1 Department of Neurology, Cleveland, Ohio 44106
2 Department of Neurosciences, Cleveland, Ohio 44106
3 Department of Visual Sciences Research Center, Ohio 44106
4 Department of Comprehensive Cancer Center, Case Western Reserve University and University Hospitals of Cleveland, Cleveland, Ohio 44106
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
skeletal muscle; hindlimb; microarray; allotype
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The muscle allotype concept arose as a framework to account for the breadth of phenotypic diversity available to skeletal muscle (23, 24). Allotype identities appear to be resident in muscle precursor cell populations, but the expression of allotype-specific traits apparently is dependent upon interactions between precursor cells of specific lineage with motoneuron pools that provide appropriate activation patterns. Three allotypes were defined on the basis of their potential to express specialized myosins: masticatory (super fast myosin), extraocular muscle (EOM; Myh13), and limb (no allotype-specific myosins). Allotype heterogeneity is, however, not limited to myosin heavy chain content. EOM, for example, is further adapted due to its eye movement role. Some central concepts in muscle biology, such as the established fiber type classification schemes and M-line/creatine kinase system, do not apply to EOM (1, 53, 63). Instead, EOM comprises six allotype-specific fiber types, including two non-twitch, multiply innervated types, and expresses embryonic, neonatal, cardiac, and tissue-specific protein isoforms that are atypical of adult skeletal muscle (27, 30, 44, 53, 55, 71, 75, 81). Recent expression profiling studies have further shown that adult EOM is fundamentally distinct from the limb and masticatory muscle allotypes (12, 18, 36, 60); these data suggest that allotype specificity may be defined by more than simple differences in myosin heavy chain expression patterns.
Myogenic mechanisms underlying allotype specificity are poorly understood. In limb, distinct myoblast populations and regulatory pathways give rise to type I and II myofibers (9, 45, 73), and hypaxial and epaxial muscle precursors differentially activate regulatory genes (16, 17). Since the allotype concept highlights differences between craniofacial and spinal muscles, rostrocaudally distributed regulatory cascades may be mechanistic in muscle divergence (46, 51, 76). As yet, few transcription factors have been linked to muscle group identities (e.g., Lbx1-forelimb extensors, Mox2-appendicular muscle, En2 and MyoR/Tcf21-masticatory muscle, and Pitx2-EOM) (14, 19, 39, 40, 42, 72). Other than these data, there has been little research geared toward understanding the full scope of developmental processes behind skeletal muscle allotypes.
Allotype heterogeneity has significant consequences. Differential responses to inherited metabolic and neuromuscular diseases are seen both between and within muscle allotypes. Most myopathies target the limb allotype but have an unexplained predilection for proximal muscles, and the rarer distal myopathies exhibit their own distinctive patterns of muscle targeting. Likewise, muscular dystrophies produce allotype-related phenotypes not predictable by current knowledge of the localization and functions of disease gene products. Since many neuromuscular diseases are not fully penetrant, and targeted muscle groups vary widely, it may be impossible to fully understand disease mechanisms without in-depth knowledge of muscle group diversity and its impact on disease.
Because of its exceptional phenotype and disease responsiveness, EOM may provide insights into the breadth, causes, and consequences of muscle-group-specific identities. EOM resistance to muscular dystrophy and particular sensitivity to myasthenia gravis have been ascribed to constitutive, rather than adaptive, differences from the limb allotype (31, 53, 60, 65). Determination of the precise mechanisms underlying such exceptional patterns of sparing or involvement in neuromuscular disease may provide important clues to pathogenesis and identify new treatment strategies.
We have proposed that the highly specialized EOM allotype is a consequence of a novel myoblast lineage interacting with extrinsic factors during a postnatal critical period of development (5, 7, 12, 57). Here, DNA microarray was used to identify conserved and muscle-group-specific transcriptional patterns during the critical period of EOM development. These data represent an important first step in determining how genetic and epigenetic factors shape the differentiated muscle allotypes.
![]() |
MATERIALS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Electron microscopy.
