Isolation of Progenitors that Exhibit Myogenic/Osteogenic Bipotency In Vitro by Fluorescence-Activated Cell Sorting from Human Fetal Muscle

Summary Fluorescence-activated cell sorting (FACS) strategies to purify distinct cell types from the pool of fetal human myofiber-associated (hMFA) cells were developed. We demonstrate that cells expressing the satellite cell marker PAX7 are highly enriched within the subset of CD45−CD11b−GlyA−CD31−CD34−CD56intITGA7hi hMFA cells. These CD45−CD11b−GlyA−CD31−CD34−CD56intITGA7hi cells lack adipogenic capacity but exhibit robust, bipotent myogenic and osteogenic activity in vitro and engraft myofibers when transplanted into mouse muscle. In contrast, CD45−CD11b−GlyA−CD31−CD34+ fetal hMFA cells represent stromal constituents of muscle that do not express PAX7, lack myogenic function, and exhibit adipogenic and osteogenic capacity in vitro. Adult muscle likewise contains PAX7+ CD45−CD11b−GlyA−CD31−CD34−CD56intITGA7hi hMFA cells with in vitro myogenic and osteogenic activity, although these cells are present at lower frequency in comparison to their fetal counterparts. The ability to directly isolate functionally distinct progenitor cells from human muscle will enable novel insights into muscle lineage specification and homeostasis.


INTRODUCTION
In mice, combinatorial surface marker analysis has been useful in enabling direct discrimination and prospective isolation of phenotypically and functionally distinct cells from skeletal muscle using fluorescence-activated cell sorting (FACS) (Cerletti et al., 2008;Kuang et al., 2007;Sacco et al., 2008;Sherwood et al., 2004;Tanaka et al., 2009). FACS has been used to purify PAX7-expressing mouse satellite cells, which exhibit self-renewal and myogenic differentiation capacities consistent with muscle stem cells (Cerletti et al., 2008;Fukada et al., 2004;Kuang et al., 2007;Montarras et al., 2005;Sacco et al., 2008;Sherwood et al., 2004;Tanaka et al., 2009). Prospective isolation of adult mouse satellite cells has also enabled studies that distinguished their myogenic differentiation potential from adipogenic/fibrogenic activities in muscle (Joe et al., 2010), revealed their contributions to muscle pathologies (Cerletti et al., 2008;Chakkalakal et al., 2012;Conboy et al., 2003;Sacco et al., 2008), and provided proof in principle that they may be useful in cell therapy approaches (Cerletti et al., 2008(Cerletti et al., , 2012Sacco et al., 2008). A similar cell-sorting approach recently allowed purification of fibroadipogenic precursors from mouse muscle and showed that these cells lack myogenic capacity (Joe et al., 2010;Uezumi et al., 2010). Together with endothelial and infiltrating immune cells, these fibroadipogenic precursors constitute the muscle stroma and play a critical role in regulating the early stages of muscle repair after damage (Wang and Rudnicki, 2012). However, in order to translate these findings to human muscle and apply them for regenerative medicine, it is essential to develop analogous strategies for prospective identification and isolation of human myogenic and adipogenic precursors. Lecourt et al. previously showed by immunofluorescence (IF) staining that cells in the satellite cell position in adult human muscle lack CD34 (Lecourt et al., 2010). Pisani et al. subsequently demonstrated that myogenic activity could be enriched in human adult muscle cells by magnetic depletion of CD34 + cells (Pisani et al., 2010b). However, as described here, negative selection for CD34 achieves only partial purification of myogenic progenitors from human fetal muscle. To establish more specific sorting strategies capable of purifying human PAX7-positive cells, we undertook a systematic study of surface markers that distinguish phenotypically and functionally distinct cells in human fetal muscle. These efforts identified a combination of seven surface markers that reliably discriminate a purified population of PAX7-expressing CD45 À CD11b À GlyA À CD31 À CD34 À CD56 int ITGA7 hi human myofiber-associated (hMFA) cells (hereafter referred to as CD34 À CD56 int ITGA7 hi cells) from infiltrating blood cells and muscle-resident adipogenic precursors, allowing direct isolation of each of these populations by FACS. Consistent with studies in the mouse, human PAX7-expressing CD34 À CD56 int ITGA7 hi cells are robustly myogenic and lack adipogenic potential. PAX7-expressing CD34 À CD56 int ITGA7 hi cells with myogenic activity in vitro are also present in adult muscle, but at lower frequency than in fetal tissue. Clonal analysis in vitro further revealed a surprising bipotency of human fetal PAX7-expressing CD34 À CD56 int ITGA7 hi cells, which exhibited both myogenic and osteogenic potential. In contrast, CD45 À CD11b À GlyA À CD31 À CD34 + fetal hMFA cells (abbreviated CD34 + cells), which exhibited potent adipogenic and osteogenic activity, lack PAX7 and show no myogenic potential. Taken together, these studies report efficient methods for the direct isolation of highly enriched human fetal bipotent myogenic/ osteogenic and adipogenic progenitors. These protocols provide tools for uncovering the cellular mechanisms and environmental interactions that sustain human skeletal muscle.

