Aldehyde Dehydrogenase Activity Identifies a Population of Human Skeletal Muscle Cells With High Myogenic Capacities

Abstract

Aldehyde dehydrogenase 1A1 (ALDH) activity is one hallmark of human bone marrow (BM), umbilical cord blood (UCB), and peripheral blood (PB) primitive progenitors presenting high reconstitution capacities in vivo. In this study, we have identified ALDH⁺ cells within human skeletal muscles, and have analyzed their phenotypical and functional characteristics. Immunohistofluorescence analysis of human muscle tissue sections revealed rare endomysial cells. Flow cytometry analysis using the fluorescent substrate of ALDH, Aldefluor, identified brightly stained (ALDH^br) cells with low side scatter (SSC^lo), in enzymatically dissociated muscle biopsies, thereafter abbreviated as SMALD⁺ (for skeletal muscle ALDH⁺) cells. Phenotypical analysis discriminated two sub-populations according to CD34 expression: SMALD⁺/CD34⁻ and SMALD⁺/CD34⁺ cells. These sub-populations did not initially express endothelial (CD31), hematopoietic (CD45), and myogenic (CD56) markers. Upon sorting, however, whereas SMALD⁺/CD34⁺ cells developed in vitro as a heterogeneous population of CD56⁻ cells able to differentiate in adipoblasts, the SMALD⁺/CD34⁻ fraction developed in vitro as a highly enriched population of CD56⁺ myoblasts able to form myotubes. Moreover, only the SMALD⁺/CD34⁻ population maintained a strong myogenic potential in vivo upon intramuscular transplantation. Our results suggest that ALDH activity is a novel marker for a population of new human skeletal muscle progenitors presenting a potential for cell biology and cell therapy.

Introduction

Aldehyde dehydrogenases (ALDHs) constitute a large and ancient family of intracellular enzymes involved in oxidation of aliphatic and aromatic aldehydes into the corresponding acids, thereby are considered as general detoxifying enzymes eliminating toxic biogenic and xenobiotic aldehydes in humans.^1,2 ALDH1 is cytosolic, ubiquitously distributed, and in particular confers resistance to anticancer drugs of the oxazaphosphorine family by their detoxification.^3,4 Fluorescent synthetic aminoacetaldehyde substrates^5,6 such as boron dipyrromethene–aminoacetaldehyde (Aldefluor) are retained intracellularly upon oxidation by ALDH, allowing the identification by flow cytometry and cell sorting of cell populations with low side scatter (SSC^lo) and high ALDH activity (SSC^lo/ALDH^br) within human bone marrow (BM),^1,5,7,8 umbilical cord blood (UCB),^{6,9,10,11,12,13,14} and peripheral blood (PB).^8,15,16 These ALDH^br/SSC^lo are rare cells, with frequencies not exceeding 3–4% of the mononucleated cellular fraction of the tissue source. Lineage-positive cell depletions and colabeling with classical hematopoietic markers such as CD34, CD38, CD45, or CD133 refined the characterizations of sub-populations and the demonstration of their differentiation capacities in vitro and in vivo,^{6,10,11,12,13,14} indicating that high ALDH activity is a hallmark of cells harbouring high myeloid, lymphoid, erythroid differentiation ability, and short and/or long-term hematopoietic reconstitution capacities in vivo.^10,11,12,14 The selection of primitive progenitors on the basis of ALDH expression has been proposed as a useful clinical tool foretelling graft efficiency in transplantation perspectives.^11,16

Alternate differentiation abilities have been attributed to BM ALDH^br cells toward mesenchymal and endothelial lineages.⁷ Meanwhile, ALDH^br cells have been described to be present in populations of human endothelial progenitors,¹⁷ cells from the stromal vascular fraction of human adipose tissues,¹⁸ normal or cancer human mammary epithelial progenitors,¹⁹ normal and malignant human colonic stem cells.²⁰ Murine ALDH^br cells are present among primitive neural stem cells from fetal and embryonic neurospheres and dissociated brain tissue and successfully engraft into mouse brain.²¹ In addition to their hematopoietic capacities, human ALDH⁺ cells have been used in murine models for revascularization of ischemic limbs as well as for liver repair.^22,23

In skeletal muscles, satellite cells have been considered as the main myogenic progenitors, allowing growth, homeostasis and regeneration.^24,25,26 These quiescent cells are located beneath the muscle fiber basal lamina²⁷ and express the surface marker CD56 in humans (neural cell adhesion molecule),²⁸ but show phenotypical heterogeneity.^25,29 However, adult stem cells of endomysial origin are now being shown to share the role of satellite cells and are becoming attractive candidates for new therapeutic strategies.^25,26,30 Side population (SP) cells are defined by their capacity to exclude Hoechst dyes and constitute a rare population of adult stem cells.^26,31 Cells migrating from the BM can also participate to muscle regeneration.^32,33 Pericytes, maybe related to mesangioblasts, wrap intramuscular vessels, express α-smooth muscle actin, CD44, CD146, NG2, alkaline phosphatase.^30,34,35,36 Perivascular cells can be isolated from several human tissues, based on their expression of CD146, NG2, and PDGF-Rb (platelet-derived growth factor receptor–β polypeptide).³⁷ Human skeletal muscle may constitute a reserve of pluripotent mesenchymal stem cells.^38,39 Myoendothelial cells express CD146 and CD31⁴⁰ or CD56, CD34 and CD144.⁴¹ Finally, circulating or muscle CD133⁺ cells may also take part in muscle regeneration.⁴² To our knowledge, the presence of progenitor cells expressing ALDH has never been documented in skeletal muscle.

In this study, we investigated the presence of ALDH-expressing populations in human skeletal muscles and their putative role as primitive progenitors of myogenic cells. Based on the measurement of ALDH activity using the fluorescent substrate Aldefluor, we have identified and further differentially phenotyped two main and well-defined sub-populations of SSC^lo/ALDH^br cells isolated from bulk dissociated muscle biopsies, which we have called SMALD⁺ (for skeletal muscle ALDH⁺) cells. Upon sorting, SMALD⁺ cells were expanded and their myogenic potential was evaluated. We show that the SMALD⁺/CD34⁻ population presents a strong myogenic potential both in vitro and in vivo.

