About 40% of our weight is formed of skeletal muscles, the hundreds of muscles in our bodies that can be voluntarily controlled by our nervous system. At the moment, the research community largely sees all these muscles as a single group whose tissues are virtually interchangeable.

Yet, skeletal muscles have highly diverse origins, shapes and roles. For example, our diaphragm is a long muscle that contracts slowly and rhythmically so we can draw breaths, while tiny muscles in our eyes generate the short and precise movements of our eyeballs. Different skeletal muscles also respond in distinct ways to injuries, drugs and diseases. This suggests that these muscles may be diverse at the genetic level.

While all the cells in our body have the same genetic information, exactly which genes are turned on and off (or ‘expressed’) changes between types of cells. On top of this ‘on or off’ regulation, the level of expression of a gene – how active it is – can also differ. However, the studies that examine the differences in gene expression between tissues usually overlook skeletal muscles.

Here, Terry et al. use genetic techniques to measure how genes are expressed in over 20 types of muscle in mice and rats. The results show that the expression levels of over 50% of all the animals’ genes vary between muscles. In fact, any two types of muscles express on average 13% of their genes differently from each other. The analyses yield further unexpected findings. For example, the expression levels in a muscle in the foot that helps to flex the rodents’ toes are more similar to those found in eye muscles than to the ones observed in limb muscles. These conclusions indicate that skeletal muscles are a widely diverse family of tissues.

The research community will be able to use the data collected by Terry et al. to explore further the origins and the consequences of the differences between skeletal muscles. This could help researchers to understand why specific groups of muscles are more susceptible to disease, or react differently to a drug. This knowledge could also be exploited to refine approaches in tissue engineering, which aims to replace damaged muscles in the body.

Gene expression atlases have made enormous contributions to our understanding of genetic regulatory mechanisms. The field of functional genomics was set in motion by the completion of the Human Genome Project and the coincident development of high throughput gene expression profiling technologies. The overriding goals of this field are to understand how genes and proteins interact at a whole-genome scale and to define how these interactions change across time, space, and different disease states. The development of SymAtlas was an early, influential effort to address these questions (Su et al., 2002). Custom microarrays were used to systematically profile mRNA expression in dozens of tissues and cell lines from humans and mice. Besides describing tissue-specific expression patterns, these data provided essential insights into the relationship between chromosomal structure and transcriptional regulation (Su et al., 2004).

Related approaches have had a similarly high impact on biomedical research. For example, microRNA expression throughout mammalian tissues was described in an expression atlas that has revolutionized the study of regulatory RNAs (Landgraf et al., 2007). Our lab has contributed to this growing literature with the creation of CircaDB, a database of tissue-specific mRNA rhythms in mice (Hughes et al., 2009; Zhang et al., 2014). Taken together, these projects demonstrate that publicly available functional genomics data have enduring value as a resource for the research community.

Nevertheless, previous gene expression atlases have largely ignored skeletal muscle. CircaDB includes gene expression data from the heart and whole calf muscle, but it does not distinguish between their constituent tissues. Similarly, SymAtlas profiles nearly 100 different tissues, but there is only a single representative sample for either cardiac or skeletal muscle. The microRNA Atlas includes over 250 human, mouse, and rat tissues; however, skeletal muscle is entirely absent from these data. More recent human gene expression atlases have similar biases (Evangelista et al., 2015; Lindholm et al., 2014; Vissing and Schjerling, 2014). Many studies have compared muscle-specific gene expression in different tissues (Porter et al., 2001), fiber-types (Chemello et al., 2011), or disease states (Chen et al., 2000; Colantuoni et al., 2001), but on the whole, there is no systematic analysis of transcriptional diversity in skeletal muscle.

This gap in the literature is problematic since skeletal muscle comprises a remarkably diverse group of tissues (Tirrell et al., 2012). Skeletal muscle groups originate from different regions of the developing embryo, and they have characteristic morphological specializations (Merrell and Kardon, 2013; Murphy and Kardon, 2011; Noden and Francis-West, 2006). Their physiological functions are similarly diversified. For example, extraocular muscles govern precise eye movements, the diaphragm drives rhythmic breathing, and limb skeletal muscles are involved in either fast bursts of motion or sustained contractions underlying posture.

These intrinsic differences contribute to differential susceptibility of muscle groups to injury and disease. For example, there are six major classes of muscular dystrophy, and each one afflicts a characteristic pattern of skeletal and cardiac muscle tissues (Ciciliot et al., 2013; Emery, 2002). Since the causative mutations underlying congenital muscular dystrophies are germline and present in all cells, this observation indicates that there are properties intrinsic to different muscle tissues that regulate their sensitivity or resistance to different pathological mechanisms.

Acquired diseases show similar specificity in which muscle tissues they affect and which they spare. Patients with chronic obstructive pulmonary disorder (COPD) typically have pronounced myopathy in their quadriceps and dorsiflexors (Barreiro and Gea, 2016; Clark et al., 2000; Gagnon et al., 2013). Critical illness myopathy, a debilitating condition caused by mechanical ventilation and steroid treatment, causes muscle weakness in limb and respiratory muscles while sparing facial muscles (Aare et al., 2011; Hermans and Van den Berghe, 2015; Latronico et al., 2012). Cancer cachexia (Acharyya et al., 2005), HIV (Serrano et al., 2008), and sepsis (Tiao et al., 1997) cause muscle wasting, typically affecting fast twitch fibers more severely than slow twitch fibers. In fact, histologically identical muscle fibers show widely divergent responses to injury and disease depending on the muscle group in which they reside (Ciciliot et al., 2013; Aravamudan et al., 2006). Taken together, these observations strongly suggest that the intrinsic diversity of skeletal muscle has important consequences for human health and disease. However, the mechanisms through which disease susceptibility and functional specialization are established are unknown.

Here we present the first systematic examination of transcriptional programming in different skeletal muscle tissues. We find that more than 50% of transcripts are differentially expressed among skeletal muscle tissues, an observation that cannot be explained by fiber type composition or developmental history alone. We show conservation of gene expression profiles across species and sexes, suggesting that these data may reveal conserved functional elements relevant to human health. Finally, we discuss how this unique data set may be applied to the study of disease, particularly regarding muscular dystrophy and regenerative medicine.

To determine which skeletal muscle tissues are of the broadest general interest, we distributed a Google poll (Figure 1—figure supplement 1) to leading investigators in the skeletal muscle field. Having recorded over 100 individual responses, we selected 11 mouse skeletal muscle tissues (Table 1) for study in order to span the functional, developmental, and anatomical diversity of skeletal muscles. To hedge against selection bias from the investigators asked to vote in the poll, we cross-correlated these results with papers indexed in NCBI’s PubMed (Figure 1—figure supplement 1). In addition to these 11 mouse skeletal muscle tissues, we also identified representative smooth and cardiac muscle tissues for collection from mouse, and two skeletal muscle tissues (EDL and soleus) from male and female rats to permit inter-species and inter-sex comparisons (Table 1).

Adult mice and rats were sacrificed and whole muscle tissues were dissected to include the entire muscle body from tendon to tendon. Each sample included six biological replicates from three animals apiece. Therefore, tissues were collected from 18 individual animals for every sample. RNA was purified from these tissues, and RNAseq was used to measure global gene expression (Materials and methods and Supplementary file 1). On average, every muscle sample was covered by greater than 200 million aligned short nucleotide reads, for a grand total of over 4.4 billion aligned reads in the entire data set (Table 1). Empirical simulations indicate that this experimental design reaches saturation with respect to identification of expressed exons (Figure 1—figure supplement 2). Pairwise comparisons of each replicate sample further indicate a high degree of reproducibility in expression level measurements among biological replicates (median R² value >0.93, Figure 1—figure supplement 3).

