|Home | About | Journals | Submit | Contact Us | Français|
Repair of adult skeletal muscle depends on satellite cells, quiescent myogenic stem cells located beneath the myofiber basal lamina. Satellite cell numbers and performance decline with age and disease, yet the intrinsic molecular changes accompanying these conditions are unknown. We identified expression of GFP driven by regulatory elements of the nestin (NES) gene within mouse satellite cells, which permitted characterization of these cells in their niche. Sorted NES-GFP+ cells exclusively acquired a myogenic fate, even when supplemented with media supporting non-myogenic development. Mutual and unique gene expression by NES-GFP+ cells from hindlimb and diaphragm muscles demonstrated intra- and inter-muscular heterogeneity of satellite cells. NES-GFP expression declined following satellite cell activation and was reacquired in late stage myogenic cultures by non-proliferating Pax7+ progeny. The dynamics of this expression pattern reflect the cycle of satellite cell self-renewal. The NES-GFP model reveals unique transcriptional activity within quiescent satellite cells and permits novel insight into the heterogeneity of their molecular signatures.
Muscle growth and repair depend on satellite cells, myogenic stem cells located between the plasma membrane and the basal lamina of the myofiber (Collins et al., 2005). During postnatal growth, satellite cells proliferate and contribute myoblasts that fuse with the enlarging myofibers. In mature muscles, satellite cells are typically quiescent, unless activated by muscle trauma. Satellite cells may fuse with existing myofibers when minute repairs are needed, and upon massive damage they can generate a larger myogenic cell pool to form new myofibers (Grounds and Yablonka-Reuveni, 1993; Hawke and Garry, 2001). Subtle muscle injuries that occur during routine muscle activity raise a continuous demand for functional satellite cells throughout life. Impaired satellite cell performance may contribute to deterioration of limb muscles during aging (Conboy et al., 2003; Shefer et al., 2006). In contrast, resident satellite cells of extraocular muscles do not appear to depreciate in their performance compared to other muscles (McLoon et al., 2004; Porter et al., 2003). Taken together, insight into the molecular characteristics of satellite cells from different anatomic origins may shed light on the diverse roles of satellite cells in muscle maintenance.
Quiescent satellite cells commonly express the paired-homeobox transcription factor Pax7, while their proliferating progeny co-express Pax7 and the muscle-specific transcription factor, MyoD (Seale et al., 2000; Shefer et al., 2006). A decline in Pax7 along with myogenin induction mark myoblasts that have entered into the differentiation phase. Emergence of cells that express Pax7, but not MyoD, in cultures derived from satellite cells may demarcate a self-renewing population (Halevy et al., 2004; Zammit et al., 2004). Quiescent satellite cells were also shown to express β-galactosidase (β-gal) when one of the Myf5 alleles was modified to direct lacZ expression in Myf5nlacZ/+ mice (Beauchamp et al., 2000). However, thus far, endogenous Myf5 expression has not been determined in quiescent satellite cells.
Cell culture and in vivo studies that identified distinctions between progeny of satellite cells from different muscles have raised the possibility that satellite cells do not necessarily encompass a homogenous population of progenitors (Zammit et al., 2006). However, these types of studies provide only indirect evidence for satellite cell heterogeneity as the progenitors per se were not analyzed. For example, differentiated satellite cell progeny derived from different muscle types were shown to express unique isoforms of sarcomeric myosins, which suggested fiber-type specificity of satellite cells (Dusterhoft and Pette, 1993; Hoh and Hughes, 1991; Kalhovde et al., 2005; Rosenblatt et al., 1996). Different proliferative patterns of myogenic progenitors in some head compared to limb muscles were also reported and may reflect the distinct ontogeny of these muscle types (McLoon et al., 2004; Noden and Francis-Wes, 2006; Pavlath et al., 1998). The modification of the mouse Pax3 locus to drive lacZ or GFP expression provided direct evidence for satellite cell heterogeneity; reporter expression was prevalent in satellite cells of diaphragm and some trunk muscles, but marginal in hindlimb muscles (Relaix et al., 2006). However, the basis for this heterogeneity is unknown and cannot merely be attributed to muscle ontogeny or fiber-type composition.
Elucidating the satellite cell genetic blueprint and gaining further insight into their heterogeneity is critical for advancing our understanding of the functional changes in satellite cells during aging and disease. For example, the varying rates at which satellite cell numbers decline in extensor digitorum longus (EDL; fast-twitch) compared to soleus (slow-twitch) muscle during aging may reflect muscle-type dependent heterogeneity of resident satellite cells (Shefer et al., 2006). Likewise, the observed accelerated differentiation of satellite cell progeny from dystrophic (mdx) muscle may reflect inherent changes in the progenitors (Yablonka-Reuveni and Anderson, 2006). However, the possibility that satellite cell performance is influenced by specific characteristics of the host muscle rather than being an outcome of intrinsic differences in quiescent satellite cells cannot be excluded without detailed molecular analysis of the progenitors.
Pinpointing the intrinsic changes in satellite cells during conditions that lead to muscle deterioration requires efficient means for isolating satellite cells from multiple muscle types. Routinely, satellite cells are released from muscle bulk by enzymatic digestion (Yablonka-Reuveni, 2004). Our previous approach for characterizing gene expression in freshly isolated satellite cells relied on sub-fractionation of the cell isolate by Percoll density centrifugation in order to remove debris and non-myogenic cells from the cell preparation (Kastner et al., 2000). Flow cytometry based on the exclusion of cells that express certain surface antigens such as CD45, CD31, and Sca1, while including cells expressing other surface antigens yielded some success toward greater purity of putative satellite cells (Sherwood et al., 2004). However, sorting approaches are limited by the varying effects of proteolytic digestion on cell surface antigens, the varying quality of antibodies used to detect surface antigens, and overlapping antigen expression between satellite cells and other co-isolated cell types (i.e., interstitial cells). Sorting of viable cells by GFP expression was used recently to isolate satellite cells from Pax3GFP/+ mice. However, the limitation of this model is its restricted applicability to only muscles where satellite cells express the reporter gene such as the diaphragm (Montarras et al., 2005).
