Human fetal skeletal muscle contains a myogenic side population that expresses the melanoma cell-adhesion molecule

Abstract

Muscle side population (SP) cells are rare myogenic progenitors distinct from satellite cells, the known tissue-specific stem cells of skeletal muscle. Studies in mice demonstrated that muscle SP cells give rise to satellite cells in vivo. Given that muscle SP cells are heterogeneous, it has been difficult to prospectively enrich for myogenic progenitors within the SP fraction, particularly from human tissue. Further, conditions that favor the expansion of human muscle SP cells while retaining their myogenic potential have yet to be reported. In this study, human fetal muscle SP and main population (MP) cells were purified based on the expression of melanoma cell adhesion molecule (MCAM), a marker we previously reported to enrich for cells with myogenic potential. To define the relationship between MCAM expression and the degree of myogenic commitment, single cells were analyzed for the expression of myogenic-specific markers. Myogenic factors strongly associated with MCAM expression in single cells, particularly Myf5. Different MCAM+ populations, including SP cells, were expanded and assayed for fusion potential in vitro and engraftment potential in vivo. All MCAM+ subpopulations fused robustly into myotubes in vitro, whereas the MCAM− subpopulations did not. Further, MCAM+ SP cells exhibited the highest fusion potential in vitro and were the only fraction to engraft in vivo, although at low levels, following propagation. Thus, MCAM can be used to prospectively enrich for myogenic muscle SP cells in human fetal muscle. Moreover, we provide evidence that human MCAM+ SP cells have intrinsic myogenic activity that is retained after propagation.

INTRODUCTION

Skeletal muscle possesses an impressive ability to regenerate in response to injury or chronic disease. This remarkable regenerative capacity is attributed to its resident mononuclear myogenic progenitors, referred to as the satellite cells (1–3). Satellite cells are defined by their proximal position to mature myofibers underneath the basal lamina (4) and by expression of Pax7 (5–8). Upon stimulation following skeletal muscle damage, satellite cells undergo cell division to replenish the satellite cell compartment and generate myogenic precursor cells, which in turn undergo multiple rounds of cell division (9,10). These derivatives of satellite cells or myoblasts express Myf5 and/or MyoD and can be expanded in vitro, undergo further differentiation and eventually fuse to each other to form new myofibers or fuse to existing fibers to repair damage (11,12). Myoblasts have been tested for their muscle repair ability in vivo in both mice and humans (13–16), and although safe, direct myoblast injections have proved inefficient (17–22). This inefficiency might be partially due to the sub-optimal conditions used to expand satellite cell-derived myoblasts prior to their injection. Studies demonstrated that satellite cells injected without prior separation from their niche retain a greater reparative capacity, compared with cultured myoblasts (23–25). However, recent studies have optimized culture conditions that mimic more closely the environment of the satellite cell niche and demonstrated effective repair and self-renewal capacity in vivo following expansion under these conditions (26).

Since optimization of cell-based therapy for muscular dystrophy would preferably deliver donor cells through a vascular route, progenitors different from satellite cells have also been isolated and tested in animal models via systemic delivery. These include murine muscle-derived stem cells (27,28), human CD133+ progenitor cells (29), human myo-endothelial cells (30), human pericytes (31) and murine skeletal muscle side population (SkM SP) cells (32–35), which are isolated by FACS on the basis of Hoechst 33342 dye exclusion (36). Murine SkM SP cells demonstrated myogenic potential in vitro when co-cultured with primary myoblasts (33,37,38). Unfortunately, SkM SP cells are heterogeneous and require further fractionation to enrich for myogenic progenitors within the population (39–41). A strategy for the enrichment of myogenic SkM SP cells was developed for adult mouse cells by selecting for the expression of ABCG2, the transporter that marks all interstitial progenitors in murine muscle, including SP cells (42), in conjunction with the cell surface markers syndecan-4 and Sca-1 (43). These ABCG2+ Sca+ syndecan-4+ cells express Pax7 and exhibit myogenic potential in vitro and in vivo (43). Unfortunately, parallel studies for the characterization of muscle SP cells from human muscles have been lagging behind. Thus far, there has been little or no indication that SP cells in human skeletal muscle have intrinsic myogenic potential and can be expanded in vitro. One problem that hindered these studies is that ABCG2 does not identify human fetal SkM SP cells (44). Therefore, alternative strategies that aid in the selection of myogenic cells within the human muscle SP fraction are desired.

Studies in our laboratory have identified melanoma cell adhesion molecule (MCAM) as a cell surface marker expressed by myogenic progenitors in human fetal skeletal muscle (45). MCAM is robustly expressed in proliferating myogenic progenitors in vitro, and its expression is significantly downregulated during myogenic differentiation and fusion (45). During skeletal muscle development, MCAM is expressed in somitic cells that specify the myotome (46). In addition, the conversion of mesoderm-like 10T1/2 cells into committed muscle cells by 5-azacytidine resulted in concomitant expression of MCAM and the myogenic transcription factor, Myf5 (46). However, whether MCAM and Myf5 were co-expressed in the same cell was not determined. Our previous studies in human fetal skeletal muscle showed that MCAM partially co-localizes in cells expressing Pax7 or MyoD (45), but the relationship between MCAM expression and the degree of myogenic commitment of distinct MCAM+ subpopulations (such as SP and non-SP cells) was not addressed. Furthermore, the in vivo engraftment potential of specific MCAM+ subpopulations remains unknown.

Here, we utilize MCAM to fractionate the SP, main population (MP) and Total (SP + MP) populations within human fetal skeletal muscle and assess each fraction's myogenic potential. We clearly demonstrate that MCAM-expressing cells are myogenic, whereas MCAM− cells are not. Specifically, MCAM+ SP and MP cells express myogenic markers, such as Myf5 and Pax7, at the single cell level, whereas MCAM− SP and MP cells do not. We also observe that expanded human MCAM+ SP, MCAM+ MP and Total MCAM+ cells form myotubes in vitro, whereas the respective MCAM− fractions do not. Assessment of the engraftment potential of expanded MCAM+ fractions by intramuscular injection shows that expanded MCAM+ SP cells engraft to a greater extent in vivo than Total MCAM+ or MCAM+ MP cells, although the overall level of engraftment only provides a proof-of-principle finding and is not therapeutically significant. Taken together, these data indicate that MCAM specifies a potent myogenic progenitor population within human fetal skeletal muscle and that these myogenic progenitors, particularly enriched in the MCAM+ SP fraction, can be expanded in vitro and retain limited engraftment potential in vivo.

