger and morphologically resembled committed myoblasts (Figure S3E). We next extracted RNA from fluorescence-activated cell sorting-purified Pax7-nGFP⁺ iMPC cultures (referred to as “Pax7-nGFP⁺”) as well as from satellite cells that were harvested directly from mice (referred to as “QSCs”) or explanted on laminin-coated plates for 7 days in myoblast medium containing horse serum and FGF (Bosnakovski et al., 2008) to stimulate their proliferation (referred to as “ASCs”). We detected a similar number of genes that were upregulated in Pax7-nGFP⁺ iMPCs and either ASCs (342 genes) or QSCs (343 genes) in comparison with MEFs (Figure 3K). Representative examples of transcripts shared between Pax7-nGFP⁺ iMPCs, and ASCs include Itga7, Myog, and Mstn, and those shared between Pax7-nGFP⁺ iMPCs and QSCs include Dmrt2 and Heyl (Figures 3L, 3M, and S3F–S3H). Additionally, we noticed that Pax7-nGFP⁺ cells have silenced MEF-associated genes (e.g., Fbln5, Thy1), indicating effective extinction of the fibroblast program in iMPCs (Figures 3M and S3F–S3H). Together, these results show that purified Pax7-nGFP⁺ cells exhibit a gene expression profile that diverges from the parental MEFs and shares key molecular characteristics with skeletal muscle stem and progenitor cell populations.

iMPCs Differentiate into Dystrophin⁺ Myofibers in Dystrophic Recipient Mice

To assess whether iMPCs have the potential to contribute to muscle regeneration in vivo, we derived iMPC lines from MEFs using the aforementioned conditions and transplanted 1 × 10⁶ cells into the tibialis anterior (TA) or gastrocnemius muscle of 12-week-old homozygous mdx dystrophic mice (Figure 4A). One month after cell injection, we consistently observed contiguous areas of dystrophin⁺ myofibers in mdx muscles transplanted with iMPCs, whereas we only detected rare dystrophin⁺ revertant myofibers in PBS-injected mdx control muscles (Figures 4B and S4A). iMPC-derived dystrophin⁺ myofibers had centrally located nuclei and varied in size, indicative of an active regenerative process involving both new fiber formation and repair of the damaged endogenous myofibers (Figures 4B, 4C, and S4A). We next transplanted 1 × 10⁶ single cell-derived iMPCs into mdx recipients to test their in vivo regeneration potential. While these subclones also had the capacity to restore dystrophin expression, the dystrophin⁺ myofibers were reduced in size and smaller in number compared with bulk iMPC-derived grafts, which is consistent with our previous observation that these subclones exhibit a more differentiated phenotype (Figure S4B). Critically, we never observed tumor formation in muscles transplanted with either bulk or clonal iMPCs (data not shown). Altogether, these results demonstrate that iMPCs exhibit not only molecular but also key functional attributes of skeletal muscle progenitors.

Transplanted iMPCs Populate the Satellite Cell Niche and Initiate Regeneration upon Serial, Acute Injury

We next determined whether transplanted iMPCs have the capacity to populate the satellite cell niche, which is a defining characteristic of bona fide skeletal muscle stem cells. Examination of the injection site of mice transplanted with bulk or clonal cell-derived iMPCs revealed numerous mononucleated Pax7⁺ cells within regenerating dystrophin⁺ myofibers (Figures S4C and S4D). However, our experimental setup did not allow us to exclude that these were host-derived satellite cells. To determine whether iMPCs can give rise to mononucleated Pax7⁺ cells in vivo, we infected iMPCs with a lentiviral vector driving constitutive expression of the nuclear H2B-RFP fusion gene, and transplanted 1 × 10⁶ cells into mdx mice that were pre-injured with cardiotoxin (CTX) 24 hr prior to transplantation to enhance cell engraftment (Figure S4E). Consistent with our previous transplantation results, we observed robust contribution of iMPCs to mdx myofibers, assessed by restored dystrophin expression and centrally located nuclear H2B-RFP signal in a subset of these regenerating myofibers (Figures 4D and 4E). Importantly, we consistently found peripherally localized Pax7⁺ cells that were also H2B-RFP⁺ and DAPI⁺ within regions of engrafted, dystrophin⁺ myofibers (Figures 4F and 4G). This result suggests that at least a subset of transplanted iMPCs harbor the potential to populate the satellite cell niche in vivo.