Here we present a novel approach for localization, isolation, and gene expression analysis of satellite cells from different muscle groups based on GFP expression driven by regulatory elements of the nestin (NES) gene. Nestin is an intermediate filament protein that was originally identified as a marker of neural progenitors. However, its expression was subsequently detected in a wide range of progenitor cells (Michalczyk and Ziman, 2005; Wiese et al., 2004). Nestin is also expressed by proliferating and differentiating progeny of satellite cells, but its expression in quiescent satellite cells has not been reported (Kachinsky et al., 1994; Shefer et al., 2004). Analysis of resident NES-GFP satellite cells demonstrates heterogeneity between genes expressed in limb and diaphragm muscles and contributes new insights regarding the molecular signature of satellite cells. Moreover, the decline in NES-GFP expression following satellite cell activation, and the reacquisition of transgene expression by non-dividing progeny sheds light on the unique state of the quiescent satellite cell and its self-renewal potential. The NES-GFP mouse provides the first model for direct isolation and characterization of satellite cells from different muscle groups regardless of their embryonic origin or fiber type composition.
Heterozygous male mice ranging in age from 1–2 months (juvenile) and 4–9 months (adult) were used in these studies. All mice were from colonies maintained at the University of Washington. The NES-GFP transgenic mouse line (C57BL/6 background) was previously described (Mignone et al., 2004). The Myf5nlacZ/+ mouse line (enriched Balb/C background; (Beauchamp et al., 2000; Shefer et al., 2006) was developed by Drs. Shahragim Tajbakhsh and Margaret Buckingham (Pasteur Institute) and was kindly provided by Dr. Michael Rudnicki (Ottawa Health Research Institute). NES-GFP mice were crossed with Myf5nlacZ/+ mice to generate double heterozygotes. The Sca1-GFP transgenic mouse line (C57BL/10 x CBA background; (Ma et al., 2002; Mitchell et al., 2005) was developed by Dr. Elaine Dzierzak (Erasmus University) and kindly provided by Dr. Grace Pavlath (Emory University). Animal care and experimental procedures were approved by the Institutional Animal Care and Use Committee at the University of Washington.
Single myofibers were isolated from the hindlimb EDL and soleus muscles after collagenase digestion and cultured individually in Matrigel-coated wells as previously described in (Shefer et al., 2006; Shefer et al., 2004; Shefer and Yablonka-Reuveni, 2005). After initial adherence, myofibers were either fixed (time 0) or supplemented with our standard myogenic growth medium (DMEM [high glucose, with L-glutamine, 110 mg/L sodium pyruvate, and piridoxine hydrochloride supplemented with 50U/ml penicillin and 50 mg/ml streptomycin; GIBCO-Invitrogen) containing 20% fetal bovine serum (Sigma-Aldrich), 10% horse serum (HyClone), 1% chicken embryo extract prepared from whole 10-day old embryos)] for the indicated periods before fixation. This growth medium supports both proliferation and spontaneous differentiation of myoblasts. To trace proliferating cells, medium was supplemented with 5 μM BrdU (Molecular Probes) for 18 hours before fixation. Clones of satellite cells were established from isolated myofibers as previously described (Shefer et al., 2006; Yablonka-Reuveni, 2004).
Populations of GFP positive and negative cells were isolated from pooled hindlimb muscles (tibialis anterior (TA), gastrocnemius, and EDL) or diaphragm. Muscles were removed from 2–4 month old NES-GFP, Sca1-GFP, or wildtype control mice, and digested with Pronase to release individual cells as reported previously (Kastner et al., 2000; Shefer et al., 2006). Harvested cells (one mouse was used for each preparation) were resuspended in complete medium containing 10 μM Hoechst 33342 (Calbiochem) and incubated for 30 min at 37°C prior to sorting. An Influx Cell Sorter (Cytopeia Incorporated) with UV (351–364nm) and 488 nm argon lasers was used to sort viable GFP-expressing cells from total viable cells. Gating of GFP positive events was set to at least 10 times the fluorescence intensity of negative events. Approximately the same numbers of negative cells were collected to match the yield of gated GFP expressing cells. Cells were harvested by low speed centrifugation and further analyzed in cell culture or processed immediately for RNA isolation.
Sorted GFP positive and negative cells were cultured at a density of 5x103 cells per well in 24-well dishes that were pre-coated with diluted Matrigel as previously described (Shefer et al., 2006). Cultures were supplemented with one of the following media: a) our standard myogenic growth medium (described above for myofiber cultures); b) medium supporting endothelial growth (Su et al., 2003) (DMEM containing 20% fetal bovine serum, 20mM HEPES, 1% nonessential amino acids (Cellgro), heparin (55U/mL, Sigma), endothelial growth supplement (100μg/mL, Sigma)); c) Serum-free medium used for maintaining neuronal progenitor growth (DFNB) (Brewer et al., 1993) (2 parts Neurobasal (Gibco), 1 part EmbryoMax DMEM (Cell and Molecular Technologies, Inc.) and 1 part F12 (Gibco) supplemented with 1% B27 (Sigma)). Media also contained antibiotics as used in the standard myogenic growth medium described above for myofiber cultures.
The following primary antibodies were used: anti-Pax7 (mouse IgG1, ascites fluid, Developmental Studies Hybridoma Bank [DSHB], 1:2000 dilution); anti-MyoD (mouse IgG1, clone 5.8A, BD Biosciences, 1:800); anti MyoD (rabbit IgG, Santa Cruz Biotechnology, 1:400); anti-β-galactosidase (mouse IgG2a, J1E7 supernatant, DSHB, 1:16); anti-desmin (mouse IgG, clone D33, Dako, 1:200); anti-nestin (mouse IgG, Rat-401 supernatant, DSHB, 1:5); anti-sarcomeric myosin (mouse IgG2b, MF20 supernatant, DSHB, 1:20); anti-BrdU (rat IgG, Abcam, 1:1000); anti-GFP (rabbit IgG, Abcam, 1:10000; anti-CD31 (rat IgG, 1:100, Pharmingen); anti-laminin (rabbit IgG, Chemicon, 1:100). Secondary antibodies used were all produced in goat and conjugated with AlexaFluor (Molecular Probes, 1:1000 dilution) and included: anti-mouse IgG1 -488, -568, anti-mouse IgG2a -568; anti-rabbit IgG -488, -568, -647; anti-rat IgG -350, -568.