Rats were perfused with physiological saline followed by 1% paraformaldehyde/2% glutaraldehyde fixative solution in 0.1 M phosphate buffer. EOMs were removed, postfixed in 4% glutaraldehyde fixative solution, followed by 1% osmium tetroxide in 0.1 M phosphate buffer, and processed into plastic resin following standard procedures (6). Some muscles were processed for visualization of neuromuscular junctions using acetylcholinesterase histochemistry. Sections were examined and photographed using a Zeiss model 10C electron microscope.
DNA microarray.
To minimize inter-litter/animal variability, muscles were pooled from multiple rats for each of three independent replicates/age/muscle group. EOM samples included all rectus and oblique muscles, while gastrocnemius and soleus muscles were pooled as representative of hindlimb muscle. Tissues were snap frozen in liquid N2 and stored at 80°C. cRNA was prepared for use on Affymetrix (Santa Clara, CA) RG-U34A arrays, as described (36, 6062, 65, 66). Briefly, RNA was extracted using TRIzol reagent (GIBCO-BRL; Invitrogen, Rockville, MD). Pellets were resuspended at 1 µg RNA/µl DEPC-treated water, and 8 µg was used in a reverse transcription reaction (SuperScript II; Life Technologies, Rockville, MD) to generate first-strand cDNA. Double-strand cDNA was synthesized and used in an in vitro transcription (IVT) reaction to generate biotinylated cRNA. Fragmented cRNA (15 µg) was used in a 300-µl hybridization cocktail containing herring sperm DNA and BSA as carrier molecules, spiked IVT controls, and buffering agents. A 200-µl aliquot of cocktail was hybridized to microarrays for 16 h at 45°C. For manufacturers standard posthybridization wash, double-stain, and scanning protocols, we used an Affymetrix GeneChip Fluidics Station 400 and Gene Array scanner.
Microarray data analysis.
Raw data from microarray scans were initially normalized and analyzed with Affymetrix Microarray Suite (MAS) 5.0. MAS evaluates sets of perfect match (PM) and mismatch (MM) probe sequences to obtain both hybridization signal values and present/absent calls for each transcript. The raw data series are housed in the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) database under series record accession number GSE903 (GSM13668 through GSM13712). We used the MAS filter only to exclude transcripts that were absent from all samples from further analysis [present calls for EOM and hindlimb were 42.6% (SD 3.7) and 39.6% (SD 4.3), respectively, of the 8,799 transcripts on U34A microarrays]. Microarray data from EOM and hindlimb then was normalized using the "robust multichip average" (RMA) algorithm (25) in ArrayAssist 2.0 (Iobion Informatics, La Jolla, CA). RMA executes a background adjustment, a quantile normalization, and a summation of individual probe set intensities using a log scale linear additive model for the log transform of background corrected/normalized PM probe intensities. RMA reportedly has higher sensitivity and specificity than either MAS 5.0 or dChip (25).
Cluster analysis of the EOM temporal series was performed with CAGED 1.1 (http://genomethods.org/caged/) (69). CAGED uses Bayesian "clustering by dynamics" to identify a statistical model of the most probable set of clusters of transcripts in a time series, without relying upon any predefined similarity threshold. Background corrected/normalized probe signal output from RMA was averaged for each transcript/age and then used as input to CAGED. To filter for transcripts differentially expressed in the EOM temporal series, fold difference threshold was set at greater than or equal to twofold. We then coded and combined the EOM and hindlimb data sets and repeated the CAGED analysis. This approach allowed determination of differentially expressed EOM transcripts that either 1) did not undergo a similar twofold change in hindlimb or 2) exhibited different temporal patterns of change in EOM and hindlimb muscle (i.e., fell into different CAGED clusters).
Self-organizing map (SOM) analysis was implemented in GeneSpring 5.0 (Silicon Genetics, Redwood City, CA), using the RMA transcript signal data from EOM and hindlimb muscle as input. For statistical comparisons of expression patterns, we used parametric testing (Welch t-test/Welch ANOVA, P 0.001), applying the Benjamini and Hochberg false discovery rate algorithm for multiple testing correction. SOM identified transcripts differing in overall temporal pattern between the EOM and hindlimb muscle series and arranged them so that similar patterns appeared in nearest neighbor clusters.
Affymetrix transcript annotations were replaced with official gene nomenclature using NCBI databases, and gene functions were assigned based upon gene ontology and other data in NCBI LocusLink, UniGene, and PubMed and Weizmann Institute of Science GeneCards (http://bioinfo.weizmann.ac.il/cards/).