Human Fetal Skeletal Muscle Contains Multiple, Distinct Cell Populations
To evaluate phenotypic and functional heterogeneity among fetal hMFA cells, we adapted previously established protocols for mouse myofiber-associated cell isolation (Conboy et al., 2003;Sherwood et al., 2004) to liberate the mononuclear cell fraction from human fetal muscle (Ehrhardt et al., 2007;Tanaka et al., 1995). Plating hMFA cells under myogenic, adipogenic, or osteogenic conditions in vitro revealed significant functional heterogeneity. Under myogenic conditions, hMFA cells formed DESMIN-expressing multinucleated myotubes. Under adipogenic conditions, hMFA cells differentiated into oil red O (ORO)-positive, lipid droplet-containing adipocytes. Under osteogenic conditions, hMFA cells produced Alizarin-red (AR)-positive calcium deposits consistent with osteogenic differentiation ( Figure 1A).
To further enrich PAX7 + cells within the CD34 À/low hMFA cell subset, we evaluated expression of additional surface markers ( Figure 1B). Differential expression of Stem Cell Reports j Vol. 2 j 92-106 j January 14, 2014 j ª2014 The Authors 93 Stem Cell Reports FACS Isolation of Progenitors from Human Muscle CD90, CD13, and CD29 did not discriminate candidate populations within the CD34 À/low hMFA cell pool, as CD90 and CD13 expression was enriched in the CD34 + cell subset and CD29 was uniformly expressed by CD34 À/low cells (data not shown). In contrast, differential expression of CD56 and ITGA7 distinguished three populations within CD34 À/low fetal hMFA cells: CD34 À CD56 hi ITGA7 low (marked in blue in Figure 1C), CD34 À CD56 int ITGA7 hi (marked in red in Figure 1C), and CD34 À/low CD56 À ITGA7 low (marked in green in Figure 1C). In addition to differences in CD56 and ITGA7 expression, variable low-level expression of CD34 was noted in CD34 À/low CD56 À ITGA7 low hMFA cells, compared to absent CD34 expression in CD34 À CD56 int ITGA7 hi and CD34 À CD56 hi ITGA7 low cells ( Figure S1H). PAX7 IF showed clear enrichment of PAX7+ cells (78.3% ± 5%, mean ± SD) in the CD34 À CD56 int ITGA7 hi subset ( Figure 2; red gate in Figure 1C). In contrast, we detected no PAX7 expression in CD34 À CD56 hi ITGA7 low cells ( Figures 2B and S2; blue gate in Figure 1C) and variable PAX7 expression in CD34 À/low CD56 À ITGA7 low cells (33.0% ± 28%, mean ± SD; Figures 2B and S2; green gate in Figure 1C).
Differences in the myogenic and adipogenic differentiation capacity of fetal CD34 À CD56 int ITGA7 hi cells and CD34 + cells remained evident under all the culture conditions used. Specifically, myogenic CD34 À CD56 int ITGA7 hi cells still formed multinucleated myotubes under adipogenic ( Figure S4A) and osteogenic (Figures 3B and S4B) conditions and never formed adipocytes ( Figures S4A and S4B). In contrast, CD34 + cells formed adipocytes under adipogenic (Figures 3 and S4A) and osteogenic ( Figure S4B) conditions and never formed myotubes (Figures 3 and S4A). Under myogenic conditions, CD34 + cells adopted fibroblastic morphology (data not shown).
Human Fetal CD34 À CD56 int ITGA7 hi Cells Exhibit Bipotent Myogenic/Osteogenic Differentiation Activity In Vitro The ability of PAX7-positive, CD34 À CD56 int ITGA7 hi human fetal cells to efficiently undergo myogenic differentiation in vitro and form AR-positive calcium deposits indicative of osteogenic differentiation is consistent with previous reports of hMFA osteogenic differentiation (Hashimoto et al., 2008;Lecourt et al., 2010;Oishi et al., 2013). To determine whether this observation reflects true bipotency, we clone-sorted CD34 À CD56 int ITGA7 hi cells into 96-well plates (one cell per well) and expanded them for parallel analysis of myogenic and osteogenic activity in vitro ( Figure 4A). The efficiency of CD34 À CD56 int ITGA7 hi fetal hMFA cell clonal expansion was variable and donor dependent, with an average seeding efficiency of 7.1% ± 5.4% (mean ± SD; Figure 4B). We succeeded in expanding 18 clones from four donors. After 14-22 days of expansion, these 18 clones were split and replated under myogenic or osteogenic differentiation conditions. Under myogenic conditions, 100% ( Figure 4B) of clones differentiated into DESMIN-positive myotubes (Figure 4C; one representative clone shown). Under osteogenic conditions, 82.5% ± 23.6% (mean ± SD; Figure 4B) of clones formed AR-positive calcium deposits ( Figure 4D; one representative clone shown). We conclude that the majority of fetal CD34 À CD56 int ITGA7 hi hMFA cells possess bipotent myogenic and osteogenic differentiation potential in these in vitro assays.
