Transcriptional and functional differences in stem cell populations isolated from extraocular and limb muscles

Abstract

The extraocular muscles (EOMs) are a distinct muscle group that displays an array of unique contractile, structural, and regenerative properties. They also have differential sensitivity to certain diseases and are enigmatically spared in Duchenne muscular dystrophy (DMD). The EOMs are so distinct from other skeletal muscles that the term “allotype” has been coined to highlight EOM group-specific properties. We hypothesized that increased and distinct stem cells may underlie the continual myogenesis noted in EOM. The side population (SP) stem cells were isolated and studied. EOMs had 15× higher SP cell content compared with limb muscles. Expression profiling revealed 348 transcripts that define the EOM-SP transcriptome. Over 92% of transcripts were SP specific, because they were absent in previous whole muscle microarray studies. Cultured EOM-SP cells revealed superior in vitro proliferative capacity. Finally, assays of the committed progenitors or satellite cells performed on myofibers isolated from EOM and limb muscles independently validated the increased proliferative capacity of these muscles. We suggest a model in which unique EOM stem cells contribute to the continual myogenesis noted in EOM and consistent with a role for their sparing in DMD. We believe the greater numbers of stem cells, their unique transcriptome, the greater proliferative capacity of EOM stem cells, and the greater number of satellite cells also offer clues for novel cell-based therapeutic strategies.

skeletal muscles comprise ∼40% of the total adult human body mass and are responsible for a number of functions including body support, force generation, and movement (20). To accomplish a diverse range of functions, specialized groups of muscles exist, exhibiting distinct properties (57). While the exact mechanism(s) pertaining to the origin of this diversity remains unclear, two main hypotheses have been proposed. The first emphasizes the instructive role played by environmental cues on a set of naive and noncommitted muscle precursors (reviewed in Ref. 51), while the second focuses on the role of lineage directives inherited by differentiating muscle cells from specialized progenitors in phenotypically distinct “allotypes” (25, 26) of muscles such as the extraocular muscles (EOMs) and limb muscles.

Indeed, the EOM allotype epitomizes functional diversity, because these specialized muscles are required to provide a variety of voluntary, saccadic, and reflex eye movements. They exhibit developmental, anatomic, metabolic, molecular, and functional properties that are at great variance from other (e.g., limb) skeletal muscles (5). Embryologically, the EOMs develop from prechordal mesoderm rather than somites (52) and are innervated by cranial nerves rather than spinal cord motoneurons. The EOM can be stimulated to twitch at extremely high frequencies, in contrast to limb muscle, in which similar frequencies of stimulation would lead to fusion and generation of tetanic contractions (4). Adult EOM can be multiply innervated and exhibit en grappe synapses, a pattern that is found in fetal rather than adult mammalian limb muscle (49). The expression pattern of myosin II isoforms in EOM is different from that seen in limb muscles; adult EOMs express an EOM-specific isoform (MYH13), continue to express the embryonic isoform MYH3, and can coexpress multiple myosin isoforms, in contrast to limb muscle (58, 59). EOM differs from limb muscle at the level of the transcriptome (15, 16, 47, 53) and the proteome (17, 18) as well. Both transcriptome-based analysis (15, 16) and cell labeling studies (40–42) have suggested that adult, uninjured EOMs have ongoing myogenesis/regeneration, in contrast to limb muscle, in which significant expression of regeneration markers or myonuclear addition would only be noted in the context of regeneration after injury or disease. But perhaps the diversity is most vividly exemplified by the differential response to diseases such as the congenital cranial dysinnervation disorders (12) and Duchenne muscular dystrophy (DMD).

DMD is the most common fatal X-linked neuromuscular disease. It is caused by mutations in the gene encoding dystrophin and affects 1 in 3,500 male newborns (24, 35). DMD is a progressive disease that presents clinically during the first decade of life, and patients become wheelchair dependent by their teens. Classical signs and symptoms include weakness and widespread muscle wasting, prominent calf enlargement, spinal deformities, and respiratory problems. Patients usually die in their thirties of respiratory and/or cardiac failure (11). While widespread muscle necrosis is pathognomonic of DMD, enigmatically, EOM function and histology are preserved in DMD patients (29, 32).