Freshly isolated and cultured myofibers, and primary cultures were fixed with 4% paraformaldehyde (containing 1% sucrose and phosphate buffered), permeabilized, blocked, and double immunolabeled with DAPI staining as performed previously (Shefer et al., 2004; Shefer and Yablonka-Reuveni, 2005). For detection of nuclei that incorporated BrdU, cultures were fixed, permeabilized, and blocked as above. They were then treated with 4N HCl for 7 min and washed thoroughly in TBS before incubation with anti-BrdU. Blue (AlexaFluor 350) secondary antibody was used for detection of anti-BrdU; DAPI staining was omitted. Acid treatment required for BrdU immunostaining can reduce GFP fluorescence intensity, and therefore in this case GFP expression was detected with anti-GFP.
For immunoanalysis of muscle tissue, various individual muscles were fixed in 4% paraformaldhyde/1% sucrose solution for 2h at room temperature. Muscles were then immersed successively in 5%, 10% and 20% sucrose solutions, each for 30min, and then sunk overnight in 30% sucrose at 4°C. Muscle was then embedded in OCT, rapidly frozen in nitrogen cooled with isopentane to prepare 5–10 μM thick cryosections. Muscle sections were double immunolabeled and counterstained with DAPI following the protocol described above for cultured cells, only omitting the Triton X-100 permeabilization step. For detection of β-gal expression in sections from NES-GFP/Myf5nlacZ/+ mice, muscles were fixed with 2% paraformaldehyde to preserve enzymatic activity and processed as above. Sections were incubated with X-gal staining solution overnight at 37°C, and observed using bright field microscopy. Since X-gal staining can result in reduced GFP fluorescence intensity, in some instances GFP expression was detected with anti-GFP. Biotinylated isolectin B4 (Griffonia simplicifolia, Vector Laboratories, 1:400) was used for vasculature staining; Streptavidin-AlexaFluor 555 (Molecular Probes, 1:1000 dilution) was used for detection.
Observations were made with an inverted fluorescent microscope (Nikon eclipse, TE2000-S). Images were acquired with a Qimaging Retiga 1300i Fast 1394 monochrome CCD camera or with CoolSNAPES monochrome CCD camera. The CCD camera drive and color acquisition were controlled by MetaVue Imaging System (Universal Imaging Corporation). Composites of digitized images were assembled using Adobe Photoshop software. All images captured in far-red were converted to red for final overlays. Due to some “bleeding” from red to far-red channels, double labeling using these channels was only used to trace epitopes with different intra- and/or extracellular localizations.
Total RNA was isolated using RNeasy Micro kit (Qiagen) according to procedure for less than 1x105 cells. NanoDrop spectrophotometry (NanoDrop Technologies) was used to determine RNA quantity. Typical yields per each mouse were between 50 and 100 ng for both GFP positive or negative cell populations used for total RNA isolation. 50ng of total RNA from each cell population was then used for cDNA synthesis using iScript reverse transcriptase (BioRad) according to manufacturer’s protocol. PCR was performed using Hot Star Taq (Qiagen) in a 25 μL total volume using 2.5 μL of cDNA and 10 pmoles of forward and reverse primers per reaction. Cycling parameters were 95°C for 15min to activate the enzyme, 13–35 cycles (depending on abundance of target message) of 95°C for 30sec, 58°C for 30 sec, 72° for 1 min, with a final extension step of 72°C for 10 min. Number of amplification cycles were 13–15 for 18S, 25 for CD31, and 30–35 for all other genes. PCR products were loaded on 1.5% agarose gels containing Sybr Green (Molecular Probes). Primer set sequences (and product sizes) were: Pax3, CCT GGA ACC CAC GAC CAC GGT GTC / AAC GTC CAA GGC TTA CTT TG (183bp) (Tamaki et al., 2002); Pax7, GAA AGC CAA ACA CAG CAT CGA / ACC CTG ATG CAT GGT TGA TGG (466bp) (Tamaki et al., 2002); Myf5, CAG CCA AGA GTA GCA GCC TTC G / GTT CTT TCG GGA CCA GAC AGG G (440bp) (Kastner et al., 2000) ; MyoD, GGA GGA GCA CGC ACA CTT CT / CGC TGT AAT CCA TCA TGC CA (464 bp); Nestin, CGG GAG AGT CGC TTA GAG G / TTG AGG TGT GCC AGT TGC (221bp); Desmin, GTG GAG CGT GAC AAC CTG AT / ATG TTC TTA GCC GCG ATG GT (335bp); c-met, TCC AGA GCT GGT CCA AGC AGT / TCT GGC AAG ACC GAA ATC AGC (505bp); Brn2, ACA GCA TCA ACA GCA ACA GC / GCT CCA GGT CGT CTG AGG TC (443bp); Sox2, ATG GGC TCT GTG GTC AAG TC / TTG GAT GGG ATT GGT GGT (369bp); Sox8, GTC CTG CGT GGC AAC CTT GG / GCC CAC ACC ATG AAG GCA TTC (277bp); Sox9, ATG ACC GAC GAG CAG GAG / CCG TTC TTC ACC GAC TTC C (529bp); CD31, AGG AGT CAG AAC CCA TCA GG / GCT ACT GGC TTT GGA GAT ACG (299bp); GFP, CTG GTC GAG CTG GAC GGC GAC G / CAC GAA CTC CAG CAG GAC CATG (629bp); 18S, ACC TGG TTG ATC CTG CCA GTA G / CGA TCG GCC CGA GGT TAT CTA (316bp).