To further investigate the potential of fetal CD34 À CD56 int ITGA7 hi cells, we determined their expression of myogenic and osteogenic lineage genes at the end of clonal expansion and after myogenic or osteogenic differentiation ( Figure 4A). All (four out of four) CD34 À CD56 int ITGA7 hi cell-derived clones analyzed by quantitative RT-PCR (qRT-PCR) expressed both myogenic (MYOD and DESMIN; Figures 4E and 4F) and osteogenic (RUNX2, OSTERIX/SP7, and OSTEOCALCIN/BGLAP; Figures 4G-4I) genes. Osteogenic lineage gene expression at the end of in vitro clonal expansion and after myogenic and osteogenic differentiation corroborated the osteogenic activity of human fetal CD34 À CD56 int ITGA7 hi cells. Expression of BGLAP increased significantly in cells that underwent osteogenic differentiation compared to proliferating clones; however, BGLAP did not increase in cells that underwent myogenic differentiation ( Figure 4I). Increased expression of RUNX2 and SP7 in CD34 À CD56 int ITGA7 hi cells after osteogenic differentiation in vitro was variable between clones and did not reach statistical significance (Figures  (red) and nuclei marked by DAPI (blue) after myogenic culture (left panels), with oil red O to mark lipid droplets after adipogenic culture (middle panels), and with Alizarin red to mark calcium deposits after osteogenic culture (right panels). (A) CD34 + cells lack myogenic activity but are adipogenic and osteogenic in vitro, while fetal CD34 À/low cells contain myogenic, adipogenic, and osteogenic activity in vitro. (B) CD34 À CD56 hi ITGA7 low cells lack myogenic, adipogenic, and osteogenic capacity in vitro. CD34 À CD56 int ITGA7 hi cells exhibit efficient myogenic activity, lack adipogenic potential, and contain osteogenic activity in vitro. CD34 À/low CD56 À ITGA7 low cells show limited myogenic activity but are adipogenic and osteogenic in vitro. Differentiation assays were performed in three biological replicates for each population. See also Figure S4.
96 Stem Cell Reports j Vol. 2 j 92-106 j January 14, 2014 j ª2014 The Authors Stem Cell Reports FACS Isolation of Progenitors from Human Muscle 4G and 4H). Importantly, at least some CD34 À CD56 int ITGA7 hi cells exposed to osteogenic differentiation conditions maintained myogenic activity as evidenced by an increase in DESMIN expression during osteogenic culture (Figures 4F) and the presence of multinucleated myotubes in these cultures (data not shown).
Human Fetal CD34 À CD56 int ITGA7 hi Cells Engraft Myofibers when Transplanted into Mouse Skeletal Muscle To assess the contributions of FACS-isolated fetal hMFA cells to muscle regeneration in vivo, we adapted previ-ously published protocols to detect engraftment of unfractionated human myogenic cells in mouse muscle (Ehrhardt et al., 2007). Freshly isolated cells were injected directly into the cardiotoxin preinjured tibialis anterior muscles of immunodeficient nonobese diabetic severe combined immunodeficiency interleukin-2g À/À mice (NSG) mice, transplanted muscles were harvested 3-8 weeks after transplantation, and engraftment of human cells was detected by staining with antibodies against the human membrane protein SPECTRIN (h-SPECTRIN). H-SPECTRIN staining was strongly positive in fetal human muscle sections ( Figure 5A, left panel) and uniformly Fold-change differences relative to whole human muscle were calculated for each gene and condition and confirm the bipotency of CD34 À CD56 int ITGA7 hi hMFA cells. This assay was replicated in four biologically independent experiments. Statistical significance was evaluated by unpaired, two-tailed t test (*p < 0.05; **p < 0.001; ns, not significant). Data are mean ± SD. See also Figure Figure 5A, right panel). To assess the in vivo myogenic activity of sorted fetal hMFA cell subsets, cells were isolated from 11 individual donors and transplanted into the cardiotoxin preinjured tibialis anterior muscles of NSG recipients. Engraftment by human cells, marked by staining with h-SPECTRIN, was detected in 6 of 28 muscles transplanted with CD34 À CD56 int ITGA7 hi fetal hMFA cells (37,000-100,000 cells injected), 8 of 25 muscles transplanted with CD34 + fetal cells (37,000-100,000 cells injected), three of ten muscles transplanted with CD34 À/low CD56 À ITGA7 low fetal cells (20,000-100,000 cells injected), and three of ten muscles transplanted with CD34 À CD56 hi ITGA7 low fetal cells (7,000-100,000 cells injected) ( Figure 5B, top panels, and Figure 5C). Similar numbers of h-SPECTRIN-positive cells were detected in muscles engrafted with human CD34 À CD56 int ITGA7 hi cells (12 ± 3; mean ± SEM). CD34 + cells (8 ± 3; mean ± SEM), CD34 À/low CD56 À ITGA7 low cells (11 ± 5; mean ± SEM), or CD34 À CD56 hi ITGA7 low cells (5 ± 2; mean ± SEM) ( Figure 5D). We next evaluated myogenic engraftment of fetal hMFA cell subsets in transplanted mouse muscles by costaining for h-SPECTRIN ( Figure 5B, top panels) and DYSTROPHIN ( Figure 5B, middle panels), a membrane protein expressed in both mouse and human muscle fibers. Costaining of h-SPECTRIN and DYSTROPHIN ( Figure 5B, bottom panels, and Figure S5) in four of six muscles engrafted with fetal human CD34 À CD56 int ITGA7 hi cells indicated that the transplanted cells underwent myogenic differentiation and contributed to the formation of mature fibers. A total of 35%-100% of h-SPECTRIN + cells in these muscles coexpressed DYSTROPHIN. In contrast, none (0%) of the h-SPECTRIN+ cells detected in muscles engrafted with fetal CD34 + cells, CD34 À/low CD56 À ITGA7 low cells, or CD34 À CD56 hi ITGA7 low cells were DYSTROPHIN + (Figure 5E). Thus, only fetal CD34 À CD56 int ITGA7 hi hMFA cells are capable of myogenic engraftment in mouse muscle.