Results

Histological and flow cytometry characterization of ALDH1⁺ cells in human skeletal muscle

Immunohistofluorescence analysis of human skeletal muscle biopsies using an anti-ALDH1 antibody (Ab) highlighted rare cells, presenting a cytoplasmic expression of ALDH1 (Figure 1a). The delineation of the muscle fibers by laminin staining indicated that cells expressing ALDH1 were located in the endomysial compartment.

Figure 1.

Immunohistological and flow cytometric characterization of ALDH⁺ cells in human skeletal muscle. (a) Immunohistofluorescence localization. An aldehyde dehydrogenase-positive (ALDH1⁺) cell (green, nucleus in blue) located outside muscle fibers delineated by laminin labeling (red, arrow heads) is in an endomysial position (merge, arrow). Bar = 10 µm. (b) Representative flow cytometric analysis of ALDH activity. Baseline fluorescence established in the presence of Aldefluor substrate and the inhibitor diethylaminobenzaldehyde. Gate 1 (G1) was defined to exclude debris (first panel). Based on G1, ALDH^br cells presented a shift of fluorescence defining the population in Gate 2 (G2) (second panel). Based on G2, SMALD⁺/CD34⁺ (G3) and SMALD⁺/CD34⁻ (G4) were clearly discriminated by the level of allophycocyanin-conjugated CD34 labeling (third and fourth panels). (c) Illustrative dot plot phenotypic characterizations of SMALD sub-populations based on G2. Percentages relate to the total number of cells in G1. SMALD, skeletal muscle aldehyde dehydrogenase.

ALDH activity was then classically analyzed in the mononucleated cell suspensions dissociated from human muscle biopsies (n = 12) using the Aldefluor substrate by flow cytometry. A small population of cells with SSC^lo and bright ALDH (ALDH^br) activity presented a shift of fluorescence (Figure 1b). This shift was not observed upon preincubation of the cells with the specific inhibitor diethylaminobenzaldehyde (Figure 1b). The median percentage of this skeletal muscle ALDH^br/SSC^lo population (abbreviated SMALD⁺) represented ~3–4% of the bulk mononucleated muscle cell suspension (Figure 1b, gate 2), with interindividual variations. Similar results were obtained using fresh, or frozen and thawed cell suspensions (data not shown).

Phenotypic characterization of SMALD⁺ populations

CD34 has been described as one main comarker of ALDH^br cells in hematopoietic tissues. Here, CD34 and ALDH colabeling clearly delineated two distinct sub-populations of SMALD⁺ cells: double positive ALDH⁺/CD34⁺ (SMALD⁺/CD34⁺) cells (Figure 1b, gate 3, median percentage 2.4 %) and simple positive ALDH⁺/CD34⁻ (SMALD⁺/CD34⁻) cells (Figure 1b, gate 4, median percentage 1.0 %). A third muscle cell population was defined as simply positive for CD34 (SMALD⁻/CD34⁺, median percentage 8.0%). Finally, cells negative for both ALDH activity and CD34 marker (SMALD⁻/CD34⁻) represented the main population contained in a dissociated muscle biopsy.

The phenotypes of these four sub-populations were further characterized by flow cytometry (Figure 1c and Table 1). It should be noted that these SMALD⁺ populations were still heterogeneous, or at different stages of maturation, because only a small percentage of cells was labeled for a given associated marker in each respective sub-population. Important interindividual variations were noted, due to the variability inherent to human biological materials. CD90 was found expressed in similar percentages by all populations, therefore was not considered as a discriminant marker. Some other markers, however, exhibited trends to specific enrichment according to the population considered.

Table 1.

Flow cytometry phenotypic characterization of dissociated muscle cells

The SMALD⁺/CD34⁻ population contained a higher proportion of cells expressing CD44 (up to 20%) than the SMALD⁺/CD34⁺ population, and a small percentage of CD105⁺ cells. The SMALD⁺/CD34⁻ population was found to be negative for other markers: CD31, CD45, CD56, and CD140b.

The double positive SMALD⁺/CD34⁺ population contained a smaller percentage of cells positive for CD44 than the SMALD⁺/CD34⁻ population. A few SMALD⁺/CD34⁺ cells also coexpressed CD140b. The SMALD⁺/CD34⁺ population was negative for CD45, CD56, CD105, and only a few cells expressed CD31.

The simple positive SMALD⁻/CD34⁺ population was especially enriched in CD31⁺ cells (up to 27%), thus suggesting a pro-endothelial commitment of a significant fraction of this population, underlined by the expression of CD105. Interestingly, CD56⁺ cells were also present in this SMALD⁻/CD34⁺ population (Table 1), suggesting a combined angiogenic and myogenic commitment of some cells, as described for myoendothelial cells.⁴¹

The double negative SMALD⁻/CD34⁻ population contained cells positive for CD45, CD56 and CD105 in low numbers. Of note, this double negative population included the highest percentage of CD56⁺ cells, reflecting their myogenic commitment, as expected from a mononucleated muscle cell preparation. Interestingly, mature circulating hematopoietic cells expressing CD45 were gathered in this double negative population, whereas they were not observed in the other ones (Table 1).

Phenotype and growth of cell-sorted populations in culture

Dissociated cells were sorted according to their ALDH activity and expression of CD34. The four populations (SMALD⁺/CD34⁺, SMALD⁺/CD34⁻, SMALD⁻/CD34⁺ and SMALD⁻/CD34⁻) were collected separately using a FACSVantage SE DiVa (Becton-Dickinson Biosciences, San Jose, CA) (Figure 2a), with a purity above 95%. The indicated purity of selection was confirmed by re-analysis of each sorted population. The selected populations were plated in myogenic proliferation medium, concurrently to a nonsorted fraction used as a quality control of muscle cell culture. Immunocytofluorescence analysis of SMALD⁺/CD34⁻ and SMALD⁺/CD34⁺ cells in vitro confirmed the initial flow cytometry data and allowed us to follow the expression of markers during the first week in culture (Figure 2b). SMALD⁺/CD34⁻ cells were found to be positive only for CD44 and negative for CD34, CD56, and CD15 within hours after seeding, and they expressed CD44, CD56 and/or CD146 in culture. Conversely, SMALD⁺/CD34⁺ cells expressed CD34 immediately after seeding, but this marker was lost in culture, whereas CD15 and/or CD44 appeared within 6 days (Figure 2b), and these cells remained negative for CD56. It should be noted that ALDH activity was sustained throughout the culture. The CD34 marker was not conserved in any sorted population in culture, nor in nonsorted fraction (data not shown).