Promoter analysis.
Here, we used the significance analysis of microarrays (SAM) algorithm to filter EOM and hindlimb microarray data for those transcripts that most clearly defined the EOM allotype. Of 117 transcripts identified by SAM, 57 were known genes (52 unique) with 3' sequence available from NCBI. RefSeq identifiers obtained for each gene from LocusLink were used to extract 500 bases of rat genomic DNA sequence upstream of transcription start sites. Match software then was used in conjunction with the TRANSFAC database (http://www.gene-regulation.com/) to locate putative transcription factor binding sites in genomic sequence of each of the 52 differentially expressed genes.
Real-time quantitative PCR.
The same samples used for microarray were used for real-time quantitative PCR (qPCR). One microgram of total RNA was reverse transcribed using oligo-dT primer (Invitrogen). One microliter of cDNA was diluted (1:6 to 1:10) and then used for qPCR. Primers used for qPCR were Myh3 (embryonic) (NM_012604), forward 5' GATGGTGGTCCATGAAAGTGA 3', reverse 5' AGGGGTTACGTGGAAATTAAGC 3'; Myh4 (IIB) (L13606), forward 5' TAAGTGAAGAGTAAGGCAGCTCTGA 3', reverse 5' GGATTAAATAGAATCACATGGGGAC 3'; Myh6 (-cardiac) (X15938), forward 5' ACACGAAGCGTGTCATCCAG 3', reverse 5' GGTCCCCTATGGCTGCAAT 3'; Myh7 (I or ß-cardiac) (X15939), forward 5' GAGTTAAATGCACTCAACGCCA 3', reverse 5' CCTGAAGCTCTTTGAGCTTCTT 3'; Myh8 (perinatal) (K02111), forward 5' CAAGTGGCTGAAGGAAAGGCA 3', reverse 5' AGTGGGAGAAAAGTAAACACGAGAG 3'; Myh13 (extraocular) (AF075250), forward 5' ATGTGGGAGGCCAGAAGAT 3', reverse 5' AGTCTCCCTCTGCTCTCCTGGA 3'. qPCR used the Roche LightCycler (Mannheim, Germany) with the LightCycle-FastStart DNA Master SYBR Green I kit, following the manufacturers protocol.
![]() |
RESULTS AND DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Postnatal morphogenesis of EOM.
Postnatal EOM morphogenesis was first characterized by electron microscopy to determine critical stages for transcriptional analysis. Unlike most neonatal skeletal muscles (33, 34, 41, 70), P0 EOMs were still composed of centrally nucleated myotubes with secondary myotubes tightly apposed to many primary myotubes (Fig. 1, A and G). Intramuscular axons were unmyelinated, and primitive neuromuscular junctions contacted only primary myotubes, activating secondary myotubes only indirectly through primary-secondary tight junctions (Fig. 1, D and G). By P7, the central nuclei characteristic of myotubes had migrated peripherally, separately innervated primary and secondary myofibers were evident, and intramuscular axons were thinly myelinated (Fig. 1, B and E). Subsequent maturation of sarcoplasmic reticulum and T-tubule networks demarcated the myofibrils, allowing recognition of the basic EOM singly (small myofibrils) and multiply innervated (large myofibrils) fiber morphologies by P14 (Fig. 1C). Intramuscular axons were more heavily myelinated by this stage, although neuromuscular junctions remained primitive in capping small myofibers until after P21 (Fig. 1, F and H). Microvascular content increased rapidly after P21. The two muscle layers and six myofiber types characteristic of adult EOM (see Ref. 53, 75) were recognizable by P28 (Fig. 1, I and J). Myofiber development between days P28 and P45 was largely restricted to increases in diameter and mitochondrial content.