The Transcriptional Signatures of Fetal hMFA Cell Subsets Are Consistent with Lineage-Specific Differences in Their Differentiation Capacities
To gain deeper insights into the molecular underpinnings of CD34 À CD56 int ITGA7 hi cells and CD34 + adipogenic precursors within the fetal hMFA cell pool, the transcriptional profile of these functionally distinct populations, as compared to unfractionated fetal hMFA cells, was evaluated. Principal component analysis (PCA; Figure 6A) and hierarchical cluster analysis ( Figure 6B) showed clustering of CD34 À CD56 int ITGA7 hi cells, CD34 + cells, and hMFA cells into three transcriptionally distinct populations. Comparison of CD34 À CD56 int ITGA7 hi cells (12.2% ± 4.3% of live fetal hMFA cells; Table S1) to unfractionated hMFA cells identified 5,686 differentially regulated probesets, and comparison of CD34 + cells (62.7% ± 7.8% of live fetal hMFA cells; Table S1) to unfractionated hMFA cells yielded 1,029 differentially regulated probesets (>1.5-fold difference up or down and p < 0.05). Notably, there was no overlap between these groups of differentially regulated genes. Ingenuity pathway analysis revealed that within the group of genes most highly upregulated in fetal CD34 À CD56 int ITGA7 hi cells as compared to CD34 + cells (>5-fold difference, p < 0.01, total 346 genes; Table S2), the 25 top-scoring functions involved muscle development, differentiation, or function (Table S3). Interestingly, within the group of genes most highly upregulated in CD34 + cells versus CD34 À CD56 int ITGA7 hi cells (>5-fold difference, p < 0.01, total 854 genes; Table S4), the seven top-scoring functions involved solid tumor malignancy (Table S5).
Adult Skeletal Muscle Shows Reduced Content of PAX7-Expressing CD34 À CD56 int ITGA7 hi hMFA Cells with Myogenic and Osteogenic Activity To determine whether the surface marker combination we identified as marking PAX7-expressing osteogenic/ myogenic progenitors in human fetal muscle would similarly mark progenitors in adult tissue, we evaluated differential expression of surface markers ( Figure 7A), PAX7 enrichment ( Figure 7B), myogenic differentiation (Figure 7C), and osteogenic differentiation ( Figure S7) in hMFA cells obtained from discarded human adult muscle.
Decreasing muscle progenitor frequency with age in human muscle is consistent with previously published findings in the mouse (Conboy et al., 2003).

