β4 Integrin Marks Interstitial Myogenic Progenitor Cells in Adult Murine Skeletal Muscle

Abstract

Skeletal muscle growth and its regeneration following injury rely on myogenic progenitor cells, a heterogeneous population that includes the satellite cells and other interstitial progenitors. The present study demonstrates that surface expression of β4 integrin marks a population of vessel-associated interstitial muscle progenitor cells. Muscle β4 integrin–positive cells do not express myogenic markers upon isolation. However, they are capable of undergoing myogenic specification in vitro and in vivo: β4 integrin cells differentiate into multinucleated myotubes in culture dishes and contribute to muscle regeneration upon delivery into diseased mice. Subfractionation of β4 integrin–expressing cells based on CD31 expression does not further enrich for myogenic precursors. These findings support the expression of β4 integrin in interstitial, vessel-associated cells with myogenic activity within adult skeletal muscle.

Skeletal muscle is a tissue capable of sustained regeneration, and this process relies on a population of myogenic precursor cells, known as satellite cells (Mauro 1961). Satellite cells are normally mitotically quiescent and localize to the periphery of myofibers, beneath the basal lamina, but following injury become activated, migrate, and proliferate extensively to produce mononuclear myoblasts that repair and replenish the damaged myofibers (Campion 1984; Grounds and Yablonka-Reuveni 1993). Several studies have described other types of cells with myogenic potential that are distinct from satellite cells (Peault et al. 2007; Mitchell et al. 2010). Among these, myoendothelial cells and vascular-associated cells have gained increased interest for their ability to contribute to muscle in vivo. However, most of these cells, such as the MDSC (Qu-Petersen et al. 2002) and the mesoangioblasts (De Angelis et al. 1999), are only isolated in retrospect after culturing muscle-derived mononuclear cells or tissue explants, respectively. Only recently, muscle progenitor cells expressing endothelial markers have been prospectively isolated from human muscle (Zheng et al. 2007).

A transcriptional profiling study comparing muscle versus bone marrow–derived progenitor cells indicated increased expression of β4 integrin in muscle (Liadaki et al. 2005). Integrins are a large family of transmembrane receptors that bind extracellular matrix proteins (collagen, laminin, fibronectin) and interact with the cell cytoskeleton. They function as heterodimers of an alpha- and a beta-subunit. In mammals, there are 18 different alpha subunits and 8 different beta subunits, and their various combinations yield a total of approximately 24 known different integrin heterodimers. A variety of cellular processes, including proliferation, differentiation, and apoptosis, are regulated by coupling of integrins with intracellular signaling pathways (Hynes 1992; van der Flier and Sonnenberg 2001), and their elimination by targeted homologous recombination often results in embryonic lethality (Hynes 1994, 1996; Fassler et al. 1996; Hynes and Wagner 1996).

α7β1 integrin is expressed on skeletal and cardiac muscle cells at specific stages during muscle development and is the major integrin receptor on adult skeletal myoblasts and muscle fibers (Song et al. 1992; Song et al. 1993; Ziober et al. 1993). Absence of α7 integrin gives rise to a muscular dystrophy phenotype (Mayer et al. 1997; Guo et al. 2006). α7 integrin expression has been used in the isolation of mouse satellite cells (Sacco et al. 2008) and human fetal myogenic progenitors (Ozeki et al. 2006).

β4 integrin (also known as CD104 antigen) forms a heterodimer only with α6 integrin, and it is best characterized as a major component of junctional adhesion complexes called hemidesmosomes, the structures that provide stable adhesion of the epidermis to the dermis (Jones et al. 1998; Litjens et al. 2006). α6β4 integrin expression has been reported on different cell types, including subsets of endothelial cells (Kennel et al. 1992; Hiran et al. 2003), immature thymocytes (Sonnenberg et al. 1990), Schwann cells (Quattrini et al. 1996), keratinocytes (Watt 2002), and a variety of carcinoma cells, where it has been proposed to enhance tumorigenesis and promote tumor angiogenesis (Kennel et al. 1989; Van Waes et al. 1991; Mechtersheimer et al. 1994; Trusolino et al. 2001; Chung et al. 2002; Chung et al. 2004; Bertotti et al. 2005; Chen and O’Connor 2005). Gene targeting studies in mice have shown that its absence results in skin fragility and blistering that can lead to perinatal death (Dowling et al. 1996; van der Neut et al. 1996; Abe et al. 2001), a phenotype that resembles the one observed in humans carrying mutations in the β4 integrin gene (Vidal et al. 1995). The expression of β4 integrin in muscle cells has not been investigated.

The present study demonstrates that β4 integrin is expressed by interstitial cells within skeletal muscle. These cells do not express myogenic markers and are associated with the vasculature and neural structures. Upon isolation and culture in vitro and following transplantation in diseased muscle in vivo, β4 integrin–positive cells demonstrate higher myogenic potential compared to cells lacking β4 integrin expression. Thus, β4 integrin marks interstitial progenitors in skeletal muscle that are distinct from satellite cells but exhibit myogenic potential both in vitro and in vivo.