Over 80% of transcripts encoded in the genome are expressed in at least one skeletal muscle (Figure 1A). Comparing the transcripts expressed in all tissues identifies a core group of ~21,000 transcripts found in every skeletal, cardiac, and smooth muscle (Figure 1B). Presumably, these mRNAs include the minimal set of genes required for a cell to generate contractile force. Differential expression analysis shows that even at extremely stringent (q < 10⁻⁶) statistical thresholds, 55.2% of mouse transcripts are differentially expressed among skeletal muscle tissues (Figure 1C). Phrased differently, over half of all transcripts are statistically different when comparing mRNA expression among the 11 mouse skeletal muscles in this study. To validate a subset of these data, we selected 10 genes differentially expressed between EDL and soleus across a range of different fold changes and performed quantitative PCR (qPCR) on independent biological samples. All 10 replicated, and manual examination of internal controls for cardiac and smooth muscle agreed with a priori expectations (Figure 1—figure supplement 4). To explore similarity between tissues, we calculated a pairwise Euclidean distance between every tissue (Figure 1—figure supplement 5). From these data, we generated a dendrogram that clusters tissues based on overall transcriptome similarity (Figure 1D). Smooth and cardiac tissues clustered together as expected, and skeletal muscles were found in two primary clusters, presumably based on the proportion of fast twitch fibers they include (i.e. the proportion of Myosin heavy chain 4 (Myh4) expression). We then calculated the proportion of differentially expressed transcripts in every pairwise tissue comparison (Figure 1E). Some surprising observations emerged from this analysis. For example, masseter, a head/neck muscle, is more similar to limb muscles like EDL than muscles that share a similar developmental history, such as the tongue. In contrast, the flexor digitorm brevis (FDB), a muscle necessary for flexing the toes, is more similar to extraocular muscles than limb muscles. On average, 13% of transcripts are differentially expressed between any two skeletal muscles, with a maximum of 36.5% and a minimum less than 1%. A list of all transcripts differentially expressed among skeletal muscles is provided in Supplementary file 2, and the analyzed expression data can be found in our online database.

Figure 1 with 7 supplements see all

Download asset Open asset

Skeletal muscle contains numerous non-muscle cells types. Therefore, a key question is whether the transcriptome diversity described above reflects genuine expression differences in muscle cells or alternatively differential cellular composition of the bulk tissues. To answer this we performed principal component analysis (Figure 2). The first principal component (PC1) accounts for nearly 80% of the variance in our data and separates skeletal muscles into two groups with striking similarity to skeletal muscle clusters 1 and 2 in Figure 1E. The second principal component (PC2) accounts for roughly 8% of variance and separates cardiac and smooth muscles from skeletal muscle (Figure 2A). We then calculated the correlation (expressed as R² values) between each transcript’s expression and PC1 (Figure 2B). A small number of transcripts are highly (R² >0.90) correlated with PC1 and are thereby predictive of skeletal muscle identity. From the literature, we identified a number of ‘marker’ genes commonly used to separate non-muscle cells from skeletal muscle in flow cytometry (Liu et al., 2015). We then compared the expression these ‘marker’ genes to PC1 to identify non-muscle cell types that may be partially responsible for the overall transcriptome diversity in our data. Pecam1 (CD31), a marker of endothelial cells, has an R² value of 0.56, reflecting modest correlation with PC1. These data suggest that different proportions of endothelial cells in skeletal muscle contribute in part to the overall mRNA diversity observed.

Figure 2 with 1 supplement see all

Download asset Open asset

Figure 2—source data 1

Gene Ontology analysis reveals that sarcomere structural components are among the most commonly enriched differential pathways among skeletal muscles.

Genes differentially expressed among skeletal muscles were identified for every pairwise comparison (q < 0.01; FC > 2; max = 3,000). These gene lists were analyzed for Gene Ontology enrichment (Huang et al., 2009a). This figure displays a group of related tables summarizing these results. For every skeletal muscle tissue, the most common enriched gene ontologies for up- and down-regulated genes are shown, along with the number and identity of tissues in which the constituent genes are differentially expressed. Key structural components of the sarcomere were among the most commonly observed differential gene ontology terms, including ‘Z disc’, ‘actin binding’, ‘M band’, and ‘myosin complex’. Additionally, ontology terms related to extracellular proteins such as collagen, integrin, and fibronectin were enriched in these data. TON = tongue, FDB = flexor digitorum brevis, SOL = soleus, EYE = extraocular eye muscles, DIA = diaphragm, EDL = extensor digitorum longus, PLA = plantaris, TAN = tibialis anterior, GAS = gastrocnemius, MAS = masseter, QUAD = quadriceps.

https://doi.org/10.7554/eLife.34613.014

Download elife-34613-fig2-data1-v2.xlsx

All other known marker genes had lower R² values than Pecam1/CD31, reflecting weaker correlation with PC1. These include markers of muscle stem cells (Vcam1, R² = 0.37), fibroblasts (Ly6a/Sca1, R² = 0.19), and blood cells (Ptprc/CD45, R² <0.01). This observation supports the interpretation that some gene expression differences observed in this study are due to differential cellular composition. Nevertheless, every one of the top ten genes most strongly correlated with PC1 (Figure 2D) encodes a characteristic skeletal muscle protein that is unlikely to have been transcribed by non-muscle cell types. To extend on this observation, Supplementary file 3 provides a list of the top 100 genes most strongly correlated with PC1. GO pathway analysis (Huang et al., 2009a) reveals that the most significantly enriched term among this list of genes is ‘muscle protein’ (enrichment 6.58, FDR < 10⁻⁴⁶). Moreover, some of these genes are positively correlated with PC1 while others are negatively correlated. This observation implies that PC1 is more than simply a gross measure of the relative contribution of muscle cell RNA in any given sample. Taken as a whole, these observations indicate that many of the gene expression differences in this study reflect bona fide differences among muscle cell transcriptional programs rather than contamination from non-muscle mRNAs.

To extend on this analysis, we performed Gene Ontology (GO) analysis (Huang et al., 2009a) on every pairwise comparison of skeletal muscle tissues. We then consolidated these results by identifying GO terms that were repeatedly enriched in pairwise comparisons. Figure 2—source data 1 summarizes these results for all 11 skeletal muscle tissues. Consistent with the principal component analysis described above, many of these GO terms were related to structural components of the sarcomere, including Z-disc, A-band, M-band, and others. We also observed enrichment in terms related to extracellular proteins such as Integrin, Collagen, and Fibronectin. More general terms were also enriched, including various pathways involved in mitochondrial function, fatty acid metabolism, and neuromuscular junction assembly. To extend on these observations, we performed co-expression analysis using WGCNA (Langfelder and Horvath, 2008). As an internal control, we first identified clusters of genes that differentiate smooth, cardiac, skeletal muscle cluster one and skeletal muscle cluster 2 (Figure 2—figure supplement 1A). As expected, GO analysis of the most statistically significant clusters of co-expressed genes (Supplementaryl file 4) included genes conventionally associated with smooth, cardiac, and skeletal muscle physiology respectively. Co-expression analysis of all 17 mouse tissues revealed a smaller number of statistically significant gene clusters (Figure 2—figure supplement 1B), and analysis of the resulting gene lists revealed enrichment in genes typically believed to be specific for skeletal muscle tissue (Supplementary file 5). These observations are consistent with the notion that many of the gene expression differences observed herein are genuine products of differential muscle cell gene expression. Moreover, close examination of the lists of genes identified by principal component, gene ontology, and co-expression analyses reveals that the most informative genes regarding skeletal muscle identity come from families of genes known for their function in skeletal muscle, such as Troponin, Tropomyosin, Calsequestrin, Myosin heavy chain (Myh), and Myosin light chain.