RESULTS

MCAM identifies myogenic cells in human fetal skeletal muscle

Previously, we have demonstrated that MCAM-expressing cells in human fetal skeletal can co-express myogenic markers such as Pax7 or MyoD (45). To investigate how robustly MCAM expression segregates with myogenicity and to determine whether early myogenic progenitors are contained within a specific subfraction of MCAM-expressing cells, such as muscle SP cells, we isolated four groups of cells on the basis of MCAM expression and Hoechst 33342 dye uptake (Fig. 1A). In 22 human fetal samples analyzed, MCAM+ cells constituted ∼35–80% of the total mononuclear cells, 19–63% of the SP and 40–81% of the MP populations; the percentage of MCAM+ cells did not trend with gestational age (Supplementary Material, Table S1).

Figure 1.

Analysis of myogenic transcripts in single cells from MCAM± SP and MCAM± MP populations from human fetal skeletal muscle. (A) Human fetal SkM cells were sorted on the basis of MCAM expression and Hoechst 33342 dye uptake. Top panels: Cells were immunostained with control mouse IgG (mIgG; left plot) or mouse anti-MCAM antibody (right plot) to identify MCAM+ and MCAM− populations. Middle panels: Cells were also stained with Hoechst 33342 dye in the presence (left plot) or absence (right plot) of reserpine to clearly define the SP and MP populations. Bottom panels: MCAM+ cells in SP (left plot) and MP (right plot) fractions. (B) Examples of myogenic transcripts amplified from single cells derived from MCAM± SP and MCAM± MP fractions. Five examples are shown for each fraction, but the complete data are provided in Supplementary Material, Table S2. Pax7, Myf5, MyoD, MCAM (control for sort purity) and β2M (control for the presence of a cell and RNA integrity) expression were assessed in single cells. NTC, no template control; −RT, no reverse transcriptase control; and +cont, unfractionated human fetal SkM cDNA-positive control. (C) Graphic summary of the results from the single-cell RT–PCR analyses. The number of cells analyzed per each fraction is shown in the legend. By Fisher's exact test, significantly more MCAM+ cells express myogenic markers than MCAM− cells (**P < 0.001; *P < 0.01).

To determine whether the expression of myogenic factors correlates with MCAM expression and whether differences in the expression of myogenic markers exist between SP and MP cells positive for MCAM, RT-PCR analyses were performed at the single cell level. Single cells from four distinct populations (MCAM+ and MCAM− SP, MCAM+ and MCAM− MP) were freshly sorted from three distinct human samples and analyzed for the expression of several genes (Fig. 1B, Supplementary Material, Table S2). As a control for the presence of a cell and RNA integrity, beta 2 microglobulin (β2M) was amplified, whereas MCAM expression was amplified as a control for FACS sort purity. Only single cells that exhibited β2M and MCAM expression when appropriate were included in this analysis. In parallel, the expression of the myogenic markers Pax7, Myf5 and MyoD were also analyzed (Fig. 1B, Supplementary Material, Table S2). Control-positive and -negative reactions were also included in each set of amplifications to ensure that all primers were amplifying properly in the presence of the proper target template and that no reagent contamination had occurred. In the MP populations, more MCAM+ cells expressed Pax7 and Myf5. In single MCAM+ MP cells, 5 out of 77 were Pax7+ (6.5%), whereas no Pax7+ cells were found in the MCAM− MP population (n = 172cells). Approximately 40% of MCAM+ MP cells were Myf5+ (31 out of 77), whereas only 1 Myf5+ out of 172 cells analyzed was found in the MCAM− MP cells (0.6%) (Fig. 1C). Of the SP populations, the MCAM+ SP fraction exhibited the highest percentage of Pax7+ single cells (8 out of 50 cells = 16%), whereas in MCAM− SP cells only 2 Pax7+ cells were detected out of 143 cells analyzed (2 out of 143 = 1.4%). Analysis of Myf5 expression revealed that 19 out of 50 cells in the MCAM+ SP fraction expressed Myf5 (38%), which was in significant contrast to the low percentage seen in the MCAM− SP population (n = 2 out of 143 = 1.4%). Furthermore, five out of the eight MCAM+ SP cells expressing Pax7 were also co-expressing Myf5, whereas four out of five single cells in MCAM+ MP fraction expressing Pax7 also co-expressed Myf5. None of the four populations exhibited a high percentage of MyoD-expressing cells (0% for MCAM+ SP, 2.8% for MCAM− SP, 5.2% for MCAM+ MP and 1.7% for MCAM− MP; Fig. 1D). These results indicate that although the expression of myogenic markers in single cells obtained from the same cell fraction is heterogeneous, all MCAM+ fractions are significantly more enriched in cells expressing myogenic factors than the MCAM− fractions.

Muscle SP cells can be expanded in vitro

Robust in vitro expansion of human muscle SP cells has not yet been optimized, hindering the investigation of the potential of this cell fraction in cell-based therapy studies. Thus, in an effort to find culture conditions that would support human muscle SP cell expansion, we isolated muscle SP cells by FACS and seeded equal numbers of cells (75 cells/well) in 12 different media (Fig. 2). Examples of seeded wells on the day of sorting are shown in Figure 2A. Analysis of approximately 200 wells from two separate experiments indicated that an average of 59 ± 10.5 cells were seeded per well (Fig. 2B) in each experiment.

Figure 2.