Materials and Methods

Animals

C57BL/6 mice (4–10 weeks), C57BL/6-Tg(ACTB-EGFP)1Osb/J heterozygotes (green fluorescent positive [GFP+]) mice (4–6 weeks), and C57Bl/6Ros-5cv (mdx^5cv) mice (4–10 weeks) (Jackson Laboratory; Bar Harbor, ME) were used. Mice were sacrificed by carbon dioxide intoxication followed by cervical dislocation. All animals were handled based on protocols approved by the animal care and use committee of Children’s Hospital, Boston, Massachusetts.

Isolation and Cell Surface Marker Staining of Muscle Cells

Mononuclear cells were isolated from the hindlimb muscles of C57BL/6, mdx^5cv, and GFP+ mice as previously described (Montanaro et al. 2004). Briefly, the hindlimb muscles were excised from sacrificed mice; cleaned from tendons, connective tissue, and adipose tissue; and thoroughly minced. Then, the muscle cells were obtained by tissue digestion with 5 mg/ml collagenase IV and 1.2 U/ml dispase II (both from Worthington Biochemical; Lakewood, NJ) for 45 min at 37C and sequential filtering through 70-µm and 40-µm cell strainers. After elimination of red blood cells, the remaining cells were resuspended in cold PBS with 0.5% bovine serum albumin (Sigma-Aldrich; St. Louis, MO), before proceeding to antibody staining.

To determine β4 integrin (CD104) expression in the skeletal muscle, mononuclear cells were incubated with rat anti-mouse CD104 (1 µg/million cells, clone 346-11A; BD Pharmingen, San Diego, CA) or its isotype control (rat IgG2a,κ; BD Pharmingen) for 25 min on ice. Cells were washed with cold PBS + 0.5% BSA and then incubated for 20 min, on ice, with PE-anti-rat Ig (BD Pharmingen) at 1 µg/million cells or FITC-anti-rat Ig (BD Pharmingen) at 2.5 µg/million cells. Following another PBS + 0.5% BSA wash, cells were resuspended in the same buffer before FACS analysis. To evaluate coexpression of β4 integrin with CD31, α6 integrin (CD49f) or α7 integrin cells were stained for β4 integrin as described above and a third 20-min incubation with PE- or APC-conjugated anti-mouse CD31 (clone MEC 13.3; BD Pharmingen) at 0.5 µg/million cells, PE-anti-human CD49f (clone GoH3; BD Pharmingen) at 0.5 µg/million cells, or FITC-conjugated anti-mouse integrin α7 (clone 334908; R&D Systems, Minneapolis, MN) at 0.25 µg/million cells, or the respective isotype controls, was performed on ice. Anti-mouse CD16/CD32 antibodies (clone 2.4G2; BD Pharmingen) were added as Fc block prior to anti-mouse integrin α7 incubation. Biotinylated Hamster anti-rat CD29 (clone Ha2/5; BD Pharmingen) or its isotype control (biotinylated Hamster IgM; Biolegend, San Diego, CA) was added in combination with anti-CD104, followed by incubation with FITC-conjugated streptavidin (BD Pharmingen). Before FACS analysis of the samples, 2 µg/ml propidium iodide (PI; Sigma-Aldrich) was added to allow exclusion of the dead cells. When samples from WT and mdx^5cv muscles were compared, at least one of each were stained and analyzed in the same experiment. PE- or APC-conjugated anti-CD45 (clone 30-F11; BD Pharmingen) or PE-anti Sca-1 (clone d7; BD Pharmingen), or the respective isotype controls, was added as a third incubation as described above. Fluorescence-activated cell sorting (FACS) analysis and sorting were performed on a three-laser FACSVantage SE or FACSaria (BD Biosciences; San Jose, CA). The results were analyzed using Flowjo software (Tree Star; Ashland, OR).

Cell culture and In Vitro Myogenic Differentiation of β4 Integrin Cells

Following staining of muscle cells with β4 integrin, CD31, and isotype control antibodies and FACS analysis, β4 integrin–positive and β4 integrin–negative cells (or the β4-positive/CD31-positive and negative subfractions) were sorted into PBS-containing Eppendorf tubes. Cells were centrifuged for 5 min at 514 × g and then resuspended in growth media designed for the proliferation of muscle cells: Ham’s F10 (Cellgro Mediatech; Manassas, VA), 20% fetal bovine serum (Atlanta Biologicals; Norcross, GA), and 200 U/ml penicillin, 0.1 mg/ml streptomycin, 2 mM glutamine, and basic Fibroblast Growth Factor (Promega; Madison, WI), at a final concentration of 2.5 ng/ml. Cells were plated at a density of 35,000 to 40,000 cells/well on eight-well plastic dishes/slides, precoated with rat tail collagen type I (Upstate; Lake Placid, NY). Cells were maintained in growth medium with changes every 2 to 3 days for a total of 11 days until they reached ~75% to 85% confluency. Growth medium was then replaced by differentiation medium: DMEM (Cellgro Mediatech), 2% horse serum (Gibco; Auckland, New Zealand), 200 U penicillin/ml, and 0.1 mg/ml streptomycin. Cells were cultured in this medium for 6 days, with daily changes of medium.