Figure 2.

Cell sorting and phenotypic analysis in vitro. (a) Flow cytometric analysis, gating, and sorting of dissociated muscle sub-populations. Cells were labeled for aldehyde dehydrogenase (ALDH) activity and CD34 expression. The shift of fluorescence defined the population in G1 (left panel) presenting high ALDH activity. The following gates defined the sub-populations on the basis of ALDH activity and allophycocyanin-conjugated CD34 labeling. G2: SMALD⁺/CD34⁺, G3: SMALD⁺/CD34⁻, G4: SMALD⁻/CD34⁺, and G5: SMALD⁻/CD34⁻ cells (right panel). G2 and G3 were first determined on G1. (b) Immunocytofluorecence analysis of SMALD⁺/CD34⁻ and SMALD⁺/CD34⁺. SMALD⁺ cells were labeled for ALDH activity (green) and extracellular markers (PE-conjugated, red) 18 hour (D0) and 6 days (D6) after plating. Nuclei were stained by DAPI (blue). SMALD⁺/CD34⁻: some cells only expressed CD44 on D0, whereas numerous cells expressed CD44, CD56, and CD146 on D6. SMALD⁺/CD34⁺: the cells only expressed CD34 on D0, whereas on D6 numerous cells expressed CD15 or CD44, but not CD34. Controls were obtained on D0 and D6 by omission of the primary antibodies. Bar = 10 µm. SMALD, skeletal muscle aldehyde dehydrogenase.

When further expanded in culture, the SMALD⁻/CD34⁺ cells remained quiescent and rapidly degenerated with time in this proliferation medium. The other cell populations adhered to the substrate, proliferated (Figure 3a, left column) and were harvested on passages 2–3, first for flow cytometry phenotypic analyses, and second for differentiation assays in vitro. Following expansion, the percentages of CD56⁺ and CD15⁺ cells (Figures 3b,c) in nonsorted-derived cells increased to 76 and 16, respectively, with high standard deviations reaching 28 and 23%, respectively. These deviations underline the intersample variations, which could be attributed to differences in the anatomical origin of the biopsies or individual status of the donors. SMALD⁺/CD34⁻-derived cell cultures produced on average 94 ± 6% of CD56⁺ cells and never expressed CD15, and were considered as a homogenous population of myoblasts. In contrast, SMALD⁺/CD34⁺-derived cells never presented a CD56⁺ phenotype, but 44 ± 19% of them expressed CD15, and were therefore considered as an heterogeneous population of CD56⁻ cells. SMALD⁻/CD34⁻-derived cells contained 51 ± 41% CD56⁺ cells as expected, and 17 ± 19% CD15⁺ cells. This was not surprising because the SMALD⁻/CD34⁻ fraction is heterogeneous and initially contained CD56⁺ cells.

Figure 3.

In vitro analysis of SMALD⁺/CD34− and SMALD⁺/CD34⁺-derived cells. (a) Phase contrast pictures of SMALD⁺/CD34⁻ and SMALD⁺/CD34⁺ cells expanded in proliferation and shifted to differentiation media. SMALD⁺/CD34⁻ derived cells differentiated into myotubes (phase contrast) expressing fast myosin heavy-chain isoform (b,e; red; nuclei stained in blue). SMALD⁺/CD34⁺ derived cells accumulated lipid droplets of various sizes (c,f; droplets stained by Oil Red). Bars = 25 µm (a–e), 8 µm (f). (b) Fluoresence-activated cell sorting phenotypic characterizations for CD56 and CD15 membrane markers on expanded SMALD⁺/CD34⁻ and SMALD⁺/CD34⁺ populations. (c) Quantitative comparisons between populations regarding CD56 or CD15 expression. Most SMALD⁺/CD34⁻-derived cells expressed CD56 in culture, but not CD15, whereas half SMALD⁺/CD34⁺-derived cells expressed CD15 but never CD56. CD56⁺ cells also originated from nonsorted and SMALD⁻/CD34⁻ fractions. n, numbers of biopsies analyzed. SMALD, skeletal muscle aldehyde dehydrogenase.

Differentiation of cell-sorted populations in vitro

After 4–5 days in myogenic differentiation medium, SMALD⁺/CD34⁻-derived cells fused and formed multinucleated myotubes expressing the fast myosin heavy-chain isoform. Placed in the same conditions, SMALD⁺/CD34⁺-derived cells were unable to fuse into myotubes and consequently stained negative for fast myosin (Figure 3a, middle column).

However, the SMALD⁺/CD34⁺-derived cells differentiated extensively into adipoblast-like cells filled with numerous lipidic vesicles. This population may therefore relate to human CD34⁺ skeletal muscle cells prone to differentiation into brown adipose tissue.⁴³ In contrast, SMALD⁺/CD34⁻-derived cells fused, differentiated into multinucleated myotubes in adipogenic medium, and never presented the adipogenic phenotype in vitro (Figure 3a, right column).

Taken together, our results suggest that SMALD⁺/CD34⁻ and SMALD⁺/CD34⁺ cells progressed in culture along two different pathways associated to myogenic or mesenchymal-like potentials, respectively.