Skeletal muscle is comprised of myofibers which are typically classified into one of four types in mice based on their Myh expression (Haizlip et al., 2015). We used our data to quantify the relative abundance of different fiber types across skeletal muscle tissues (Figure 3A). These results were cross-correlated with legacy data from histological studies, showing close agreement (Figure 3—figure supplement 1). One hypothesis is that fiber type composition (i.e. the relative amounts of fast versus slow twitch fibers) establishes the observed diversity of gene expression profiles. To test this, we clustered skeletal muscle tissues based on similarity of Myh expression. The resulting dendrograms are grossly similar to those based on global gene expression (Figure 3B; compare with Figure 1D), as predominantly fast twitch muscles expressing high levels of Myh4 (Type IIB fibers) tend to cluster together. Nevertheless, clustering within the two major skeletal muscle groups reveals important differences, such as the high similarity of diaphragm and FDB based on Myh expression, but their dissimilarity based on global gene expression (Figure 1E). Similarly, the clustering of tongue and extraocular eye muscles varies considerably depending on whether Myh or global gene expression establishes the pairwise Euclidean distance.

Figure 3 with 1 supplement see all

Download asset Open asset

To explore these observations further, we made use of a single-fiber microarray study that identified genes enriched in slow (Type I) versus fast twitch (Type IIB) skeletal muscle fibers (Chemello et al., 2011). These legacy data reveal that Myostatin (Mstn) is almost exclusively expressed in fast twitch fibers. Therefore, if fiber type composition determines gene expression patterns, we would expect close correlation between Mstn and Myh4. In actuality, there is only moderate correlation (R² = 0.47, Figure 3C). Expanding on this result, we find weak correlation between genes enriched in fast twitch fibers and Myh4 (median R² = 0.132, Figure 3D). The low correlation between fiber type composition and gene expression signatures is also seen in slow twitch fibers (Figure 3E and F). Taken together, we conclude that fiber type based on Myh expression contributes to tissue-specific gene expression but is insufficient to establish the diversity of transcriptional patterns we observe.

Alternatively, developmental history may play an essential role in defining gene expression patterns in adult skeletal muscle. Hox genes are a family of transcription factors that establish anterior/posterior patterning and skeletal muscle specification during development (Krumlauf, 1994). We therefore examined the expression of Hox genes in muscle tissues (Figure 4A). Clustering of tissues based on Hox gene expression reveals that the head and neck muscles cluster more closely with cardiac tissues than other skeletal muscles. Similarly, the diaphragm is more similar to the aorta than limb skeletal muscle. This is in marked contrast to clustering by the entire transcriptome, where each skeletal muscle is more similar to other skeletal muscles than any cardiac or smooth tissue. Hox gene expression in muscle recapitulates the developmental history of these tissues with respect to anterior/posterior axis formation. But, as these clusters are significantly different from those seen when comparing whole transcriptome expression patterns (Figure 1D), we conclude that Hox genes are insufficient to explain the mRNA diversity among adult skeletal muscle tissues.

Figure 4

Download asset Open asset

Similar to Hox family members, many developmentally significant genes maintain expression in adult muscle tissues. These include the myogenic regulatory factor Myf6 that is expressed in, and exclusive to, skeletal muscle (Figure 4B). Related differentiation factors, Myod1, Myf5, and Myog, are also expressed in adult muscle, albeit at roughly 10-fold lower levels than Myf6. Hotair is a noncoding RNA that regulates Hox gene expression in development (Tsumagari et al., 2013); as expected, it is expressed in all limb muscles (Figure 4C). In contrast, Pitx2 is a transcription factor with a key role in driving extraocular muscle development (Zhou et al., 2012), and Lhx2 is a transcription factor important for masseter muscle development (Buckingham and Rigby, 2014). The expression of both genes is consistent with regulatory activity that continues into adulthood and may contribute to maintaining proper mRNA expression patterns (Figure 4D and E). Notably, Lhx2 is also involved in limb muscle development (Hobert and Westphal, 2000); it is of interest to determine how and why its expression is maintained in adult masseter but not adult limb tissues.

We speculate that computational modeling of transcription factor (TF) networks may resolve the mechanisms underlying tissue-specific mRNA profiles, particularly since neither developmental history nor fiber type composition are sufficient to explain the diversity of the skeletal muscle transcriptome. As a first step to this end, we used Ingenuity Pathway Analysis (IPA) (Krämer et al., 2014) to predict TFs upstream of differentially expressed genes among skeletal muscles. We then consolidated these predictions by restricting our analysis to TFs that are predicted to drive differential gene expression in at least five of the 11 total skeletal muscle tissues. A summary of these results is provided in Table 2, which includes 20 TFs predicted to contribute to skeletal muscle specialization. A simplified visual display of these data is provided in Figure 5, which highlights a complex web of 1) predicted upstream transcription factors, 2) numerous differentially expressed genes that we predict act as effectors of tissue specialization (including Myostatin and Vegfa), and 3) downstream processes involved in skeletal muscle disease and physiology. We note that Smarca4 is predicted to be upstream of tissue-specific gene expression in nine of 11 tissues, and that it ranked as the most statistically confident prediction by IPA in four of these tissues. As Smarca4 (Brg1) is involved in early transcriptional patterning of skeletal muscle (Albini et al., 2015), we believe it is a promising candidate for establishing transcriptional specialization in skeletal muscles.

Figure 5

Download asset Open asset

The generation of tissue-specific promoter lines in transgenic mice will facilitate testing these candidates. Therefore, we identified transcripts expressed in a tissue-specific fashion in skeletal muscle. Plotting tissue specificity versus average expression reveals a bimodal distribution among all transcripts (Figure 1—figure supplement 6A and B). The majority of transcripts in skeletal muscle are expressed evenly across most skeletal muscle tissues, consistent with the notion that there is a minimal set of genes required to form sarcomeres and to generate contractile force. Nevertheless, a smaller set of transcripts (~5%) show nearly exclusive expression in a single skeletal muscle tissue. Manual examination of these tissue-specific genes reveals that many are found in head and neck muscles that have the most divergent developmental history of tissues in this study. Nonetheless, we found examples of tissue-specific genes in the diaphragm (Figure 1—figure supplement 6C) and the FDB (Figure 1—figure supplement 6D). A list of the most specific genes for any given muscle tissue is presented in Supplementary file 6. We note that one of the most specific genes for soleus, Myh7, has been validated previously using measures of protein expression (Burkholder et al., 1994). Nevertheless, we caution readers that independent validation using alternative approaches is recommended for the observations herein.

The presence of tissue-specific gene expression is consistent with a skeletal muscle transcriptome that is considerably more complex than previously appreciated. Moreover, the unprecedented depth of sequencing coverage in muscle permits the discovery of heretofore-unannotated transcripts. Manual examination of ‘gapped reads’ spanning two or more loci in the genome indicates that hundreds to thousands of splicing events occur in muscle that have not been previously described (Figure 1—figure supplement 7A). The majority of these novel splicing events result in either inclusion of novel exons or the exclusion of exons previously thought to be constitutive (Figure 1—figure supplement 7B). To validate this observation, we designed oligonucleotide primers (Figure 1—figure supplement 7C) specific for two novel exons of Myosin light chain kinase 4 (Mylk4), a gene with high expression (>200 FPKM) in many skeletal muscle tissues. PCR amplification and molecular cloning confirmed that these putative exons are included in full-length Mylk4 transcripts in two different skeletal muscle tissues, EDL and soleus (Figure 1—figure supplement 7D). As the previously canonical splicing event linking exons 2 and 3 is detected by RNAseq in every muscle sample in this study, albeit at levels below the threshold for detection in RT-PCR, we conclude that transcripts including these putative exons are in actuality the predominant species of Mylk4 mRNA. A list of all novel splice junctions is provided in Supplementary file 7 and 8.