Immunostaining of Freshly Isolated Cells and Cultured β4 Integrin Cells

Following staining of muscle cells with β4 integrin antibody and FACS analysis, β4-positive and β4-negative cells were sorted and transferred onto glass slides by centrifugation in a cytospin centrifuge (Thermo Shandon Electron Corp.; Waltham, MA). Slides were either fixed in methanol for 3 min at room temperature (for desmin staining) or cold 4% paraformaldehyde (PFA) in PBS for 20 min at 4C, followed by permeabilization for 3 min at room temperature in PBS + 0.5% Triton X-100 (for MyoD and Pax7 staining). Blocking solutions contained PBS + 10% horse serum + 0.1% Triton X-100 when methanol was used for fixation or PBS + 10% horse serum + 0.5% Triton X-100 when PFA was used for fixation. Slides were incubated for 1 hr at room temperature with blocking solution and then incubated overnight at 4C with primary antibodies, washed three times for 10 min each at room temperature with PBS + 0.1% Triton X-100, and then incubated with secondary antibodies overnight at 4C, followed by 3 × 10-min washes with PBS + 0.1% Triton X-100. The antibodies used were the following: monoclonal mouse anti-human desmin (clone D33; DAKO, Carpinteria, CA) 1:100 or mouse anti-chicken Pax7 (Developmental Studies Hybridoma Bank; Iowa City, IA) 1:100, each followed by Texas Red anti-mouse-IgG 1:100 (Jackson Immunoresearch; West Grove, PA) or rabbit anti-mouse MyoD 1:50 (C-20; Santa Cruz Biotechnology, Santa Cruz, CA) in blocking buffer followed by Texas Red anti-rabbit-IgG 1:100 (Jackson Immunoresearch). Some slides were co-stained with rat anti-mouse CD104 (1:100 in blocking solution), followed by FITC-anti-rat-IgG (Jackson Immunoresearch) 1:100 in blocking solution, to definitively demonstrate the presence of double-positive cells for the markers tested. The slides were mounted with Vectashield medium for fluorescence with a nuclear counterstaining by 4′,6-diamidino-2-phenylindole (DAPI; Vector Labs, Burlingame, CA) and were examined under a Nikon Eclipse E1000 microscope.

Immunostaining was also performed on cultured cells following differentiation using a similar protocol. Cells were stained with either desmin (as mentioned before) or mouse anti-chicken myosin heavy chain (MyHC, clone MF20; Developmental Studies Hybridoma Bank, Iowa City, IA) 1:25. The fusion index (expressed as number of nuclei in MyHC-positive myotubes) and the number of mononuclear cells positive for MyHC were counted in 12 representative fields (six with the highest and six with the lowest number of positive cells) and were represented as percentages of the number of total nuclei in the selected field. Quantification was performed in a total of six independent cultures for each subpopulation, and the results were analyzed using the Wilcoxon rank sum test.

Detection of β4 Integrin–Expressing Cells in Tissue Sections

Skeletal muscle was dissected from C57BL/6 mice and snap-frozen in isopentane chilled in liquid nitrogen. Then, 7-µm muscle sections were fixed in methanol, blocked for 1 hr at room temperature in blocking buffer (PBS + 10% horse serum + 0.1% Triton X-100), incubated with primary antibodies for 1 hr at room temperature, and then washed three times for 10 min in PBS + 0.1% Triton X-100 at room temperature and incubated with secondary antibodies for 45 min at room temperature before three new 10-min washes. Each section was incubated with rat anti-mouse CD104, in combination with either of the following antibodies: rabbit anti-mouse dystrophin polyclonal antibody (6/10 antibody; Lidov et al., 1990) 1:500, rabbit anti-laminin (Sigma-Aldrich) 1:300, rabbit anti-NG2 proteoglycan (Millipore; Billerica, MA) 1:100, or goat anti-CD31 (C-20; Santa Cruz Biotechnology) 1:50, followed by Alexa Fluor 568–conjugated goat anti-rat-IgG (Molecular Probes; Eugene, OR) 1:1000 or FITC-donkey anti-rat IgG 1:100, as well as FITC-donkey or Alexa Fluor 568–goat anti-rabbit-IgG (Molecular Probes) 1:100 and 1:1000, respectively, or Alexa Fluor 568–donkey anti-goat-IgG (Molecular Probes) 1:1000. The slides were mounted with Vectashield medium with DAPI as described above.

Intramuscular Injections

Mononuclear cells were prepared from skeletal muscle isolated from GFP+ mice and stained with β4 integrin antibody followed by FACS analysis. GFP-positive/β4 integrin–positive and GFP-positive/β4 integrin–negative cells were sorted and resuspended in 20 µl PBS, following 5-min centrifugation (514 × g at 4C). Equal numbers of each cell type (80,000–200,000 cells) were injected intramuscularly into the tibialis anterior (TA) muscles of mdx^5cv mice (6–8 weeks old).