Optimization of conditions that favor human SP cells expansion while preventing differentiation in vitro. (A) Examples of three wells from a plate that was fixed on the day of seeding and stained with PI to determine the average number of cells deposited by FACS per well across the plate. (B) Quantification of the variation in number of cells seeded by FACS. Blue points indicate individual well cell counts; the red point indicates the average cell count of approximately 200 wells from two separate experiments. The calculated average of cells seeded/well was 59 ± 10.5. (C) Human fetal SkM SP cells were seeded and cultured in 12 different media for 10 days and assessed for the extent of proliferation and differentiation relative to the 20% FBS/DMEM condition. Medium conditions: (1) 20% fetal bovine serum (FBS)/DMEM, (2) Invitrogen StemPro^® MSC SFM, (3) 20% FBS/F10, (4) 20% FBS/F10:DMEM (1:1), (5) 20% FBS/F12, (6) 20% FBS/F12:DMEM (1:1), (7) Invitrogen MesenPRO RS™, (8) Invitrogen Knockout™ SR, (9) Invitrogen StemPro^® hESC SFM, (10) 20% FBS/RPMI, (11) 10% FBS/OptiMEM and (12) 20% FBS/αMEM. Scale bar: 400 µm. After 10 days of culture in vitro, qualitative analysis showed that differentiation of SP cells occurred in several of the media conditions. Notably, StemPro^® MSC SFM (condition 2) was the medium that supported the propagation of human fetal skeletal muscle SP cells with minimal differentiation as evidenced by the lack of fused myotubes (green: desmin; blue: DAPI). (D) Quantitative analysis of cell growth indicated that cells proliferated 3.2 to 11.5 times better in several media compared with 20% FBS/DMEM (conditions 2–7, 11–12), whereas no significant growth was apparent in three of the medium conditions tested (conditions 8–10). The data represent the scaled average results from three different experiments (**P < 0.01; *P < 0.025).

Muscle SP cells were cultured for 10 days in vitro and assessed for proliferation under each medium condition, accompanied by an assessment for minimal differentiation (i.e. low or no myotube formation) (Fig. 2C and D). Propagation of unfractionated human fetal SkM SP cells was significantly higher in several media relative to 20% FBS/DMEM, which was used as a standard (Fig. 2C), while three media did not support the proliferation of SP cells to any significant extent in vitro (Fig. 2C and D). In several media conditions, human muscle SP cells had also differentiated significantly and formed fused myotubes (Fig. 2C). Thus, under the tested conditions, StemPro^® MSC SFM was the medium that seemed to best support the propagation of human fetal SkM SP cells with minimal differentiation, as evidenced by the lack of fused myotubes (Fig. 2C, condition 2). StemPro^® MSC SFM is a serum-free medium and is thus completely defined, which should allow for better consistency from culture to culture, and removes the effect of different serum lot compositions on resulting cultures.

Expression of myogenic markers is retained in propagated MCAM+ populations

We next tested whether StemPro^® MSC SFM could support the expansion of MCAM+ and MCAM− cell fractions in vitro and whether the myogenic potential of propagated cells might be retained and/or acquired. To do this, six cell fractions (MCAM+ SP, MCAM− SP, MCAM+ MP, MCAM− MP, Total MCAM+ and Total MCAM−) were purified by FACS, plated at equal number and propagated in parallel (Fig. 3A). After 10 days, all cell populations had propagated and adhered to the plate surface, displaying an elongated morphology (Fig. 3A). Sufficient cell numbers of each population, including the SP populations, could easily be obtained for subsequent RT-PCR and western blot analyses.

Figure 3.

Expression of myogenic markers following parallel expansion of MCAM± SP, MP and Total cells in vitro using Stempro^® MSC SFM medium. (A) MCAM± SP, MP and Total cells were sorted in parallel and propagated in vitro using Stempro^® MSC SFM. Images of cultured cells at Day 1 and Day 10 after cell sorting are shown. All six fractions were capable of expansion from 7500–20 000 freshly isolated cells to millions of expanded cells within 10 days of culture. Scale bar: 200 µm. (B) RNA was harvested from each expanded cell population to assess expression of myogenic transcription factors. Real-time quantitative RT–PCR analysis indicated that the MCAM+ populations expressed Pax7, Myf5, MyoD and Myog, whereas the MCAM− populations did not. Additionally, the cultured MCAM+ SP population expressed Pax7, Myf5 and MyoD to a significantly greater extent than the cultured MCAM+ MP population. (C) Western blot analysis of the cultured populations confirmed the data obtained by real-time quantitative RT–PCR (***P < 0.001; **P < 0.01; *P < 0.05).

To determine whether these cell populations retained (in the case of the MCAM+ populations) or acquired (in the case of the MCAM− populations) myogenic potential after expansion in vitro, real-time quantitative RT–PCR analyses were performed to determine the expression of the myogenic factors Pax7, Myf5, MyoD and myogenin, as well as β2M, which was used as an internal gene control. In agreement with the single-cell data collected from freshly isolated cells, all cultured MCAM+ populations expressed myogenic markers, whereas the expanded MCAM− populations did not (Fig. 3B). Notably, the cultured MCAM+ SP population expressed higher mRNA levels of Pax7, Myf5 and MyoD than the cultured MCAM+ MP population. These results were confirmed by western blot analysis on protein lysates obtained from all six cultured populations (Fig. 3C). Thus, cultured MCAM+ cells retained the expression of myogenic proteins, whereas the MCAM− populations did not gain the expression of myogenic markers upon expansion.

