Direct reprogramming of fibroblasts into skeletal muscle progenitor cells by transcription factors enriched in undifferentiated subpopulation of satellite cells

Satellite cells comprise a functionally heterogeneous population of stem cells in skeletal muscle. Separation of an undifferentiated subpopulation and elucidation of its molecular background are necessary to identify the reprogramming factors to induce skeletal muscle progenitor cells. In this study, we found that intracellular esterase activity distinguishes a subpopulation of cultured satellite cells with high stemness using esterase-sensitive cell staining reagent, calcein-AM. Gene expression analysis of this subpopulation revealed that defined combinations of transcription factors (Pax3, Mef2b, and Pitx1 or Pax7, Mef2b, and Pitx1 in embryonic fibroblasts, and Pax7, Mef2b and MyoD in adult fibroblasts) reprogrammed fibroblasts into skeletal muscle progenitor cells. These reprogrammed cells formed Dystrophin-positive mature muscle fibers when transplanted into a mouse model of Duchenne muscular dystrophy. These results highlight the new marker for heterogenous population of cultured satellite cells, potential therapeutic approaches and cell sources for degenerative muscle diseases.


Results
In the previous study, we analyzed intracellular Ca 2+ levels ([Ca 2+ ] i ) using fluorescent Ca 2+ indicator, Fluo-4 AM 18 . Interestingly, we observed stronger Fluo-4 fluorescence in differentiated C2C12 myotubes compared with that in undifferentiated cells even before stimulation with Ca 2+ ionophores or agonists (data not shown). We first hypothesized that this basal differences in the fluorescence intensity of Fluo-4 were due to the differences in [Ca 2+ ] i between differentiated and undifferentiated cells. However, unexpectedly, we also observed the similar differences in the fluorescence intensity using [Ca 2+ ] i -independent cell-staining reagents, such as Calcein-AM, fluorescein diacetate (FDA), BCECF-AM and carboxyfluorescein succinimidyl ester (CFSE) (Supplementary Fig. 1). These results suggested that the differences in the fluorescence intensity were originated from the common property of used fluorescent reagents. These reagents have a common chemical structure of acetoxymethyl ester (AM). AM is used for wide variety of reagents to enhance the membrane permeability 19 . AM-based cell-staining reagents are digested by unspecific esterases after which cells are labeled with fluorescence 19 , suggesting that differences in fluorescence intensity were derived from the differences in the esterase activity between differentiated and undifferentiated cells.
In this study, we found a relationship between SC heterogeneity and intracellular esterase activity. Calcein-AM is a widely used cell-staining reagent that has a chemical structure of AM. We treated primary cultured SCs (cSCs) with calcein-AM 7 days after isolation by single fiber culture, when undifferentiated stem cells and differentiated myotubes existed simultaneously. We observed strong calcein fluorescence in differentiated myotubes (Fig. 1a, upper left). Interestingly, we also observed heterogeneous calcein fluorescence among undifferentiated round cells, suggesting a relationship between esterase activity and heterogeneity of cSCs (Fig. 1a, upper right and lower). We therefore separated subpopulation of cSCs based on calcein intensity by fluorescence-activated cell sorting (FACS). Dynamic range of calcein intensity in cSCs was wider than C2C12 mouse muscle cell line, indicating the heterogeneous esterase activity in cSCs (Fig. 1b upper). The lower, middle, and upper 10-15% of the population were defined as Calcein low , Calcein middle and Calcein high cSCs, respectively (Fig. 1b lower). The expression of Pax7 was 3-fold lower in Calcein high cSCs compared with the other two groups (Fig. 1c). Consistent with this, expression of Myogenin and Desmin were also higher in Calcein high cSCs; notably, expression of Myogenin was 10-fold higher in Calcein high cSCs compared with the other two groups, suggesting that differentiating or differentiated myocytes were enriched in Calcein high cSCs. Calcein high cSCs also showed 3-fold lower and 4-fold higher expression levels of Myf5 and MyoD compared with Calcein low cSCs, respectively. We assessed cell proliferation ability by analyzing time-dependent changes of the cell numbers (Fig. 1d,e). Proliferation of Calcein high cSCs was lower than that of the other two groups. Calcein low cSCs showed intermediate proliferation. Furthermore, the percentage of bromodeoxyuridine (BrdU)-positive cells was highest in Calcein middle cSCs and lowest in Calcein high cSCs (Fig. 1f,g). We transplanted these different subpopulations into mdx mice, a mouse model of Duchenne muscular dystrophy, and counted the numbers of Dystrophin-positive engrafted fibers. We observed the highest number of Dystrophin-positive fibers in Calcein low cSCs-transplanted mdx mice (Fig. 1h,i). Overall, these results indicated that the esterase activity was increased with the differentiation of cSCs, and: 1) differentiating, non-proliferating cells were enriched in Calcein high cSCs; 2) vigorously growing cells were enriched in Calcein middle cSCs; and 3) relatively undifferentiated cSCs, which showed low proliferation and high transplantation efficiency, were enriched in Calcein low cSCs.