Recipient mouse muscle was harvested and analyzed for GFP and dystrophin expression 45 days after cell injections as previously described (Liadaki et al. 2007). Briefly, excised TA muscles were fixed overnight in 4% PFA solution at 4C, then soaked in 5% sucrose in PBS for 6 hr followed by 20% sucrose in PBS overnight. Muscle samples were snap- frozen in liquid nitrogen–chilled isopentane and stored at −80C until sectioning. For analysis of GFP expression, the whole TA muscle was sectioned into 10-µm-thick cross sections, and slides were examined under a microscope to identify sections with the highest number of GFP-positive fibers. These GFP-positive fibers (opposed to auto-fluorescent fibers) were scored only when detected on the fluorescein filter and not on the rhodamine filter. Selected serial sections were subsequently analyzed for dystrophin expression. Slides were fixed in methanol for 3 min and were incubated in blocking buffer (PBS + 10% horse serum + 0.1% Triton X-100) overnight at 4C. Sections were incubated with primary dystrophin antibody (rabbit anti-mouse antibody, 6/10; Lidov et al. 1990) and diluted 1:500 in blocking buffer overnight at 4C, followed by an overnight incubation with Texas Red–conjugated anti-rabbit IgG, diluted 1:100 in blocking buffer (Jackson Immunoresearch) at 4C. After three washes of 10 min each with PBS + 0.1% Triton X-100, slides were cover-slipped as described above and examined under a Nikon Eclipse E1000 microscope, using the following emission filters: bandpass 535/50 for GFP detection, bandpass 605/55 for Texas Red detection, and bandpass 460/50 for DAPI detection. Images were collected from the same microscopic field with separate filters for GFP, Texas Red, and DAPI signals using a CCD camera (Orca-ER, Hamamatsu, Japan) and overlaid using Improvision OpenLab software (version 5.5.0).

Results

Identification and Characterization of β4 Integrin–Expressing Cells in Skeletal Muscle

The expression of specific members of the integrin family on the cell surface of hindlimb muscle mononuclear cells was studied using FACS analysis. Most mononuclear cells in skeletal muscle tissue express β1 integrin on their surface (81.1 ± 9.2%, mean ± SD, n=4; see the gating strategy in Suppl. Fig. S1). α7 integrin, one of the integrins that dimerizes with β1 integrin, is detected on the surface of 24.4 ± 13.4% (mean ± SD, n=5) of mononuclear muscle cells, a subpopulation probably enriched in satellite cells (Sacco et al. 2008). Integrin β4 is expressed by a smaller population of cells: The mean percentage of β4 integrin–positive cells among mononuclear cells in limb skeletal muscle of wild-type C57BL/6 mice (WT) as detected by FACS was 7.7 ± 3.7 (mean ± SD, n=70). As expected, these cells coexpress α6 integrin (92.1 ± 6.5%, mean ± SD, n=3; Fig. 1A). However, β4 integrin–positive cells can also express other integrin dimers, such as α6β1 (59.1 ± 12.6% of β4 integrin–positive cells coexpress β1 integrin, mean ± SD, n=4; Fig. 1B) or α7β1 (34.1 ± 12.6% coexpress α7 integrin, mean ± SD, n=5; Fig. 1C).

Figure 1.

Expression of β4 integrin and other laminin-binding integrins in murine skeletal mononuclear cells as detected by fluorescence-activated cell sorting (FACS). Representative experiment of a flow cytometry analysis of combinations of integrin chains in wild-type (WT) hindlimb skeletal muscle mononuclear cells. (A) Detection of β4 and α6 integrins. β4 integrin is detected on the surface of 5% to 10% of mouse muscle mononuclear cells. Almost all of them coexpress α6 integrin, but they represent a small subpopulation of the α6 integrin–positive cells. (B) Co-detection of β4 and β1 integrins. Around 80% of muscle mononuclear cells are positive for β1 integrin as well as 50% to 60% among the β4 integrin–positive cells. (C) Detection of β4 and α7 integrins. α7 integrin is detected on 10% to 30% of muscle mononuclear cells and on approximately one-third of β4 integrin–positive cells. Numbers indicate percentages of gated cells compared to the whole parent population. For the gating strategy, see Supplemental Figure S1.

In TA muscle sections, stained with a combination of antibodies against β4 integrin, laminin (Fig. 2A,B), and dystrophin (Fig. 2C,D), β4 integrin was detected mostly in interstitial cells in between myofibers and outside their basal lamina (although often surrounded by the basal lamina of blood vessels; arrows in Fig. 2A,B). It was also detected in structures resembling nerve bundles (consistent with its expression by Schwann cells, not shown). β4 integrin was never detected beneath the basal lamina of myofibers, where satellite cells are located, and did not co-localize with Pax7, a satellite cell marker (not shown). Co-staining of β4 integrin with CD31, a marker known to stain endothelial cells and blood vessels, demonstrated that β4 integrin is sometimes coexpressed with CD31 (Fig. 2E,F, arrows), but each can be expressed independently (shown for β4 integrin in Fig. 2F with an arrowhead). The latter finding is in agreement with previous studies that demonstrated that α6β4 integrin is expressed by a subset of vessels (Kennel et al. 1992; Mechtersheimer et al. 1994; Hiran et al. 2003) and with our FACS studies of coexpression of β4 integrin and CD31 in a subgroup of cells (Fig. 2G). Vessel associated β4 integrin–positive cells can be in close contact with NG2 proteoglycan-expressing cells, probably pericytes (Ozerdem et al. 2001), but do not co-localize in the same cell (Fig. 2H).