MCAM+ cells undergo myogenic differentiation in vitro

To evaluate the ability of expanded MCAM+ cell populations, including muscle SP cells, to fuse in vitro, equal numbers (7500cells/well in 96-well plates) of expanded MCAM± SP, MP and Total (SP + MP) cells were placed in differentiation medium and assessed for their ability to form myotubes. A duplicate plate was fixed and stained on Day 0 to confirm equal plating density (Fig. 4A). After 3–4 days post-induction, all three MCAM+ populations were capable of myogenic differentiation in vitro, whereas the MCAM− populations were not (Fig. 4A, Day 3). Furthermore, cultured MCAM+ SP cells fused significantly better than the cultured MCAM+ MP or Total MCAM+ cell populations (76.2 ± 1.4, 41.4 ± 12.1 and 50.1 ± 1.2%, respectively; Fig. 4A and C). These results were further confirmed by assessment of myogenin expression, which is found in differentiating myocytes and myotubes (47) (Fig. 4B and D). Specifically, 53.2 ± 2.9% of differentiated MCAM+ SP nuclei were myogenin+, whereas 43.1 ± 14.8 and 46.4 ± 4.2% of MCAM+ MP and Total MCAM+ nuclei, respectively, were myogenin+ (Fig. 4D). In contrast, all three MCAM− populations exhibited <7% myogenin+ nuclei (6.6 ± 4.9% for MCAM− SP, 0.2 ± 0.09% for MCAM− MP and 0.7 ± 0.7% for Total MCAM−). These results indicate that expanded MCAM+ cell fractions can undergo myogenic differentiation in vitro, whereas MCAM− cells cannot. Additionally, MCAM+ SP cells appear to possess a higher fusion potential than MCAM+ MP or Total MCAM+ cells following expansion.

Figure 4.

Expanded MCAM+ cell fractions exhibit robust myogenic differentiation potential in vitro. (A) Cultured MCAM± SP, MP and Total cell populations were plated at 7500 cells/well in 96-well plates and induced to differentiate. A duplicate plate was fixed and stained on Day 0 to confirm equal plating density. The presence of mononuclear desmin+ cells was assessed in the populations prior to induction of differentiation. After 3–4 days of differentiation in vitro, cells were again stained for desmin (green) to visualize mononuclear myogenic cells and fused myotubes. Cell nuclei were visualized with DAPI (blue). Myotubes are clearly apparent in all MCAM+ populations, whereas few or no desmin+ cells or myotubes were observed in the MCAM− fractions. Scale bar: 100 µm. (B) Duplicate plates were also fixed and stained with an antibody against myogenin (green). Nuclei were counterstained with DAPI (blue). Again, the MCAM+ populations exhibited a high percentage of myogenin+ nuclei, whereas the MCAM− populations did not. (C) Fusion indices were calculated 3–4 days post-induction for all six cultured populations as the number of fused nuclei (nuclei contained in fused desmin+ myotubes) divided by the total number of nuclei in the field. All MCAM+ populations fused to a much greater extent than the MCAM− populations (for example, the MCAM+ SP cells fused 76.2 ± 1.4%, whereas the MCAM− SP cells fused only 1.9 ± 2.5%). MCAM+ SP cells exhibited a greater fusion potential than the MCAM+ MP or Total MCAM+ cells. (D) Percentage of myogenin-expressing cells after 3–4 days of differentiation in vitro. All MCAM+ populations exhibited a significantly higher percentage of myogenin+ nuclei compared with their MCAM− counterparts. Graphs represent the combined results of four experiments; ***P < 10⁻⁴; **P < 10⁻³; *P < 5 × 10⁻².

Since these studies indicated that MCAM expression on freshly isolated cells is a strong predictor of myogenic activity, we sought to determine whether expanded cells would also maintain this positive correlation. To do this, we assayed all expanded cell fractions for the expression of MCAM by FACS and western blot analyses. Interestingly, these assessments indicated robust MCAM levels in all six cultured populations, regardless of whether the populations were originally sorted as MCAM+ or MCAM− (Supplementary Material, Fig. S1). Given that the cultured MCAM− populations did not acquire myogenic potential, these results indicate that MCAM is highly specific for myogenic progenitor cells only in primary tissue and is not indicative of myogenicity upon culture in vitro.

Expanded MCAM+ SP cells are capable of engraftment in vivo

Given that expanded MCAM+ cell fractions exhibited myogenic differentiation potential in vitro, we next assessed whether they could also engraft in vivo. In total, 100 000 expanded MCAM+ SP, MCAM+ MP or Total MCAM+ cells were injected intramuscularly into the TA of NOD/Rag1^null mdx^5CV mice in a solution of BaCl₂ in order to damage host muscle and induce myogenic regeneration. The NOD/Rag1^null mdx^5CV mouse is a model of Duchenne muscular dystrophy and does not express dystrophin; furthermore, the mouse is immunodeficient as it lacks T, B and NK cells and therefore it is a suitable host to receive human donor cells (48). Four to five mice were injected for each cell fraction (cultured MCAM+ SP, MCAM+ MP and Total MCAM+). Cell fractions from two unrelated individuals were tested in independent experiments.

One month after injection, muscles were harvested and assessed for engraftment by immunostaining tissue sections with antibodies specific to dystrophin and human spectrin (Fig. 5). Successful engraftment was defined as a myofiber that co-expressed dystrophin and human spectrin. Dystrophin and human spectrin double-positive myofibers were detected in mice injected with cultured MCAM+ SP cells (28 ± 13 myofibers; n = 4 mice), whereas significantly fewer were detected in Total MCAM+ cell-injected mice (3 ± 5; n = 4 mice) and no double-positive myofibers were observed in MCAM+ MP-injected mice (0; n = 5 mice) (Table 1). Overall, despite the substantial in vitro myogenic activity exhibited by the expanded MCAM+ fractions, the engraftment yield in vivo was disappointing and too low to be considered effective. Nevertheless, these studies support the proof of principle that human MCAM+ muscle SP cells can be expanded in vitro and are capable of engraftment in vivo at yields higher than MCAM+ MP and Total MCAM+ cells tested in parallel.

Figure 5.

Transplantation of expanded MCAM+ SP, MP and Total cells into dystrophic muscle in vivo. Expanded MCAM+ SP, MP and Total cells (100 000 cells) were injected in a solution of 1.2% BaCl₂ in PBS and fluorescent beads (green; asterisk) into the TA of mdx^5cv NOD/Rag1^null mice. After 1 month, muscles were harvested and analyzed for expression of dystrophin (red) and human-specific spectrin (green). Cell nuclei were stained with DAPI (blue). Myofibers that were double-positive for human spectrin and dystrophin were detected in mice injected with cultured MCAM+ SP cells (arrows). Human spectrin+/Dys+ myofibers were not or were rarely observed in mice injected with cultured MCAM+ MP and Total MCAM+ cells, respectively.