|
|
|
Genes associated with transcription, cell signaling, cell cycle/cell death, cell surface/cell adhesion, cytoskeleton, extracellular matrix, and protein/nucleic acid metabolism were found at considerably higher frequency in group II (Fig. 3B), indicating progressive downregulation of these functions as myogenesis is completed. Only cluster II.1 contained transcripts with a precipitous decline after birth and plateau after P14 (e.g., Tacstd1, Lgals7, and Cd24, which function in cell surface/cell adhesion), although some other clusters showed this trend to a more modest extent (e.g., clusters II.2, II.3, and II.5). Of 28 cell cycle/death transcripts dynamically regulated in postnatal EOM (e.g., Ccnd2, Cdk4, and Ccnl), 21 were in clusters II.4, II.6, II.7, and II.8, which have very similar temporal patterns of gradual downregulation, and all but 3 of the genes in this functional class were in group II (only Ccng1 and two probes for Gadd45a, a positive regulator of apoptosis, were in group I). Correspondingly, these same four group II clusters also contained the majority (50 of 68 or 74%) of transcripts functioning in nucleic acid/protein structure and metabolism. Representative transcripts in this class play roles in DNA metabolism or packaging (e.g., Hmgb2, Top2a, Nap1l1, Prim1, and Apex), ribosomal structure (e.g., Rpl18 and Rps10), or RNA processing (e.g., Ptb and Sfrs10) that would be expected to parallel cell cycle/death functions. By contrast, protein metabolism transcripts were more likely to be upregulated or to exhibit little postnatal change (e.g., Stnl, Pcmt1, Eef2, and Eif4ebp1 in groups I or III). Finally, cytoskeletal (e.g., Tuba1, Actg2, and Krt118) and extracellular matrix (e.g., fibril forming, Col5a1, and nonfibril forming, Col12a1, collagens) genes were primarily distributed in group II clusters (95% and 80%, respectively), with the majority showing very gradual decreases in expression (55% in clusters II.4, II.5, and II.6). The principal fibril-forming collagens (types I and III) did not meet the criteria for differential regulation from days P0 to P45.
Energy metabolism transcripts were more frequently found in group I clusters (84% of category) (Fig. 3B), indicating induction as myotubes transition to functional myofibers. Clusters I.1, I.2, and I.3 had the most rapid increases in expression and leveled off after P14. The 15 known genes in these clusters include adult contractile protein isoforms (Myh4 and Myh13), channels/transporters (Pva, Atp1a2, Atp1b1, Slc2a4, and Fabp3), and genes encoding energy metabolism enzymes (Ckmt2, Acadm, and Got1). At least three of these are either EOM specific (Myh13) or expressed at higher levels in EOM than other skeletal muscles (Pva and Ckmt2) (10, 63). By contrast, a significant fraction (66%) of all dynamically regulated metabolism transcripts were distributed across a restricted set of clusters (clusters I.6, I.7, and I.8) with very gradual increases in expression level. The majority of enzymes in the glycolytic, tricarboxylic acid, respiratory, and ß-oxidation of fatty acid pathways were in these three clusters, suggesting tight coregulation of energetic mechanisms in developing EOM. Taken together, microarray findings correlated with morphological indicators of energetic status (mitochondria/microvasculature) that mature after P14 (see Figs. 1 and 2A).
Ion channel and transporter genes required by excitable tissues also were more likely to be found in group I (58% of category) (Fig. 3B). Most members of this class were not among the CAGED clusters that were rapidly up- or downregulated in postnatal EOM. Instead, muscle ion channels and binding proteins were modestly upregulated (e.g., two probes for Cacna1s in clusters I.6 and I.8 and S100a1 in cluster I.8) or showed transient spikes in expression (Kcnj11 and Vdac1 in cluster III.2). The expression patterns of these transcripts are consistent with the temporal appearance and elaboration of the T-tubule and sarcoplasmic reticulum networks and with reported postnatal changes in EOM calcium content and calcium-ATPase content and activity (58).