Expression of Myogenin in Calcein middle cSCs was 2 to 3 times higher than that in the Calcein low cSCs, though no difference in expression of Pax7 was detected between these subpopulations (Fig. 1c). Together with the differences in proliferation ability ( Fig. 1d-g) and transplantation efficiency (Fig. 1h,i), these results implied that Calcein low cSCs had the highest stemness. We analyzed the molecular signature of these cells by genome-wide gene expression analysis (Supplementary Table 1). Several genes related to muscle development and structural components (i.e., myofibril, muscle contraction) were enriched in Calcein high cSCs compared with Calcein low cSCs (Supplementary Table 2). Interestingly, genes related to muscle development and structural components were also enriched in Calcein middle cSCs compared with Calcein low cSCs, further supporting the idea that undifferentiated stem cells were enriched in the Calcein low fraction.
Given these results, we hypothesized that TFs enriched in Calcein low cSCs include those with the ability to maintain the undifferentiated state, which might have the potential to induce non-muscle cells, such as fibroblasts, into the myogenic lineage. We investigated TFs (Meox1, Meox2, Mef2b, Twist1, Twist2, Pitx1, and Hoxc12) with higher expression levels in Calcein low cSCs compared with Calcein middle cSCs ( Supplementary Fig. 2), and exogenously expressed these TFs in mouse embryonic fibroblasts (MEFs), together with Pax3, Pax7, and MyoD, using retrovirus. We observed colonies of round cells with morphology similar to cSCs (Fig. 2a), and designated these as iSkM progenitor cells. We removed the overexpressed TFs one by one to identify which TFs were essential, and identified either Pax3, Mef2b, and Pitx1, or Pax7, Mef2b, and Pitx1 as essential TFs for the induction of iSkM progenitor cells from MEFs (Fig. 2b,c), while MyoD was not required. We quantified the number and size of colonies. Both number and size of colonies was much higher in Pax7-expressing lines compared with that in Pax3-expressing lines (Supplementary Table 3). We also used doxycycline (dox)-inducible lentivirus in which expression of exogenes was upregulated by addition of dox with stable expression of orange fluorescent protein Kusabira-Orange 20,21 (Fig. 2b). Dox-induced expression of these TFs resulted in formation of iSkM progenitor cells in a concentration-dependent manner (Fig. 2d). Furthermore, leukemia inhibitory factor (LIF) and basic fibroblast growth factor (bFGF), well-known growth factors that inhibit differentiation of SCs 22,23 , enhanced production efficiency of iSkM progenitor cells (Fig. 2d,e). We defined iSkM progenitor cells that were established from each dish as a cell line, rather than a clone, because each line was composed of a heterogeneous cell population, as described below.