Table 1.

Summary of myofibers expressing dystrophin and human-specific spectrin following IM injection

Cell population	Dys+ fibers	Spec+ fibers	Dys+/Spec+ fibers
MCAM+ SP
No. 1	54 ± 6	41 ± 5	34 ± 1
No. 2	49 ± 6	54 ± 24	35 ± 9
No. 3	86 ± 19	103 ± 17	41 ± 5
No. 4	24 ± 09	18 ± 6	10 ± 2
MCAM+ MP
No. 1	8 ± 1	7 ± 2	0
No. 2	1 ± 1	3 ± 2	0
No. 3	1 ± 1	2 ± 1	0
No. 4	5 ± 3	0 ± 1	0
No. 5	6 ± 1	5 ± 4	0
Total MCAM+
No. 1	0	1 ± 1	0
No. 2	2 ± 1	3 ± 2	0 ± 1
No. 3	31 ± 6	35 ± 11	12 ± 5
No. 4	4 ± 2	3 ± 2	1 ± 1

Summary of analyses of mice injected with cultured MCAM+ SP, MCAM+ MP or Total MCAM+ cells. The entire muscle was sectioned with 1 in 15 sections collected per slide and 2–3 duplicate slides per muscle. For analysis, the average number and standard deviation of positive myofibers were assessed in two to six sections per mouse. The number of dystrophin+ (Dys+), human spectrin+ (Spec+) and Dys+/human Spec+ double-positive myofibers in each section was quantified. Mice injected with cultured MCAM+ SP cells exhibited a larger number of engrafted double-positive myofibers than mice injected with cultured MCAM+ MP or Total MCAM+ cells.

DISCUSSION

The presence of progenitor cells within adult tissues is crucial for postnatal growth and regeneration, particularly in a highly dynamic tissue such as skeletal muscle. Satellite cells are the cell type responsible for efficient repair of damaged muscle throughout postnatal life (1–3). Other progenitors with myogenic potential have been identified (49), but their role in muscle maintenance and their potential for cell therapy studies have been debated (50–52). In particular, mouse muscle SP cells first gained interest for their ability to fuse and express dystrophin, following intravenous injection in mdx mice (32,33). Studies demonstrated that muscle SP cells are present in normal proportion in Pax7 knockout mice, which exhibit severe postnatal muscle growth and lack satellite cells (5). In addition, murine muscle SP cells have been shown to generate satellite cells in vivo and express myogenic-specific factors (37,43). Altogether, these studies supported the idea that muscle SP cells could be progenitors that contribute to muscle during development, although they are inactive or quiescent in postnatal muscle. Studies on muscle SP cells have been hindered by their low frequency, heterogeneity and the inability to efficiently expand these cells in vitro, particularly from human skeletal muscle. Based on our previous observations that MCAM is expressed in myogenic progenitors within human fetal muscle (45), we hypothesized that muscle SP cells with myogenic potential might be enriched via positive selection for MCAM and Hoechst dye exclusion stain. Our analysis of single cells freshly isolated from MCAM± SP and MP fractions demonstrates that MCAM+ SP and MCAM+ MP cells are more likely to express myogenic-specific markers than the respective MCAM− populations.

The myogenic potential of MCAM+ cells from SP and MP fractions was confirmed following cell propagation in vitro. Our culture conditions utilized a serum-free medium and did not necessitate the use of feeder cells, both of which are desirable features for large-scale expansion of cells to be used for cell therapy applications. Despite the enrichment for myogenic cells achieved via MCAM selection in both SP and MP populations, which was confirmed by the in vitro fusion studies, the purified cell fraction is still heterogeneous. For example, the expression of Pax7, Myf5 or MyoD was detected only in ∼50% of single freshly isolated cells, although the remaining were indeed positive for MCAM, but negative for the expression of myogenic markers. Thus, identification of additional markers is necessary to further enrich prospectively for myogenic human muscle SP cells. Previous studies in adult murine SkM SP cells did not detect Pax7+ cells (37,38) and only a more recent enrichment based on co-expression of ABCG2, syndecan-4 and Sca-1 demonstrated strong myogenic potential of this cell fraction in vitro and in vivo (43). Unfortunately, this isolation strategy might not be adapted to isolate human fetal SkM SP cells, given that ABCG2 was not detected (44). Differences in marker consistency between species are not uncommon and it has been shown for several murine satellite cell markers, including CD34 and M-cadherin (49).

Among the myogenic markers analyzed, Myf5 expression has the highest association with MCAM expression in both SP and MP single cells. Myf5 clearly plays important roles in developing myogenic progenitors (53–55) and its expression is upregulated when quiescent satellite cells become activated (10). In addition to myogenic cells, transient expression of Myf5 has also been reported in non-muscle mesoderm (56). Thus, Myf5 expression in human fetal muscle SP and MP cells could result from a myogenic or non-myogenic developmental origin of these cells. Studies in avian somites have reported the expression of MCAM in myogenic progenitors (46). In addition, specification of the mesenchymal cell line 10T1/2 cells toward the myogenic lineage is characterized by upregulation of MCAM expression, which is also accompanied by an increase in Myf5 expression (46). Collectively, these previous studies point to the possibility that the expression of MCAM and Myf5 occur at similar times, but whether they are co-expressed in the same cells was not addressed. Our single-cell analysis demonstrates that MCAM and Myf5 are indeed co-expressed in human fetal SkM SP and MP cells, and that these cells exhibit strong myogenic potential in vitro compared with cells that do not express MCAM. Whether MCAM expression is directly regulated by Myf5 or, conversely, whether MCAM triggers a signaling cascade that induces downstream expression of Myf5 is not clear from the current data. In metastatic melanoma, MCAM has been shown to positively regulate the expression of Id1 via binding of ATF-3 to the Id-1 promoter (57). In myogenic cells, Id-1 has been shown to suppress MyoD expression (58,59). Thus, although not yet proven, it is possible that regulation of Myf5 expression can be mediated by MCAM and Id-1. This hypothesis could partially explain why ∼50% of single freshly isolated MCAM+ SP cells do not co-express other myogenic factors.