We recently demonstrated that progeny of satellite cells express nestin (Shefer et al., 2004) and hypothesized that NES-GFP expression could provide a means for distinguishing proliferating myoblasts from their quiescent progenitors. Satellite cells and their progeny were monitored in isolated myofiber cultures from EDL and soleus muscles of young and adult NES-GFP mice. Unexpectedly, the satellite cells themselves, each situated on the myofiber, exhibited intense GFP fluorescence. When investigating myofibers from NES-GFP/Myf5nlacZ/+ mice, the majority of GFP+ satellite cells were also positive for β-gal (Figs. 1A-A” and B-B”) or Pax7 (Figs. 1C-C” and D-D”). Many satellite cells exhibited short membrane extensions of various appearances (Figs. 1A’–D’), although the cell nucleus occupied the vast majority of the cytoplasm.
We quantified the number of satellite cells within their niche in individual myofibers from NES-GFP mice based on GFP and Pax7 expression (Fig. 2A, Table 1). Regardless of age, the vast majority of satellite cells were positive for both GFP and Pax7. However, myofibers from juvenile (1–2 month old) mice displayed a greater number of single-labeled cells (GFP or Pax7 only) compared to myofibers from adult (4–10 month old) mice. As shown in Table 1, the proportion of double-labeled, GFP+/Pax7+ satellite cells was about 90 and 98% in myofibers from juvenile and adult mice, respectively. Furthermore, results showed a higher average number of satellite cells per myofiber in juvenile mice than in adults, and a wider range of satellite cell numbers per soleus myofibers, in accordance with our previous report (Shefer et al., 2006). Satellite cells in EDL myofibers from adult NES-GFP/Myf5nlacZ/+ mice exhibited a greater agreement between Pax7 and GFP expression (94%) than between β-gal and GFP (80%) (Fig. 2B, Table 1). Likewise, littermates from the NES-GFP x Myf5nlacZ/+ breeding pairs that carried only the Myf5nlacZ allele demonstrated a reduced agreement between β-gal and Pax7 expression (data not shown). Consistent among all mouse models used here, the number of satellite cells that expressed Pax7 or GFP was always greater than that which expressed β-gal. No β-gal+/GFP− or β-gal+/Pax7− cells were recorded (Fig. 2B, Table 1). Even in a larger scale analysis of 89 myofibers from four adult Myf5nlacZ/+ mice, only one β-gal+/Pax7− cell was noted per 546 Pax7+ satellite cells recorded. Taken together, these data demonstrate that NES-GFP expression provides an excellent marker for resident satellite cells, and indicates some molecular heterogeneity within satellite cells from juvenile and adult muscles.
Satellite cells in freshly isolated myofibers contained intense NES-GFP fluorescence, but GFP was no longer detected by day 4 in culture even when myofibers were reacted with an anti-GFP antibody. To determine more specifically if the decline in NES-GFP intensity correlated with satellite cell proliferation, cultures were pulsed with BrdU for 18 hours in order to identify cells that enter S-phase during this initial time period (Figs. 3A-A”). The majority of the cells entered proliferation between 24–48 hours in myofiber cultures (Figs. 3B-B”). GFP fluorescence intensity remained strong throughout the first day in culture even for cells that had entered the proliferative phase (Figs. 3A-A”’). However, after 2 days in culture, GFP intensity decreased in proliferating, MyoD+ progeny of satellite cells that remained on the fiber (Figs. 3B-B”’) and in those which had emanated from the myofiber (Figs. 3C-C”’). After 4 days in culture, GFP was undetectable. Primary myoblast cultures also exhibited this same decrease in GFP fluorescence (data not shown).
We investigated limb, head, and trunk muscles for the expression of NES-GFP by satellite cells to evaluate if the transgene is expressed in vivo regardless of host muscle type. TA, EDL, quadriceps, tongue, masseter, extraocular, diaphragm, and intercostal muscles were removed for histological examination from juvenile and adult NES-GFP/Myf5nlacZ/+ and NES-GFP mice. In all muscles examined, as depicted in EDL and tongue (Figs. 4A–C’), satellite cells were positive for NES-GFP. Satellite cells were identified by GFP fluorescence combined with Myf5-lacZ expression (detected by X-gal reaction) and location beneath the myofiber basal lamina (detected by laminin immunolabeling). These findings are in accordance with our observations in isolated myofibers. As shown in EDL muscle, GFP+ satellite cells were associated with a range of myofiber sizes (Figs. 4A-A” and B-B”), and this range likely reflected different fiber types.
We also observed NES-GFP fluorescence in gross muscle bulk during digestions to release either single myofibers or single cells. GFP expression was detected throughout the muscle in structures that resembled dense capillary networks in which intensely expressing cell bodies had processes connected to one another (Figs. 5A and B, depicting EDL and diaphragm muscles, respectively). Immunolabeling of muscle cross sections for the endothelial cell marker CD31 (Figs. 5C-C”) and parallel staining with isolectin B4 (Figs. 5D-D”) demonstrated that the GFP expressing cells were endothelial cells, and confirmed that GFP expression was found in capillary networks situated in close contact with adjacent myofibers.
Previous studies demonstrated that stem cell antigen-1 (Sca1) was expressed in capillaries of muscle and other tissues, but not in satellite cells (Kotton et al., 2003; Zammit and Beauchamp, 2001). We hypothesized that the Sca1-GFP transgenic mouse could be used as a positive control for GFP expression within endothelial cells to compare to expression levels observed in capillaries of NES-GFP muscle. Indeed, networks of GFP+ capillaries were also evident in muscles from Sca1-GFP mice. Partially digested muscles (Figs. 5E–F) demonstrated Sca1-GFP+ microvaculature and sections labeled for CD31 (Figs. 5H-H”) and isolectin B4 (Figs. 4I-I”) further demonstrated expression of Sca1-GFP specifically by endothelial cells. The greater intensity of Sca1-driven GFP expression in the microvasculature compared to nestin-driven GFP enabled refined detection of endothelial cell structure that corroborated closely with GFP+ capillaries from NES-GFP mice. No GFP+ cells were observed beneath the myofiber lamina from Sca1-GFP muscle sections (Figs. 5G and G’), in contrast to the expression of GFP by satellite cells from NES-GFP mice (Figs. 1, ,44 and and5).5). Furthermore, out of 68 freshly isolated myofibers (prepared from EDL muscles of 2 adult Sca1-GFP mice), not one myofiber displayed GFP+ cells that could be categorized as satellite cells (data not shown). Thus, within the context of the muscles analyzed in the present study, Sca1-GFP transgene expression was displayed by endothelial cells and not by satellite cells.