DISCUSSION
Recent advances enabling the prospective isolation of mouse satellite cells have facilitated mechanistic analyses of their myogenic function. For example, the ability to clonally sort satellite cells with high purity made possible the demonstration that these cells can undergo asymmetric division (Kuang et al., 2007;Rocheteau et al., 2012) and repopulate the satellite cell niche in vivo (Cerletti et al., 2008). While findings in mouse models are often extrapolated to human biology, whether mouse and human myogenic cells exhibit fully equivalent properties may still be questioned, particularly given significant phenotypic discrepancies in several mouse models of human muscle disease (Bulfield et al., 1984). All of these issues can be addressed through the establishment of robust methods for direct purification of human muscle progenitors.
Previous work by Pisani et al. demonstrated the utility of the sialomucin CD34 to enrich for myogenic cells within the CD34 À subset of magnetically separated cells in adult muscle (Pisani et al., 2010b), consistent with immunohistochemical studies reporting the absence of CD34 in adult human muscle cells located in the satellite cell position (Lecourt et al., 2010). Pisani et al. also noted mixed myogenic and adipogenic activity within CD34 + adult muscle cells, which showed differential expression of CD56 (Pisani et al., 2010a). Findings from our study confirm that CD34 distinguishes myogenic and nonmyogenic cells within the nonhematopoietic, nonendothelial (CD45 À CD11b À GlyA À CD31 À ) hMFA cell pool in both fetal and adult tissue: CD34 + cells are PAX7-negative, adipogenic cells that do not possess any myogenic activity, whereas within the CD45 À CD11b À GlyA À CD31 À CD34 À/low subset, selection of CD56 int ITGA7 hi cells yields a highly enriched population of PAX7-expressing, robustly myogenic progenitors. Yet, it is important to note that these FACS-based strategies pertain to cells isolated from fresh muscle only. Sorted cells may undergo marked changes in their surface marker profiles during ex vivo culture, and it is unclear if our protocols are applicable to cells that have undergone expansion/differentiation in culture.
Fluorescence-activated cell-sorted fetal human CD34 À CD56 int ITGA7 hi cells engraft in mouse muscle to form new myofibers, albeit at low efficiency ( Figures 5B-5F). Low-level engraftment of human cells into mouse tissue is not unexpected, as similar outcomes have been observed for other human, tissue-specific stem and progenitor cells upon transfer into immune-compromised mice (Doulatov  Racki et al., 2010). Unfortunately, given the relatively sparse presence of human myofibers in this system, we were unable to establish conditions to reliably detect costaining for PAX7 and human species-specific nuclear antigens on engrafted mouse muscle sections. We therefore were unable to determine if fluorescence-activated cell sorted fetal human CD34 À CD56 int ITGA7 hi cells can repopulate the PAX7-expressing satellite cell pool in vivo. We also were unable to test the in vivo engraftment potential of fluorescence-activated cell-sorted progenitors from adult human muscle, given the low yield of cells that could be obtained from adults ( Figure 7E).
Interestingly, enrichment of PAX7 expression and myogenic activity within the CD34 À compartment in human muscle stands in contrast to immunophenotyping studies in mouse muscle, which localize myogenic activity to the CD34 + subset of mouse MFA cells (Beauchamp et al., 2000;Conboy et al., 2010;Montarras et al., 2005;Sacco et al., 2008;Sherwood et al., 2004). Species-specific differences in CD34 expression have also been noted in other somatic stem cell populations, including hematopoietic stem cells (HSCs), which are CD34 + in adult human bone marrow and CD34 À in adult mouse bone marrow (Okuno et al., 2002;Osawa et al., 1996). Such differences appear to arise from the presence of species-specific upstream regulatory elements, which differentially regulate CD34 gene transcription in mouse and human cells (Okuno et al., 2002).