There are significant differences in fiber type composition between analogous muscle tissues of different mammals. For example, mouse soleus is a mixture of fast- and slow fibers (Figure 3A), while rat soleus is almost entirely slow twitch (Figure 6A). Because of this, we asked whether gene expression differences in mice are conserved across species. Expression of orthologous genes in mouse versus rat tissues has moderate levels of correlation (R² >0.6, Figure 6B), despite the difference in fiber type composition noted above. This reinforces the observation that tissue identity, rather than fiber-type composition, drives transcriptome diversity in muscle. Moreover, the vast majority of genes differentially expressed between EDL and soleus in both mice and rats changed in the same direction (Figure 6C). Taking this observation one step further, the fold change of all genes differentially expressed in mouse EDL compared to soleus (Figure 6D) is largely consistent with the fold change of same genes in rat EDL versus soleus (Figure 6E). Sex differences did not dramatically influence differential gene expression between EDL and soleus (Figure 6—figure supplement 1). Fewer than 3% of transcripts were differentially expressed between male and female rat EDL (2.7%) and male and female rat soleus (1.9%). Of these differentially expressed genes, most are up-regulated in males. This observation agrees with previous studies (Roth et al., 2002) and is consistent with the possibility that androgen response elements influence sex-specific gene expression differences in skeletal muscle. These results indicate that differentially expressed genes are largely conserved between mice and rats and suggest that these data may predict gene expression in related species.

Figure 6 with 1 supplement see all

Download asset Open asset

Taken as a whole, there is considerable variance in gene expression profiles among skeletal muscle tissues, in stark disagreement with the assumptions of previous gene expression atlases. In the remainder of this paper, we will explore several analyses illustrating the utility of these data as resource for generating testable hypotheses related to tissue specialization.

Based on the likely conservation of gene expression patterns in humans, we speculate that genes associated with disease and those encoding drug targets will be of particular importance to follow-up studies. Of the roughly 23,000 genes encoded by the mouse genome, over 50% are differentially expressed among mouse skeletal muscle tissues. Of these, 3370 differentially expressed genes have human orthologs associated with disease, and 556 of those encode molecular targets of drugs on the market today (Figure 7A). These genes may contribute to the molecular mechanisms underlying differential disease susceptibility and pharmaceutical sensitivity in skeletal muscle tissues. As a resource for investigators, we provide a list of all differentially expressed genes specifically involved in human skeletal muscle disorders (Supplementary file 9).

Figure 7 with 2 supplements see all

Download asset Open asset

To give one example of differential disease susceptibility, the aberrant expression of an embryonic isoform of Pyruvate kinase (Pkm) is involved in the mechanism of myotonic dystrophy (Gao and Cooper, 2013). Our data show that isoforms of Pkm are up-regulated by several standard deviations in EDL compared to all other muscle groups (Figure 7B). Myotonic dystrophy disrupts normal splicing and pathologically elevates Pkm, which in turn disrupts normal metabolism, decreasing oxygen consumption and increasing glucose consumption. Based on these observations, elevated expression of Pkm in adult tissues is hypothesized to be a critical step in the pathology of myotonic dystrophy (Gao and Cooper, 2013). Since Pkm is expressed much higher in EDL than all other muscle types (Figure 7B), we predict that EDL would be more sensitive to degeneration than other muscles. As myotonic dystrophy most dramatically affects certain subsets of muscle tissues, these observations suggest testable hypotheses regarding the underlying mechanism of disease susceptibility. Consistent with this possibility, we note with great interest that in mouse models, EDL is considerably more susceptible to muscle weakness than either diaphragm or soleus (Moyer et al., 2011).

In addition to disease susceptibility, these data may help explain differential drug sensitivity in muscle. To give one example, the drug chlorzoxazone (brand name: Lorzone) is used to treat muscle spasms. It is thought to act on the central nervous system (CNS) by regulating a potassium channel encoded by the gene Kcnma1 (Dong et al., 2006). Although Kcnma1 is expressed throughout the central nervous system (ENCODE Project Consortium, 2012), it is also found at comparable levels in most skeletal muscle tissues (Figure 7C). The two exceptions are extraocular eye muscles and the soleus that have greater than three-fold higher Kcnma1 expression than other muscle tissues. The differential abundance of chlorozoxazone’s target could influence either the efficacy of this drug or the severity of its side effects in different muscles. The expression in skeletal muscle of mRNAs encoding chlorzoxazone’s protein target calls into question the assumption in the literature that this drug acts exclusively through the CNS. Moreover, given that some commonly prescribed drugs, such as glucocorticoids, cause muscle wasting in specific subsets of muscle tissues (Schakman et al., 2008), we speculate that this data set will be a valuable resource for exploring the mechanisms underlying differential drug sensitivity among skeletal muscles.

Skeletal muscles are endocrine tissues, secreting numerous hormones that influence the physiology and metabolism throughout the body. Termed ‘myokines’, these signaling molecules are key regulators of human health and disease (Lightfoot and Cooper, 2016). The total number of myokines is currently unknown. Our data contribute to this field by comprehensively defining mRNA expression of candidate myokines for future study (Figure 7—figure supplement 1 and Supplementary file 10). Dozens of genes encoding secreted proteins are differentially expressed among skeletal muscles at relatively high levels (N = 42 unique genes, FPKM >10). One notable example is Vegfa (q-value ~10⁻³⁰), a gene involved in angiogenesis, cardiac disease, cancer progression, and many other normal and pathological processes (Smith et al., 2015). Average expression of its predominant isoform is in the top 98^th percentile of all transcripts expressed by skeletal muscle. Moreover, maximal expression of Vegfa in the diaphragm is nearly 10-fold greater than its minimal expression in the FDB, indicating that there are muscle-specific mechanisms for regulating Vegfa levels in particular and myokine levels in general. The serum concentration of VEGF-A in healthy adult humans is considerably greater than serum levels of IL6, a canonical myokine (Lightfoot and Cooper, 2016). Given that roughly 40% of the human body is comprised of skeletal muscle expressing high levels of Vegfa, we speculate that muscle may be a heretofore under-appreciated source of VEGF-A in circulation. Whether Vegfa expression by skeletal muscle has endocrine as well as paracrine functions is unknown.

Regenerative medicine has made considerable progress in generating skeletal muscle from stem cells (Qazi et al., 2015). Nevertheless, engineered tissues have important deficiencies in generating sufficient force and forming appropriate neuromuscular synapses (Juhas and Bursac, 2013). Given the extensive diversity observed among skeletal muscle tissues, we speculate that the inability to form proper synaptic connections in engineered tissue may be due to the expression of inappropriate or incomplete transcriptional programs. In other words, differentiating stem cells into generic skeletal muscle may not recapitulate all the cues necessary for muscle-specific synapse formation. Therefore, we examined the differential expression of genes implicated in synapse assembly (Figure 7—figure supplement 2). Dozens of transcripts involved in synapse formation are expressed at high levels (FPKM >10), suggesting that they are skeletal muscle mRNAs, rather than contamination from nearby neurons. One illustrative example, Fbxo45, is an E3 ligase involved in synapse formation. Mouse knock-outs of Fbxo45 have disrupted neuromuscular junction (NMJ) formation in the diaphragm, resulting in early lethality (Saiga et al., 2009). Moreover, the C. elegans ortholog, FSN-1, also disrupts NMJ formation, with some synapses being over-developed while others are under-developed (Liao et al., 2004). We speculate that this protein and its orthologs may play a conserved role in tissue-specific NMJ assembly.

General textbook discussions of skeletal muscle typically focus on developmental patterning, neuromuscular synapse physiology, or the biophysics of contractile functions (Alberts, 2014). This reflects the generally held belief that adult skeletal muscle is interesting only as a mechanical output of the nervous system. Functional genomics studies have acted on this assumption to the extent that every gene expression atlas generated to-date has selected at most one skeletal muscle as representative of the entire family of tissues. As such, the null hypothesis of this study was that gene expression profiles would be largely similar among skeletal muscle tissues.

However, as more than 50% of transcripts are differentially expressed among skeletal muscles (Figure 1C), and 13% of transcripts are differentially expressed between any two skeletal muscle tissues on average (Figure 1E), the data are entirely inconsistent with the null hypothesis. These results indicate that there is no such thing as a representative skeletal muscle tissue. Instead, skeletal muscle should be viewed as a family of related tissues with a common contractile function but widely divergent physiology, metabolism, morphology, and developmental history. Based on these antecedents, it comes as no surprise that the transcriptional programs maintaining skeletal muscle specialization in adults are highly divergent as well.