We next asked if NES-GFP expression could be used as a means for direct isolation of satellite cells by FACS. Cells were released from pooled hindlimb muscles of NES-GFP mice by our standard protocol of enzymatic digestion followed by shearing of the muscle bulk to release myogenic progenitors. Cells were then sorted based on GFP intensity and viable Hoechst stain to discriminate GFP+ and GFP− cells from dead cells and debris (Figs. 6A and A’). Cells isolated from wildtype muscle served as negative control (Figs. 6B and B’). Based upon Hoechst and GFP gating, only a marginal population of GFP+ cells appeared to be in G2 (Figs. 6A’ and B’, gate R4).
To determine the potential of NES-GFP positive and negative cells, sorted cells were seeded in established media conducive to myoblast, endothelial, and neuronal growth. We reasoned that some non-myogenic cell types that express GFP (such as endothelium as demonstrated above by histology) could have possibly been co-isolated and grown with myoblasts derived from GFP+ satellite cells. Also, previous studies showed the presence of GFP+ neural progenitors in NES-GFP mice, although these were isolated from brain and have not been identified in muscle (Mignone et al., 2004). Regardless of medium type, cultures derived from GFP+ cells yielded highly enriched myogenic cultures as shown by Pax7 and MyoD expression and fusion into myotubes after 7 days in culture (Figs. 7A–C’). In contrast, the vast majority of GFP− cells did not contribute myogenic cells in the different media as shown by the lack of myotubes (Figs. 7D–F) and immunostaining for Pax7 and MyoD (data not shown). Interestingly, GFP+ cells seeded into neuronal growth medium rapidly fused into myotubes and individual myogenic cells were rare by day 7 in culture (Figs. 7C and C’). Overall, cultured GFP+ cells exclusively entered myogenesis regardless of the supplemented media.
To characterize gene expression in sorted GFP+ satellite cells, we collected positive and negative cells from hindlimb muscles for total RNA preparation. Since we observed strong expression of the Sca1-GFP transgene in the endothelium, we reasoned that sorted Sca1-GFP cells could be used as a control for determining the efficiency of endothelial cell recovery by our primary myogenic cell preparation protocol. Despite stronger GFP expression within capillaries in Sca1-GFP muscle than in NES-GFP muscle (see Fig. 5F versus 5B, respectively), the proportion of recovered Sca1-GFP+ cells from adult hindlimb muscle was somewhat lower than that of NES-GFP+ cells when using the same gating parameters (Fig. 8A). These sorted Sca1-GFP+ cells were used as a positive control for CD31 gene expression (Fig. 8B).
As shown by RT-PCR analysis, NES-GFP+ cells isolated from hindlimb muscles specifically expressed Pax3, Pax7, and Myf5, while desmin and c-met (HGF receptor) were expressed in both GFP positive and negative populations, albeit at higher levels in the GFP+ fraction (Fig. 8B). Very faint MyoD expression was also detected in NES-GFP+ cells (not seen at amplification cycle shown in Fig. 8B), which is in accordance with infrequent MyoD+ cells (i.e., activated satellite cells) on freshly isolated myofibers (data not shown). NES-GFP+ cells sorted from the diaphragm (Fig. 8C) demonstrated higher expression of Pax3 relative to Pax7 and Myf5 that displayed similar expression levels to those observed in GFP+ cells from hindlimb muscles. Low-level expression of Pax7 in GFP− cells from NES-GFP muscle is in agreement with our identification of some Pax7+/GFP− satellite cells in single myofibers (Fig. 2).
In the Sca1-GFP sort, Pax3, Pax7, and Myf5 transcripts were expressed only in GFP− cells (Fig. 8B). This further demonstrated that satellite cells do not express Sca1-GFP. Additionally, in the Sca1-GFP sort, CD31 expression was localized strictly to GFP+ cells (Fig. 8B). Taken together with Sca1-GFP expression in endothelial cells by histological examination (Fig. 5), CD31 expression indicated that some endothelial cells survived from our isolation protocol. Notably, GFP+ and GFP− sorted cells from NES-GFP muscle expressed a similar level of CD31, but this level was far lower than CD31 expression observed in Sca1-GFP cells (Fig. 8B). Thus, some CD31 expressing cells (presumably endothelial cells) were present in the NES-GFP+ cell fraction, but their contribution was far lower compared to the Sca1-GFP+ cell population. Overall, our analysis demonstrated that the sorted NES-GFP+ population is enriched for satellite cells while the Sca1-GFP+ population is not.
Preceding reports demonstrated that the second intron of rat nestin gene present in the NES-GFP construct contained a neural enhancer element responsible for driving nestin expression in neural progenitors (Johansson et al., 2002; Tanaka et al., 2004; Zimmerman et al., 1994). However, our analysis of single fibers and muscle tissue demonstrated previously unreported NES-GFP expression in satellite cells. RT-PCR showed that endogenous nestin was expressed both in NES-GFP positive and negative cells (Figs. 8B and D). As predicted from sorted cells, the transgene was strongly expressed in the GFP+ population, with a basal mRNA expression level in the GFP− population (Fig. 8D). We reasoned that transcription factors previously shown to promote nestin expression by binding to the second intron enhancer region of the nestin gene may account for directing the nestin transgene expression in satellite cells. Specifically, levels of Brn2, a POU domain transcription factor, and Sox2, a sry-HMG box related gene, were analyzed in GFP positive and negative sorted cells. Brn2 was detected in NES-GFP− but not in NES-GFP+ cells, while Sox2 expression was below detection levels in both populations (Fig. 8D). We also analyzed the Sox genes 8 and 9, both reported to be expressed in satellite cells (Schmidt et al., 2003). Sox8 and Sox9 were expressed in both NES-GFP positive and negative sorted cells, indicating that these two transcription factors are not exclusive to satellite cells (Fig. 8D).