The current study demonstrates that MCAM expression strongly correlates with myogenic activity in human fetal muscle. Cells selected for MCAM expression within different fractions (i.e. SP and MP cells) show a high degree of myogenic activity in vitro following expansion, with the expression of multiple myogenic-specific markers, extensive fusion and formation of mature myotubes. Interestingly, the strong correlation between MCAM expression and myogenic activity only applies to freshly isolated cells. In fact, cells that were originally sorted as MCAM− cells acquire MCAM expression following in vitro expansion. Despite gaining MCAM expression, cells originally sorted as MCAM− do not gain myogenic potential. Thus, myogenic enrichment based on MCAM expression must be performed on non-cultured cells only. Under these conditions, MCAM is a prospective robust marker for myogenic activity, as it was confirmed for MCAM+ SP, MCAM+ MP cells and Total MCAM+ fractions.

Muscle SP cells, particularly from humans, have been infrequently studied, partially because of the scarce number of these cells in primary tissue and the inability to expand them in vitro. Here, we report conditions that favor human fetal muscle SP cell expansion in a serum-free medium. These conditions do not require co-culture with myoblasts as had been described earlier for adult murine muscle SP cells (33,37,38). Although the ability of in vitro expansion of human muscle SP cells is viewed overall as a positive finding and expanded cells demonstrated high potential for myogenic differentiation in vitro, the engraftment yield in vivo was very low. Other MCAM+ cell fractions expanded in parallel also exhibited strong myogenic activity in vitro, but did not engraft at all (MCAM+ MP) in vivo. Thus, additional optimization conditions are needed to improve the efficacy of expanded SP cells in vivo and bring them into consideration as a valuable source for cell-based therapy. Despite the low overall engraftment efficacy, the fact that only expanded muscle SP cells or Total MCAM+ cells (which contain both SP and MP MCAM+ cells) could engraft points to the possibility that the long-term myogenic activity linked to MCAM in human fetal muscle is enriched in muscle SP cells. Thus, as suggested for other tissues (36,60), human MCAM+ muscle SP cells are likely to contain more primitive myogenic progenitors than the MCAM+ MP population. Given the results of the present study, MCAM appears to be a good candidate for the isolation of prospective myogenic progenitors from human fetal SkM SP cells. Additional studies are required to demonstrate that the expansion of these cells can be achieved together with high engraftment potential in vivo.

MATERIALS AND METHODS

Tissue preparation

Mononuclear cells were isolated from de-identified, discarded human fetal muscle samples of gestational age 15–20 weeks. These samples were collected under a protocol approved by the Committee of Clinical Investigation at Brigham and Women's Hospital and Boston Children's Hospital. Limb skeletal muscle tissue was removed of skin and bone prior to being minced into fine pieces and enzymatically dissociated with 0.5 mg/ml of collagenase (Worthington BioChemicals) and 0.6U/ml dispase II (Roche Applied Science) for 45–60min at 37°C as previously described (61,62). Cell suspensions were frozen and stored at −150°C for later use.

Flow cytometry

Human fetal muscle cells were thawed 8–24h prior to analysis/cell sorting. Both floating and adherent cells were collected for FACS analysis and sorting. Cells were washed once with 1× Hank's balanced salt solution (HBSS) and adherent cells were lifted using cell dissociation buffer (Invitrogen GIBCO). Samples were filtered and resuspended at 10⁶ cells/ml in warm 0.5% bovine serum albumin (BSA)/HBSS and then incubated for 1h and 15min at 37°C with 3.5 µg/ml Hoechst 33342 (Sigma) in the presence or absence of 5 µm reserpine (Sigma). For MCAM staining, samples were then washed and resuspended in ice-cold 0.5% BSA/HBSS at a concentration of 10⁶ cells/ml and then incubated on ice for 30min with unconjugated MCAM antibody (1:100; Millipore). After washing with 0.5% BSA/HBSS, samples were incubated with AF488-conjugated goat anti-mouse antibody (1:100; Invitrogen) for 30min on ice. Finally, samples were washed, resuspended in 2 µg/ml propidium iodide (PI)/0.5% BSA/HBSS and filtered through a 40 µm filter prior to FACS analysis/sorting.

Flow cytometric analysis and cell sorting were performed at the Dana Farber Cancer Institute/Hematologic Neoplasia Flow Cytometry Core Facility, using a BD FACSAria II SORP flow cytometer equipped with 355 nm (20 mW UV) and 488 nm (100 mW) lasers, a temperature-controlled collection chamber and a plate collection attachment option. The following filters were utilized for Hoechst visualization: 450 ± 25 nm band pass (Hoechst Blue) and 660 ± 20 nm band pass with 635 nm long-pass dichroic mirror (Hoechst Red). Live cells were first selected using side scatter (SSC) versus forward scatter gating, followed by gating on the PI− population. Live cells were then viewed on a Hoechst Blue versus Hoechst Red plot for SP gating. The reserpine sample, in which the SP profile is absent, was used to accurately set the SP gate. For MCAM selection, gating of live cells was performed on an AF488 versus SSC plot.

Single-cell RT–PCR analysis on freshly isolated cells

Single cells were directly sorted by FACS into wells of a 384-well small-volume plate (Greiner) containing 5 µl of lysis buffer solution [0.15% IGEPAL (Sigma), Protector RNase Inhibitor (Roche), 2× RT-PCR Buffer (Superscript III Platinum One-Step RT-PCR System w/Platinum Taq; Invitrogen)]; the plate was kept at 4°C at all times. After sorting, an additional 4 µl of lysis buffer solution was added to each well containing a single cell. The plate was then sealed with sealing film (Applied Biosystems) and shock-frozen on dry ice. Samples were stored at −80°C for subsequent analysis.