Given the myogenic potential of SMALD⁺/CD34⁻ population in vitro, we analyzed the initial ratio (termed predictor percentage, PP) of SMALD⁺/CD34⁻ cells over the whole SMALD⁺ cells contained in muscle biopsies (n = 20), and observed a positive correlation with the yield of long-term cultures of bulk dissociated muscle biopsies (i.e., without any selection of populations) in myogenic conditions (Figure 4a). In nine of the biopsies the final number of CD56⁺ cells was greater than that of CD15⁺ cells, whereas in 11 of the biopsies the final yield of CD15⁺ cells was higher than that of CD56⁺ cells (Figure 4a). The PP presented farther median distribution values (33.8 % for the first group and 13.8% for the second group) which were significantly different between the groups (P < 0.01, analysis of variance test). Using Bayes law, the probabilities of cell culture outcomes were calculated for given PP values, and illustrated (Figure 4b). As indicated, when the threshold PP value is <25.5%, there is a higher probability to obtain a cell culture containing less CD56⁺ cells than CD15⁺ cells. Above this 25.5% threshold PP value, the probability is higher to obtain a cell culture containing more CD56⁺ cells than CD15⁺ cells. Above 40%, the probability for obtaining more CD56⁺ than CD15⁺ cells is 100%. Overall, the analysis of ALDH-expressing cells and concomitant CD34 labeling provides a predictive indication on the myogenic potential of cell culture issued from human muscle tissue. Our results suggest that SMALD⁺/CD34⁻ cells play an important role in culture outcome, i.e., in the final yield of CD56⁺ myogenic cells, because the initial proportion of these cells correlates with the final outcome. Consequently, SMALD⁺/CD34⁻ cells may become an important player in the field of muscle cell production in vitro. However, whether the acute condition of muscle dissociation set up in cell culture reflects the physiological muscle repair in course of homeostasis or acute regeneration is not known.

Figure 4.

Establishment and use of the predictor percentage (PP). (a) Initial fluoresence-activated cell sorting analysis of Aldefluor and CD34-allophycocyanin labeling on dissociated muscle cells (n = 20 biopsies) enabled calculation of PP value for each biopsy, before expansion in culture. Upon phenotyping (CD56, CD15), cultures were divided in two groups according to respective proportions of markers: [56⁺ > 15⁺] and [56⁺ < 15⁺]. In each group, biopsies (circles) were ranked as functions of their initial PP, with median PP indicated by black dash. The difference between groups was significant (P < 0.01). At the threshold PP value of 25.5% (dashed line) the outcome of a culture was equivalent regarding the yield in CD56⁺ and CD15⁺ cells. (b) Illustration of calculated outcome probabilities at established PP values. Below 25.5%, the probability is higher to produce more CD15⁺ cells than CD56⁺ cells. Above 25.5%, the probability is higher to produce more CD56⁺ cells.

In vivo differentiation abilities of sorted populations

To assess the in vivo myogenic potential of the SMALD⁺ populations, freshly sorted human cells or control myoblasts from third passage cultures were injected into muscles of immunodeficient SCID mice in a model of muscle injury which allows us to evaluate the participation of the injected cells to skeletal muscle repair.⁴⁴ Tibialis anterior muscles were first irradiated to block the regenerative abilities of recipient cells, including satellite cells, and favor the implantation of donor cells. Then, myonecrosis was induced in the Tibialis anterior muscles using notexin, to trigger a local cycle of muscle degeneration and regeneration. Finally, donor cells were injected in these pretreated recipient muscles.⁴⁴ Implantation success was estimated using histological identification of human antigens within the murine host tissue (nuclear envelope lamin A/C, mitochondrial enzyme COX2), and localization of the cells relatively to structural proteins such as laminin located in muscle fiber basal lamina, and dystrophin expressed at the inner part of cytoplasmic muscle fiber membrane. Identification of lamin A/C allows one to follow the fate of human nuclei even after the fusion of the human cells with myotubes or muscle fibers. But identification of human nuclei inside muscle fibers delineated by laminin only was not considered to be sufficient because cells could have penetrated the basal lamina, remaining intact without contributing to muscle formation. Then, identification of human COX2 protein within muscle fibers indicates the transfer of donor mitochondria into recipient muscle fibers, thus illustrating a true fusion event between donor cell and recipient fiber (Figure 5, asterisks). In addition, the localization of cells relative to dystrophin delineates their position, hence a potential function (i.e., in a satellite position, inside the muscle fiber, or in an endomysial position; Figure 6).

Figure 5.

Participation of SMALD⁺/CD34− and SMALD⁺/CD34⁺ sorted cells to muscle regeneration in vivo. Numerous SMALD⁺/CD34⁻ cells were identified at respective injection sites by human nuclei (lamin A/C, red), human mitochondrial COX2 (green) and basal lamina delineation (laminin, blue) on transverse (a–r) sections. In s-u (oblique sections), blue DAPI staining of nuclei was illustrated instead of laminin staining. Only a few SMALD⁺/CD34⁺ cells were observed at injection sites (a–c, g–i) as labeled for COX2 (asterisks) and lamin A/C (arrows), and remained in the endomysium. In contrast, SMALD⁺/CD34⁻ cells participated to muscle formation as illustrated by human mitochondria labeling inside muscle fibers indicative of the fusion between donor cells and recipient fibers or the formation de novo of muscle fibers (d–f, j–u, asterisks). Human nuclei were observed inside fibers (arrowheads) or in the endomysium (arrows). Bars = 50 µm (a–f), 10 µm (g–o, s–u), 20 µm (p–r). SMALD, skeletal muscle aldehyde dehydrogenase.

Figure 6.

Localization of SMALD⁺/CD34− cells upon transplantation. SMALD⁺/CD34⁻ cells were identified at injection sites by human nuclei (lamin A/C, red), dystrophin (green), and basal lamina (laminin, blue) delineations on transverse sections. Human nuclei were observed inside fibers (asterisks) or in the endomysium (arrows; a–f, j–l). Some fibers were surrounded by laminin, but did not express detectable levels of dystrophin, and may be regenerating fibers (d–f, j–l). A few human nuclei were located below basal lamina of muscle fibers (stained by laminin), but still outside them (delineated by dystrophin), hence in a putative satellite cell position (arrowheads; g–i, m–o). Bars = 15 µm (a–i), 8 µm (j–o). SMALD, skeletal muscle aldehyde dehydrogenase.