Several hypotheses have been proposed to explain EOM sparing in DMD; however, the exact mechanism(s) of sparing remain unclear (cf. discussion). On the basis of cell labeling (40–42) and expression profiling (15, 16) studies, we and others have proposed the hypothesis that the EOM may have a greater myogenic/regeneration capacity; however, how the greater myogenic capacity can be achieved has yet to be addressed. Adult skeletal muscle has postpartum growth and regenerative capacity due to the presence of muscle progenitors, including committed progenitors or satellite cells (SCs) that are capable of forming muscle (39, 75) as well as uncommitted progenitors or stem cells that are capable of forming a number of tissue types including muscle. A variety of cell populations, such as side population (SP) cells (22), pericytes (8), CD133⁺ progenitor cells (1), and mesoangioblasts (61), have all been suggested to form the muscle stem cell pool(s). However, it remains an open question whether the stem cell content, type, and myogenic capacity of EOMs are different compared with limb muscles.

To address this question, we quantified and characterized one stem cell population present in muscle, the SP cells, using a variety of cellular and molecular methods. In this study, we tested the hypothesis that an increased number of stem cells present in EOM may explain the continuous regeneration and subsequent sparing of these muscles in DMD.

MATERIALS AND METHODS

Animals.

Twelve- to sixteen-week-old C57BL/10ScSnJ, C57BL/10ScSn-Dmd mdx/J (Jackson Laboratory, Bar Harbor, ME), and C57BL/6-Tg(ACTB-EGFP)1 Osb/J GFP (gift from Dr. Alan Flake, Children's Hospital of Philadelphia) mice were used for SP experiments. Animal experiments were approved by the University of Pennsylvania Institutional Animal Care and Use Committee and conformed to the relevant regulatory standards.

SP cell preparations.

SP cells were fractionated from EOM and tibialis anterior (TA) muscles (n = 5 mice) by flow cytometry as previously described (44). Briefly, TA, quadriceps, and EOM were dissected, weighed, minced, and placed in skeletal muscle medium (SKM) [Ham's F-10 with 20% FBS in 0.5% penicillin-streptomycin (pen/strep)] at 4°C until digestion (all reagents from GIBCO-Invitrogen, Grand Island, NY, unless specified). Enzymatic digestion was performed with 1.2 U/ml dispase and 5 mg/ml collagenase IV (Worthington, Lakewood, NJ) for 45 min at 37°C and then quenched with SKM. Cell suspensions were filtered through 70-μm and 40-μm cell strainers, centrifuged (514 g) at 4°C, and resuspended in 3 ml of medium.

Red blood corpuscle/ cells were lysed with NH₄Cl, and the remaining cells were resuspended at 10⁶ cells/ml in PBS with 0.5% BSA (Sigma). Verapamil (100 μM, Sigma) was added to one aliquot of cells and incubated 5 min at 37°C before 5 μg/ml of Hoechst 33342 (Sigma) was added to all samples and incubated 60 min at 37°C. Propidium iodide (PI, Sigma; 2 μg/ml) was added to the samples before fluorescence-activated cell sorting (FACS) with a BD LSRII cell sorter (BD Biosciences, San Jose, CA). Flow cytometries were analyzed with BD CellQuest Pro version 5.2 (BD Biosciences). Details of SP cell preparation and gating strategy and yields are provided in Supplemental Fig. S1 and Supplemental Table S1, respectively.1 The amount of SP cells per gram of dry tissue was calculated with the following equation: SP cell number per gram = number of mononuclear cells per gram × (%viability/100) × (%SP cells/100).

Sca-1 and CD45 expression.

Cells were incubated with 2 μg/10⁶ cells of primary or isotype control MAb (BD Biosciences Pharmingen) for 15 min immediately after Hoechst staining, then cells were washed and resuspended in 0.5 ml of cold PBS-BSA, and PI was added before FACS analysis.

SP cell microarray hybridization and analysis.