For nested, multiplexed RT–PCR, cell lysates were thawed on ice and transferred to 96-well PCR plates (VWR) with the addition of gene-specific external primer pairs (Supplementary Material, Table S2), RNase inhibitor and RT–PCR reaction mix as specified by the manufacturer. Samples were reverse-transcribed and subsequently amplified in an initial round of PCR amplification. Expected PCR product sizes ranged from 250 to 300 bp. PCR product from this reaction was then diluted 1:2 with water, and 1 µl of diluted product was subjected to a second round of PCR amplification using nested, gene-specific internal primer pairs (Supplementary Material, Table S1) and Taq-PRO Red Complete reaction mix (Denville) according to the manufacturer's instructions. Expected second round PCR product sizes ranged from 120 to 170 bp. PCR products were visualized using gel electrophoresis on a 2% agarose gel.

Media testing assay for expansion of human muscle SP cells

Human fetal skeletal muscle SP cells were directly sorted into 384-well Small Volume™ µclear^® plates (Greiner BioOne); triplicate plates were seeded with 75 cells/well. Plates were pre-coated with 0.15% gelatin for 1h at 37°C, washed once with 1× HBSS and loaded with 5 µl/well medium. Sixteen wells on each plate were seeded per medium condition and cells were allowed to settle at 37°C for 1–3h before an additional 15 µl of medium was gently added to each well. One plate was immediately fixed as described below and stained with 2 µg/ml PI for analysis of seeding variability between wells. For the two remaining plates, cells were propagated in vitro for 10 days, with media replenished every other day as follows: 10 µl of medium was gently removed and replaced with 15 µl of fresh medium. Cells were maintained in a humidified 5% CO₂ incubator at 37°C.

After 10 days of culture in vitro, wells were fixed as follows: (i) medium from each well was removed such that ∼5 µl of medium remained per well; (ii) 20 µl of 5% paraformaldehyde/1× phosphate buffered saline (PBS) was gently added to each well; (iii) cells were fixed for 20min at room temperature; and (iv) cells were washed three times gently with 1× PBS. Cells were then permeabilized with 0.5% Triton X-100/PBS for 3min, and subsequently blocked with 10% FBS/0.1% Triton X-100/PBS (blocking solution) for 30min at room temperature. For primary antibody staining, cells were incubated overnight at 4°C with a mouse anti-desmin (1:100; Dako D33) antibody diluted in blocking solution. Cells were then washed three times with PBS and incubated for 1h at room temperature with DyLight488-conjugated donkey anti-mouse antibody (1:1000; Jackson ImmunoResearch) diluted in blocking solution. Cells were washed three times again in PBS before storage at 4°C in DAPI/PBS.

Images of each plate (before and after in vitro culture) were obtained at 4× magnification using a Nikon Eclipse TS100 microscope fitted with a Spot RT3 camera and appropriate filter sets (Day 0 plate: Texas Red/PI; Day 10 plates: ultraviolet/DAPI, and FITC/GFP). Images were analyzed using ImageJ (NIH free software) with the Cell Counter plugin. Eight of the 16 initially seeded wells were assessed for each medium condition, and three experiments were performed on different human fetal skeletal muscle cell isolations.

Data analysis

Overall growth efficiency was corrected for by summing the number of cells in the eight wells for each medium condition in a given experiment and then calculating a global rescaling factor for each experiment by fitting the best α to the following: log(N_j) = log(N_i) + log(α_j), where N is the total cell count for a given medium condition in a given experiment, α is the global rescaling factor, j is the experiment being normalized and i is the ‘standard’ experiment to which the other two experiments are being normalized. The global rescaling factor was then utilized to normalize the total cell count for each medium condition for the given experiment. Second, the average fold change in growth compared with the 20% FBS/DMEM condition was calculated by taking the logarithm of the total normalized cell count for each medium condition (log₂ N_normalized) in a given experiment and then averaging the log₂ N_normalized values for each medium condition from all experiments. The standard deviation of the fold change in growth was also calculated from the log₂ N_normalized values. Finally, statistical significance in growth rate in a given medium condition when compared with the 20% FBS/DMEM condition was calculated using Welch's t-test on the log₂ N_normalized values.

Cell expansion

Sorted MCAM± SP, MP and Total cells were plated at 7500–20 000 cells per well in 48-well culture plates coated with 0.15% gelatin. Cells were grown in 1% penicillin–streptomycin–glutamine (PSG; Invitrogen Gibco)/StemPro^® MSC SFM (Invitrogen). Cells were passaged at 60–70% confluency onto successively larger plates and maintained in vitro for ∼2 weeks. After propagation, samples were harvested for RNA and protein analyses (real-time quantitative RT–PCR and western blot, respectively), used for in vitro fusion assays or intramuscular injections or frozen at −150°C for future use.

Real-time quantitative RT–PCR analysis on expanded cells

Total RNA from cultures of sorted human fetal skeletal muscle cells was obtained using the RNeasy Mini Kit (Qiagen) as described by the manufacturer. After RNA quantification, 5 µg of total RNA was reverse-transcribed using oligo(dT) primers and the Superscript III First-Strand Synthesis System (Invitrogen), according to the manufacturer's instructions. Resulting cDNA product was diluted 1:4 with water and then subjected to real-time qPCR analysis using gene-specific primers (Supplementary Material, Table S1; internal primer pairs only), SYBR green PCR master mix (Applied Biosystems) as described by the manufacturer and a 7900HT fast real-time PCR system (Applied Biosystems). Expected product sizes ranged between 120 and 170 bp. β2M was utilized as an internal control for each sample. Data were analyzed using the 2^−ΔΔC_T method as previously described (63). dC_T values for three independent experiments were calculated for each myogenic gene, using β2M as an internal gene control, and then averaged prior to calculation of average 2^−ΔΔC_T values. All 2^−ΔΔC_T values were calculated using the MCAM− MP population as the normalization control. Statistical significance of average gene expression was calculated using Welch's t-test.

In vitro fusion assays

Samples were plated at 7500 cells per well in 96-well culture dishes as described above and maintained in StemPro^® medium overnight. Fusion was induced using differentiation medium (2% horse serum/1% PSG/low-glucose DMEM), which was subsequently changed every 24h. Time points were taken at 0, 1, 2, 3 and 4 days post-induction and processed for immunocytochemistry as below.