Figure 2.

Localization of β4 integrin–expressing cells in mouse skeletal muscle tissue sections. (A, B) Tissue cross sections stained with β4 integrin (red, Alexa Fluor 568) and laminin (green, FITC). Nuclei are stained in blue with DAPI (A–F, H). (C, D) Tissue sections stained with β4 integrin (red, Alexa Fluor 568) and dystrophin (green, FITC). β4 integrin–positive cells were found in the interstitial space and never beneath a myofiber basal lamina. β4 integrin–positive cells in the interstitium could be found both surrounded by a basal lamina (arrows, A, B) or not (arrowhead, B). (E–F) Longitudinal muscle section stained with CD31 (red, Alexa Fluor 568) and β4 integrin (green, FITC). β4 integrin can co-localize with the endothelial cell marker CD31 (arrows) in blood vessels, but each is also expressed independently (arrowhead showing a β4 integrin–positive/CD31-negative cell). (G) Fluorescence-activated cell sorting (FACS) analysis of muscle mononuclear hindlimb skeletal muscle cells of C57BL/6 mice stained with β4 integrin and CD31, confirming the coexpression of both markers by a subset of cells. Numbers indicate percentages of gated cells relative to the parent population. (H) Longitudinal muscle section stained with β4 integrin (red, Alexa Fluor 568) and NG2 proteoglycan (green, FITC). Cells positive for the pericyte marker NG2 (arrowheads) are located nearby β4 integrin–expressing cells (arrow). Scale bars, 20 µm.

To study whether the expression of β4 integrin is altered in diseased muscle, β4 integrin–positive cells were analyzed in healthy and dystrophic mice. Mdx^5cv mice are a mouse model of human Duchenne muscular dystrophy: They carry a mutation in the dystrophin gene that leads to severe muscle degeneration and show a very low frequency of spontaneous revertant, dystrophin-expressing fibers (Chapman et al. 1989; Danko et al. 1992; Im et al. 1996). No difference in the proportion of β4 integrin–positive cells was observed among mononuclear cells from mdx^5cv and WT muscles between 4 and 10 weeks of age (7.4 ± 3.7% in mdx^5cv, n=7, vs 8.8 ± 4.7% in WT, n=6, mean ± SD; see Fig. 3A,B for a representative experiment).

Figure 3.

Fluorescence-activated cell sorting (FACS) analyses of mononuclear cell populations in skeletal muscles of C57BL/6 and mdx^5cv mice. Similar percentages of β4-positive cells are observed in wild-type (WT; A) and mdx^5cv (B) muscles. A lower proportion of mononuclear muscle cells are positive for SCA-1 in mdx^5cv (D) muscle than in WT muscle (C), whereas for the percentage of CD45-positive cells, a 5-fold increase is observed in mdx^5cv muscle (F) compared to WT muscle (E). Numbers indicate percentages of gated cells compared to the whole parent population. Among the β4 integrin–positive subpopulation (G–J), more than 50% of the cells express SCA-1 both in WT muscle (G) and in mdx^5cv muscle (H). A small percentage of β4 integrin–positive cells coexpress CD45 in WT (I) and mdx^5cv (J) muscles.

The distribution of β4 integrin–positive cells in relation to the expression of either SCA-1 or CD45 was also studied in skeletal muscle of 8-week-old WT and mdx^5cv mice. SCA-1 (stem cell antigen–1) was first described as a hematopoietic stem cell marker (Spangrude et al. 1988) and was subsequently found to be expressed by subsets of cells in skeletal muscle (Gussoni et al. 1999; Lee et al. 2000; Torrente et al. 2001; Kafadar et al. 2009). SCA-1 was detected on the surface of 78.4 ± 6.7% of skeletal muscle mononuclear cells in WT and 42.4 ± 8.8% in mdx^5cv (mean ± SD, n=5, p<0.01, Wilcoxon rank sum test; Fig. 3C,D). In both conditions, more than half of β4 integrin–expressing cells are similarly positive for SCA-1 (55 ± 5.3% in WT and 63.5 ± 7.6% in mdx^5cv, mean ± SD, n=5, not significant; Fig. 3G,H). CD45 is a cell surface tyrosine phosphatase found on all nucleated cells of hematopoietic origin. CD45 is also found on a subset of muscle cells that proliferate in response to acute injury (Polesskaya et al. 2003). Staining of muscle cells with both β4 integrin and CD45 antibodies showed that >95% of β4 integrin–expressing cells are negative for CD45 both in WT and mdx^5cv muscles (the percentage of CD45-positive cells among them was 2.31 ± 0.52% in WT and 4.64 ± 0.52% in mdx^5cv; Fig. 3I,J). Overall, these results demonstrate that expression of β4 integrin in skeletal muscle is observed in a population of muscle interstitial cells that is heterogeneous for CD31 and SCA-1 expression. β4 integrin–positive cells in WT skeletal muscle are also heterogeneous for expression of CD34, a cell surface adhesion glycoprotein expressed by satellite cells and by other previously described interstitial myogenic progenitors (Tamaki et al. 2002; Sacco et al. 2008; Mitchell et al. 2010; Suppl. Fig. S2).