As shown in Fig. 3, NES-GFP fluorescence diminished in proliferating progeny of satellite cells in myofiber culture. Myofiber cultures typically develop an elaborated network of myotubes following 10–14 days, but also contained residual mononucleated Pax7+/MyoD− cells (Shefer et al., 2006). In view of our hypothesis that non-proliferating Pax7+/MyoD− cells represent the renewed progenitor population, we asked if they might also reacquire NES-GFP expression in long-term myofiber cultures and myogenic clones.
We observed reappearance of GFP expression following 3 weeks in culture. Cells were barely distinguishable at 2 weeks, but became more apparent as they developed stronger GFP fluorescence intensity and increased in number by 3–4 weeks in culture (Fig. 9). GFP+ cells uniformly displayed elongated cell bodies with thin extensions at each end and contained oblong-shaped nuclei that were clearly discernable from relatively round nuclei within myotubes. Such GFP+ cells were found in regions of well-developed, dense myotubes where they occupied spaces between the myotubes (Fig. 9). Only clones that grew to large sizes with a high density of myotubes developed GFP+ cells.
GFP+ cells ranged from just a few cells in clones to 20–50 in myofiber cultures and expressed Pax7 (Figs. 9A and A’). These Pax7+/GFP+ cells were not proliferating according to BrdU labeling with an 18h pulse (data not shown; less than 0.5% of the cells were BrdU+ in 21 day old cultures and none of the GFP+ cells incorporated BrdU). These cells typically did not express MyoD (Figs. 9B and B’), but in some cases low levels were detected in faint GFP+ cells (Fig. 9B, arrow). Furthermore, GFP+ cells did not express the differentiation markers, myogenin (Figs. 9C-C’) or sarcomeric myosin (Figs. 9D-D”). Cultures of isolated myofibers and clones from Myf5nlacZ/+ mice also show expression of β-gal within these oblong-shaped nuclei (data not shown). These NES-GFP+ cells expressed the intermediate filament proteins nestin and desmin. Immunostaining for nestin showed that it was typically concentrated around one side of the nucleus (Figs. 9E-E’), while desmin was more uniformly distributed throughout the cell (Figs. 9F-F’).
Here we investigated quiescent satellite cells and their self-renewing population, which we were able to identify in live myogenic cultures based on NES-GFP expression. In addition, gene expression analysis of satellite cells following their isolation and in culture provided new insight regarding transcriptional activity and molecular heterogeneity of quiescent satellite cells.
Analysis of satellite cells in isolated myofibers from NES-GFP mice for Pax7 expression demonstrated some intramuscular variation between satellite cells. Most satellite cells coexpressed GFP and Pax7, while some were only positive for Pax7 or GFP. The number of these single-labeled satellite cells was higher in juvenile myofibers, and thus, the coexpression of Pax7 and NES-GFP may reflect satellite cell maturation.
The present study sheds light on Myf5 expression by quiescent satellite cells. First, RT-PCR analysis demonstrated a relatively low-level expression of Pax7 and no Myf5 expression in the sorted GFP− population, indicating the presence of some Pax7+/Myf5− cells in the GFP− fraction. This Pax7+/Myf5−/GFP− sub-population is validated by the observation that all satellite cells positive for β-gal are also GFP+ in NES-GFP/Myf5nlacZ/+ myofibers. Second, the immunolabeling analysis demonstrated that 15–20% of the GFP+ satellite cells were β-gal− in myofibers from NES-GFP/Myf5nlacZ/+ mice. This indicates the presence of a dominant Myf5+ population and a minor Myf5− population within the GFP+ satellite cells. Lastly, when analyzing satellite cells in myofibers from the Myf5nlacZ/+ mouse strain, only 1 Pax7−/β-gal+ cell was detected compared to 546 Pax7+ cells recorded; of these Pax7+ cells, 6% wereβ-gal−. Based on this observation that essentially all β-gal+ cells were Pax7+ in the Myf5nlacZ/+ mouse, we infer that all β-gal+/GFP+ cells observed in NES-GFP/Myf5nlacZ/+ myofibers also express Pax7.
Collectively, this study validated the presence of a dominant population that expresses all three genes examined (i.e., Pax7+/Myf5+/GFP+ cells) and a minor population that express Pax7, but not Myf5, and in which GFP expression is not a uniform property of all cells (i.e., Pax7+/Myf5−/GFP±). It is conceivable that this phenotypic range represents the dynamic state of the satellite cells even within the presumably quiescent population. Satellite cells in Myf5nlacz/+ myofibers also exhibited some variation based on coexpression of β-gal, M-cadherin and CD34 (Beauchamp et al., 2000). Intramuscular heterogeneity among satellite cells may be reflective of different compartments cued for renewal or differentiation. To investigate this hypothesis we are currently developing means for separating Myf5+ and Myf5− populations (based on β-gal expression) in order to investigate the possible role of Myf5 in satellite cell renewal.
The cell isolation and sorting approach described in this study resulted in enrichment of satellite cells as shown by the far higher expression levels of recognized markers of satellite cells in the GFP+ versus GFP− populations. Furthermore, sorted NES-GFP+ cells gave rise exclusively to myogenic cells when cultured in different cell culture media that support the growth of non-myogenic cell types. Based on further immunohistological characterization and RT-PCR for CD31 expression, we concluded that some endothelial cells within the muscle vasculature also express the NES-GFP transgene. This is in agreement with other studies that utilized the same transgenic mouse founder (Amoh et al., 2005). However, the contribution of surviving endothelial cells to the sorted NES-GFP+ population was marginal as shown by a much lower CD31 expression level in NES-GFP+ compared to Sca1-GFP+ cells. Furthermore, our finding that CD31 was exclusively expressed in the Sca1-GFP+ population, while it was expressed in both NES-GFP+ and NES-GFP− populations, suggested that fewer endothelial cells express NES-GFP compared to Sca1-GFP. Sca1-GFP+ cells served as an excellent control for isolation of endothelial cells based on our observation of strong Sca1-GFP fluorescence in capillaries, with absence of expression in satellite cells.