Cell surface markers that have proved useful for the isolation of mouse PAX7 + satellite cells include ITGA7 (Pasut et al., 2012;Sacco et al., 2008), CXCR4 and b1 INTEGRIN (Sherwood et al., 2004), and VCAM1 (Chakkalakal et al., 2012). Our studies indicate that PAX7-expressing cells in fetal and adult human muscle coexpress ITGA7 and CD56, although neither of these markers alone is sufficient to distinguish these cells. This is consistent with ITGA7 and CD56 expression by human myogenic progenitors derived from PAX7-expressing induced pluripotent stem cells (Darabi et al., 2012). Clear expression of CXCR4 was observed in adult and fetal PAX7-expressing cells ( Figure S6A), and b1-INTEGRIN was detected on 90% of hMFA cells (Figure 1B). Additional surface marker analyses, focusing particularly on those previously linked to mouse and/or human myogenic precursors (Cerletti et al., 2012;Darabi et al., 2012;Lecourt et al., 2010;Zheng et al., 2007), revealed increased expression of MCAM (CD146) and CD44 in human fetal CD34 À CD56 int ITGA7 hi cells as compared to CD34 + cells ( Figure S6B). Finally, our microarray analyses suggest that PROMININ1 (CD133) and PDGFRA (CD140a; previously reported to mark adipocyte precursors; Berry and Rodeheffer, 2013) could also be useful in distinguishing human myogenic progenitors, as they are differentially expressed in fetal CD34 À CD56 int ITGA7 hi cells and CD34 + cells ( Figure S6C; Table S6).
MFA cells obtained from mouse and human muscle were previously shown to exhibit osteogenic activity (Glass et al., 2011;Hashimoto et al., 2008;Lecourt et al., 2010;Oishi et al., 2013). In our studies, both fetal and adult PAX7-expressing CD34 À CD56 int ITGA7 hi cells formed ARpositive calcium deposits and expressed osteogenic lineage genes under osteogenic conditions, consistent with prior reports of osteogenic differentiation potential within the pool of muscle cells containing PAX7 + progenitors (Hashimoto et al., 2008;Ozeki et al., 2006). Clonal assays revealed that the osteogenic activity of human fetal CD34 À CD56 int ITGA7 hi hMFA cells is unlikely to be attributable to contamination by other cells, as the majority of clonesorted cells exhibited bipotent myogenic and osteogenic activity. Future studies are needed to investigate adult muscle progenitor bipotency and delineate the events that trigger possible osteogenic differentiation of human CD34 À CD56 int ITGA7 hi cells, their possible contributions to normal bone regeneration, and their relationship to other mesenchymal precursor cells.
In contrast to fetal CD34 À CD56 int ITGA7 hi cells, which exhibit robust myogenic activity and lack adipogenic potential, human fetal CD34 + cells are adipogenic and lack myogenic capacity. While our studies evaluated the ability of these cells to form white adipocytes, a previous report indicates that CD34 + cells in fetal and adult human muscle contain brown adipogenic activity as well (Crisan et al., 2008). Gene expression profiling confirms profound differences in the transcriptional signatures of fetal human CD34 À CD56 int ITGA7 hi and CD34 + hMFA cells, with increased expression of muscle lineage genes in CD34 À CD56 int ITGA7 hi cells and increased expression of adipogenic genes in CD34 + cells. Given their differentiation profile, we speculate that CD34 + hMFA cells may represent the human counterparts of the fibroadipogenic precursor (FAP) population in mouse muscle (Joe et al., 2010;Uezumi et al., 2010), an important subset of nonmyogenic cells that appears to enhance muscle regenerative capacity (Joe et al., 2010).
(C) Myogenic differentiation assays showed that in vitro myogenic activity is highly enriched in adult CD34 À CD56 int ITGA7 hi cells and absent from CD34 + and CD34 À/low CD56 À ITGA7 low cells. (D-F) Numbers of hMFA cells (D) or CD34 À CD56 int ITGA7 hi cells (E) per gram of tissue, and frequency of CD34 À CD56 int ITGA7 hi cells (F), were compared for adult and fetal muscle. Statistical significance was evaluated by unpaired, two-tailed t test. Data are mean ± SD. See also Figure S7.