To further characterize these cells, the expression of myogenic markers was studied on freshly isolated β4 integrin–positive muscle cells. The activation of myogenic cells is a highly regulated process that includes downregulation of the satellite cell-specific marker Pax7 and the upregulation of the myogenic-specific factors Myf5 and MyoD (Rudnicki et al. 1993; Rudnicki and Jaenisch 1995; Olguin and Olwin 2004; Zammit et al. 2004; Kuang et al. 2006; Zammit et al. 2006; Olguin et al. 2007). Freshly isolated β4 integrin–positive cells were transferred onto slides by cytospin and stained by immunofluorescence for the expression of the myogenic markers Pax7 and MyoD. Staining for Pax7 was performed in three independent preparations of β4 integrin–expressing cells. A total of 10,000 β4-positive cells were analyzed, but none was found positive for Pax7 (Suppl. Fig. S3A–F). Expression of Pax7 was demonstrated in a small percentage of β4-negative cells (less than 1%), consistent with the expectation that a small fraction of quiescent satellite cells should be present within this population. Less than 3% of β4 integrin–expressing cells were positive for the early myogenic transcription factor MyoD (Suppl. Fig. S3G–J). In summary, more than 95% of freshly isolated β4-expressing cells are negative for the myogenic markers Pax7 and MyoD, suggesting limited intrinsic myogenic potential.

β4 Integrin Cells Are Capable of Myogenic Differentiation In Vitro

To study whether in vitro culture of β4-expressing cells would allow their specification toward the myogenic lineage, equal numbers of β4 integrin–positive and β4 integrin–negative cells (35,000–40,000) were plated on collagen-coated slide chambers and cultured in growth medium commonly used to support culture of primary muscle cells (Rando and Blau 1994). Both β4 integrin–positive and β4 integrin–negative populations required 2 to 3 days for stable adherence to the culture dishes and initiation of growth. After a week in culture, β4 integrin–positive cells appeared to have a more refringent, rounded, or elongated morphology (myogenic phenotype), whereas β4-negative cultures, which are a mixture of myoblasts, fibroblasts, and other interstitial cells, contained more flattened “fibroblast-like” cells (Fig. 4A,B). In less than 2 weeks, β4 integrin–positive and β4 integrin–negative cultures had reached 75% to 85% confluency. Cultures were switched to differentiation medium for 6 days, and the formation of multinucleated myotubes was examined by staining for MyHC and desmin (Fig. 4). β4 integrin–positive cells were able to form elongated multinucleated myotubes (commonly 8–10 fused nuclei) that showed striations and contracted spontaneously, whereas β4 integrin–negative cells contained mainly di- and trinucleated myotubes (Fig. 4C–F). β4 integrin–positive cells showed a statistically significant increase of fused cells; the average fusion index was 16.5% for β4-positive cells and 4.7% for β4-negative cells (p value of 0.0022). There were numerous MyHC-positive mononuclear cells in both types of culture, but, when considering both fused and unfused nuclei, MyHC(+) nuclei were more frequent in the β4 integrin–positive cell cultures (24.3 ± 6.0% vs 10.6 ± 4.0%, mean ± SD, p<0.01). Interestingly, when the cells were analyzed under growth conditions, either before the induction of differentiation or after passaging, more than 80% of the cells with rounded refringent morphology in the culture expressed both MyoD and Pax7 (not shown). These studies indicate that β4-positive cells are capable of myogenic specification following culture in vitro and exhibit increased myogenic differentiation potential compared to the β4 integrin–negative population.

Figure 4.

Cell culture and in vitro myogenic differentiation of muscle mononuclear cells isolated based on β4 integrin expression. (A, B). Phase-contrast pictures of β4 integrin–positive (A) and β4 integrin–negative (B) cells after 6 days of culture. (C, D). Myosin heavy chain expression of β4 integrin–positive (C) and β4 integrin–negative (D) cells after 6 days of differentiation. (E, F). Desmin expression of β4 integrin–positive (E) and β4 integrin–negative (F) cells after 6 days of differentiation. Following differentiation, β4 integrin–positive cells form long multinucleated myotubes, whereas β4 integrin–negative cells form only di- or trinucleated myotubes. (G, H) Myosin heavy chain staining of β4 integrin–positive/CD31-positive (G) and β4 integrin–positive/CD31-negative (H) cells after 6 days of differentiation. No differences in morphology or fusion rate were observed. Scale bars = 100 µm.