SMALD⁺/CD34⁻ cells participated with high efficiency to muscle regeneration (Figure 5 d–f,j–u) because as few as 6–20 × 10³ cells generated 3–4 × 10⁴ human nuclei throughout the muscle (Table 2). This indicates that the number of cells had increased in vivo during the 1 month interval. However, the experimental schedule does not allow one to identify the extent, mechanism, or timing of the increase, nor the cell type involved. Approximately half of the human nuclei were found located beneath muscle fiber basal lamina (Figure 5, arrowheads), and several fibers contained human mitochondria expressing COX2, indicating that these cells directly participated in muscle formation. The other half of human nuclei were located within the endomysium (Figure 5, arrows). Several fibers expressing COX2 did not contain a human nucleus, at the level of a given cryostat section (Figure 5, asterisks), suggesting that human mitochondria diffuse within the muscle fiber sarcolemma over longer distances than that simply defined by the plane of a human nucleus. Comparatively, 5–20 × 10³ control myoblasts generated 0.2–4.5 × 10⁴ human nuclei, each injected cell providing ~1.05 nuclei. Co-staining for dystrophin confirmed and refined the contribution of SMALD⁺/CD34⁻ cells to muscle fiber regeneration (Figure 6). The use of a nonspecies-specific anti-dystrophin Ab was preferred over a human-specific Ab. Indeed, using species (human)-specific antibodies, weak or discontinuous expression of human dystrophin may be observed, which may be attributed to the size of the dystrophin expression domain, the intracellular localization of the human nucleus and its transcriptional activity, the competition with murine dystrophin at the membrane, and the timing of human cell penetration relative to the time of observation (i.e., the time required for human dystrophin to be expressed in sufficient amounts in the large murine muscle fiber). These results may lead to difficulties in interpretation of the results, and failure to unambiguously localize the position of injected nuclei inside the muscle fibers. The use of a nonspecies-specific Ab immediately allows delineation of the muscle fiber inner membrane, and therefore the position of the injected cells. Human lamin A/C⁺ nuclei were observed inside the muscle fibers delineated both by dystrophin and laminin (Figure 6, asterisks). Some lamin A/C⁺ nuclei were also observed inside fibers delineated by laminin, but expressing few or no dystrophin: these may be necrotic or regenerating fibers or myotubes formed by the fusion of donor cells in the process of maturation. Moreover, a few lamin A/C⁺ nuclei were observed in a putative satellite position, i.e., below laminin and above dystrophin (Figure 6, arrowheads). It should be noted that it is difficult to ascertain the position of laminin in this model, because the animals have undergone irradiation and myonecrosis. Indeed, basal lamina have been damaged and remodeled by these treatments, and their complete regeneration is hampered by the blockade of resident muscle cells induced by irradiation. This constitutes one limitation of the model, which on the other hand has been proved very useful to evaluate the contribution of exogenous cells to muscle regeneration.

Table 2.

Implantation and regeneration efficiencies in vivo of initially sorted muscle cell populations

In contrast, SMALD⁺/CD34⁺ cells did not successfully integrate into the host muscle tissue, as underlined by the small number of human cells identified by immunohistofluorescence (Figure 5 a–c,g–i) and consequently the poor ratio of labeled nuclei to initially injected cells (Table 2). Moreover, the cells only contributed marginally to muscle regeneration because only 3% of these were located beneath the basal lamina.

Therefore, the phenotypical differences described above between SMALD⁺/CD34⁻ and SMALD⁺/CD34⁺ cells are reflected by functional differences in vivo.

Discussion

Identification of ALDH^br cells in human skeletal muscle

From classically dissociated muscle biopsies of various anatomical origins, we directly (i.e., without any step of cell culture) identified, characterized and sorted two main populations of SSC^lo cells expressing high levels of ALDH (SMALD⁺ cells), presenting similar cytometric features as populations isolated from human BM, UCB, PB, and murine brain. SMALD⁺ cells represented ~2–4% of the mononucleated cells obtained upon enzymatic dissociation of the muscle biopsies. The use of fluorescent substrates of ALDH such as Aldefluor offers several advantages: the methodology is independent of the strategy used for dissociation of the tissue and would not be sensitive to enzymatic surface digestion. It does not require cell permeabilization. Upon oxidation, the fluorescent probe is trapped inside the cells, thus allowing sorting of viable cells. Analysis and selection of the cells do not require the use of intercalating agents, nor light excitation in the ultraviolet range. The characterization relies robustly on a cellular function rather than on a membrane marker, whose turn-over may introduce biases as previously suggested.⁸ Finally, the reagent may be provided at clinical grade,¹⁰ and the methodology remains reliable using frozen and thawed stocks of dissociated cells, thus widening its usefulness in a clinical perspective.^10,45

CD34 labeling of dissociated muscle cells allowed a further refinement of the characterization of the SMALD⁺ populations. We then discriminated SMALD⁺/CD34⁺ and SMALD⁺/CD34⁻ cells. The muscle biopsy also contained SMALD⁻/CD34⁺ and SMALD⁻/CD34⁻ cells. These populations were clearly different, in terms of expression of extracellular markers of commitment, and of their potential to differentiate both in vitro and in vivo.

SMALD⁺ cells belong to two sub-populations with distinct phenotypical and functional characteristics

Although some markers, such as CD90, were expressed by all populations at different levels, others were restricted to a given population, suggesting a specific stage of commitment (CD31, CD44, CD45, CD56, CD105, CD140b). Based on these distributions and the classical functional attributions of these markers, SMALD⁺ cells would be devoid of already engaged endothelial (CD31), hematopoietic (CD45), and satellite (CD56) cells. Some SMALD⁺/CD34⁻ cells were enriched in CD44⁺ and CD105⁺ cells which are involved in cell adhesion, migration, and myogenesis.⁴⁶ Noteworthy, these populations were still heterogeneous. CD133, which is expressed by important fractions of ALDH^br cells from other sources, was not detected on SMALD⁺ cells. This absence may be the hallmark of muscle specificity, or due to the technical issue raised by labeling of enzymatically dissociated cells. Indeed, CD133 labeling is generally observed on nonadherent cells and its sensitivity to collagenase or trypsin treatments is unknown. However, at the present time, no enzyme-free methodology allows a rapid and complete dissociation of skeletal muscle tissue.