Images were taken using a 20× objective on a Nikon Eclipse TS100 microscope fitted with a Spot RT3 camera. Fusion indices were calculated as the number of nuclei in fused myotubes over the total number of nuclei in a given field. The percentage of myogenin+ nuclei was calculated as the number of myogenin+ nuclei over the total number of nuclei in a given field. For each experiment, four fields from two wells for each sample were assessed, resulting in 1500–2000 total nuclei counted per sample. Standard deviation was calculated for the two wells for each sample in a given experiment. The experiment was independently performed four times, using different human fetal samples.

To combine and compare the fusion index results of the individual experiments, the fusion indices across the different experiments were first determined to be proportional to each other, using the Pearson correlation method (r = 0.96, P ≤ 0.002). Since they were found to be proportional, fusion indices from one experiment can be rescaled to the levels of another experiment (called ‘standard experiment’), using a proportionality constant α. After rescaling, average fusion index results for each cell population as well as the average standard deviation were calculated using standard propagation of error calculations.

To combine and compare the percentage of myogenin+ nuclei of the individual experiments, myogenin expression was determined to depend on the fusion index in a Michaelis–Menten way, with the dissociation constant K₀ being experiment-independent. Thus, myogenin expression results, M, were rescaled using the Michaelis–Menten relation between the myogenin expression and the fusion index: M_scaled = M × (α_F /(α_F + K₀))/(F/(F + K₀)), where F is the fusion index of a given experiment and α is the proportionality constant between the fusion indices of the given experiment and the ‘standard experiment’.

P-values for pairwise comparisons between each of the cell populations were calculated using Welch's t-test method and tested for multiple hypothesis false discovery, using the Benjamini–Hochberg method (B-H). For the B-H method: (i) uncorrected P-values were rank-ordered from smallest to largest values, (ii) P × (i/N) was calculated for each comparison, where N is the number of comparisons (15 in this case), i is the rank order and P is the significance level (0.05 in this case), and (iii) P × (i/N) < 0.05 was tested for each comparison. Eleven of 15 comparisons for the fusion indices and 9 of the 15 comparisons for the percentage of myogenin+ nuclei passed the B-H method test. All P-values were thus corrected by multiplying N/i for each comparison.

Immunocytochemistry of cultured cell populations

Slides or plates were washed once in PBS and fixed for 20min in 4% paraformaldehyde in PBS at room temperature. Cells were then permeabilized with 0.5% Triton X-100/PBS for 3min and subsequently blocked with 10% FBS/0.1% Triton X-100/PBS for 30min at room temperature. For primary antibody staining, cells were incubated overnight at 4°C with anti-desmin (1:100; Dako) or anti-myogenin (1:50; Dako) antibodies diluted in blocking solution. Cells were then washed three times with PBS and incubated for 1h at room temperature with DyLight488-conjugated donkey anti-mouse antibody (1:1000; Jackson ImmunoResearch) diluted in blocking solution. Cells were washed again in PBS before storing samples at 4°C in DAPI/PBS.

IM injection of mdx^5cv NOD/Rag1-null mice

MCAM+ SP, MCAM+ MP and Total MCAM+ cells were isolated by FACS, cultured for 2–3 weeks as described above and then frozen and stored at −150°C until 2–5 days prior to injection. At this time, cells were thawed onto 0.15% gelatin-coated plates in 1% PSG (Invitrogen Gibco)/StemPro^® MSC SFM (Invitrogen) and allowed to recover. On the day of injection, cells were gently removed from the plate using cell dissociation buffer (Invitrogen GIBCO), counted and resuspended in 1.2% BaCl₂ in sterile PBS, to stimulate muscle fiber necrosis and subsequent regeneration at the injection site. Yellow-green fluorescent beads diluted 1:10 000 (v/v; Molecular Probes Invitrogen) were also added to the solution to locate the site of injection in subsequent histological analyses. In total, 100 000 cells were injected intramuscularly into the right tibialis anterior (TA) of mdx^5cv NOD/Rag1-null mice; four to five mice were injected per cell population. Mice were maintained and treated according to procedures approved by the Animal Care and Use Committee at Boston Children's Hospital. TA muscles were harvested 1 month after injection, snap-frozen in liquid-nitrogen-cooled isopentane and stored at −80°C for later analysis by immunohistochemistry as below.

Immunohistochemistry

Frozen muscle samples were sectioned on a cryostat at a thickness of 9 µm and collected on slides (Tissue Tack Microscope slides; Polysciences). The entire muscle was sectioned with 1 in 15 sections collected and 2–3 duplicate slides per muscle. Slides were then fixed for 3min in ice-cold methanol, washed once in PBS and blocked in 10% FBS/PBS for 30min at room temperature. Slides were stained as mentioned above with polyclonal rabbit anti-dystrophin (CAP6–10; 1:2000) (64,65) and monoclonal mouse anti-human spectrin (1:100; Vector Labs) diluted in 10% FBS/PBS. For secondary antibody staining, DyLight488-conjugated anti-mouse (1:1000; Jackson ImmunoResearch) and AlexFluor568-conjugated anti-rabbit (1:1000; Invitrogen) antibodies diluted in 10% FBS/PBS were used. Images were taken using a 20× objective lens on a Nikon Eclipse E1000 microscope fitted with a Hamamatsu ORCA-ER CCD camera. For analysis, the average number and standard deviation of positive myofibers were assessed for up to six sections per mouse.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported by a grant from National Institutes of Health (NINDS 2R01NS047727 to E.G) and by a grant from the Muscular Dystrophy Association (grant 199642 to E.G). We thank the Dana Farber Hematologic Neoplasia Flow Cytometry Core and the IDDRC Stem Cell Flow Cytometry Core at Boston Children's Hospital, supported by the National Institutes of Health (grant P30 HD018655). The Pax7 antibody (Developmental Studies Hybridoma Bank, University of Iowa) was developed by Atsushi Kawakami under the auspices of the NICHD.