FACS Isolation of Progenitors from Human Muscle
In summary, we report functionally distinct cell populations within the hMFA cell pool in fetal and adult muscle and provide a specific method for the prospective isolation of purified PAX7 + cells from human muscle. We anticipate that this technology will facilitate novel insights into human muscle homeostasis, aging, and disease. Phenotypically distinct cells with myogenic progenitor function also represent promising targets for muscle regenerative cell therapy and could conceivably be used to treat a variety of diseases, including muscular dystrophy and muscle injuries.

Human Skeletal Muscle Specimens
Human fetal muscle was obtained from 20-to 23-week-gestation fetuses and adult muscle from deceased volunteers or discarded during surgery (Table S1). Use of human tissues was approved by relevant institutional review boards.
Isolation of hMFA Cells hMFA cells were isolated by two-step enzymatic digestion and mechanical dissociation as per previously published protocols (Sherwood et al., 2004).

Antibody Staining and FACS
Primary and secondary antibodies used for FACS are listed in Table  S7. All cell populations were sorted twice to maximize purity.

PAX7 IF and Quantification
hMFA cell subsets were sorted directly into 40 ml of PBS spotted on a glass slide (5 3 10 3 cells per slide) according to protocols adapted from (Ema et al., 2006). Sorted cells were stained with PAX7 (DSHB, 10 mg/ml).

Adipogenic Differentiation Assay
hMFA cell subsets were sorted in 96-well plates, expanded in adipogenic growth medium until confluent (13-14 days), transitioned into adipogenic induction medium for 3 days, placed in differentiation medium for 4 days, and fixed and stained with ORO (Sigma).

Clonal Cell Culture
Fetal hMFA cell subpopulations were sorted at 1 cell per well in 96-well plates in myogenic growth medium. After 9-10 days, the number of wells with visible cell growth was determined. Cells were kept in myogenic growth conditions until confluent and passaged into myogenic or osteogenic differentiation assays.

Transplantation Studies
NSG mice (Jackson Laboratory) were bred and maintained at Joslin Diabetes Center under Institutional Animal Care and Use Committee-approved protocols. The tibialis anterior (TA) muscle of 6-to 8-week-old NSG mice was conditioned 24 hr prior to transplantation by injection of 25 ml (0.03 mg/ml) of Naja mossambica mossambica cardiotoxin (CTX, Sigma). Recipient muscles were harvested 3-8 weeks after transplantation. Engraftment was evaluated by IF staining of 7 mm sections for h-SPECTRIN (human species-specific) and DYSTROPHIN. Tissue was blocked using Papain-digested RAM antibodies supplemented with goat Fc antibodies at 5 mg/ml and 5% fetal bovine serum according to previously published protocols (Ehrhardt et al., 2007).

Microarray Analysis
Total RNA was extracted using TRIzol labeled and hybridized to Affymetrix microarrays (Human Genome U133 Plus 2.0). Raw data were normalized and differentially regulated probesets were identified using GenePattern. Hierarchical clustering was performed in GenePattern (Broad Institute), PCA using 3D-PCA, and pathway analysis using Ingenuity.

PCR
Total RNA was extracted using TRIzol, reverse transcribed using Superscript III First-Strand Synthesis System (Invitrogen). qRT-PCR was performed using an AV7900 PCR system (Applied Biosystems) and TaqMan Gene Expression Assays (Invitrogen).

Statistics
Statistical analysis was performed using two-tailed Student's t test for unpaired data when appropriate. p values are indicated with asterisks (*p < 0.05, **p < 0.001, and ***p < 0.0001) and NS (not significant).
For additional details regarding experimental procedures, please see the Supplemental Experimental Procedures.

ACCESSION NUMBERS
The microarray data reported in this paper were deposited in the NCBI Gene Expression Omnibus under accession number GSE44227.

Supplemental Information includes Supplemental Experimental
Procedures, seven figures, and seven tables and can be found with this article online at http://dx.doi.org/10.1016/j.stemcr. 2013.12.006.