SMALD⁺/CD34⁺ cells presented some features of mesenchymal-like cells in vitro and in vivo. Upon culture these cells did not give rise to CD56⁺ cells, but mainly to an heterogeneous population of CD56⁻ cells, they were unable to form myotubes but differentiated into adipogenic or osteogenic-like cells (K. Vauchez and J.-T. Vilquin, unpublished results) in respective differentiation media. Upon intramuscular injection, sorted cells survived poorly, remained within the endomysial compartment, and did not participate to muscle regeneration.

Conversely, SMALD⁺/CD34⁻ cells proliferated rapidly in the myogenic proliferation medium used in this study and rapidly expressed CD56, the marker of satellite cells and myoblasts. Upon differentiation, these cells formed myotubes in myogenic medium, and osteoblastic-like cells in osteogenic medium (K. Vauchez and J.-T. Vilquin, unpublished results), however they did not produce adipocytes in adipogenic medium. Finally, sorted SMALD⁺/CD34⁻ cells participated to muscle regeneration in vivo with high efficiency, as assessed by the presence of thousands of human nuclei within the SCID mouse muscle tissue, and the labeling of human COX2 protein inside muscle fibers. A small number of cells were observed in a putative satellite position; however, further experiments will be required to confirm their functionality, e.g., submitting the injected muscles to repeated cycles of degeneration and regeneration and evaluating the contribution of injected cells to new rounds of regeneration in vivo. The number of human nuclei estimated at the time of sacrifice exceeded by two- to fivefold the number of injected cells, representing an indication that the number of cells had increased during the 1 month interval. This value may be largely underestimated because cell death or disappearance upon injection generally decreases the efficacy of cell transplantation. The mechanism, extent and timing of survival and/or proliferation cannot be assessed in this study, which was focused on initial and final time points dedicated to establish a first proof-of-concept, whereas survival and proliferation are generally considered as early occurrences after cell transplantation.

Comparison between SMALD⁺ cells and ALDH^br cells from other sources

The comparison between ALDH^br cells originating from hematopoietic compartments and skeletal muscle reveals differences in absolute and relative numbers, the nature and proportions of associated markers, and the lineage differentiation abilities of the different fractions. The frequency of SMALD⁺ cells is ~10–50 times that of ALDH^br cells in PB.¹⁶ Similar proportions of ALDH^br cells express CD34⁺ in muscle, PB and BM,^7,15 and the proportion is slightly or significantly higher in UCB.^6,11,12,13 SMALD⁺ cells differ from BM and UCB cells by the differential expression of CD45 and CD31. In the present study, CD45 was not expressed by SMALD⁺ cells, but only by a few SMALD⁻/CD34⁻ cells, although it labels 65 and 99% of BM and PB ALDH^br cells, respectively.^7,11 Here also, CD31 was not expressed by SMALD⁺ cells, but only by SMALD⁻/CD34⁺ cells, whereas it labels 99% of UCB cells¹¹ and is depleted in BM ALDH^br fraction.⁷ SMALD⁺ resembles BM ALDH^br fraction regarding the expression of CD105 and CD90.⁷ SMALD⁺ cells do not express CD133, which is a hallmark of ALDH^br primitive hematopoietic progenitors detected in BM, UCB, and PB, although this marker may be cautiously interpreted in our labeling conditions as indicated above. Hematopoietic studies attributed the multilineage capacities to Lin⁻/ALDH^br/CD34⁺, Lin⁻/ALDH^br/CD133⁺, and/or Lin⁻/ALDH^br/CD34⁺/CD133⁺ cells, whereas no significant hematopoietic activity could be demonstrated in ALDH^br/CD34^– fractions selected from UCB and PB.^9,13,15 In our hands, the ALDH^br/CD34⁻ fraction (i.e., SMALD⁺/CD34⁻) presented the highest tissue-specific differentiation capacity (i.e., myogenic in muscle). Taken together, these observations suggest that, in humans, ALDH^br populations from various sources may differ in composition and may have different functions. Interestingly, in rodents, a fraction of BM hematopoietic cells containing c-kit⁺ immature myelomonocytic precursors is able to contribute to muscle fiber formation after intramuscular injection.⁴⁷ However, the human counterpart of this murine population has not been characterized yet.

Comparison between SMALD⁺ cells and classical myogenic progenitors

The analysis of the PP ratio of SMALD⁺/CD34⁻ to SMALD⁺ cell content in muscle biopsies confers a predictive value as to the fate of long-term cultures in myogenic conditions. The use of such a predictive marker has been proposed in a retrospective study of hematopoiesis in humans, where the implantation yield depended on the ratio of ALDH^br cells versus the total CD34⁺ cells.⁸ Here, our in vitro results suggest a role played by SMALD⁺ cells in the production of myogenic cells, at least in vitro. Whether this ratio is of physiological relevance is unknown yet, as is the relationship between satellite cells and SMALD⁺/CD34⁻ cells in vivo. Further studies are necessary to correlate this prognostic feature to the anatomical origin of muscle, or to donor parameters.

On the basis of histological, phenotypical, and functional characterizations, our results reveal differences between SMALD⁺ cells and previously described muscle progenitors.²⁵ They first suggest that SMALD⁺ are not classical satellite cells because they are not located beneath the muscle fiber basal lamina and do not express CD56.²⁸ Moreover, satellite cells were identified within the ALDH-negative muscle-derived fractions. The so-called muscle SP cells are heterogeneous.⁴⁸ Overlaps between the SP- and ALDH-defined cell populations have been described in both BM- and UCB-derived cells within the CD34⁺ cell population.⁴⁹ In our hands, however, SMALD⁺/CD34⁺ and SMALD⁻/CD34⁺ fractions, which contain the CD34⁺ populations, do not undergo myogenic differentiation. Furthermore, SP cells classically require coculture with myoblasts to undergo myogenic specification,³¹ whereas SMALD⁺/CD34⁻ cells undergo rapid myogenic differentiation without coculture. Therefore, the SMALD⁺/CD34⁻ cell population would not correspond to the SP cells. A direct back-to-back comparison between SMALD⁺ and muscle SP cells will be mandated in future experiments. Human myoendothelial cells initially harbor CD31, CD34, CD56, CD144, and CD146 antigens.^40,41 SMALD⁺ cells, either CD34⁺ or CD34⁻, do not express CD56 nor CD31 and are, therefore, unlikely to be myoendothelial cells. In the present study, cells sharing myoendothelial characteristics were observed in the SMALD⁻/CD34⁺ fraction upon sorting, but not in the SMALD⁺ fraction. The direct comparison between SMALD⁺ cells, human mesangioblasts, and pericytes, is made difficult due to methodological differences. Indeed, although SMALD⁺ cells are isolated without prior cultivation, muscle mesangioblasts/pericytes are classically obtained through cell culture.³⁴ Although freshly isolated SMALD⁺ cells did not share the pericytic characteristics, one cannot exclude that some particular sub-population of SMALD⁺ cells express the phenotype of pericytes when grown in vitro. Finally, a muscle cell population expressing CD34⁺/CD133⁺ has myogenic capacities and has been used in clinical trial.⁴² SMALD⁺ cells did not express CD133, with all proper reserves regarding this analysis cited above, and moreover SMALD⁺/CD34⁺ cells were unable to undergo myogenic differentiation. This suggests that SMALD⁺ cells would not be equivalent to the CD34⁺/CD133⁺ muscle progenitors.