Despite some expression of CD31 in freshly isolated NES-GFP+ cells, the absence of endothelial cell growth in cultures from NES-GFP+ cells can potentially be due to the inability of sorted endothelial cells to survive following our isolation protocol. Preliminary studies of cultures from Sca1-GFP+ sorted cells demonstrated some non-myogenic cell growth, which is presumably the progeny of endothelial cells. Thus, it would be expected that at least some non-myogenic cells should be detected in NES-GFP+ cultures if the CD31 signal detected in sorted NES-GFP+ cells reflects endothelial cell contribution. However, it is also conceivable that CD31 expression in the sorted, NES-GFP+ population reflects expression within the myogenic cells themselves as shown recently in human fetal myoblasts (Cerletti et al., 2006). Earlier studies also identified other endothelial cell markers to be expressed by satellite cells (De Angelis et al., 1999; Beauchamp et al., 2000; Zammit and Beauchamp, 2001).
While it is generally accepted that satellite cells do not express Sca1, some studies have identified rare Sca1-expressing cells that can enter myogenesis (Gussoni et al., 1999; Mitchell et al., 2005; Qu-Petersen et al., 2002; Tamaki et al., 2002). In accordance with these reports, we observed a very low level of Myf5 expression in sorted Sca1-GFP+ cells at higher PCR amplification cycles (i.e., 35–40). Also, in agreement with Mitchell et al. (2005), we noted enhanced Sca1-GFP expression in some proliferating myoblasts emanating from isolated myofiber cultures; nevertheless, quiescent satellite cells were always Sca1-GFP negative.
Pax7 was expressed at comparable levels in NES-GFP+ cell populations from both hindlimb and diaphragm, whereas much higher levels of Pax3 were expressed in cells from the diaphragm. The observed low Pax3 expression in NES-GFP sorted cells from the hindlimb may represent a sub-population of satellite cells expressing this gene, while in the diaphragm Pax3-expressing satellite cells are most prevalent. These data, focusing on endogenous Pax3 gene expression, are in accordance with studies demonstrating that Pax3-driven lacZ or GFP expressing cells were abundant in diaphragm and not in hindlimb muscles (Montarras et al., 2005; Relaix et al., 2006).
It remains unclear what the significance of increased expression of Pax3 is within satellite cells. No differences in the proliferative or differentiation potential of satellite cells from hindlimb versus diaphragm muscles were identified by means of transplantation into host muscle (Montarras et al., 2005), or by immunofluorescence analysis of differentiation and fusion into myotubes in primary cell cultures (Yablonka-Reuveni et al., 1999). Different satellite cell origins such as distinct myotomal subdomains, various inductive events or environmental signals within resident muscles, proliferative history, and rates of recruitment are all factors that may contribute to the disparity of Pax3 expression. Higher Pax3 expression levels by some satellite cell populations could also reflect the evolutionary emergence of specialized muscle groups required to maintain essential functions (i.e., diaphragm needed to support breathing through lungs). Pax3 may play a role in priming satellite cells for ongoing repair of such constantly utilized muscles.
The expression of Myf5 by mitotically inactive progenitors has been under debate since the expression of β-gal in satellite cells of Myf5nlacZ/+ mice was not in accordance with the report that Myf5 is expressed only upon satellite cell activation (Cornelison and Wold, 1997). We demonstrate here that quiescent satellite cells isolated from hindlimb and diaphragm muscles express Myf5. Minimal expression of MyoD in NES-GFP+ sorted cells verified that Myf5 expression in such cells cannot be merely attributed to activated satellite cells. Myf5 gene expression by satellite cells is in excellent agreement with β-gal expression of Myf5nlacZ/+ mice. It is possible that quiescent satellite cells express lower levels of Myf5 compared to proliferating cells, such that the method used previously for gene expression analysis in quiescent satellite cells was not sensitive enough to identify the low abundance of message (Cornelison and Wold, 1997).
Expression of desmin at relatively high levels in NES-GFP+ cells, in parallel with Pax7 and Myf5, suggests that quiescent satellite cells express desmin. Early studies demonstrated that desmin is expressed upon myogenic differentiation in all species analyzed and in proliferating myoblasts of some species (Kaufman and Foster, 1988; Yablonka-Reuveni and Nameroff, 1990; Yablonka-Reuveni et al., 1999). However, desmin expression in quiescent satellite cells has not been previously reported. Desmin expression in the differentiated state is regulated by MyoD (Paulin and Li, 2004). However, desmin expression in quiescent satellite cells cannot be attributed to MyoD in view of the negligible expression of MyoD in sorted NES-GFP+ cells discussed above. Desmin expression was also evident in the sorted NES-GFP− population, albeit at a lower level than in NES-GFP+ cells. Most likely this expression in NES-GFP− cells was contributed by vascular smooth muscle cells that are typically released from skeletal muscle during enzymatic digestion and which also express desmin (Costa et al., 2004; Yablonka-Reuveni et al., 1998).
The expression of desmin and nestin in sorted NES-GFP+ as shown by RT-PCR, and in re-emerging GFP+ cells in long-term cultures by immunostaining, suggest a role for these intermediate filaments in quiescent satellite cells. Decreased desmin expression in myoblasts from mice lacking lamin A/C (Frock et al., 2006), and our observation of “polarized” expression of nestin to one side of the nucleus in NES-GFP+ cells may suggest possible interactions between intermediate filaments and nuclear structure (Banwell, 2001). Previous reports with other cells types also showed nonuniform localization of nestin proein (Michalczyk and Ziman, 2005; Sjoberg et al., 1994; Wroblewski et al.,1997). Interestingly, the renewing satellite cells have distinct nuclear morphology compared to nuclei within myotubes. We are currently pursuing the potential structural requirements that may lead to this morphology as NES-GFP+ cells leave and re-enter quiescence.
We also demonstrated expression of c-met (HGF receptor) in satellite cells (GFP+ population) as anticipated (Charge and Rudnicki, 2004). However, in contrast to Pax7, c-met was expressed at a comparable level in the non-myogenic GFP− population. This confirmed our previous findings with rodent satellite cells, fractioned by Percoll density centrifugation, where both myogenic and non-myogenic cell fractions expressed c-met (Kastner et al., 2000). Therefore, while c-met expression marks satellite cells within the context of the isolated myofiber, it cannot be considered as a satellite cell specific marker in the context of whole muscle. Additionally, we could not confirm restricted expression of Sox8 and Sox9 in satellite cells as previously reported (Schmidt et al., 2003) as the two transcription factors were expressed at similar levels in both NES-GFP positive and negative populations.