SP cell preparations were washed, pelleted, and resuspended in 20 μl of PBS and stored at −80°C until processing for microarray analysis according to manufacturer's instructions and previously described methods from our laboratory (6, 15). Total RNA was isolated with the RNeasy micro kit (Qiagen, Valencia, CA). The purity, concentration, and integrity of RNA were determined with a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE) and an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). RNAs were subjected to linear amplification with the GeneChip two-cycle target-labeling protocol (Affymetrix, Santa Clara, CA).

Six independently separated EOM and six TA SP cell preparations were used for microarray analysis with Affymetrix Mouse 430 version 2.0 GeneChip arrays. One EOM and two TA SP cell GeneChip hybridizations failed to meet quality control standards and were discarded from further analysis. Raw intensities for each probe set were stored in electronic formats with GeneChip Operating System version 1.1, and expression summaries were calculated with Microarray Suite version 5.0 (MAS5) algorithms. All data were normalized by GC-RMA algorithm with Genespring version 7.3.1 software (Agilent) and filtered separately to obtain at least one present call for each transcript in any of nine conditions by using MAS5 output. Statistical analysis was applied on 28,987 of 45,101 probe sets that passed the filtering. Differentially expressed genes were identified by applying the two-class unpaired data settings in Significant Microarray Analysis version 2.21 software at the level of 0% false discovery rate (FDR). Verification of the accuracy of the analysis of relationships between the samples was done by principal component analysis (PCA) and visualization of the relative expression level of each transcript with Pearson correlation-based hierarchical clustering of samples using Genespring version 7.3.1 software (Agilent).

Validation of gene expression by quantitative PCR.

Quantitative PCR (qPCR) was performed as previously described (6, 15) in order to validate differential expression of Dspg3, Fbln1, Mmp23, Cd36, and Utrn on three independent EOM and TA SP cell preparations. GAPDH expression was used for normalization. RNA from each sample was reverse transcribed into cDNA with oligo(dT) primers and amplified with an ABI 7900HT real-time PCR system (Applied Biosystems, Foster City, CA) and gene-specific primers (Idaho Technology, Salt Lake City, UT). Sequence Detection Software (ABI, Version 2.2) was used for analysis and the ΔΔC_t method (where C_t is threshold cycle) for computing gene expression levels as described previously (37). Primer sequences and PCR conditions are provided in Supplemental Table S2.

In vitro muscle formation assays.

SP cells were cultured in vitro to form myotubes with standard methods. Briefly, equal numbers of EOM and TA SP cells [from green fluorescent protein (GFP) mice] were resuspended in Rubinstein complete medium [69% DMEM, 17.3% M199, 9.7% horse serum, 0.97% chick embryo extract (SLI, Bolney, UK), 1.04% glutamine, 1.04% pen/strep, 1.04% Fungizone] and cultured alone or over a mdx myoblast feeding layer on glass-bottomed 35-mm petri dishes or eight-well culture chamber slides coated with Matrigel (BD Biosciences, Bedford, MA). The feeder layer (2.7 × 10⁴/cm² myoblasts) was plated 1 h before SP cells were plated. GFP⁺ cells and myotubes were counted with a Nikon Eclipse TE 2000-U at 48–72 h and 13–18 days after plating. Myotube formation was documented on 2-wk-old cultures stained with Hoechst 33342.

Statistical analysis.

GraphPad Prism version 4.0, (GraphPad Software, San Diego, CA) was used for statistical analysis, and a P value <0.05 was considered significant.

All primary microarray data are MIAME compliant, have been deposited in NCBI's Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/), and are accessible through GEO Series accession number GSE9294.

RESULTS

EOM contains a 15.7× greater number of SP cells per gram than limb muscle.

We evaluated the uncommitted stem cell compartment by quantifying the numbers of SP cells based on their ability to exclude Hoechst 33342 (21, 22, 44). This strategy allows the purification of living SP cells from a muscle mononuclear cell pool by FACS-based fractionation, without having to expose the cells to culture conditions. As shown in Fig. 1, we found that EOM (291,929 ± 267,346) contains 15 times more SP cells per gram compared with TA (22,799 ± 23,282; n = 7; P < 0.05). Over 95% of EOM-SP (96.62 ± 4.02) and TA-SP (97.06 ± 2.26) were CD45⁻ (n = 13).