Given the heterogeneity in surface marker expression detected in β4 integrin–positive cells, further subfractionation of this population was attempted for a better characterization of the myogenic progenitors within it. β4 integrin–positive cells were subfractionated according to CD31 expression and sorted through FACS. β4 integrin–positive/CD31-positive and β4 integrin–positive/CD31-negative cells were plated and cultured as described above. No marked difference in adherence and proliferation was noted between the two subpopulations. When induced to differentiate, both subpopulations fused into myotubes of comparable morphology (Fig. 4G,H) and at a similar rate (fusion index for β4 integrin–positive/CD31-positive was 23.9 ± 6.6% vs 22.1 ± 3.5% for β4 integrin–positive/CD31-negative cells, mean ± SD, based on three independent experiments). These results suggest that subfractionation of β4 integrin±positive skeletal muscle cells based on CD31 expression does not further enrich in myogenic precursor cells, as detected by an in vitro differentiation assay. β4 integrin–positive/CD45-positive skeletal muscle cells were also isolated by FACS, but these cells did not grow under the myoblast culture conditions used (data not shown).

β4 Integrin Cells Participate in Skeletal Muscle Repair In Vivo

To determine the in vivo potential of β4 integrin cells to participate in skeletal muscle repair, β4-positive and β4-negative cells were isolated from wild-type GFP+ mice and injected into the TA muscle of mdx^5cv mice. GFP+ mice express enhanced GFP (eGFP) under the control of a ubiquitous beta-actin promoter (Okabe et al. 1997).

Four independent experiments were performed, in which mice were injected with equal numbers of freshly isolated β4 integrin–positive or β4 integrin–negative cells (80,000–200,000 cells) that originated from the same FACS preparation. This process controls for variation between donor mice, antibody staining, and FACS analysis and allows pairwise comparisons of the engraftment efficiencies of the two subpopulations. Forty-five days after injections, recipient mice were sacrificed, and the injected TA muscles, as well as the non-injected (contralateral muscles), were sectioned entirely and analyzed for GFP fluorescence (Fig. 5). GFP-positive fibers were detectable with the injection of both β4 integrin–positive and β4 integrin–negative cells. The specificity of GFP fluorescence was confirmed by ensuring that the signal could be detected only on the fluorescein filter and not on the rhodamine filter, thereby excluding the possibility of auto-fluorescence. Staining with an antibody against GFP yielded similar results (not shown). For each injected mouse, dystrophin expression was evaluated in the sections with the highest number of GFP-expressing muscle fibers. Fibers exhibiting very strong GFP fluorescence were observed adjacent to fibers with less intense GFP fluorescence. The level of GFP expression varied between positive muscle fibers, possibly depending on the number of donor nuclei fused and the distance that GFP diffuses within the cytoplasm of a myofiber. For example, muscle fibers with weak GFP fluorescence in a given section were found to exhibit strong GFP fluorescence in other sections approximately 1000 µm apart. Interstitial GFP-positive cells were also detected in the β4 integrin–positive cell injected muscles (not shown). No GFP fluorescence was detected in the non-injected contralateral TA muscles (data not shown).

Figure 5.

β4 integrin cells isolated from green fluorescent protein (GFP)–expressing mice participate in muscle repair in mdx^5cv mice. (A, B) Representative engraftment of GFP-positive/β4 integrin–positive cells (A) and GFP-positive/β4 integrin–negative cells (B) in the tibialis anterior (TA) muscle of mdx^5cv mice. Scale bar = 200 µm. (C) The percentage of GFP-positive myofibers per section demonstrates that the contribution to regenerating myofibers was significantly higher for β4 integrin–positive cells than for β4 integrin–negative cells (*p=0.022, two-tailed t test, n=4). Error bars, standard error of the mean. (D–F) Representative engraftment of GFP-positive/β4 integrin–positive cells in the TA muscle of mdx5cv mice. GFP detection (D), expression of dystrophin (red, Texas Red, E), and merged images of GFP, dystrophin expression, and DAPI-stained nuclei (blue, F). Scale bar = 30 µm.

A representative example of the engraftment of injected β4 integrin–positive cells is shown in Figure 5A,D–F and for the β4 integrin–negative cells in Figure 5B. The percentage of GFP-positive myofibers, relative to the total number of fibers in the section, was significantly higher for β4 integrin–positive cells than for β4 integrin–negative cells (with a mean ± SEM of 5.50 ± 1.55% vs 1.09 ± 0.77%, p=0.022, two-tailed paired t test; Fig. 5C), and it reached up to 9.33% in one experiment (Suppl. Table S1). The number of GFP-positive fibers that were also dystrophin-positive was quantified, and their percentage among all GFP-positive fibers ranged from 25% to 53.3% (Suppl. Table S1). The finding that not all GFP-positive fibers express dystrophin in a section was not surprising because the nuclear domain of membrane-bound dystrophin differs from that of cytosolic GFP (Pavlath et al. 1989). In addition, revertant fibers were easily identified as GFP-negative/dystrophin-positive fibers and thus were excluded from the numbers listed in Supplemental Table S1. In conclusion, β4 integrin–expressing cells in skeletal muscle are interstitial cells that exhibit in vitro and in vivo myogenic potential.