This study is the first systematic examination of transcriptome diversity in skeletal muscle. At greater than 200 million aligned short nucleotide reads per tissue and six biological replicates apiece, this data set is unprecedented in its scope, accuracy, and reproducibility (Figure 1—figure supplement 4, Figure 1—figure supplement 4, and Figure 3—figure supplement 1). Moreover, the depth of sequencing allows the detection of previously unannotated transcripts that may play a role in muscle physiology (Figure 1—figure supplement 2 and Figure 1—figure supplement 7, Supplementary file 7 and 8).

Besides establishing that skeletal muscles have considerable differences in their transcriptomes, the key significance of this paper will be as a resource for future studies. Therefore, we have made our analyzed data freely available (http://muscledb.org), and all raw data may be downloaded from NCBI’s GEO. This resource will allow investigators to perform analyses beyond the scope of this paper, such as generating muscle-specific Cre-recombinase mouse strains for genetically manipulating specific muscle groups. Most importantly, these data will provide the foundation for computational modeling of transcription factor networks, a method we believe will uncover the genetic mechanisms that establish and maintain muscle specialization. To this end, we have used principal component analysis (Figure 2) and related approaches (Figure 1—figure supplement 6 and Figure 1—figure supplement 7) to show that expression of key skeletal muscle genes including different versions of Troponin, Tropomyosin, and Calsequestrin are highly predictive of skeletal muscle identity. Moreover, we used pathway analysis (Figure 5 and Table 2) to identify 20 candidate transcription factors that may drive transcriptional specialization in muscle cells.

One potential criticism is that gene expression does not always predict protein level and therefore function. We acknowledge that many processes besides steady-state mRNA levels regulate protein expression. Nevertheless, regulation of mRNA expression is unquestionably of biological importance in muscle cells, and transcriptional profiling predicts the majority of protein expression levels even in highly dynamic settings (Robles et al., 2014). Moreover, the larger dynamic range of RNAseq measurements permits a more comprehensive description of expression profiles than would be possible with existing proteomic technology. We look forward to proteomic studies making use of these mRNA data, especially the identification of novel spliceforms, to generate improved catalogs of protein expression in skeletal muscle.

We further acknowledge that the samples collected and analyzed in this study are bulk tissues rather than single-fiber preparations, and that contaminating tissues such as vasculature and immune cells may influence some gene expression measurements. However, our data agrees with the few single-fiber profiling papers in the literature (Chemello et al., 2011), indicating that this potential bias is unlikely to confound the major observations herein. Consistent with this, the genes whose expression is most predictive of skeletal muscle identity are almost unanimously canonical skeletal muscle genes (Supplementary file 3). Furthermore, we emphasize that the majority of studies in the field use bulk tissues rather than single cell preparations. As such, our experimental design yields the greatest possible consistency with previous and future studies. In short, we believe whole tissue expression profiling provides a critical reference point and the rationale for follow-up work examining single-fiber gene expression.

Acquired and genetic diseases show remarkable selectivity in which muscles they affect and which muscles they spare. For example, Duchenne muscular dystrophy severely affects the diaphragm and proximal limb extensors, while oculopharyngeal dystrophy causes weakness in the neck, facial, and extraocular muscles (Emery, 2002). At present, there is no satisfying explanation for how this occurs. A reasonable hypothesis is that intrinsic properties of muscle cells, such as gene expression, determine their sensitivity to different pathological mechanisms. These data are a starting point for future studies on how specialized transcriptional programs in muscles are maintained and how they ultimately influence disease. As an illustrative example, we note that the naturally elevated expression of Pkm in EDL may in part explain how this muscle is most dramatically affected in mouse models of myotonic dystrophy (Figure 7).

Finally, skeletal muscle is an endocrine organ that regulates many normal and pathological processes, including sleep, bone health, diabetes, cancer, and cardiovascular disease (Giudice and Taylor, 2017; Iizuka et al., 2014; Karsenty and Olson, 2016). At roughly 40% of an adult human’s body weight, skeletal muscle has an enormous capacity to influence other tissues through the expression of local or systemic signaling molecules. This study reveals extensive differential expression of putative myokines with largely unexplored functional significance (Figure 7—figure supplement 1 and Supplementary file 10). As such, we predict these data will be instrumental in future studies of the endocrine mechanisms through which skeletal muscle regulates health and disease.

RNA Sequencing data have been deposited in GEO under accession code GSE100505. Analyzed data are available on http://muscledb.org.

1. 1. Terry EE
   2. Zhang X
   3. Hoffmann C
   4. Hughes LD
   5. Lewis SA
   6. Li J
   7. Wallace MJ
   8. Riley LA
   9. Douglas CM
   10. Gutierrez-Montreal MA
   11. Lahens NF
   12. Gong M
   13. Andrade F
   14. Esser KA
   15. Hughes ME

   (2018) Transcriptional Profiling Reveals Extraordinary Diversity Among Skeletal Muscle Tissues

   Publicly available at the NCBI Gene Expression Omnibus (accession no: GSE100505).

   http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE100505

1. 1. Aare S
   2. Ochala J
   3. Norman HS
   4. Radell P
   5. Eriksson LI
   6. Göransson H
   7. Chen YW
   8. Hoffman EP
   9. Larsson L

   (2011) Mechanisms underlying the sparing of masticatory versus limb muscle function in an experimental critical illness model

   Physiological Genomics 43:1334–1350.

   https://doi.org/10.1152/physiolgenomics.00116.2011

   • PubMed
   • Google Scholar
2. 1. Acharyya S
   2. Butchbach ME
   3. Sahenk Z
   4. Wang H
   5. Saji M
   6. Carathers M
   7. Ringel MD
   8. Skipworth RJ
   9. Fearon KC
   10. Hollingsworth MA
   11. Muscarella P
   12. Burghes AH
   13. Rafael-Fortney JA
   14. Guttridge DC

   (2005) Dystrophin glycoprotein complex dysfunction: a regulatory link between muscular dystrophy and cancer cachexia

   Cancer Cell 8:421–432.

   https://doi.org/10.1016/j.ccr.2005.10.004

   • PubMed
   • Google Scholar
3. Book

   1. Alberts B

   (2014)

   Molecular Biology of the Cell (Sixth)

   New York, NY: Garland Science.

   • Google Scholar
4. 1. Albini S
   2. Coutinho Toto P
   3. Dall'Agnese A
   4. Malecova B
   5. Cenciarelli C
   6. Felsani A
   7. Caruso M
   8. Bultman SJ
   9. Puri PL

   (2015) Brahma is required for cell cycle arrest and late muscle gene expression during skeletal myogenesis

   EMBO Reports 16:1037–1050.

   https://doi.org/10.15252/embr.201540159

   • PubMed
   • Google Scholar
5. 1. Aravamudan B
   2. Mantilla CB
   3. Zhan WZ
   4. Sieck GC

   (2006) Denervation effects on myonuclear domain size of rat diaphragm fibers

   Journal of Applied Physiology 100:1617–1622.

   https://doi.org/10.1152/japplphysiol.01277.2005

   • PubMed
   • Google Scholar
6. 1. Barreiro E
   2. Gea J

   (2016) Molecular and biological pathways of skeletal muscle dysfunction in chronic obstructive pulmonary disease

   Chronic Respiratory Disease 13:297–311.

   https://doi.org/10.1177/1479972316642366

   • PubMed
   • Google Scholar
7. 1. Buckingham M
   2. Rigby PW

   (2014) Gene regulatory networks and transcriptional mechanisms that control myogenesis

   Developmental Cell 28:225–238.

   https://doi.org/10.1016/j.devcel.2013.12.020

   • PubMed
   • Google Scholar
8. 1. Burkholder TJ
   2. Fingado B
   3. Baron S
   4. Lieber RL