Fig. 1.

Extraocular muscle (EOM) contains 15 times more side population (SP) cells than tibialis anterior (TA). Total mononuclear cells from EOM and TA were labeled with Hoechst 33342 in the presence and absence of verapamil. SP cell population was selected with a third log cutoff between main population and SP in the density plots (see Supplemental Fig. S1). Values are mean ± SD numbers of SP cells per gram of tissue in EOM compared with TA; n = 7. *P < 0.05.

Expression profiling of EOM vs. limb muscle SP cells.

To identify the expression profile of SP cells fractionated from EOM versus limb muscle we performed microarray-based screening using Affymetrix Mouse Genome 430 2.0 Array GeneChips. We performed hierarchical clustering of nine samples to determine overall similarities within and between the EOM-SP and TA-SP samples. Branch-length analysis of data (Fig. 2A, top, red and cyan lines) demonstrated that the five EOM-SP samples were more similar to each other than to TA-SP samples, and, vice versa, the four TA-SP samples were more similar to each other than to EOM-SP. Further analysis of the profiling with a FDR cutoff set at a highly stringent 0% revealed 348 differentially expressed transcripts (0.77% of 45,101 evaluated transcripts) in the profile, of which 229 were upregulated in EOM-SP and 119 upregulated in TA-SP. Figure 2A, bottom, shows the heat map representation of hierarchical clustering of the entire profile. The complete list of 348 transcripts is provided in Supplemental Table S3.

Fig. 2.

Microarray heat map, principal component analysis (PCA), and scatter graph. A: hierarchical clustering analysis of all EOM-SP and TA-SP samples. Unsupervised clustering in Genespring 7.2 version was used to analyze similarities among replicate samples across sorted EOM-SP vs. TA-SP cells. Similarities between samples are indicated by the length of the lines that are at top of each sample column. Relative expression levels of 348 statistically significant transcripts are shown as horizontal colored bars. Red color refers to higher and blue color refers to lower expression as shown on scale bar. Replicate sample names with purple and cyan color codes are indicated at bottom. B: PCA: 3-dimensional plot showing distribution of EOM-SP (red) and TA-SP (cyan) transcripts. The 3 axes correspond to the components with the highest variance between the groups. C: microarray scatter graph of genes differentially expressed between EOM-SP and TA-SP. Black, normalized expression values of 313 transcripts (205 upregulated in EOM and 108 in TA) on a logarithmic scale after a 2-fold cutoff between EOM-SP and TA-SP. Gray, transcripts that did not pass the cutoff. The names of the highest-expressed transcripts in EOM (Dspg3) and TA (CD36) are shown.

To evaluate the overall relationship between the multivariate profiles of SP cells isolated from EOM and TA, we performed a PCA (62). Figure 2B shows the distribution of EOM-SP and TA-SP chip data in three-dimensional space, where the three axes correspond to the three components with highest variability in our data set. As can be seen in Fig. 2B, EOM-SP chips (red) and TA-SP chips (cyan) cluster as groups on different coordinates, demonstrating that the biggest variance is due to the tissue from which the SP cells were isolated.

To identify and graphically depict all the genes that were differentially expressed at greater than twofold levels, we performed a scatter graph analysis after imposing the fold change cutoff. Even with this limitation, a majority of transcripts (313 of 348 or 89.9%) were found to be differentially expressed: 205 upregulated in EOM-SP and 108 upregulated in TA-SP, as shown in Fig. 2C, which shows the two transcripts (Dspg3 and Cd36) found to be differentially expressed with the greatest magnitude of difference in the profile.

To validate the transcriptome analysis we performed qPCR on five transcripts (3 upregulated in EOM-SP and 2 upregulated in TA-SP) on RNA extracted from three independent EOM-SP and TA-SP preparations. As shown in Table 1, qPCR validated differential expression of Dspg3, Fbln1, Mmp23, Cd36, and Utrn in SP cells fractionated from EOM and limb muscle. Furthermore, as shown in Table 1, the magnitude of change was consistent with levels of differential expression revealed by microarray analysis.

Table 1.