Few muscle-specific transcriptional regulators met the criteria for dynamic regulation in postnatal EOM, and all were downregulated (Myog in cluster II.6, Gata6 in cluster II.5, and Csrp3 in cluster III.1). Conservative microarray analytic tools, such as the RMA algorithm used here, tend to exclude genes expressed at very low levels. Two transcripts that are signal transducers for myoblast-to-myotube differentiation were induced (Mapk12 in cluster I.8 and Vamp5 in cluster I.6). Muscle structure and development transcripts were broadly distributed among groups I (n = 9), II (n = 5), and III (n = 6). EOM-specific (Myh13) and type IIB (Myh4) myosins were in cluster I.2, with rapid postnatal increases and plateau after P14. Transcripts representing the -cardiac (Myh6) and ß-cardiac (Myh7) myosins and sarcomeric structural components, Mybph and Myl3, were in clusters I.8 and III.3, with little postnatal increase in expression level. Contractile regulatory proteins, Tpm1, Tnnt2, and Tpm3, showed changes opposite to myosin heavy chain transcripts; these were found in group II clusters with slow repression after birth. Since the majority of muscle developmental regulators and contractile proteins did not meet the
2-fold threshold for dynamically regulated transcripts, postnatal elaboration of distinctive muscle traits must be considered as a gradual process with few sudden developmental shifts. Rapid upregulation of an EOM-specific trait, Myh13, contrasts with this pattern and supports the appearance of a strong inductive signal in EOM by P7. This concurs with other data on Myh13 expression (6) and may explain this transcripts sensitivity to postnatal visual/vestibular deprivation (5, 7).
EOM is uniquely spared in the dystrophin-glycoprotein complex-based muscular dystrophies (29, 32, 38, 59, 64, 67, 68). Although normal EOM expresses all known components of this complex (2, 37, 65), only two muscular dystrophy-related transcripts, Capn3 (3 different probes in clusters I.6 and I.8) and Cav3 (cluster III.1), met criteria for developmental regulation in postnatal EOM. EOM status in Capn3 and Cav3 knockout mice has not yet been assessed. Likewise, only two of many transcripts known to participate in neuromuscular junction formation and maintenance, Agrn (cluster II.7) and Chrne (cluster I.7), were dynamically regulated in postnatal EOM, despite concurrent changes in neuromuscular junction morphology. Transient expression behavior of myelin protein transcripts (Mbp and Mpz in cluster III.2) or myelin signaling/processing transcripts (Mapk12 and Ugt8 in clusters I.8 and III.2) coincided with intramuscular nerve development.
EOM exhibits higher capillary content than most skeletal muscles due to its highly oxidative energetics (75). Vascular signaling transcripts (Vegf in cluster I.6 and Egln3 in cluster I.9) closely paralleled the morphogenesis of EOM fiber types and postnatal increase in energy metabolism transcripts (which predominate in cluster I.6). By contrast, markers of differentiated endothelial cell and smooth muscle phenotypes exhibited an inverse relationship (Acta2, Actg2, and Tagln in cluster II.2, Csrp2 in cluster II.4, and Edg2 in cluster II.6), falling at approximately the same rate as the rise in metabolism transcripts.
The transcriptional profile identified here for postnatal EOM using the CAGED algorithm includes genes that met an induction/suppression threshold (2-fold) and therefore are most likely to contribute toward emergence of the novel phenotype. To compare morphophysiological and microarray data, gene expression profiles are shown graphically by functional category in Fig. 2B. Data are consistent with 1) the slow, progressive downturn of cellular processes associated with myoblast proliferation, cell-cell contact and fusion, and general cell growth that would be expected following completion of myofiber formation; 2) the completion of secondary myogenesis by
P7 and recognition of the major two classes of differentiated EOM fiber types, singly and multiply innervated, by P14; and 3) the induction of energy metabolism transcripts and ion channels and various transporters accompanying the initiation of myofiber function. The dynamic changes in gene expression patterns were nearly complete by P14 to P21, indicating that eyelid opening and the onset of purposeful fixation and targeting eye movements represent important landmarks in EOM maturation (see Fig. 2).
Overview of expression differences between postnatal EOM and hindlimb muscle.
To identify developmental features unique to EOM, we generated an expression profile for the limb allotype and compared these profiles using two strategies. Combined gastrocnemius and soleus muscles were selected for this comparison. They represent all traditional skeletal muscle fiber types, like EOM with an overall bias toward fast-twitch fibers [an estimate of the mass of muscle from each fiber type for gastrocnemius is 1% type I, 3% type IIA, 28% type IIX/D, and 68% type IIB; soleus is 86% type I, 6% type IIA, 8% type IIX/D, and 0% type IIB (15)]. First, we repeated the CAGED analysis with the combined EOM and hindlimb data to identify transcripts that either were dynamically regulated only in the EOM series or exhibited temporal pattern differences between the two muscle groups (i.e., appeared in different CAGED clusters). Second, SOM was used to identify patterned differences in gene expression levels between the EOM and hindlimb profiles, without the CAGED restriction to dynamically regulated transcripts only.