   (1994) Relationship between muscle fiber types and sizes and muscle architectural properties in the mouse hindlimb

   Journal of Morphology 221:177–190.

   https://doi.org/10.1002/jmor.1052210207

   • PubMed
   • Google Scholar
9. 1. Chemello F
   2. Bean C
   3. Cancellara P
   4. Laveder P
   5. Reggiani C
   6. Lanfranchi G

   (2011) Microgenomic analysis in skeletal muscle: expression signatures of individual fast and slow myofibers

   PLoS One 6:e16807.

   https://doi.org/10.1371/journal.pone.0016807

   • PubMed
   • Google Scholar
10. 1. Chen YW
    2. Zhao P
    3. Borup R
    4. Hoffman EP

    (2000) Expression profiling in the muscular dystrophies: identification of novel aspects of molecular pathophysiology

    The Journal of Cell Biology 151:1321–1336.

    https://doi.org/10.1083/jcb.151.6.1321

    • PubMed
    • Google Scholar
11. 1. Ciciliot S
    2. Rossi AC
    3. Dyar KA
    4. Blaauw B
    5. Schiaffino S

    (2013) Muscle type and fiber type specificity in muscle wasting

    The International Journal of Biochemistry & Cell Biology 45:2191–2199.

    https://doi.org/10.1016/j.biocel.2013.05.016

    • PubMed
    • Google Scholar
12. 1. Clark CJ
    2. Cochrane LM
    3. Mackay E
    4. Paton B

    (2000) Skeletal muscle strength and endurance in patients with mild COPD and the effects of weight training

    European Respiratory Journal 15:92–97.

    https://doi.org/10.1183/09031936.00.15109200

    • PubMed
    • Google Scholar
13. 1. Colantuoni C
    2. Jeon OH
    3. Hyder K
    4. Chenchik A
    5. Khimani AH
    6. Narayanan V
    7. Hoffman EP
    8. Kaufmann WE
    9. Naidu S
    10. Pevsner J

    (2001) Gene expression profiling in postmortem rett syndrome brain: differential gene expression and patient classification

    Neurobiology of Disease 8:847–865.

    https://doi.org/10.1006/nbdi.2001.0428

    • PubMed
    • Google Scholar
14. 1. Dong DL
    2. Luan Y
    3. Feng TM
    4. Fan CL
    5. Yue P
    6. Sun ZJ
    7. Gu RM
    8. Yang BF

    (2006) Chlorzoxazone inhibits contraction of rat thoracic aorta

    European Journal of Pharmacology 545:161–166.

    https://doi.org/10.1016/j.ejphar.2006.06.063

    • PubMed
    • Google Scholar
15. 1. Durinck S
    2. Spellman PT
    3. Birney E
    4. Huber W

    (2009) Mapping identifiers for the integration of genomic datasets with the R/Bioconductor package biomaRt

    Nature Protocols 4:1184–1191.

    https://doi.org/10.1038/nprot.2009.97

    • PubMed
    • Google Scholar
16. 1. Emery AEH

    (2002) The muscular dystrophies

    The Lancet 359:687–695.

    https://doi.org/10.1016/S0140-6736(02)07815-7

    • Google Scholar
17. 1. ENCODE Project Consortium

    (2012) An integrated encyclopedia of DNA elements in the human genome

    Nature 489:57–74.

    https://doi.org/10.1038/nature11247

    • PubMed
    • Google Scholar
18. 1. Evangelista D
    2. Avino M
    3. Tripathi KP
    4. Guarracino MR

    (2015)

    Lecture Notes in Computer Science

    273–284, A Web Resource on Skeletal Muscle Transcriptome of Primates, Lecture Notes in Computer Science, Cham, Springer.

    • Google Scholar
19. 1. Fisher A
    2. Caffo B
    3. Schwartz B
    4. Zipunnikov V

    (2016) Fast, exact Bootstrap principal component analysis for p > 1 million

    Journal of the American Statistical Association 111:846–860.

    https://doi.org/10.1080/01621459.2015.1062383

    • PubMed
    • Google Scholar
20. 1. Gagnon P
    2. Maltais F
    3. Bouyer L
    4. Ribeiro F
    5. Coats V
    6. Brouillard C
    7. Noël M
    8. Rousseau-Gagnon M
    9. Saey D

    (2013) Distal leg muscle function in patients with COPD

    COPD 10:235–242.

    https://doi.org/10.3109/15412555.2012.719047

    • PubMed
    • Google Scholar
21. 1. Gao Z
    2. Cooper TA

    (2013) Reexpression of pyruvate kinase M2 in type 1 myofibers correlates with altered glucose metabolism in myotonic dystrophy

    PNAS 110:13570–13575.

    https://doi.org/10.1073/pnas.1308806110

    • PubMed
    • Google Scholar
22. 1. Giudice J
    2. Taylor JM

    (2017) Muscle as a paracrine and endocrine organ

    Current Opinion in Pharmacology 34:49–55.

    https://doi.org/10.1016/j.coph.2017.05.005

    • PubMed
    • Google Scholar
23. 1. Grant GR
    2. Farkas MH
    3. Pizarro AD
    4. Lahens NF
    5. Schug J
    6. Brunk BP
    7. Stoeckert CJ
    8. Hogenesch JB
    9. Pierce EA

    (2011) Comparative analysis of RNA-Seq alignment algorithms and the RNA-Seq unified mapper (RUM)

    Bioinformatics 27:2518–2528.

    https://doi.org/10.1093/bioinformatics/btr427

    • Google Scholar
24. 1. Greising SM
    2. Medina-Martínez JS
    3. Vasdev AK
    4. Sieck GC
    5. Mantilla CB

    (2015) Analysis of muscle fiber clustering in the diaphragm muscle of sarcopenic mice

    Muscle & Nerve 52:76–82.

    https://doi.org/10.1002/mus.24641

    • PubMed
    • Google Scholar
25. 1. Haizlip KM
    2. Harrison BC
    3. Leinwand LA

    (2015) Sex-based differences in skeletal muscle kinetics and fiber-type composition

    Physiology 30:30–39.

    https://doi.org/10.1152/physiol.00024.2014

    • PubMed
    • Google Scholar
26. 1. Hermans G
    2. Van den Berghe G

    (2015) Clinical review: intensive care unit acquired weakness

    Critical Care 19:274.

    https://doi.org/10.1186/s13054-015-0993-7

    • Google Scholar
27. 1. Hobert O
    2. Westphal H

    (2000) Functions of LIM-homeobox genes

    Trends in Genetics 16:75–83.

    https://doi.org/10.1016/S0168-9525(99)01883-1

    • PubMed
    • Google Scholar
28. 1. Hochberg Y
    2. Benjamini Y

    (1990) More powerful procedures for multiple significance testing

    Statistics in Medicine 9:811–818.

    https://doi.org/10.1002/sim.4780090710

    • PubMed
    • Google Scholar
29. 1. Huang daW
    2. Sherman BT
    3. Lempicki RA

    (2009a) Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources

    Nature Protocols 4:44–57.

    https://doi.org/10.1038/nprot.2008.211

    • PubMed
    • Google Scholar
30. 1. Huang daW
    2. Sherman BT
    3. Lempicki RA

    (2009b) Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists

    Nucleic acids research 37:1–13.

    https://doi.org/10.1093/nar/gkn923

    • PubMed
    • Google Scholar
31. Software

    1. Hughes L

    (2017) Database of expression of muscle tissue, version c500624

    Github.

    https://github.com/flaneuse/muscleDB
32. 1. Hughes LD
    2. Lewis SA
    3. Hughes ME

    (2017) ExpressionDB: An open source platform for distributing genome-scale datasets

    PLoS One 12:e0187457.

    https://doi.org/10.1371/journal.pone.0187457

    • PubMed
    • Google Scholar
33. 1. Hughes ME
    2. DiTacchio L
    3. Hayes KR
    4. Vollmers C
    5. Pulivarthy S
    6. Baggs JE
    7. Panda S
    8. Hogenesch JB