Validation of expression profiling

Affy ID	Gene Symbol	Microarray Fold	qPCR Fold	qPCR P Value
1421114_a_at	Dspg3	338.4	259.3	3.99916E-07
1439688_at	Fbln1	13.7	5.0	1.64621E-05
1417282_at	Mmp23	7.6	2.3	1.95926E-06
1450884_at	Cd36	−69.9	−26.0	3.09794E-06
1452222_at	Utrn	−2.7	−9.0	1.50034E-07

Five of 348 differentially expressed transcripts revealed by the microarray were independently validated by quantitative PCR (qPCR). Fold change of expression in the microarray and in the qPCR are shown.

Bioinformatic analysis of EOM and limb muscle SP cell transcriptomes.

We used different bioinformatic strategies to further analyze the function of the genes that were found to be differentially expressed in the EOM and limb muscle SP cell transcriptomes. The bioinformatic program DAVID (version 2.1) was used to functionally cluster the entire list of differentially expressed genes. Interestingly, DAVID-based analysis led to clustering of 93 genes (from the 348 transcripts submitted), and the largest number of genes were clustered in biological processes that are associated with muscle formation/development, such as cell adhesion, morphogenesis, organ development, cell differentiation, cell growth, and regulation of growth and tissue development (Fig. 3A and Supplemental Table S4). To address whether the differences in gene expression noted among the SP cell transcriptomes were due to the previously reported differences between EOM and limb muscle transcriptomes (15, 16, 47, 53) rather than SP cells, we compared the previous “EOM vs. limb muscle” significant gene list with the new list from the EOM-SP vs. limb-SP comparison; we converted the original rat transcripts into their mouse orthologs (total number 585), and then we intersected the 348 transcripts from the SP cell microarray. We observed that of the 348 differentially expressed genes (i.e., upregulated in EOM-SP and upregulated in TA-SP) in the SP microarray, 322 genes were SP specific and only 26 genes were present in the previous (whole muscle) microarray comparison (Fig. 3B, Supplemental Table S5), demonstrating that the vast majority of genes identified in this study reflected differences in the transcriptomes of EOM-SP and TA-SP cells rather than of the tissue from which they were fractionated.

Fig. 3.

Biological process (DAVID) and Venn diagram. A: biological processes of genes differentially expressed between EOM-SP vs. TA-SP. With DAVID 2.1, 93 of 322 genes that are available in DAVID database are grouped based on Gene Ontology (GO; http://www.geneontology.org). Each biological process is ranked based on its P value, indicating its overexpression. B: Venn diagram representing stem cell-specific (EOM-SP vs. TA-SP cells) differentially expressed transcripts against EOM vs. limb muscles (EOM vs. TA) differentially expressed transcripts with 26-transcript overlap (92% concordance, 8% discordance in expression levels).

Differences in myogenic potential of SP cells fractionated from EOM and limb muscles.

To ensure that the SP cells had myogenic potential and were capable of forming myotubes in vitro, we fractionated EOM-SP and TA-SP from GFP⁺ mice. Equal numbers of these were plated on a myoblast feeding layer made from mdx mice, and we evaluated them after 13 days in culture. To determine the relative myogenic potential in vitro, EOM-SP and TA-SP were cultured in parallel and the total number of GFP⁺ cells cultured per well was evaluated at two time points: early (48–72 h) and late (2 wk). We found that both EOM-SP and TA-SP were able to fuse and form multinucleated myotubes in vitro as shown in Supplemental Fig. S3. (In one case, fusion of EOM-SP-derived GFP⁺ cells to twitching myotubes was also observed, see Supplemental Movie S1; also see Supplemental Figs. S4–S6 for more examples of cultured EOM-SP and TA-SP cells.) As shown in Fig. 4B, at the early time point the number of GFP⁺ EOM-SP cells was higher than TA-SP (110.5 ± 16.3 vs. 35.0 ± 15.6; n = 2, P < 0.0001). Similarly, at the late time point (Fig. 4, A and C) the number of GFP⁺ EOM-SP cells was higher than TA-SP (528.3 ± 453.7 vs. 47 ± 25.36; n = 3, P < 0.05). As independent validation of the increased proliferative capacity of these muscles we have included Supplemental Fig. S2, which shows SC counts performed on myofibers isolated from EOM and limb muscles.