Discussion

β4 integrin was previously identified as an important component of hemidesmosomes in stratified squamous epithelium, providing stable attachment of the epidermis to the dermis (Jones et al. 1998; Litjens et al. 2006). In addition to epithelial sites, β4 integrin expression has been reported on the endothelium of medium-sized blood vessels of several organs in the mouse, including lung, thymus, spleen, and liver (Kennel et al. 1992), and in Schwann cells (Sonnenberg et al. 1990; Quattrini et al. 1996). In this study, we present the novel expression of β4 integrin in skeletal muscle as a promising alternative marker for interstitial cells with myogenic potential.

Increased expression of β4 integrin mRNA was previously identified in muscle-derived progenitor cells compared to bone marrow–derived muscle progenitors (Liadaki et al. 2005) and in myoblasts derived from WT and myoD –/– mice (Seale et al. 2004). Whether expression of β4 integrin protein could be used to prospectively isolate myogenic progenitor cells was not known. Here we show that β4 integrin is expressed by a small proportion of cells in skeletal muscle tissue and that the integrin heterodimer α6β4 can be coexpressed with other laminin-binding integrins such as the α6β1 and α7β1 dimers. β4 integrin–expressing cells form a heterogeneous population with respect to their tissue localization and the expression of cell surface proteins such as SCA-1 and CD31. Interestingly, however, they are consistently localized in the muscle interstitium and never in a satellite cell position beneath a myofiber basal lamina. When freshly isolated, β4 integrin–positive cells do not express the satellite cell marker Pax7, and the majority (>95%) is also negative for MyoD. However, in vivo and in vitro studies demonstrate that β4 integrin–positive cells are capable of myogenic specification. The possibility of a semi-clonal expansion of a very small population of myogenic cells among the β4 integrin–positive cell population cannot be experimentally excluded, although most likely these cells should be Pax7-negative/MyoD-positive upon isolation and will gain Pax7 expression in culture, suggesting that they are a population distinct from satellite cells. On the other hand, our subfractionation studies of β4 integrin–positive cells based on the expression of CD31 demonstrated that myogenic activity is equally present in both cultured β4 integrin–positive/CD31-positive and β4 integrin–positive/CD31-negative cells, thus not providing additional enrichment.

Myogenic progenitors residing in the skeletal muscle interstitium have been previously described using different strategies that require either cell culture selection or FACS. Flow cytometric sorting of these cells is based on the exclusion of Hoechst dye or the use of a combination of cell surface markers that can be detected by immunofluorescence. Several of the myogenic interstitial cell populations that have been described differ in their marker expression (Peault et al. 2007), and their possible association to the vasculature has been proposed. A partial overlap between β4 integrin–positive cells and cells obtained with these different isolation strategies cannot be excluded. For example, mesoangioblasts (Minasi et al. 2002) have been associated with a pericyte localization (Dellavalle et al. 2007) and should be NG2-positive/CD31-negative when freshly isolated. Likewise, Pw1-positive interstitial progenitors, which are isolated by FACS among the CD34(+)SCA-1(MED) skeletal muscle cell fraction, do not co-localize with the endothelial marker CD31 (Mitchell et al. 2010). Muscle cells isolated by Hoechst dye exclusion comprise a much smaller population and lack intrinsic myogenic potential in vitro (Asakura et al. 2002). Multipotent interstitial cells among the CD34(+)CD45(–) skeletal muscle cell fraction, which corresponds to ~50% of dissociated muscle fiber–associated mononuclear cells, contribute to both myofibers and blood vessels after transplantation and do not express myogenic transcription factors or CD31 upon isolation (Tamaki et al. 2002). Interestingly, progenitors with the same multilineage potential have been found among CD34(–)CD45(–) cells (Tamaki et al. 2003). On the contrary, cells that express both endothelial and myogenic markers have been prospectively isolated from human skeletal muscle and shown to regenerate myofibers when transplanted to injured muscle more efficiently than CD56(+) myogenic cells (Zheng et al. 2007). Prospective isolation of cells based on β4 integrin expression gives rise to a heterogeneous population containing cells that overlap with some of the above-described interstitial progenitors: CD31(+) and CD31(–) cells that are myogenic in vitro without need to co-culture them with primary myoblasts or cell lines; in addition, following intramuscular delivery into dystrophin-deficient mice, these cells persist for at least 45 days both as mononuclear interstitial cells and contribute to myofibers that produce dystrophin without the need of preconditioning the recipient muscle using X-irradiation or chemically induced damage. The physiological role of the different interstitial myogenic progenitors in muscle regeneration is still unclear, but they are being actively studied for their potential in cell therapy. A reduced physiological participation in muscle regeneration or the high heterogeneity among the β4 integrin–positive cell population could explain the absence of difference in the proportion of these cells in diseased muscle.

In summary, this study reports the identification of β4 integrin as a potential novel marker for interstitial muscle cells. Although these cells do not express myogenic markers upon isolation, they exhibit myogenic potential upon culture in vitro and in vivo, suggesting that they might be enriched for uncommitted myogenic precursors. Future studies will address whether increased incorporation of these cells in dystrophic muscles can be achieved by culturing the cells prior to injection or by modifying the methods of cell delivery in vivo.