To purify and characterize the iSkM progenitor cells derived from MEFs, we analyzed the several surface markers by FACS. iSkM progenitor cells were purified based on their expression of M-cadherin, a SC and myogenic cell marker 24 (Fig. 3a), and were used in subsequent experiments. Expression of CD3e, CD11b, CD31, CD34, CD36, CD45, CD45R, CD68, CD93, CD135, Flk-1, PDGFRα, Ly-6G and TER-119 in iSkM progenitor cells was comparable to that in MEFs ( Supplementary Fig. 3). The number of M-cadherin-positive cells was significantly higher in Pax7-expressing lines, though the incidence of colonies in Pax3-expressing lines was comparable to that in Pax7-expressing lines (Fig. 3b), suggesting the overlapping but distinct function between Pax3 and Pax7 25 . Myf5 was expressed in these cells, though MyoD and Myogenin were rarely detected (Fig. 3c). In addition to exogenous expression of Pax3 or Pax7, Mef2b, and Pitx1 (Fig. 3d), endogenous expression of Myf5, but not MyoD, was comparable to cSCs (Fig. 3e). Endogenous expression of Pax7 was not observed. Endogenous expression of Pax3 was slightly higher than cSCs, though its differences were not statistically significant. Together with the low expression level of Pax3 in limb muscle 25 , these results suggested that iSkM progenitor cells maintained their myogenic properties by endogenous expression of Myf5 and exogenous expression of Pax3 or Pax7. Expression levels of the differentiation markers, Myogenin, Mef2c, Mrf4, and Desmin in iSkM progenitor cells were lower than those in differentiated myotubes (Fig. 3e). Bisulfite genomic sequencing demonstrated that the promoter regions of Myf5 were demethylated overall in iSkM progenitor cells ( Supplementary Fig. 4), while both demethylated and methylated sites were observed simultaneously in some cell lines, suggesting the existence of a heterogeneous population. We further evaluated the myogenic potential of iSkM progenitor cells by analyzing their ability to differentiate into myotubes in vitro. Multi-nucleated myosin heavy chain (MyHC)-positive myotubes were detected 4 days after induction with removal of dox (Fig. 3f). Co-culture of iSkM progenitor cells with cSCs isolated from GFP mice resulted in the formation of both Kusabira-Orange-and GFP-positive myotubes (Fig. 3g), suggesting that iSkM progenitor cells formed myotubes by fusing with endogenous muscle cells. Although brown adipocytes and skeletal muscle cells originate from the same Myf5-positive cells, iSkM progenitor cells did not differentiate into white or brown adipocytes 26 (Supplementary Fig. 5). We transplanted iSkM progenitor cells derived from MEFs into mdx mice, and observed a significant increase in Dystrophin-positive myofibers (Fig. 3h,i).  Round iSkM progenitor cells were induced from adult tail-tip fibroblasts (TTFs) by exogenous expression of Pax7, Mef2b, and Pitx1. However, we failed to expand these cells, suggesting that exogenous expression of these TFs was not sufficient to induce stable iSkM progenitor cells from adult fibroblasts (data not shown). Similar difficulties in inducing cardiomyocytes from TTFs, which were more resistant to reprogramming, have also been reported 27 . Therefore, we expressed these factors exogenously with MyoD, the first identified reprogramming factor to induce myocytes from fibroblasts 4 , which was not required for induction of iSkM progenitor cells from MEFs. iSkM progenitor cells were induced from TTFs by exogenous expression of Pax7, Mef2b, Pitx1, and MyoD (Fig. 4a). Pitx1 was not required in the presence of MyoD, suggesting that Pitx1 and MyoD have overlapping functions (Fig. 4b,c). iSkM progenitor cells derived from TTFs were positive for M-cadherin (Fig. 4d), and immunocytochemistry revealed that these cells expressed Pax7, Myf5, and MyoD, but not Myogenin (Fig. 4e). Gene expression analysis showed the endogenous expression of Myf5 and MyoD in iSkM progenitor cells derived from TTFs (Fig. 4f,g). These cells formed myotubes in vitro (Fig. 4h). Finally, we transplanted iSkM progenitor cells derived from TTFs into mdx mice, and observed a significant increase in Dystrophin-positive myofibers (Fig. 4i,j). Overall, we achieved direct reprogramming of fibroblasts into skeletal muscle progenitor cells by defined factors enriched in an undifferentiated subpopulation of cSCs ( Supplementary Fig. 6).

Discussion
Loss of the undifferentiated state and subsequent differentiation are associated with intracellular metabolic systems such as anaerobic glycolysis and oxidative phosphorylation in SCs and hematopoietic stem cells 28 . In this study, we identified esterase activity as a novel hierarchical marker of cSCs. Further studies are needed to determine the molecular mechanisms linking esterase activity or its metabolites to heterogeneity of cSCs. The relationship between esterase activity and stemness in more superior stem cells, such as quiescent SCs and other tissue stem cells, is also worthy of consideration.
We identified the combination of Pax3, Mef2b, and Pitx1 or Pax7, Mef2b, and Pitx1 in embryonic fibroblasts, and Pax7, Mef2b and MyoD in adult fibroblasts as essential TFs for inducing skeletal muscle progenitor cells. The spatiotemporal pattern of Mef2b expression in developing skeletal muscle lineages differs from that of other Mef2 genes that are required for adult muscle regeneration 29,30 . In accordance with these previous studies, we found that only Mef2b was upregulated in Calcein low cSCs, while others were upregulated in Calcein high cSCs (Supplementary Table 1), suggesting the unique character of Mef2b. In addition to Mef2b, the other Calcein low cSCs-enriched TFs included noteworthy genes from the past findings. Pitx1 was required for hindlimb patterning and development by upregulating Tbx4 31,32 . Meox1 and Meox2 were expressed in developing limb buds, and their disruption resulted in impaired expression of Pax3, Pax7, and Myf5 in mouse embryos 33,34 . Twist, in co-operation with Notch, negatively regulated muscle differentiation in Drosophila 35,36 . Analysis of Calcein low cSCs-enriched genes other than TFs may help to elucidate the underlying system for the maintenance of heterogeneous populations and their stemness in SCs.