Fig. 4.

EOM-SP cells proliferate better in vitro than TA-SP cells. A: representative culture of EOM and TA green fluorescent protein-positive (GFP⁺) SP cells after 2 wk of culture over a myoblast feeding layer. Equal numbers of GFP⁺ SP cells were plated on a mdx-myoblast feeding layer. After 18 days, cells were counted. Left: EOM-SP. Right: TA-SP. From top to bottom: GFP, Hoechst, phase, and overlaid (Ov) images. B: a higher number of EOM-SP cells than TA-SP cells was present in culture 48–72 h after sorting. Cells were counted with a fluorescence microscope. Means ± SD are shown; n = 2. ***P < 0.0001. C: differences in the number of EOM and TA GFP⁺ cells in cultures increased over time. After 13–18 days, cells were counted with a fluorescence microscope. Means ± SD are shown; n = 3. *P < 0.05.

DISCUSSION

In adults, stem cells are known to reside in tissues undergoing rapid turnover such as the hematopoietic system, gut lining, and epidermis. Adult skeletal muscle is not associated with rapid cell turnover compared with tissues such as the gut; however, it has a limited ability to respond to physiological stimuli such as increased workload and pathophysiological insults such as trauma, disease, or toxins. SCs are critical for the ability to regenerate and repair muscle; however, uncommitted stem cell progenitors such as SP cells are also capable of considerable myogenesis.

While previous studies have compared uncommitted SP cells with main population cells (3, 46, 68) and SP cells from different tissues and/or organs (22, 45, 56), no previous study has compared the content of SP cells between different skeletal muscle groups. In this study we found that the number of SP cells per gram in EOM is 15 times the number in limb muscle. Transcriptome analysis using Affymetrix GeneChips revealed that 348 transcripts were found differentially expressed in EOM-SP vs. TA-SP cells out of 45,101 evaluated transcripts (which include >34,000 well-characterized mouse genes). Thus the vast majority (∼92%) of transcripts identified in this study reflected transcriptome level differences between SP cells obtained from different tissues rather than from the tissue from which they were isolated (15, 16, 53). However, a small fraction (∼8% or 26 of 348) of differentially expressed transcripts could potentially be ascribed to differences in EOM and limb muscle transcriptomes per se. This group included three forms of Gst (Gst m4, Mgst 1, and Gst a4) and two isoforms of UDP (33Galt1 and Galntl1) that were upregulated in EOM-SP cells. The presence of high levels of chemoprotective enzymes such as Gst and Udp offers an advantage in overcoming oxidative stress; resistance to such stress has been described in other stem cells (2, 63). Additionally, the bicoid class of homeodomain transcription factor Pitx-2 was found to be upregulated. Pitx-2 is an upstream activator of the myogenic regulatory factors Myf5 and MyoD that plays an important role in EOM and craniofacial muscle development. Indeed, mutations of Pitx-2 lead to Axenfeld-Rieger syndrome associated with craniofacial malformation and ocular and EOM dysgenesis (65).

Bioinformatic analysis of the transcriptomes revealed a number of interesting differences between the EOM and limb muscle stem cell transcriptomes. Genes found upregulated in the EOM transcriptome included Tgfbi, Fn1, Aebp1, Dpt, Tro, Alcam, and Col14a1 (23, 34, 69), which have been shown to be expressed by stem cells in different tissues. Biological process clustering with DAVID revealed differential expression of a group of 26 cell adhesion molecules, including the endothelial markers CD36 (10), Scarb1 (74), CD93 (76), Vwf (67), Myh9 (27), and Nrp2 (13), which were all downregulated in EOM-SP. Interestingly, CD36 (14), Scarb1(74), and CD93(76) have also been associated with fatty acid transmembrane transport and clearance of apoptotic cells in different populations of stem cells. The genes Nrp2, Kitl (9), Itga (73), and Dll1 (33) were downregulated in EOM-SP cells and have been previously associated with a proliferative stimulus for stem cells. Additionally, two tumor suppressor genes, Jup (71) and Stim2 (50, 60), were downregulated in the EOM-SP cells. Consistent with the increasingly important role ascribed to the interaction between stem cells and extracellular matrix in formation and maintenance of stem cell niches (64), as well as the role played by Tcf4, a downstream effector of Wnt-β-catenin signaling pathway in chicken limb bud muscle pattern formation (30), we found Dspg3, a small leucine-rich repeat protein greatly upregulated in EOM-SP.