CAGED identified 314 transcripts that were dynamically regulated (mean value for three replicates/age/muscle 2-fold) in EOM, but not in hindlimb, 130 that were dynamically regulated in hindlimb, but not in EOM, and 440 that were dynamically regulated in both muscles, between days P0 and P45 (Supplemental Fig. S1). Here, we focused upon the 314 transcripts dynamically regulated only in EOM (Supplemental Table S2) and 291 transcripts that were dynamically regulated in EOM and hindlimb, but fell into different CAGED clusters (Supplemental Table S3 and Supplemental Fig. S1), indicating that their postnatal expression patterns were dissimilar. The transcripts identified here represent potential causes or consequences of EOM divergence from the traditional skeletal muscle allotype. Transcripts were assigned to functional categories (Supplemental Tables S2 and S3 and Fig. 4). Other than ESTs and unclassified (other) transcripts, the energy metabolism and cell signaling categories best distinguished the EOM and limb allotypes.
|
Taken together, CAGED and SOM analyses suggest that the number of transcripts truly unique to the EOM allotype is relatively small. However, nearly one-third of transcripts either detected as differentially regulated by CAGED only in the EOM series or with significantly higher expression in EOM by SOM were ESTs. This finding suggests that there may be more genes with expression confined to EOM than are currently known. Based upon the known genes, data thus far are consistent with the hypothesis that a limited range of tissue-specific mechanisms interact with more subtle differences in generalized skeletal muscle mechanisms to determine the novel EOM phenotype.
Patterned differences between EOM and hindlimb muscles from CAGED and SOM.
We focused here upon identities and functions of those transcripts with EOM-biased expression patterns from CAGED, SOM, or both analyses. Transcription factors are essential to the emergence of tissue-specific identities. For the developmental stages examined here, several transcriptional regulators were dynamically regulated only in EOM (Gata6, Hif1a, Jun, Neurod1, Nifb, Nr2f1, Sox10, Thrsp, Tieg, and Zpf36l1) or were expressed at constitutively higher levels in EOM (Gtf3c1 and Pitx2). While any of these may play a key role in regulation of the EOM allotype, only one has been tested to date. Pitx2 deletions cause EOM agenesis, showing that it plays a critical role in early muscle development (19, 39). The maintained expression of Pitx2 throughout EOM maturation further suggests that it may be essential in the emergence and maintenance of EOM-specific properties.
The energy metabolism category contained the largest number of transcripts identified by CAGED as dynamically regulated only in EOM (n = 45; Fig. 4). SOM identified additional metabolism transcripts with constitutive expression levels for EOM > hindlimb. The transcriptional regulator, Hif1a, is essential for coordinating adaptive responses to hypoxia/ischemia, by activating genes for glycolytic enzymes, glucose transport, and angiogenesis. A prior study indicating that Hif1a expression is essential for embryogenesis of cephalic mesenchyme and vasculature (26), coupled with our gene expression data, suggests that it may be a determinant of energetic mechanisms in the EOMs. Moreover, our observation of the coordinate, dynamic upregulation of Hif1a and Egln3, which stabilizes and thereby prolongs Hif1a activity, and the EOM > hindlimb expression of Hif1a downstream targets/effectors (Vegfb, Hmox2, Orp150, and Tf) collectively support an important role for Hif1a signaling in EOM development.
Since prior studies also identified substantial adult EOM-hindlimb differences in energy metabolism (18, 36, 60), we mapped the normalized microarray signal levels for all enzymes onto glycogen and glycolytic pathways (Fig. 5). Developmental patterns in rate-limiting enzyme transcripts in glycogen anabolic (glycogen synthase) and catabolic (phosphorylase) pathways were consistent with our failure to detect glycogen deposits in postnatal EOM and with prior data showing that EOM does not use glycogen as a key energy store (18, 36, 60, 75). Nonmuscle isoforms of phosphorylase (Pygl and Pygb) were, however, transiently high in neonatal EOM, a finding that correlates with high glycogen levels in prenatal EOM (unpublished data). Likewise, several glycolytic enzymes and regulators of glycolysis exhibited EOM-hindlimb differences, including EOM utilization of non-skeletal muscle enzyme isoforms for lactic dehydrogenase (Ldhb), enolase (Eno2), and aldolase (Aldoc) and its low expression of 3-phosphoglycerate kinase and the skeletal muscle isoform of a key glycolytic regulator, 6-phosphofructo-2-kinase/fructose 2,6-bisphosphatase. The adaptive value of this divergence for EOM, including any efficiency gain from non-muscle enzyme isoforms, is unknown. Finally, genes encoding many lipid transporters and fatty acid ß-oxidation enzymes were either dynamically upregulated only in EOM (e.g., Cd36, Fabp4, Facl2, Cpt1b, and Acadl) or exhibited more dramatic induction in postnatal EOM vs. hindlimb (e.g., Hadhb, and Acdml). Two additional fatty acid transport/metabolism transcripts (Decr1 and Apoe) were detected at higher constitutive levels in EOM by SOM. These data suggest that fatty acids may provide an alternative energy source to glycogen in EOM, much like they do in other highly active muscles such as heart and insect flight muscle (21).
|
The divergence of EOM from the more traditional skeletal muscles with respect to energy metabolism mechanisms may be directly responsible for its novel response to metabolic myopathies. For example, low reliance upon specific transcripts correlates with EOM mild or absent responses to glycogen storage disease type 1 (G6pt1), glycogenosis type IX myopathy (Pgk1), and various hereditary and acquired myopathies due to purine nucleotide cycle defects (Ampd1). By contrast, the high mitochondrial content of EOM is associated with substantial allotype sensitivity to accumulation of mitochondrial DNA mutations and the resultant mitochondrial myopathies (28, 47). Collectively, these correlations of gene expression patterns with disease sensitivity support the hypothesis that allotype-specific traits are determinants of the severity of response to a variety of muscle diseases.
Differences in myosin heavy chain isoform expression were first used to define skeletal muscle allotypes. We used qPCR to validate myosin heavy chain isoform expression patterns detected in postnatal EOM and hindlimb muscle using RMA (Supplemental Figs. S4 and S5). qPCR data were highly correlated with expression patterns obtained from microarray and both were in agreement with prior findings of the developmental regulation of Myh3, Myh8, and Myh13 in adult EOM (6, 81).
Intrinsic and extrinsic mechanisms in EOM divergence.
Given a subset of genes that are preferentially expressed in EOM, promoter analysis can be used to obtain a better understanding of upstream regulatory mechanisms for the novel EOM allotype. We filtered the DNA microarray temporal series for those genes that most clearly distinguished EOM, using the SAM algorithm (117 transcripts; Supplemental Table S5). We then analyzed cis-regulatory genomic sequence (500 bases upstream of transcription start sites) that was available for 52 known genes in this group. Conserved transcription factor binding motifs among these genes represent putative regulatory mechanisms. Binding sites for 86 different transcription factors were found in 3' sequence of EOM-enriched genes (Supplemental Tables S5 and S6). Binding sites for ubiquitously expressed factors, such as heat shock transcription factor (found upstream of transcription start sites for 96% of the 52 genes) and C/EBP (upstream of 44% of genes), were commonly found, whereas binding motifs for transcription factors known to be active in skeletal muscle were prominent, but encountered less frequently (e.g., Sp1, 31%; Msx1, 23%; Pou2f1, 23%; Usf, 8%, Myog, 6%, and Myod1, 2%).
Consistent with EOM expression of multiple cardiac muscle traits (12, 36, 60), binding sites for transcription factors known to be active in heart were detected among EOM-enriched genes (e.g., Pax4, 62%; Pitx2, 15%; Nkx2.5, 13.5%; Supplemental Table S6). Binding sites for transcription factors that are involved in eye development and/or are responsive to retinoic acid (e.g., Pax2, 25%; Msx1, 23%; Hoxa3, 21%; Znf42, 19%; Pitx2, 15%; Tfap2a, 10%; Pbx1, 8%; Hnf3b, 6%) also were detected at moderate frequency among these EOM-enriched genes. This finding is consistent with both present data and prior reports (12, 35, 36, 60) of enhanced expression of retinoic acid-sensitive and ocular development-related genes in EOM. Prior studies also have shown that Pitx2 is essential for heart and eye development and EOM myogenesis (19, 39) and may regulate the EOM-specific Myh13 (3). Here, we provide additional evidence of a role for Pitx2 in EOM myogenesis by showing its maintained expression into adult EOM and identifying several additional EOM-enriched transcripts with upstream Pitx2 motifs (Ace, Atp1a1, Cd9, Cds1, Dcn, Gpx3, LOC64312, and Scya11).
Ultimately, the EOM allotype is shaped by myoblast lineage-specific traits interacting with epigenetic mechanisms, such as motoneuron activity patterns, diffusible trophic factors, circulating hormones, and other features of the local environment. Studies using recently derived immortalized cell lines suggest that EOM myoblasts themselves are unique (unpublished data), while in vivo studies point to novel patterns of regulation of myogenic transcripts in EOM (46, 50, 80). Disruption of oculomotor motoneuron maturation alters the molecular profile of EOM, including suppression of Myh13 (5, 7, 11). EOM perinatal expression patterns of Pitx2, a potential regulator of Myh13, then may also be activity dependent. Specific neurotrophic mechanisms do function in EOM development, as spinal motoneurons cannot substitute for oculomotor motoneurons in supporting EOM maturation in organotypic nerve-muscle coculture (57). Finally, local interactions, including those between neural crest and EOM precursor cells, likely influence gene expression patterns in developing EOM (13, 49, 78).
In typical skeletal muscle, there is no apparent molecular "master regulator" of specific fiber types (74). Instead, regulatory mechanisms are multifactorial with contractile and fatigue properties independently determined by and tailored to the functional demands placed upon specific muscle fiber types and/or groups. Based upon our present data, it is probable that the EOM allotype also is the consequence of combinatorial activity of a variety of transcriptional mechanisms, with individual signaling pathways responsible for only a narrow range of traits. Our promoter analysis of a subset of genes with EOM-enriched expression helps narrow the field of candidates that may be involved in allotype regulation.
Conclusions
We have used morphological and DNA microarray analyses to establish an integrated postnatal profile for EOM development. Based upon the coordinate regulation of these traits with key events in visual, vestibular, and oculomotor system development, we propose a model that the EOM phenotype is a consequence of extrinsic factors that are unique to its local environment and sensory-motor control system, acting upon a novel myoblast lineage. Using the CAGED and SOM algorithms, we identified a broad spectrum of differences in postnatal transcriptional patterns of the EOM and limb muscle allotypes along with several differentially regulated transcription factors that may signal allotype specificity. Here, we have highlighted allotype energetics differences but identified divergence in many other properties that remain to be further explored.
Collectively, this analysis shows that emergence of the EOM allotype is the consequence of tissue-specific mechanisms that direct expression of a limited number of EOM-specific transcripts and broader, incremental differences in transcripts that are conserved across the two allotypes. We propose that transcriptional mechanisms that are shared by EOM and cardiac muscle (e.g., Pitx2 and Nkx2.5) or EOM and the eye proper (e.g., Pitx2, Pax2, and Tbx15) represent the best candidates for regulation of the EOM allotype. Comparing expression profiles of entire skeletal muscle groups may partially mask the divergence of the EOM and hindlimb by failing to reveal unique combinations of transcripts expressed in specific fiber types. We recently have shown that even the two distinct EOM layers differ in gene expression profiles (35). Single fiber type expression analysis then is likely to be critical in understanding the heterogeneity of skeletal muscle allotypes. Taken together, these data represent an important first step in dissecting allotype-specific regulatory mechanisms that subsequently may explain the differential sensitivity of the skeletal muscle allotypes to metabolic and neuromuscular diseases.
![]() |
GRANTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
ACKNOWLEDGMENTS |
---|
![]() |
FOOTNOTES |
---|
Address for reprint requests and other correspondence: J. D. Porter, Dept. of Neurology, Case Western Reserve Univ., 11100 Euclid Ave., Cleveland, OH 44106-5040 (E-mail: john.porter{at}case.edu).
10.1152/physiolgenomics.00222.2003
1 The Supplementary Material for this article (Supplemental Tables S1S6 and Supplemental Figs. S1S5) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00222.2003/DC1.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|