    (2009) Harmonics of circadian gene transcription in mammals

    PLoS genetics 5:e1000442.

    https://doi.org/10.1371/journal.pgen.1000442

    • PubMed
    • Google Scholar
34. 1. Iizuka K
    2. Machida T
    3. Hirafuji M

    (2014) Skeletal muscle is an endocrine organ

    Journal of Pharmacological Sciences 125:125–131.

    https://doi.org/10.1254/jphs.14R02CP

    • PubMed
    • Google Scholar
35. 1. Juhas M
    2. Bursac N

    (2013) Engineering skeletal muscle repair

    Current Opinion in Biotechnology 24:880–886.

    https://doi.org/10.1016/j.copbio.2013.04.013

    • PubMed
    • Google Scholar
36. 1. Karsenty G
    2. Olson EN

    (2016) Bone and muscle endocrine functions: unexpected paradigms of inter-organ communication

    Cell 164:1248–1256.

    https://doi.org/10.1016/j.cell.2016.02.043

    • PubMed
    • Google Scholar
37. 1. Krumlauf R

    (1994) Hox genes in vertebrate development

    Cell 78:191–201.

    https://doi.org/10.1016/0092-8674(94)90290-9

    • PubMed
    • Google Scholar
38. 1. Krämer A
    2. Green J
    3. Pollard J
    4. Tugendreich S

    (2014) Causal analysis approaches in Ingenuity Pathway Analysis

    Bioinformatics 30:523–530.

    https://doi.org/10.1093/bioinformatics/btt703

    • PubMed
    • Google Scholar
39. 1. Landgraf P
    2. Rusu M
    3. Sheridan R
    4. Sewer A
    5. Iovino N
    6. Aravin A
    7. Pfeffer S
    8. Rice A
    9. Kamphorst AO
    10. Landthaler M
    11. Lin C
    12. Socci ND
    13. Hermida L
    14. Fulci V
    15. Chiaretti S
    16. Foà R
    17. Schliwka J
    18. Fuchs U
    19. Novosel A
    20. Müller RU
    21. Schermer B
    22. Bissels U
    23. Inman J
    24. Phan Q
    25. Chien M
    26. Weir DB
    27. Choksi R
    28. De Vita G
    29. Frezzetti D
    30. Trompeter HI
    31. Hornung V
    32. Teng G
    33. Hartmann G
    34. Palkovits M
    35. Di Lauro R
    36. Wernet P
    37. Macino G
    38. Rogler CE
    39. Nagle JW
    40. Ju J
    41. Papavasiliou FN
    42. Benzing T
    43. Lichter P
    44. Tam W
    45. Brownstein MJ
    46. Bosio A
    47. Borkhardt A
    48. Russo JJ
    49. Sander C
    50. Zavolan M
    51. Tuschl T

    (2007) A mammalian microRNA expression atlas based on small RNA library sequencing

    Cell 129:1401–1414.

    https://doi.org/10.1016/j.cell.2007.04.040

    • PubMed
    • Google Scholar
40. 1. Langfelder P
    2. Horvath S

    (2008) WGCNA: an R package for weighted correlation network analysis

    BMC Bioinformatics 9:559.

    https://doi.org/10.1186/1471-2105-9-559

    • PubMed
    • Google Scholar
41. 1. Latronico N
    2. Tomelleri G
    3. Filosto M

    (2012) Critical illness myopathy

    Current Opinion in Rheumatology 24:616–622.

    https://doi.org/10.1097/BOR.0b013e3283588d2f

    • PubMed
    • Google Scholar
42. 1. Li J
    2. Grant GR
    3. Hogenesch JB
    4. Hughes ME

    (2015) Considerations for RNA-seq analysis of circadian rhythms

    Methods in Enzymology 551:349–367.

    https://doi.org/10.1016/bs.mie.2014.10.020

    • PubMed
    • Google Scholar
43. 1. Liao EH
    2. Hung W
    3. Abrams B
    4. Zhen M

    (2004) An SCF-like ubiquitin ligase complex that controls presynaptic differentiation

    Nature 430:345–350.

    https://doi.org/10.1038/nature02647

    • PubMed
    • Google Scholar
44. 1. Lightfoot AP
    2. Cooper RG

    (2016) The role of myokines in muscle health and disease

    Current Opinion in Rheumatology 28:661–666.

    https://doi.org/10.1097/BOR.0000000000000337

    • PubMed
    • Google Scholar
45. 1. Lindholm ME
    2. Huss M
    3. Solnestam BW
    4. Kjellqvist S
    5. Lundeberg J
    6. Sundberg CJ

    (2014) The human skeletal muscle transcriptome: sex differences, alternative splicing, and tissue homogeneity assessed with RNA sequencing

    FASEB journal : official publication of the Federation of American Societies for Experimental Biology 28:4571–4581.

    https://doi.org/10.1096/fj.14-255000

    • PubMed
    • Google Scholar
46. 1. Liu L
    2. Cheung TH
    3. Charville GW
    4. Rando TA

    (2015) Isolation of skeletal muscle stem cells by fluorescence-activated cell sorting

    Nature Protocols 10:1612–1624.

    https://doi.org/10.1038/nprot.2015.110

    • PubMed
    • Google Scholar
47. 1. Merrell AJ
    2. Kardon G

    (2013) Development of the diaphragm -- a skeletal muscle essential for mammalian respiration

    FEBS Journal 280:4026–4035.

    https://doi.org/10.1111/febs.12274

    • PubMed
    • Google Scholar
48. 1. Moyer M
    2. Berger DS
    3. Ladd AN
    4. Van Lunteren E

    (2011) Differential susceptibility of muscles to Myotonia and force impairment in a mouse model of myotonic dystrophy

    Muscle & Nerve 43:818–827.

    https://doi.org/10.1002/mus.21988

    • PubMed
    • Google Scholar
49. 1. Murphy M
    2. Kardon G

    (2011) Origin of vertebrate limb muscle: the role of progenitor and myoblast populations

    Current Topics in Developmental Biology 96:1–32.

    https://doi.org/10.1016/B978-0-12-385940-2.00001-2

    • PubMed
    • Google Scholar
50. 1. Noden DM
    2. Francis-West P

    (2006) The differentiation and morphogenesis of craniofacial muscles

    Developmental Dynamics 235:1194–1218.

    https://doi.org/10.1002/dvdy.20697

    • Google Scholar
51. 1. Porter JD
    2. Khanna S
    3. Kaminski HJ
    4. Rao JS
    5. Merriam AP
    6. Richmonds CR
    7. Leahy P
    8. Li J
    9. Andrade FH

    (2001) Extraocular muscle is defined by a fundamentally distinct gene expression profile

    PNAS 98:12062–12067.

    https://doi.org/10.1073/pnas.211257298

    • PubMed
    • Google Scholar
52. 1. Qazi TH
    2. Mooney DJ
    3. Pumberger M
    4. Geissler S
    5. Duda GN

    (2015) Biomaterials based strategies for skeletal muscle tissue engineering: existing technologies and future trends

    Biomaterials 53:502–521.

    https://doi.org/10.1016/j.biomaterials.2015.02.110

    • PubMed
    • Google Scholar
53. 1. Robles MS
    2. Cox J
    3. Mann M

    (2014) In-vivo quantitative proteomics reveals a key contribution of post-transcriptional mechanisms to the circadian regulation of liver metabolism

    PLoS Genetics 10:e1004047.

    https://doi.org/10.1371/journal.pgen.1004047

    • PubMed
    • Google Scholar
54. 1. Roth SM
    2. Ferrell RE
    3. Peters DG
    4. Metter EJ
    5. Hurley BF
    6. Rogers MA

    (2002) Influence of age, sex, and strength training on human muscle gene expression determined by microarray

    Physiological Genomics 10:181–190.