Analysis with KEGG (http://www.genome.ad.jp/kegg/pathway.html) allowed identification of a number of candidate pathways related to Notch, Vegf, Follistatin, and BMP signaling as differentially regulated in the SP transcriptomes. Transcriptional evidence of Notch pathway inhibition in EOM-SP comes from the downregulation of DLL1, Notch 1, Nrarp, and Evi1 and upregulation of the inhibitor HDAC4. Previous studies (48, 66) have demonstrated that Notch inhibits muscle differentiation and blockade of Notch signaling either by g-secretase inhibition or by Numb overexpression causes myotube hypertrophy by recruiting reserve cells that do not normally fuse (33). Vegf, VEGFR1, and Cdc42 were found to be downregulated in EOM for reasons that are unclear since the EOM are highly vascular; however, this may reflect heterogeneity of different stem cell populations in muscle. FST, Fstl-1, Tmeff2, and HDAC4 (7, 19, 28, 36, 43) were upregulated, suggesting mechanisms by which EOM could have more efficient myogenesis. Bmp1, Bmpr1b, Htra1, Twsg1, and Rgmb genes were upregulated in EOM-SP cells and are interesting in the context of FST and myogenesis since the metalloprotease Bmp1 is known to cleave the myostatin propeptide and activate latent myostatin (72). Consistent with the transcriptome-level differences identified here, it has been reported that SP cells fractionated from skin and muscle differ in their capacity to restore dystrophin, suggesting that inherent differences exist even among the same stem cell populations isolated from different tissues (45). The limited numbers of surviving cells in these reports provide an important impetus toward efforts to identify and study myogenic progenitors from different tissue sources to increase the likelihood of making these approaches therapeutically effective in DMD. The differential expression of these genes may reflect mechanisms enabling EOM-SP cells to undertake the more efficient myogenesis previously noted in EOM (15, 16, 40–42). We found that EOM-SPs were able to proliferate more efficiently compared with TA-SPs, after equal numbers of them were cultured up to 2 wk in vitro, despite identical matrix/culture conditions. Together, these findings support a role of lineage directives inherited by differentiating muscle cells from specialized progenitors in phenotypically distinct “allotypes” such as the EOM.

From a disease perspective, several hypotheses have been proposed to explain the enigmatic EOM sparing in DMD, including smaller surface-to-volume ratio resulting in less mechanical stress (31), higher capacity to regulate intracellular calcium in EOM than other skeletal muscles (32), higher antioxidant capacity in EOM (55), overexpression of dystrophin-related protein (DRP)/utrophin (38, 54), increased elasticity due to differential expression of M bands (70), and increased myogenesis (15, 16, 40–42). However, independent tests of these hypotheses either have not been performed or have been equivocal, e.g., DRP/utrophin upregulation where utrophin upregulation has not been confirmed at the protein or mRNA level (15, 16, 32, 53). Our identification in EOM of a unique stem cell content with greater numbers of SP cells with a specific expression profile coupled with the greater capacity of these cells for in vitro myogenesis provides a model (Fig. 5) and support for the hypothesis that an increased stem cell content may underlie the efficient and continual myogenesis and therefore may contribute to sparing of adult EOM in DMD.

Fig. 5.

Model for the role of stem cells in EOM sparing in Duchenne muscular dystrophy (DMD). Flow chart summarizes findings from stem cell quantitation, microarray, and in vitro studies. Left: EOM. Right: TA. We believe that an increased and unique stem cell content underlies the continuous myogenesis observed in normal EOM muscle and may explain the EOM sparing in DMD.

GRANTS

This work was supported by National Institutes of Health Grants EY-013862 and AR-051696 to T. S. Khurana.