Conclusion and perspectives

This study, to our knowledge, demonstrates for the first time the presence of two cell populations expressing high levels of ALDH within the human skeletal muscle tissue. The SMALD⁺/CD34⁻ population presents the abilities to undergo myogenic differentiation in vitro and in vivo, thus adding a potential new player in the field of muscle homeostasis. However, the alternate differentiation capacities of the SMALD⁺/CD34⁺ and SMALD⁺/CD34⁻ populations along different lineages remain to be explored, and the relationship with ALDH^br cells of other tissues deserve further direct studies. Indeed, we recently identified ALDH^br populations within murine skeletal and cardiac muscle tissues (K. Vauchez and J.-T. Vilquin, unpublished results). In the medical field, correlations between clinical or physiopathological status of human donors and the presence of SMALD⁺ cells mandate large-scale studies. Finally, the myogenic properties of SMALD⁺/CD34⁻ progenitors suggest their potential for the treatment of muscle diseases.

Materials and Methods

Human muscle biopsies. Frozen and fresh human muscle biopsies were obtained for immunohistological analysis and cell extraction, respectively, in agreement with the French Regulatory Health Authorities and our Ethics Committee via the Tissue Bank for Research of the French Association against Myopathies. These were 0.3–4 g res nullus specimen from orthopaedic surgery. The patients were free of muscular pathologies.

ALDH characterization by immunohistochemistry. Seven micrometer cryostat sections were fixed (cold acetone, 10 minutes, −20 °C), permeabilized [0.03% triton in phosphate-buffered saline, 10 minutes, room temperature (RT)] and nonspecific labeling was blocked [10% defined fetal bovine serum (Hyclone-Perbio, Brebieres, France) in phosphate-buffered saline, 30 minutes, RT]. Human ALDH1 was labeled using anti-ALHD1 mouse monoclonal Ab (mAb) (1/50, overnight, 4 °C, Becton-Dickinson, Le Pont de Claix, France) followed by goat anti-mouse immunoglobulin G (IgG) Ab (Alexa fluor 568, 1/1,000, 1 hour, RT, Invitrogen, Cergy-Pontoise, France). Then, sections were incubated with a rabbit polyclonal anti-laminin Ab (1/200, 1 hour, RT, Abcam, Cambridge, UK), followed by goat anti-rabbit IgG Ab (Alexa fluor 488, 1/1,000, 1 hour, RT, Invitrogen) to delineate skeletal muscle fibers. Nuclei were labeled with 4′-6-Diamidino-2-phenylindole (DAPI) in mounting medium (Vectashield hard set + DAPI, AbCys, Paris, France). Negative controls were obtained by omitting primary Ab. Fluorescence imaging was assessed using Zeiss microscope and pictures captured using a Sony CCD cooled camera (Roper Scientific, Evry, France). The images were combined using Metamorph software and assembled under Adobe Photoshop.

Muscle processing. Fresh muscle fragments were minced and digested using collagenase (Liberase, 1 hour, 37 °C, Roche, Meylan, France), then trypsin–EDTA (0.25%, Hyclone-Perbio, 20 minutes, 37 °C) as described.⁵⁰ Following filtration through 100 then 40 µm cell strainers (Becton-Dickinson), cells were counted under haemacytometer, and either used directly or frozen for later use.⁵⁰

Flow cytometry phenotyping and sorting. Red cells were depleted from initial fresh bulk using lysis buffer (5 minutes, Beckman Coulter, Roissy, France). Cells were incubated in Aldefluor assay buffer containing the ALDH substrate (1 µmol/l, 20 minutes, 37 °C, Stemcell Technologies, Grenoble, France). Controls were obtained by prior incubation of cells with 50 mmol/l of the specific ALDH inhibitor diethylaminobenzaldehyde. Extracellular markers were detected by incubations with allophycocyanin (CD34, Becton-Dickinson), or PE (CD31, CD44, CD45, CD56, CD90, CD105, CD106, CD140b, CD146 from Becton-Dickinson; CD133 from Miltenyi Biotec, Paris, France) -conjugated mAb (1/20, 15 minutes, 4 °C). Cells were centrifuged, suspended in Aldefluor kit buffer, and analyzed by flow fluorocytometry (Facscalibur, Becton-Dickinson) using the Cell Quest Software (10⁴ events analyzed). Populations were sorted using a Vantage SE DiVA cell sorter on the basis of both ALDH activity and expression of CD34.

Cell culture. Unsorted and sorted cells were seeded and expanded in the proliferation medium containing 80% modified custom-made MCDB medium (Hyclone), 20% defined fetal bovine serum, Gentamicine (50 µg/ml, Panpharma, Fougères, France), 10 ng/ml human recombinant bFGF (R&D systems, Lille, France), 10⁻⁶ mol/l dexamethasone (Merck, Clermont-Ferrand, France). The medium was changed on the following day. The cultures were grown for 7 to 14 days to reach 60% confluence, then cells were harvested by trypsinization and further expanded before reaching 80% confluence for two or three passages. An aliquot was analyzed as above for expression of CD56, CD34 and CD15 (Becton-Dickinson). An aliquot was seeded for differentiation studies (see below). In view of cell transplantation studies, a batch of myoblasts was prepared from unsorted cells as above.