To corroborate the conclusion that some of the transplanted iMPCs have acquired properties of muscle stem cells, we assessed their ability to initiate muscle regeneration using a serial, acute injury model (Hardy et al., 2016) (Figure 4H). In brief, we treated immunodeficient Foxn1^nu (referred to as nude mice) with 50 μM (1.2%) barium chloride (BaCl₂) to trigger muscle damage and enable engraftment of transplanted cells in non-dystrophic hosts (Figure S4E). Two days later, we transplanted either 1 × 10⁶ bulk iMPCs or 7.5 × 10⁴ satellite cells into the TA of recipient animals. Satellite cells were isolated using a previously reported protocol based on surface markers (Cerletti et al., 2008) (Figure S4F). We transplanted an excess of bulk iMPCs relative to purified satellite cells because of our prior observation that only 5%–10% of mononucleated iMPCs are Pax7⁺/MyoD⁻ (Figure 3H) or Pax7-nGFP⁺ (Figure 3I). After 26 days, we imaged transplants to confirm successful engraftment before subjecting muscles to another round of BaCl₂-induced injury (Figures 4I and S4G). Whereas analysis of individual transplants showed a noticeable loss or reduction of RFP signal 3 days after reinjury, we detected a robust RFP signal 31 days later, indicating successful regeneration by both iMPCs and satellite cells (n = 4 animals per cell type) (Figures 4I–4K, S4G, and S4H). Collectively, these results show that iMPCs have the potential to (1) engraft in dystrophic and non-dystrophic, injured muscles, (2) give rise to Pax7⁺ cells within transplants, and (3) contribute to muscle regeneration in a serial injury model.

iMPC Growth Is Driven by Pax7⁺ Muscle Stem-like Cells that Recapitulate Myogenesis In Vitro

To determine a possible lineage hierarchy within iMPC subsets in vitro, we first examined the expression of the surface marker VCAM-1, which has been associated with both QSCs and ASCs (Fukada et al., 2007). We noticed that the majority of mononucleated iMPCs were positive for VCAM-1 expression upon flow-cytometric analysis, which is consistent with a satellite cell or myoblast identity (Figure S5A). Consistently, immunofluorescence analysis showed that myofibers were negative while mononucleated cells were positive for VCAM-1 expression within heterogeneous iMPC cultures (Figure S5B). Critically, purified VCAM1⁺ iMPCs initially lacked MyHC expression by qRT-PCR, yet upregulated MyHC expression and gave rise to proliferative and contractile colonies upon culturing for 9 days (Figures S5C and S5D). However, VCAM-1⁻ iMPCs did not give rise to proliferating cells or contractile myotubes, which indicates a differentiated phenotype (Figures S5C and S5D). These results suggest a differentiation hierarchy between undifferentiated VCAM-1⁺ stem/progenitor-like cells and VCAM-1⁻ differentiated progeny akin to myogenic cells in vivo.

To further explore the hierarchical relationship among iMPC subsets and their possible resemblance to satellite cells, we generated iMPCs from MEFs carrying a satellite cell-specific Pax7-CreER allele (Murphy et al., 2011) as well as an ROSA26-LSL-EYFP reporter (Figure 5A). We failed to detect EYFP⁺ cells in MEFs exposed to 4OHT, ruling out the presence of contaminating Pax7⁺ myogenic or neural cells (Figure S5E). Moreover, expression of MyoD alone or in combination with single molecules was insufficient to activate the reporter in the presence of 4OHT in these MEFs, confirming our earlier observations (Figure S5F). By contrast, induction of MyoD in bulk MEFs, Thy1⁺ MEFs, or TTFs in the presence of bFGF and F/R or F/R/G activated the EYFP reporter in 3%–5% of cells after 6–10 days of MyoD induction and exposure to 4OHT (Figures 5B and S5G–S5J). Treatment of established, dox-independent bulk or clonal iMPC cultures with 4OHT for 2 days gave rise to a similar range of EYFP⁺ cells (1%–7%) (Figures S6A and S6B) and this fraction progressively increased to over 80% upon continuous passaging in the presence of 4OHT (Figures 5C, S6C, and S6D; Videos S6A and S6B). Importantly, we detected EYFP signal not only in contracting mononucleated cells but also in multinucleated myofibers 1–2 weeks after 4OHT treatment (Figure 5D). We further noticed that the vast majority of iMPCs that were EYFP⁺ after a 3-day 4OHT pulse were also VCAM-1⁺ (93%) (Figure S6E), which is consistent with our finding that VCAM-1 positivity corresponds to an immature myogenic state (Figures S5A–S5D). Together, these assays corroborate the conclusion that iMPC cultures contain undifferentiated Pax7⁺ stem-like cells, which self-renew and differentiate into mature, contracting myotubes over time, thus recapitulating aspects of normal myogenesis ex vivo.