The nestin gene contains enhancer elements within the first and second introns that where shown to drive selective expression to muscle, endothelial, and neuronal lineages (Aihara et al., 2004; Mignone et al., 2004; Zimmerman et al., 1994). However, the intense NES-GFP fluorescence observed in satellite cells was unanticipated. First, whereas progeny of satellite cells were shown to express nestin, such expression was not apparent in resident satellite cells (Kachinsky et al., 1994; Shefer et al., 2004; Sjoberg et al., 1994; Vaittinen et al., 2001). Second, the regulatory elements driving the NES-GFP transgene in the mouse line used in this study were shown previously to be neural-specific, with strong expression particularly in the subventricular zone and dentate gyrus (Mignone et al., 2004).
Specific and conserved binding sequences for the POU domain transcription factor Brn2 is positioned adjacent to its potential partner, the transcription factor Sox2, in the nestin neural enhancer (Johansson et al., 2002; Tanaka et al., 2004). We therefore hypothesized that these may participate in NES-GFP expression in satellite cells. However, our RT-PCR results did not support a role for Brn2 and Sox2 in regulating NES-GFP expression by satellite cells as expression levels of these transcription factors were below detection. It also seemed reasonable to assume that Sox8 and Sox9, which are expressed in satellite cells at relatively high levels, may contribute to NES-GFP expression. Nevertheless, as both Sox8 and Sox9 are also expressed in NES-GFP− cells, additional (or altogether different) factors may be required to enhance NES-GFP transgene expression in satellite cells. Such factors may be commonly expressed by a variety of other adult progenitors shown to express NES-GFP (Amoh et al., 2005; Davidoff et al., 2004; Gleiberman et al., 2005; Li et al., 2003).
As shown here, NES-GFP fluorescence was lost in cultures derived from satellite cells. This decline in NES-GFP expression compared to persistent endogenous nestin expression in myoblasts and myotubes indicates that the regulation of the NES-GFP transgene operates differently than the endogenous nestin promoter. Intriguingly, strong GFP expression reappeared in some Pax7+ mononucleated cells following 3 weeks in culture. These GFP+ cells were non-proliferating, negative for MyoD, and did not express markers of myogenic differentiation. The same GFP+/Pax7+ cells were also detected by 10–14 days in culture in high density primary cultures of sorted NES-GFP+ cells, which was very similar to dense myofiber cultures containing well developed myotubes (data not shown). The incompatibility of GFP expression with MyoD may suggest that activators of MyoD expression also function to diminish transcription from the NES-GFP transgene. Alternatively, MyoD may suppress NES-GFP expression directly through binding to putative consensus sites that appear to be present in the transgene. Altogether, Pax7+/MyoD− cells that reacquire the quiescent satellite cell phenotype also regain the correct transcriptional machinery to express NES-GFP transgene.
There has been much debate regarding how the satellite cell pool is renewed. Some studies proposed that the circulation and interstitial muscle cells (i.e., non-satellite cell sources) may contribute to replenishment of satellite cells (Zammit et al., 2006). However, single myofibers from donor muscle transplanted into injured host muscle clearly established that satellite cells are able to contribute to both myofiber repair and self-renewal (Collins et al., 2005). Nevertheless, the mechanisms that control satellite cell self-renewal have remained unknown. We suggest that the regain of NES-GFP expression levels detected in culture, is indicative of a “maturation” process toward satellite cell self-renewal. We only observed such GFP+ cells in myofiber and clonal cultures containing dense myotubes. Thus, the environment generated by myotubes appears to provide for the renewal of NES-GFP+ cells. This may reflect the in vivo process whereby growing and regenerating myofibers direct myoblasts to enter the satellite cell compartment. According to this model, the niche may direct some myoblasts to become new satellite cells. Quiescent satellite cells, even before entering proliferation, may already be specified toward renewal, differentiation, or both. The molecular heterogeneity among satellite cells within and between muscles could be reflective of such a specification process. Clonal variation with respect to robust myotube formation and parallel emergence of GFP+ cells suggests that not all individual satellite cells can contribute to the renewal process. Our hypothesis that the environment generated by myotubes directs myoblasts toward satellite cell renewal does not preclude the contribution of intracellular mechanisms, such as differences in proliferative rates (Schultz, 1996), or asymmetric cell division (Shinin et al., 2006), in regulating satellite cell renewal.
We are presently developing means for isolating cells that regain NES-GFP to further study their function. This is a challenging undertaking because of the relative small number of NES-GFP cells that emerge from long-term cultures. NES-GFP expression enables live imaging of the satellite cell renewal pathway from a single cell and permits investigation of how this process is controlled by both intrinsic and extrinsic factors. By investigating myogenic clones from growing, adult and aging muscle of NES-GFP mice we may determine if the proportion of satellite cells that can give rise to NES-GFP cells is influenced by age. Such a decline in renewal potential may contribute to the diminishing number of satellite cells with age (Shefer et al. 2006). Overall, unraveling the molecular basis for NES-GFP expression will lead to a better understanding of unique transcriptional networks active within quiescent satellite cells.
We are grateful to T. Pham and P. Rabinovitch for their valuable assistance with cell sorting (performed at the core facility of the University of Washington Nathan Shock Center of Excellence) and to our team member, I. Kirillova, for her important input. We also thank M. Rudnicki (Ottawa Health Research Institute) and S. Tajbakhsh / M. Buckingham (Pasteur Institute) for the Myf5nlacZ/+ mouse strain and to G. Pavlath (Emory University) and E. Dzierzak (Erasmus University) for the Sca1-GFP mouse strain. This work was supported by grants to Z.Y.R. from the National Institute on Aging (AG021566 and AG013798) and the USDA Cooperative State Research, Education and Extension Service (NRI, 99-35206-7934). K.D. was supported by the Genetic Approaches to Aging Training Program. G.E. was supported by the Ira Hazan Fund and the Seraph Foundation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.