    https://doi.org/10.1152/physiolgenomics.00028.2002

    • PubMed
    • Google Scholar
55. 1. Saiga T
    2. Fukuda T
    3. Matsumoto M
    4. Tada H
    5. Okano HJ
    6. Okano H
    7. Nakayama KI

    (2009) Fbxo45 forms a novel ubiquitin ligase complex and is required for neuronal development

    Molecular and Cellular Biology 29:3529–3543.

    https://doi.org/10.1128/MCB.00364-09

    • PubMed
    • Google Scholar
56. 1. Schakman O
    2. Gilson H
    3. Thissen JP

    (2008) Mechanisms of glucocorticoid-induced myopathy

    Journal of Endocrinology 197:1–10.

    https://doi.org/10.1677/JOE-07-0606

    • PubMed
    • Google Scholar
57. 1. Serrano AL
    2. Jardi M
    3. Suelves M
    4. Klotman PE
    5. Munoz-Canoves P

    (2008) HIV-1 transgenic expression in mice induces selective atrophy of fast-glycolytic skeletal muscle fibers

    Frontiers in Bioscience 13:2797–2805.

    https://doi.org/10.2741/2886

    • PubMed
    • Google Scholar
58. 1. Smith GA
    2. Fearnley GW
    3. Harrison MA
    4. Tomlinson DC
    5. Wheatcroft SB
    6. Ponnambalam S

    (2015) Vascular endothelial growth factors: multitasking functionality in metabolism, health and disease

    Journal of Inherited Metabolic Disease 38:753–763.

    https://doi.org/10.1007/s10545-015-9838-4

    • PubMed
    • Google Scholar
59. 1. Su AI
    2. Cooke MP
    3. Ching KA
    4. Hakak Y
    5. Walker JR
    6. Wiltshire T
    7. Orth AP
    8. Vega RG
    9. Sapinoso LM
    10. Moqrich A
    11. Patapoutian A
    12. Hampton GM
    13. Schultz PG
    14. Hogenesch JB
    15. Ai S

    (2002) Large-scale analysis of the human and mouse transcriptomes

    PNAS 99:4465–4470.

    https://doi.org/10.1073/pnas.012025199

    • PubMed
    • Google Scholar
60. 1. Su AI
    2. Wiltshire T
    3. Batalov S
    4. Lapp H
    5. Ching KA
    6. Block D
    7. Zhang J
    8. Soden R
    9. Hayakawa M
    10. Kreiman G
    11. Cooke MP
    12. Walker JR
    13. Hogenesch JB
    14. Ai S

    (2004) A gene atlas of the mouse and human protein-encoding transcriptomes

    PNAS 101:6062–6067.

    https://doi.org/10.1073/pnas.0400782101

    • PubMed
    • Google Scholar
61. 1. Tiao G
    2. Lieberman M
    3. Fischer JE
    4. Hasselgren PO

    (1997) Intracellular regulation of protein degradation during Sepsis is different in fast- and slow-twitch muscle

    American Journal of Physiology-Regulatory, Integrative and Comparative Physiology 272:R849–R856.

    https://doi.org/10.1152/ajpregu.1997.272.3.R849

    • PubMed
    • Google Scholar
62. 1. Tirrell TF
    2. Cook MS
    3. Carr JA
    4. Lin E
    5. Ward SR
    6. Lieber RL

    (2012) Human skeletal muscle biochemical diversity

    Journal of Experimental Biology 215:2551–2559.

    https://doi.org/10.1242/jeb.069385

    • PubMed
    • Google Scholar
63. 1. Tsumagari K
    2. Baribault C
    3. Terragni J
    4. Chandra S
    5. Renshaw C
    6. Sun Z
    7. Song L
    8. Crawford GE
    9. Pradhan S
    10. Lacey M
    11. Ehrlich M

    (2013) DNA methylation and differentiation: HOX genes in muscle cells

    Epigenetics & Chromatin 6:25.

    https://doi.org/10.1186/1756-8935-6-25

    • PubMed
    • Google Scholar
64. 1. Vissing K
    2. Schjerling P

    (2014) Simplified data access on human skeletal muscle transcriptome responses to differentiated exercise

    Scientific Data 1:140041.

    https://doi.org/10.1038/sdata.2014.41

    • PubMed
    • Google Scholar
65. 1. Zhang B
    2. Horvath S

    (2005) A general framework for weighted gene co-expression network analysis

    Statistical Applications in Genetics and Molecular Biology 4:Article17.

    https://doi.org/10.2202/1544-6115.1128

    • PubMed
    • Google Scholar
66. 1. Zhang R
    2. Lahens NF
    3. Ballance HI
    4. Hughes ME
    5. Hogenesch JB

    (2014) A circadian gene expression atlas in mammals: implications for biology and medicine

    PNAS 111:16219–16224.

    https://doi.org/10.1073/pnas.1408886111

    • PubMed
    • Google Scholar
67. 1. Zhou Y
    2. Gong B
    3. Kaminski HJ

    (2012) Genomic profiling reveals Pitx2 controls expression of mature extraocular muscle contraction-related genes

    Investigative Opthalmology & Visual Science 53:1821–1829.

    https://doi.org/10.1167/iovs.12-9481

    • PubMed
    • Google Scholar

Author details

1. Erin E Terry

   Division of Pulmonary and Critical Care Medicine, Washington University School of Medicine, St. Louis, United States

   Contribution

   Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Project administration, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1334-4238

2. Xiping Zhang

   Department of Physiology and Functional Genomics, University of Florida College of Medicine, Gainesville, United States

   Contribution

   Resources, Investigation, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

3. Christy Hoffmann

   Division of Pulmonary and Critical Care Medicine, Washington University School of Medicine, St. Louis, United States

   Contribution

   Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

4. Laura D Hughes

   Department of Integrative, Structural and Computational Biology, The Scripps Research Institute, La Jolla, United States

   Contribution

   Resources, Data curation, Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

5. Scott A Lewis

   Division of Pulmonary and Critical Care Medicine, Washington University School of Medicine, St. Louis, United States

   Contribution

   Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

6. Jiajia Li

   Division of Pulmonary and Critical Care Medicine, Washington University School of Medicine, St. Louis, United States

   Contribution

   Resources, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

7. Matthew J Wallace

   Division of Pulmonary and Critical Care Medicine, Washington University School of Medicine, St. Louis, United States

   Contribution

   Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

8. Lance A Riley

   Department of Physiology and Functional Genomics, University of Florida College of Medicine, Gainesville, United States

   Contribution

   Resources, Investigation, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

9. Collin M Douglas

   Department of Physiology and Functional Genomics, University of Florida College of Medicine, Gainesville, United States

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

10. Miguel A Gutierrez-Monreal

    Department of Physiology and Functional Genomics, University of Florida College of Medicine, Gainesville, United States

    Contribution

    Investigation, Methodology

    Competing interests

    No competing interests declared

11. Nicholas F Lahens

    Institute for Translational Medicine and Therapeutics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States

    Contribution

    Conceptualization, Formal analysis, Investigation, Methodology, Writing—review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-3965-5624

12. Ming C Gong

    Department of Physiology, University of Kentucky School of Medicine, Lexington, United States

    Contribution

    Conceptualization, Supervision, Methodology, Writing—review and editing

    Competing interests

    No competing interests declared

13. Francisco Andrade

    Department of Physiology, University of Kentucky School of Medicine, Lexington, United States

    Contribution

    Conceptualization, Supervision, Funding acquisition, Methodology, Writing—review and editing

    Competing interests

    No competing interests declared

14. Karyn A Esser

    Department of Physiology and Functional Genomics, University of Florida College of Medicine, Gainesville, United States

    Contribution

    Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Validation, Investigation, Methodology, Project administration, Writing—review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-5791-1441

15. Michael E Hughes

    Division of Pulmonary and Critical Care Medicine, Washington University School of Medicine, St. Louis, United States

    Contribution

    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing

    For correspondence

    michael.hughes@wustl.edu

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-8828-3732

• 8,190

  views

• 1,078

  downloads

• 106

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.