Distinct from the exogenous expression of MyoD alone 4 , the induction of iSkM cells included a progenitor-cell state, allowing these cells to be purified and their proliferation maintained. Importantly, iSkM progenitor cells did not show high endogenous expression of Pax3 or Pax7 (Figs 3e and 4g), suggesting that iSkM progenitor cells were not "induced muscle stem cells". Induction of these muscle stem cells might require upstream TFs responsible for inducing the endogenous expression of Pax3 or Pax7. Because overexpression of MyoD exerts anti-proliferation effects, requirement of MyoD to generate proliferating muscle progenitor cells from TTFs seems to be controversial. As proliferating cSCs express MyoD, some TFs, such as Pax7, might inhibit the anti-proliferating effects of MyoD. Overall, we investigated the molecular background of cSC subpopulations and identified TFs essential for inducing skeletal muscle progenitor cells from fibroblasts. These results may contribute to the development of new therapeutics for muscle degenerative diseases, such as Duchenne muscular dystrophy.

Materials and Methods
Animals. Twelve-to sixteen-week-old male C57BL/6 mice were purchased from Nihon CREA. C57BL/6-GFP transgenic mice were kindly provided by Dr. Masaru Okabe (Osaka University, Osaka, Japan). C57BL/6-mdx mice were a kind gift from Dr. Toshikuni Sasaoka (National Institute for Basic Biology, Aichi, Japan). All mice were housed at the institutional animal facility. All of the animal procedures were approved by the Experimental Animal Care and Use Committee at the National Center of Neurology and Psychiatry (NCNP, Tokyo, Japan). All of the experimental methods were performed in accordance with approved guidelines.

FACS analysis.
Calcein-AM-treated cSCs 4 or 7 days after isolation were dissociated by 0.05% trypsin/EDTA (Thermo Fisher Scientific) at 37 °C with 5% CO2 for 5 min. Cell sorting was performed on a FACS ARIA (BD Biosciences). The lower, middle, and upper 10-15% populations of calcein-AM-treated cSCs were defined as Calcein low , Calcein middle , and Calcein high cSCs, respectively, based on the fluorescence intensity of calcein. For surface-marker analysis and purification by M-cadherin antibody, cells were dissociated in Cell Dissociation Buffer (Thermo Fisher Scientific) at 37 °C with 5% CO2 for 30 min, followed by vigorous pipetting. Proliferation assay. After FACS sorting, 1.0 × 10 4 Calcein low , Calcein middle , and Calcein high cSCs, respectively, 4 days after isolation were seeded onto 24-well plates and the numbers of cells per field were counted every day. For BrdU assay, cSCs 4 days after isolation were treated with 10 μM BrdU (Sigma Aldrich) in DMEM (high glucose, sodium pyruvate, and GlutaMAX supplement) supplemented with 20% FBS, 1% CEE, and 1% PS at 37 °C with 5% CO2 for 2 h. The cells were then treated with 50 nM calcein-AM followed by FACS sorting. Sorted Calcein low , Calcein middle , and Calcein high cSCs were fixed with 4% paraformaldehyde/phosphate-buffered saline (PBS). After washing, fixed cells were treated with 0.1% Triton X-100/PBS for 10 min for permeabilization, followed by 2 N HCl for 30 min at room temperature, blocking with 5% goat serum (Cedarlane) in 2% bovine serum albumin (BSA)/PBS for 15 min, and incubation with anti-BrdU (1:400, clone: BU1/75 (ICR1); AbD Serotec) in 2% BSA/PBS at 4 °C overnight. After washing, the cells were incubated with Alexa Fluor 488-labelled secondary antibody (1:1000; Thermo Fisher Scientific) in 1% BSA/PBS. After several washings, nuclei were stained with DAPI (Dojindo). Immunofluorescent-staining images were evaluated by fluorescence microscopy (Olympus IX71), and the percentage of BrdU-positive cells was counted.