Cytofluorescence studies of SMALD⁺ populations. Upon dissociation and sorting, 500 SMALD⁺/CD34⁺ or SMALD⁺/CD34⁻ cells were seeded in 24-well dishes, and observed 18 hour or 6 day thereafter. Cells were incubated with anti- CD15, CD34, CD44, CD56, CD146 Ab (1/20, 30 minutes, RT, Becton-Dickinson) in culture medium, followed by goat anti-mouse IgM (CD15 analysis) or IgG Abs (all other CDs; Alexa fluor 568, 1/200, 20 minutes, RT). Cells were rinsed in Aldefluor kit buffer and incubated in Aldefluor substrate (20 minutes, 37 °C) supplemented with DAPI (30 µg/ml, Sigma, L'Isle d'Abeau, France), then rinsed in buffer and observed immediately using a fluorescence Olympus IX10 inverted microscope. Controls were obtained by omission of the primary antibodies. Captured images (Sony CCD cooled camera) were treated and assembled using Metaview and Photoshop softwares, respectively.

Prognostic value of SMALD⁺/CD34− versus total SMALD⁺ cells in culture. Initial flow cytometry analysis of Aldefluor and CD34-allophycocyanin labeling on dissociated muscle cells (n = 20 biopsies) enabled calculation of PP (percentage of SMALD⁺/CD34⁻ cell number in the total SMALD⁺ cell number, PP) for each biopsy, which then underwent expansion in culture. Then, cultures were phenotyped (CD56, CD15) and divided into two groups according to respective proportions of markers: [56⁺ > 15⁺] and [56⁺ < 15⁺]. After ensuring that values of both groups were distributed according normal laws, we performed extrapolation of these distributions and determined the threshold value when both groups [56⁺ > 15⁺] and [56⁺ < 15⁺] presented equal probability according to Bayes law.

In vitro differentiation assays. Both sub-populations were subjected to both myogenic and adipogenic differentiation conditions in culture.

Myogenesis: Fifty thousand expanded cells were seeded into 12-wells plates. Differentiation was induced 24 hour later using Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% Horse serum (AbCys). Four to five days later, cells were permeabilized (methanol −20 °C, 10 minutes). The expression of fast myosin heavy-chain in differentiated myotubes was assessed using antifast myosin heavy-chain mAb (clone MY32, 1/400, 1 hour, RT, Sigma) followed by goat anti-mouse IgG Ab (Alexa Fluor 568, 1/1,000, 1 hour, RT). The nuclei were labeled with DAPI in mounting medium (Vectashield + DAPI, AbCys).

Adipogenesis: Hundred thousand expanded cells were seeded into 6-well plates. Differentiation was induced 48 hour later using Dulbecco's modified Eagle's medium (1 g/l glucose; Invitrogen) supplemented with 10% fetal bovine serum, 0.5 mmol/l isobutyl-methylxanthine (Sigma), 60 µmol/l indomethacin (Sigma), 10⁻⁶ mol/l dexamethasone (Merck) and 5 µg/ml insulin (Sigma). The medium was replaced every 3–4 days for 21 days. Cells were fixed in Baker buffer and incubated in Oil Red reagent (15 minutes, 37 °C), rinsed, counter-stained with hematein and mounted. The cells containing lipid vacuoles were photographed under phase contrast.

In vivo muscle pretreatment and cell transplantation. All the procedures were conducted according to the Guide for the Care and Use of Laboratory animals (DHAW publication no. 85-23 Office of Science and Health Reports, DRR/NIH, Bethesda MD 20892), and under anesthesia (80 mg/kg Ketamine, 16 mg/kg Xylazine). Tibialis anterior muscles of two-month-old female SCID mice (Charles River, L'Arbresle, France) received 18 Gy cobalt irradiation,⁴⁴ followed by the injection of notexin (5 ng, Sigma), respectively 2–4 day and 1 day before cell implantation. Freshly sorted human cells or long-term myoblast culture at third passage were counted, suspended in phosphate-buffered saline containing 0.5% BSA (Sigma) and kept on ice until transplantation. Cells (numbers indicated in Table 2) were injected in 8–10 sites under 10 µl using a custom-made glass needle.

Analysis and quantification of cell implantation in vivo. The animals were killed 4 weeks after transplantation. Cryostat sections (7 µm) were performed throughout whole muscles, from tendon to tendon, in order to ensure a complete overview of human cells distribution. Implantation of human cells in host muscles was evaluated by species-specific labeling of the human proteins lamin A/C and COX2, whereas laminin staining delineated the muscle fibers basal lamina, and dystrophin staining delineated the inner membrane of muscle fibers. Sections were incubated with rabbit polyclonal anti-COX2 Ab (1/1,000, 1 hour, RT, kind gift of Dr. Anne Lombès; Figure 5) or antidystrophin IgG1 mAb (clone NCLDys2, 1/20, 1 hour, RT, NovoCastra Laboratories; Figure 6) followed by goat anti-rabbit IgG Ab (Alexa Fluor 488, 1/1,000, 1 hour, RT) or by goat anti-mouse IgG1 Ab (Alexa Fluor 488, 1/300, 1 hour, RT) respectively. After washings, sections were incubated with antihuman lamin A/C IgG2b mAb (1/100, 1 hour, RT, AbCys), followed by goat anti-mouse IgG2b Ab (Alexa Fluor 568, 1/300, 1 hour, RT). Then, sections were incubated with rabbit polyclonal anti-laminin Ab (1/200, 1 hour, RT) followed by goat anti-rabbit IgG Ab (Alexa Fluor 350, 1/500, 1 hour, RT). Slides were mounted in Mowiol. Negative controls were obtained by omitting primary Ab. Image captures were performed as above.

Eight to ten representative sections were selected at regular intervals from proximal to distal tendon to count the numbers of human nuclei located outside or inside muscle fibers (×25 magnification). Then inferred number of human nuclei (IN), percentage of nuclei included in muscular fibers (% NIF), and ratio of IN on initial number of injected cells (RNC) were calculated as described below: