Using a 3D virtual muscle model to link gene expression changes during myogenesis to protein spatial location in muscle
© Waardenberg et al; licensee BioMed Central Ltd. 2008
Received: 08 May 2008
Accepted: 22 October 2008
Published: 22 October 2008
Myogenesis is an ordered process whereby mononucleated muscle precursor cells (myoblasts) fuse into multinucleated myotubes that eventually differentiate into myofibres, involving substantial changes in gene expression and the organisation of structural components of the cells. To gain further insight into the orchestration of these structural changes we have overlaid the spatial organisation of the protein components of a muscle cell with their gene expression changes during differentiation using a new 3D visualisation tool: the Virtual Muscle 3D (VMus3D).
Sets of generic striated muscle costamere, Z-disk and filament proteins were constructed from the literature and protein-interaction databases. Expression profiles of the genes encoding these proteins were obtained from mouse C2C12 cells undergoing myogenesis in vitro, as well as a mouse tissue survey dataset. Visualisation of the expression data in VMus3D yielded novel observations with significant relationships between the spatial location and the temporal expression profiles of the structural protein products of these genes. A muscle specificity index was calculated based on muscle expression relative to the median expression in all tissues and, as expected, genes with the highest muscle specificity were also expressed most dynamically during differentiation. Interestingly, most genes encoding costamere as well as some Z-disk proteins appeared to be broadly expressed across most tissues and showed little change in expression during muscle differentiation, in line with the broader cellular role described for some of these proteins.
By studying gene expression patterns from a structural perspective we have demonstrated that not all genes encoding proteins that are part of muscle specific structures are simply up-regulated during muscle cell differentiation. Indeed, a group of genes whose expression program appears to be minimally affected by the differentiation process, code for proteins participating in vital skeletal muscle structures. Expression alone is a poor metric of gene behaviour. Instead, the "connectivity model of muscle development" is proposed as a mechanism for muscle development: whereby the closer to the myofibril core of muscle cells, the greater the gene expression changes during muscle differentiation and the greater the muscle specificity.
Skeletal muscle exhibits organisation and uniformity in anatomical structure between vertebrate species whilst displaying characteristic differences between muscle types as a functional requirement. This structural organisation is due to the sarcomere, the basic contractile unit and core structural component responsible for the striated appearance of skeletal and cardiac muscle fibres [1, 2]. Sarcomeres overlap longitudinally (at the Z-disk) and are joined transversely (at the M-line and the Z-disk) into bundles known as myofibrils. Myofibrils join transversely to the sarcolemma via the costamere, which has been suggested to exist as a skeletal muscle specific extension of the focal adhesions expressed in non-muscle cells and essentially forms the interface between the intracellular and extra-cellular spaces of muscle cells [3–5].
Myogenesis is an ordered process whereby mono-nucleated muscle precursor cells (myoblasts) fuse into multi-nucleated myotubes that mature into the different classes of myofibre [6, 7]. Successful skeletal muscle development relies upon the correct cyto-architectural arrangement of the various muscle components and abnormal arrangements can result in devastating effects on muscle function. Indeed, mutations in the genes encoding these components are often associated with muscular diseases including dystrophies [8, 9]. One of the major challenges in muscle research is to understand how a muscle precursor cell with a very different organisation develops the highly organised structure responsible for the sophisticated mechanics of muscle. Understanding this fundamental process is of high interest to medical science for treatment of muscle diseases and also animal science for exploring body growth and meat quality determinants. Although muscle development is well-studied, the molecular mechanisms underlying this developmental process remain mostly unknown. Gene expression profiles of myogenesis have primarily focused on the up-regulation of muscle specific genes, for example, genes coding the filamin proteins change from non-muscle-specific to muscle-specific [10–12]. However, restricting analyses to muscle-specific or dynamically changing gene expression sets does not provide an integrated model for understanding the overall function of muscle structural proteins. This is especially true for understanding the transition of broadly expressed to structurally restricted proteins which needs to be underpinned by an understanding of their structure-function relationship.
We have developed the virtual muscle (VMus3D), an animated 3D computer graphics model to visualise the relative locations of the products of muscle gene expression which characterises striated muscle structure at molecular resolution. To test the novelty of the VMus3D in terms of its ability to contribute to the understanding of myogenesis, we have used it to analyse a number of publicly available muscle gene expression datasets.
Gene Expression Spatial Visualisation of muscle differentiation using Virtual Muscle and a generic set of costamere, Z-disk and filament proteins
The set of generic mammalian genes encoding the costamere, Z-disk and filament system proteins and their paralogs was built using the published literature and protein-protein interaction databases including BIND (Biomolecular Interaction Network Database; http://www.bind.ca/) and HPRD (Human Protein Reference Database; http://www.hprd.org). The paralogs were included to enable the analysis and address the issue of paralog swapping. Gene expression data for this set of genes was obtained from a mouse C2C12 in vitro muscle cell line before and after differentiation from the NCBI GEO (Gene Expression Omnibus) database .
Gene Expression Profiles of Costamere, Filament system and Z-disk
The limited change in gene expression for the majority of costamere proteins and some Z-disk proteins may have been due to an already increased expression of these genes in the C2C12 myoblasts rather than a genuine consistency of expression level between muscle and undifferentiated cells. To address this question the skeletal muscle specificity (SPi: see Materials and Methods: Data Input to the Virtual Muscle) of the expression of each of the genes was ascertained by interrogating the SymAtlas mouse GNF1M dataset which contains a tissue survey of mouse gene expression (see Additional File 1). The expression profiles of the genes encoding the components of the costamere, filament system and Z-disk of muscle, as well as their functionally related paralogs (see Additional File 1) were plotted in three dimensions to include (1) the expression level of myoblasts, (2) the expression level of myotubes and (3) muscle specificity (see Additional File 2). In general, those genes with increased gene expression from myoblast to myotube were also highly muscle specific. The distribution of expression patterns of the genes encoding costamere proteins (and paralogs) remained distinct from those of the Z-disk and the filament systems of muscle. This analysis confirmed that the genes encoding the majority of costamere proteins were not expressed in a muscle-specific manner.
Two-Component Mixture Model
To further interrogate the observed results, a two-component mixture of bi-variate normal distributions was fitted to the data. The mixture model allowed us to identify the clusters, or components in the mixture, that encapsulated those genes that were simultaneously differentially expressed from myoblasts to myotubes and/or showed skeletal muscle specificity. The two component mixture model analysis successfully defined two distinct clusters (see Additional Files 1 and 3), where clusters 1 and 2 included 57 and 93 genes respectively of the 150 genes analysed. Cluster 1 represented genes that were dynamically expressed from myoblast to myotube and demonstrated muscle specificity whereas cluster 2 did not represent these features. Furthermore, each structural location could then be categorised as having a probability of belonging to each cluster. The filament genes had a 67.5% and 32.5% probability of belonging to cluster 1 and 2 respectively, the costamere genes had a 21.3% and 78.7% probability of belonging to cluster 1 and 2 respectively, while the Z-disk genes had a 46.3% and 53.7% probability of belonging to cluster 1 and 2 respectively.
Paralog exchange during differentiation
Gene Expression Profile and Skeletal Muscle Specificity of Costamere, Filament system and Z-disk during Myogenesis
Clustering of gene expression data followed by grouping on the basis of gene ontology terms or functional annotation is a standard process in the analysis of gene expression data. However this approach tends to neglect those genes that are not differentially expressed and typically categorised as being unimportant, although they may be functionally relevant to the particular biological process under investigation. In our approach we have first grouped genes on the basis of function, in this case on the basis of the complexes which the proteins encoded by the genes are components of, and subsequently analysed changes in gene expression of the muscle system within the context of wider gene expression profiles. Our visualisation of myoblast to myotube differentiation gene expression data using the VMus3D tool resulted in the visualisation of gene expression as a function of time and space during the development of muscle and its relationship to tissue specificity. By observing spatial changes of gene expression during myogenesis, two predominant expression profiles were observed: one profile for which sub-components of the Z-disk and the majority of the costamere did not undergo major changes from myoblast to myotube; and another profile for which the majority of myofibril filament system proteins were substantially up-regulated (Figure 1).
Incorporation of multiple tissue gene expression data into the analysis demonstrated an association of tissue specificity with changes of gene expression from myoblast to myotube. This was supported and refined by a two-component analysis of these genes and their paralogs which demonstrated that two clusters were clearly present as a function of myogenesis and skeletal muscle tissue specificity (Figure 4). By including all available functionally related paralogs for each structural gene in the two-component analysis, this not only improved the statistical power, but also put a biologically relevant level of variance into each of the possible gene states, thus accounting for temporal gene variation whilst enhancing the boundaries of this clustering approach. This inclusion also demonstrated the variability between the different systems and the costamere demonstrated a general lack of this paralog exchange mechanism.
Connectivity Model of Muscle Development
Many of the costamere proteins have been described as components of the non-muscle specific dystroglycan/sarcoglycan, spectrin and focal adhesion complexes. The Z-disk also appears to form as an early Z-body (initially non-muscle specific) promoting the development of organised sarcomeres, which later integrate muscle-specific proteins to structurally support expanding and contracting myofibres [11, 22, 23]. Our observations suggest that although the encoded proteins are incorporated into different structures in muscle, the induction of the muscle differentiation process has limited impact on the expression program of many of the genes encoding these proteins. This recruitment of non-muscle specific structures into muscle-specific roles could be coordinated by the presence of the muscle specific components of the costamere which demonstrated up-regulation of gene expression upon differentiation observed herein (Figure 3); Caveolin-3, Syncoilin, Sarcolgycans alpha/gamma and the Beta1 Integrin binding proteins 2 and 3. Each of these components have already been independently recognised as having some involvement in muscle development due to their gene expression profiles [15, 24–28], but have not been connected simultaneously to a probable role in the development of the costamere. It is proposed that these genes could represent an initial group of structural proteins bordering intrinsic/extrinsic capability during myotube formation. Evidence has already suggested that the development of pre-myofibril Z-bodies originate from the cell membrane of spreading cardiomyocytes, and that these Z-bodies are aligned to eventually organise into mature Z-discs [23, 29]. Therefore, following on in support of previous models, the "connectivity model of muscle development" is proposed: whereby the closer to the myofibril core (i.e. the filament system) of muscle cells, the greater the gene expression changes during muscle differentiation and the greater the muscle specificity index. In addition, the remainder of the genes coding for the costamere proteins that did not demonstrate these activities (i.e. displaying smaller differential and skeletal muscle gene expression changes) leading to the hypothesis that the transcriptome state of the costamere genes in the myoblast may not be different from many other cells in the body.
There are four major mechanisms for the modification of the structures of multi-subunit complexes: (1) recruitment of entirely new proteins; (2) paralog exchange; (3) alternative splice variants; and (4) modification of existing proteins (such as phosphorylation). Construction of the costamere and Z-disk clearly involves the first process, and in some cases the second, although this is most strongly associated with the filament proteins. The relationship of paralog exchange with location also occurs in developed muscle, whereby the filament system also uses paralog exchange as a fundamental mechanism for altering muscle characteristics and hence function; for example via differential incorporation of myosin heavy/light chains and other proteins to drive fibre type changes. In some circumstances, these paralogs maybe co-expressed within an adult cell during fibre type transitions (due to the hybrid nature of myosin heavy/light chains) but the repertoire of paralogs is reduced compared to those expressed and selected for or against during the transition of cell state during early development, as observed herein . In addition, the microarray platform used in these experiments does not enable the resolution of alternative splice variants due to the difficulty in deconvulution of probe sets with regards to individual oligo variances and gene coverage of these oligos in this circumstance being undesirable. Nonetheless, alternative splicing could be contributing to the lack of overall gene expression change if underpinned by changes in splicing patterns during myogenesis. This appears to be a likely explanation for the lack of large differential expression observed for the Alpha and Beta Integrins, which are known to have developmentally regulated alternative splice variants expressed during myogenesis . Finally, gene expression data resulting from microarray hybridizations does not allow exploration of the potential role of protein modification in the structure of multi-subunit complexes.
In general, the construction of the costamere and to a lesser extent the Z-disk, during muscle differentiation involves the recruitment of broadly expressed proteins. However a small number of genes encoding costamere proteins and more Z-disks proteins are induced during differentiation and these proteins may serve to nucleate the construction of muscle specific structures. Our approach using VMus3D and combining gene expression datasets provides a spatial method of gene expression analysis at molecular resolution in the context of the core structural components of a skeletal muscle cell and these findings highlight the utility of this approach.
The mouse C2C12 muscle cell line (a myoblast cell line maintained in growth medium for proliferation that, upon transfer to an appropriate stimulant medium, differentiates into myofibres), and in vitro differentiation have been used in a number of studies [7, 13, 32]. The gene expression dataset of C2C12 cells differentiated in vitro was retrieved from NCBI GEO ; dataset accession number GDS2412 . The user-submitted data (NCBI SOFT format) was used for analysis and the normalisation procedure for the Affymetrix GeneChip Mouse Genome 430 2.0 dataset can be accessed via the accompanying publication . Briefly, the experiment comprised of six microarray hybridisations with two triplicates: one for proliferating myoblasts and another for differentiated myotubes. For each annotated gene and in each of the two conditions (myoblasts and myotubes), the normalised mean expression was calculated from its average log2base transformed expression across the triplicate.
In addition, a mouse GNF1M (MAS5 compressed) dataset was downloaded from the GNF SymAtlas Download portal http://wombat.gnf.org/index.html. This data was combined with the Affymetrix microarray GNF1M annotation file to associate each probe with its corresponding gene symbol. The median expression level of the log2base normalised means of all 61 tissues sampled excluding skeletal muscle and the log2base normalised mean skeletal muscle gene expression level was calculated for each probe.
To eliminate unnecessary relational querying, these two datasets were combined into a single dataset on the basis of matching gene symbols. The criterion chosen (due to multiple probes possibly representing the same gene) was: for every gene symbol from the C2C12 differentiation dataset to include every matching gene symbol and its relative tissue expression level from the processed SymAtlas dataset. Therefore, all possible tissue expression profiles for any of the probes representing a gene were included. Multiple gene expression profiles for a gene were reduced to a single profile for each gene symbol by selecting only the highest normalised mean gene expression value from C2C12 differentiation dataset and tabulated tissue specificity index.
Construction of the VMus3D
AC3D version 6 was used to build de-convoluted 3D models of the structural proteins and their arrangements within skeletal muscle . The data used for the construction of these simplified structures was restricted to vertebrate evidence due to skeletal muscle structural similarity and included where possible: protein-protein interaction data from public data repositories (HPRD and BIND), published experimental localisation data (e.g. protein localisation), structural protein data (e.g. domain structure: which was also used to determine interaction sites for arrangements of complexes where possible) and previously published diagrammatic representations of the contractile apparatus [5, 14–19].
Data Input to the Virtual Muscle
The VMus3D, in addition to its accompanying relational database schema also has scripts for transforming and selecting appropriate gene expression values based upon a standard gene set corresponding to the structures within the model for developed adult muscle. For each gene in i, two parameters were explored as follows: (1) MMi: its differential expression computed from the log ratio of the normalised gene expression values from myoblast to myotube and; (2) SP i : its skeletal muscle specificity computed from the difference between the normalised mean expression in skeletal muscle and the median expression in the remaining tissues (see Additional File 1). Both MM i and SP i were inputs for representation in the VMus3D (Figure 3). The genes included in the VMus3D analysis were also filtered to eliminate genes that were considered to have not been expressed (gene expression levels less than 50, or log2base of ~5.64, in either mean myoblast or myotube) from the C2C12 differentiation dataset. The VMus3D used a colour coded representation of the log ratio thresholds; purple was greater than two, blue was greater than five, yellow was less than two.
3D graphing and paralogs
To combine changes in gene expression during C2C12 differentiation relative to muscle specificity, genes were plotted in 3-dimensions (using a X3D graphing application developed for use with the VMus3D): 2 dimensions belonging to time (myoblast and myotube: GDS2412) and another dimension belonging to the difference between the normalised mean expression of a gene in skeletal muscle and its median gene expression across the remaining tissues sampled (GNF SymAtlas). Available paralogs were automatically extracted from the datasets via a script that retrieved all numerical variants of a gene and selected alphabetical subtypes. The retrieved genes and their paralogs were categorised and colour coded into three groups corresponding to the costamere (blue), Z-disk (red) and filament system (green).
Simultaneous Modelling of Differential Gene Expression and Muscle Specificity
A simulation was also performed by generating 100,000 elements drawn from the estimated two-component mixture model. Genes lacking either skeletal muscle specificity or myogenesis differential expression data were appropriated a probability of belonging to cluster 1 or 2 from their average p-values of the components from the 100,000 simulation that fell into a range of +/-0.1 from the log ratios (see Additional File 1). The VMus3D was then used to display the spatial distribution of the components from the two clusters adjusted using the 100,000 simulation (+/-0.1 myoblast-myotube & +/-0.1 skeletal muscle index) for finer cluster resolution (Figure 4).
Andy Coulbourne 3D
Biomolecular Interaction Network Database
Gene Expression Omnibus
Affymetrix Mouse GeneChip Microarray
Human Protein Reference Database
Scene Authoring Interface
Virtual Muscle 3D
The authors wish to thank Evgeny Glazov for valued discussions and the anonymous reviewers for their feedback. Ashley Waardenberg is supported by postgraduate scholarships from Griffith University and the CSIRO Food Futures Flagship.
- Oota S, Saitou N: Phylogenetic relationship of muscle tissues deduced from superimposition of gene trees. Mol Biol Evol. 1999, 16: 856-867.View ArticlePubMedGoogle Scholar
- Scott W, Stevens J, Binder-Macleod SA: Human skeletal muscle fiber type classifications. Phys Ther. 2001, 81: 1810-1816.PubMedGoogle Scholar
- Wu JC, Sung HC, Chung TH, DePhilip RM: Role of N-cadherin- and integrin-based costameres in the development of rat cardiomyocytes. J Cell Biochem. 2002, 84: 717-724. 10.1002/jcb.10092View ArticlePubMedGoogle Scholar
- Morris EJ, Fulton AB: Rearrangement of mRNAs for costamere proteins during costamere development in cultured skeletal muscle from chicken. J Cell Sci. 1994, 107 (Pt 3): 377-386.PubMedGoogle Scholar
- Ervasti JM: Costameres: the Achilles' heel of Herculean muscle. J Biol Chem. 2003, 278: 13591-13594. 10.1074/jbc.R200021200View ArticlePubMedGoogle Scholar
- Golson ML, Sanger JM, Sanger JW: Inhibitors arrest myofibrillogenesis in skeletal muscle cells at early stages of assembly. Cell Motil Cytoskeleton. 2004, 59: 1-16. 10.1002/cm.20017View ArticlePubMedGoogle Scholar
- Tomczak KK, Marinescu VD, Ramoni MF, Sanoudou D, Montanaro F, Han M, Kunkel LM, Kohane IS, Beggs AH: Expression profiling and identification of novel genes involved in myogenic differentiation. Faseb J. 2004, 18: 403-405.PubMedGoogle Scholar
- Sorimachi H, Kinbara K, Kimura S, Takahashi M, Ishiura S, Sasagawa N, Sorimachi N, Shimada H, Tagawa K, Maruyama K, et al.: Muscle-specific calpain, p94, responsible for limb girdle muscular dystrophy type 2A, associates with connectin through IS2, a p94-specific sequence. J Biol Chem. 1995, 270: 31158-31162. 10.1074/jbc.270.13.7437View ArticlePubMedGoogle Scholar
- Tagawa K, Taya C, Hayashi Y, Nakagawa M, Ono Y, Fukuda R, Karasuyama H, Toyama-Sorimachi N, Katsui Y, Hata S, et al.: Myopathy phenotype of transgenic mice expressing active site-mutated inactive p94 skeletal muscle-specific calpain, the gene product responsible for limb girdle muscular dystrophy type 2A. Hum Mol Genet. 2000, 9: 1393-1402. 10.1093/hmg/9.9.1393View ArticlePubMedGoogle Scholar
- Bains W, Ponte P, Blau H, Kedes L: Cardiac actin is the major actin gene product in skeletal muscle cell differentiation in vitro. Mol Cell Biol. 1984, 4: 1449-1453.PubMed CentralView ArticlePubMedGoogle Scholar
- Lin JJ, Lin JL: Assembly of different isoforms of actin and tropomyosin into the skeletal tropomyosin-enriched microfilaments during differentiation of muscle cells in vitro. J Cell Biol. 1986, 103: 2173-2183. 10.1083/jcb.103.6.2173View ArticlePubMedGoogle Scholar
- Gunning P, Hardeman E, Wade R, Ponte P, Bains W, Blau HM, Kedes L: Differential patterns of transcript accumulation during human myogenesis. Mol Cell Biol. 1987, 7: 4100-4114.PubMed CentralView ArticlePubMedGoogle Scholar
- Chen IH, Huber M, Guan T, Bubeck A, Gerace L: Nuclear envelope transmembrane proteins (NETs) that are up-regulated during myogenesis. BMC Cell Biol. 2006, 7: 38- 10.1186/1471-2121-7-38PubMed CentralView ArticlePubMedGoogle Scholar
- Bader GD, Donaldson I, Wolting C, Ouellette BF, Pawson T, Hogue CW: BIND–The Biomolecular Interaction Network Database. Nucleic Acids Res. 2001, 29: 242-245. 10.1093/nar/29.1.242PubMed CentralView ArticlePubMedGoogle Scholar
- Brancaccio M, Guazzone S, Menini N, Sibona E, Hirsch E, De Andrea M, Rocchi M, Altruda F, Tarone G, Silengo L: Melusin is a new muscle-specific interactor for beta(1) integrin cytoplasmic domain. J Biol Chem. 1999, 274: 29282-29288. 10.1074/jbc.274.41.29282View ArticlePubMedGoogle Scholar
- Clark KA, McElhinny AS, Beckerle MC, Gregorio CC: Striated muscle cytoarchitecture: an intricate web of form and function. Annu Rev Cell Dev Biol. 2002, 18: 637-706. 10.1146/annurev.cellbio.18.012502.105840View ArticlePubMedGoogle Scholar
- Davies KE, Nowak KJ: Molecular mechanisms of muscular dystrophies: old and new players. Nat Rev Mol Cell Biol. 2006, 7: 762-773. 10.1038/nrm2024View ArticlePubMedGoogle Scholar
- Li J, Mayne R, Wu C: A novel muscle-specific beta 1 integrin binding protein (MIBP) that modulates myogenic differentiation. J Cell Biol. 1999, 147: 1391-1398. 10.1083/jcb.147.7.1391PubMed CentralView ArticlePubMedGoogle Scholar
- Peri S, Navarro JD, Amanchy R, Kristiansen TZ, Jonnalagadda CK, Surendranath V, Niranjan V, Muthusamy B, Gandhi TK, Gronborg M, et al.: Development of human protein reference database as an initial platform for approaching systems biology in humans. Genome Res. 2003, 13: 2363-2371. 10.1101/gr.1680803PubMed CentralView ArticlePubMedGoogle Scholar
- Whalen RG, Butler-Browne GS, Gros F: Identification of a novel form of myosin light chain present in embryonic muscle tissue and cultured muscle cells. J Mol Biol. 1978, 126: 415-431. 10.1016/0022-2836(78)90049-9View ArticlePubMedGoogle Scholar
- Gomer RH, Lazarides E: Switching of filamin polypeptides during myogenesis in vitro. J Cell Biol. 1983, 96: 321-329. 10.1083/jcb.96.2.321View ArticlePubMedGoogle Scholar
- Berendse M, Grounds MD, Lloyd CM: Myoblast structure affects subsequent skeletal myotube morphology and sarcomere assembly. Exp Cell Res. 2003, 291: 435-450. 10.1016/j.yexcr.2003.07.004View ArticlePubMedGoogle Scholar
- Dabiri GA, Turnacioglu KK, Sanger JM, Sanger JW: Myofibrillogenesis visualized in living embryonic cardiomyocytes. Proc Natl Acad Sci USA. 1997, 94: 9493-9498. 10.1073/pnas.94.17.9493PubMed CentralView ArticlePubMedGoogle Scholar
- Brancaccio M, Fratta L, Notte A, Hirsch E, Poulet R, Guazzone S, De Acetis M, Vecchione C, Marino G, Altruda F, et al.: Melusin, a muscle-specific integrin beta1-interacting protein, is required to prevent cardiac failure in response to chronic pressure overload. Nat Med. 2003, 9: 68-75. 10.1038/nm805View ArticlePubMedGoogle Scholar
- Li J, Rao H, Burkin D, Kaufman SJ, Wu C: The muscle integrin binding protein (MIBP) interacts with alpha7beta1 integrin and regulates cell adhesion and laminin matrix deposition. Dev Biol. 2003, 261: 209-219. 10.1016/S0012-1606(03)00304-XView ArticlePubMedGoogle Scholar
- McCullagh KJ, Edwards B, Poon E, Lovering RM, Paulin D, Davies KE: Intermediate filament-like protein syncoilin in normal and myopathic striated muscle. Neuromuscul Disord. 2007, 17: 970-979. 10.1016/j.nmd.2007.06.004View ArticlePubMedGoogle Scholar
- Mermelstein CS, Martins ER, Portilho DM, Costa ML: Association between the muscle-specific proteins desmin and caveolin-3 in muscle cells. Cell Tissue Res. 2007, 327: 343-351. 10.1007/s00441-006-0296-zView ArticlePubMedGoogle Scholar
- Duggan DJ, Hoffman EP: Autosomal recessive muscular dystrophy and mutations of the sarcoglycan complex. Neuromuscul Disord. 1996, 6: 475-482. 10.1016/S0960-8966(96)00388-4View ArticlePubMedGoogle Scholar
- Wang J, Shaner N, Mittal B, Zhou Q, Chen J, Sanger JM, Sanger JW: Dynamics of Z-band based proteins in developing skeletal muscle cells. Cell Motil Cytoskeleton. 2005, 61: 34-48. 10.1002/cm.20063PubMed CentralView ArticlePubMedGoogle Scholar
- Pette D, Peuker H, Staron RS: The impact of biochemical methods for single muscle fibre analysis. Acta Physiol Scand. 1999, 166: 261-277. 10.1046/j.1365-201x.1999.00573.xView ArticlePubMedGoogle Scholar
- Kaariainen M, Nissinen L, Kaufman S, Sonnenberg A, Jarvinen M, Heino J, Kalimo H: Expression of alpha7beta1 integrin splicing variants during skeletal muscle regeneration. Am J Pathol. 2002, 161: 1023-1031.PubMed CentralView ArticlePubMedGoogle Scholar
- Koh GY, Klug MG, Soonpaa MH, Field LJ: Differentiation and long-term survival of C2C12 myoblast grafts in heart. J Clin Invest. 1993, 92: 1548-1554. 10.1172/JCI116734PubMed CentralView ArticlePubMedGoogle Scholar
- Barrett T, Suzek TO, Troup DB, Wilhite SE, Ngau WC, Ledoux P, Rudnev D, Lash AE, Fujibuchi W, Edgar R: NCBI GEO: mining millions of expression profiles–database and tools. Nucleic Acids Res. 2005, 33: D562-566. 10.1093/nar/gki022PubMed CentralView ArticlePubMedGoogle Scholar
- Su AI, Wiltshire T, Batalov S, Lapp H, Ching KA, Block D, Zhang J, Soden R, Hayakawa M, Kreiman G, et al.: A gene atlas of the mouse and human protein-encoding transcriptomes. Proc Natl Acad Sci USA. 2004, 101: 6062-6067. 10.1073/pnas.0400782101PubMed CentralView ArticlePubMedGoogle Scholar
- AC3D Manual.http://www.ac3d.org/ac3d/download/ac3dman6b.pdf
- McLachlan GJ, Bean RW, Peel D: A mixture model-based approach to the clustering of microarray expression data. Bioinformatics. 2002, 18: 413-422. 10.1093/bioinformatics/18.3.413View ArticlePubMedGoogle Scholar
- McLachlan GJ, Bean RW, Jones LB: A simple implementation of a normal mixture approach to differential gene expression in multiclass microarrays. Bioinformatics. 2006, 22: 1608-1615. 10.1093/bioinformatics/btl148View ArticlePubMedGoogle Scholar
- Reverter A, Wang YH, Byrne KA, Tan SH, Harper GS, Lehnert SA: Joint analysis of multiple cDNA microarray studies via multivariate mixed models applied to genetic improvement of beef cattle. J Anim Sci. 2004, 82: 3430-3439.PubMedGoogle Scholar
- Efron B, Tibshirani R, Storey JD, Tusher V: Empirical Bayes Analysis of a Microarray Experiment. J Am Stat Assoc. 2001, 96: 1151-1160. 10.1198/016214501753382129.View ArticleGoogle Scholar
- Reverter A, Hudson NJ, Wang Y, Tan SH, Barris W, Byrne KA, McWilliam SM, Bottema CD, Kister A, Greenwood PL, et al.: A gene coexpression network for bovine skeletal muscle inferred from microarray data. Physiol Genomics. 2006, 28: 76-83. 10.1152/physiolgenomics.00105.2006View ArticlePubMedGoogle Scholar