To determine whether Pax7-expressing cells are crucial for the establishment of iMPCs, we generated MEFs carrying Pax7-CreER and ROSA26-LSL-DTA alleles and induced MyoD in the presence or absence of F/R/G and 4OHT (Figure 5E). Exposure of these cells to 4OHT is expected to ablate Pax7-expressing cells by inducing diphtheria toxin (DTA)-mediated apoptosis. MEFs expressing exogenous MyoD readily gave rise to myotubes that activated endogenous MyoD and Myog in the presence or absence of 4OHT (Figure 5F). By contrast, 4OHT treatment blunted the generation of Pax7 and Myf5 expressing iMPCs, yet did not affect the formation of MyoD/Myog-expressing myotubes, indicating that Pax7⁺ cells are indeed essential for the formation of iMPCs (Figures 5F–5H).

iMPC Maintenance Requires the Satellite Cell Master Regulator Pax7

We next tested whether the establishment or maintenance of iMPC cultures is dependent upon Pax7 itself. To this end, we intercrossed Pax7^+/− mice (Mansouri et al., 1996) to obtain both Pax7^−/− experimental and Pax7^+/− control MEFs (Figure 5I). These cells were infected with lentiviral vectors expressing M2rtTA and tetOP-MyoD and subsequently exposed to either transdifferentiation (MyoD) or reprogramming (MyoD + F/R/G) conditions. MyoD expression alone yielded multinucleated myotubes from both Pax7^+/− and Pax7^−/− MEFs, indicating that Pax7 is dispensable for the direct conversion of fibroblasts to myotubes (Figure 5J, left). Moreover, we detected proliferative and contractile colonies in both Pax7^+/− and Pax7^−/− cultures upon overexpression of MyoD and treatment with dox and F/R/G, suggesting that Pax7 is also dispensable for the establishment of iMPC-like cells (data not shown). However, we were unable to maintain these Pax7^−/− iMPC-like cultures despite the presence of F/R/G treatment (Figure 5J, bottom right). Specifically, Pax7^−/− cultures ceased to proliferate and contract over time, leaving behind only post-mitotic myotubes that lacked Pax7 or Myf5 expression (Figures 5J–5L). We conclude that iMPC propagation in vitro relies on the same transcriptional regulator as satellite cell maintenance in vivo, providing further mechanistic evidence that these two cell states are related.

Derivation of MPCs from Muscle Tissue Using Small Molecules Alone

Our finding that concomitant MyoD expression and small-molecule treatment endows fibroblasts with a myogenic progenitor cell (MPC) state raises the possibility that small molecules alone are sufficient to capture an MPC-like state in primary muscle cells, which already express endogenous MyoD. To test this hypothesis, we explanted muscle tissue from Pax7-CreER;ROSA26-LSL-EYFP mice that had been treated with a pulse of tamoxifen to label satellite cells, isolated cells through mechanical and enzymatic digestion, and cultured these in iMPC medium (DMEM, KOSR, FBS, bFGF, F/R or F/R/G) without 4OHT (Figure 6A). We were indeed able to establish MPC-like colonies that could be propagated for several passages and were consistently EYFP⁺, demonstrating their origin from the satellite cell lineage (Figures 6B–6D; Videos S7A and S7B). Quantification of cells expressing Pax7 and MyoD within these satellite cell lineage-derived MPCs (SC-MPCs) revealed a similar proportion of Pax7⁺/MyoD⁺ and Pax7⁺/MyoD⁻ subsets (Figures 6E and 6F) compared with MEF-derived iMPCs (Figure 3H). Moreover, SC-MPCs engrafted and differentiated into dystrophin⁺ myofibers with centrally located nuclei and EYFP fluorescence following transplantation into mdx mice, documenting their differentiation and regeneration potential in vivo (Figures 6G–6I). Notably, we detected rare EYFP⁺/Pax7⁺ cells within dystrophin⁺ areas, suggesting that transplanted SC-MPCs may also replenish the endogenous satellite cell pool (Figure 6I). However, we acknowledge that serial injury or serial transplantation experiments would be required to assess whether MPC-derived Pax7⁺ cells are similar to bona fide satellite cells. These results show that our small-molecule cocktail not only facilitates the reprogramming of MEFs expressing MyoD into iMPCs but also allows capture of MPCs from the satellite cell lineage (i.e., satellite cells or their progeny).