The hypothesis was tested that differential, coregulated transcriptional adaptations of various cellular pathways would occur early with increased mechanical loading of atrophied skeletal muscle and relate to concurrent damage of muscle fibers. Atrophy and slow-to-fast fiber transformation of rat soleus muscle was provoked by 14 days of hindlimb suspension (HS). Subsequent reloading of hindlimbs caused a fourfold increase in the percentage of muscle fibers, demonstrating endomysial tenascin-C staining. Five days after reloading, when 10% of the fibers were damaged, the normal muscle weight and slow-type fiber percentage were reestablished. Microarray analysis revealed major, biphasic patterns of gene expressional alterations with reloading that distinguish between treatments and gene ontologies. Transcript levels of factors involved in protein synthesis and certain proteasomal mRNAs were increased after 1 day of reloading and correlated to the percentage of fibers surrounded by tenascin-C. By contrast, levels of gene messages for fatty acid transporters, respiratory chain constituents, and voltage-gated cation channels were transiently reduced after 1 day of muscle loading and associated with the number of damaged fibers and the regain in muscle weight. This coregulation points toward important retooling of oxidative metabolism and the T- and SR-tubular systems with rebuilding of slow fibers. The observations demonstrate that early nuclear reprogramming with reloading of atrophic soleus muscle is coordinated and links to the processes involved in mechanical damage and regeneration of muscle fibers.
- expression profile
- mechanical stress
muscle deconditioning is a consequence of long-term reductions in load and metabolic activity of skeletal muscle that occur with inactivity due to prolonged bed rest (28, 46) or microgravity (26, 30). The hallmarks of muscle deconditioning are a loss of muscle mass (atrophy), reduced muscle strength, and diminished resistance to fatigue (24, 28, 34, 62). Increasing mechanical load of skeletal muscle is the obvious and efficient measure to induce recovery of mass and functional performance of atrophied muscle (62).
Hindlimb suspension (HS) is a rodent model in which to investigate the structural and functional events involved in atrophy of antigravitational muscles as induced by mechanical unloading and subsequent muscular recovery by reloading (8, 20, 23, 48, 61). HS of rats provokes a significant reduction in mass and cross-sectional area of soleus muscle fibers and a transformation from a slow-oxidative toward a fast-glycolytic phenotype (4, 22, 30, 38, 52, 57). These myocellular changes are accompanied by parallel changes in satellite cells and in the interstitial space involving capillaries and connective tissue (21, 43, 45). The time course of changes indicates that most of the important contractile, metabolic, and cellular adaptations occurring with HS-induced atrophy of rat musculus soleus (m. soleus) become manifest after 1–2 wk of HS (1, 17, 22, 29, 38, 52, 61).
By contrast, the cellular events involved in recovery of atrophic muscle are much less understood. This information would be of clinical importance to plan effective countermeasures for prolonged bed rest or spaceflight. The current data document that access of 35-day HS rats to normal cage activity permits spontaneous recovery of cross-sectional area, fiber type distribution, and aerobic capacity of atrophied m. soleus (23). During the early phase of muscle reloading, mechanical stress and consequent damage of some fibers are apparent. This is indicated by ectopic endomysial expression of the marker of mechanical stress and muscle damage, tenascin-C, and the appearance of central nuclei that are a sign of regenerative processes in damaged muscle fibers (7, 32, 45, 56). Regeneration of damaged fibers, as well as hypertrophy and transformation of muscle fiber types, are therefore part of the normal course of reestablishment of “normal characteristics” of atrophied soleus muscle (41).
With regard to gene regulatory events, it has recently been demonstrated that coincidental expressional changes of multiple gene ontologies, i.e., factors involved in the same cellular pathways, underlie the adaptation to muscle unloading in this model. For instance transcript levels of genes related to metabolism, protein turnover, excitation-contraction coupling, contraction, and cell regulation were altered after HS (3, 11, 38, 42, 57, 58, 60, 64). With reference to the transcriptional mechanisms involved in the mechano-dependent recovery process from muscle atrophy, little information exists. The available data indicate that reloading of atrophied muscle provokes transcript level alterations related to the cytoskeletal-extracellular matrix axes (7, 32, 35, 45, 56), proteolysis (59) as well as changed metabolic protein expression and an increase in total RNA content (15, 40). However, the understanding of the interactions of transcriptional adjustments of the biological processes governing mechano-dependent recovery of atrophied muscle is still sketchy with regard to knowledge on coregulation and identity of affected gene ontologies. Major compositional alterations during muscle fiber transformation and regeneration (23, 25, 45, 59) suggest that expressional alterations of the gene products encoding the individual subunits of remodeled organelle complexes may be coordinated with reloading of atrophied muscle.
The purpose of this study was to characterize the transcriptional reprogramming within the window of the first five days of reloading of 14-day hindlimb suspended rat m. soleus. It was assumed that transcriptional adaptations of cellular pathways whose adjustment is a characteristic of a stable atrophy phenotype are established after 2 wk of HS. With subsequent reloading, we hypothesized that coregulated transcriptional adjustments of gene ontologies would correspond to certain phenotypical alterations of muscle involving myosin-based fiber types, endomysial tenascin-C expression, appearance of central nuclei, and regrowth of atrophied muscle. We tested transcript-phenotype associations with supervised cluster algorithms a posteriori.
Muscle unloading and reloading.
Hindlimbs of young female pathogen-free Wistar rats were unloaded in individual cages by hanging the rats by their tails as described (23). Before the suspension, animals were acclimatized to housing in single cages for 5 days. One set of animals had their hindlimbs suspended for 14 days (termed HS14). After 2 wk of HS, another set of animals was allowed to perform normal cage activity for 1 or 5 days, therefore permitting reloading of the m. soleus (termed HS-R1 and HS-R5, respectively). Other details have been published elsewhere (32). At the end of the HS protocol and the reloading period, rats were weighed and anesthetized by inhalation of 1.5% halothane (oxygen mixture introduced by a face mask), and the m. solei of both hindlimbs were rapidly dissected. The muscles were immediately weighed, frozen in melting-isopentane, and stored at −75°C. Then the soleus-to-body weight ratio was calculated for each animal. Six separate animals each were used per treatment and given time point. Animals of the same age as the rats after HS served as experimental controls (termed C14). The animal experiments were approved by the Institutional Animal Care and Use Committee and carried out in the animal facilities of the Faculté de Médecine, Université Lyon I in Lyon, France, according to the newest guiding principles for research (2).
Fiber type analysis.
Consecutive cross cryosections from the belly portion of the contralateral m. soleus used for microarray analysis were subjected to myosin ATPase staining as described (23). Muscle fibers were classified into major types (I, II) and hybrid type I/II fibers. The histochemical method used was based on the observed difference in pH lability of the myosin ATPase activity of the isomyosins in the different fibers. We counted 1,200 fibers per muscle from five individual animals per C14, HS14, and HS-R5 treatment. Significance of differences in the fiber type distribution between treatments was verified with a one-way ANOVA with the post hoc test of Fisher for least-significant difference [Statistica software package 6.1; StatSoft (Europe), Hamburg, Germany]. Alternatively, immunohistochemical determination of the type I, type II and hybrid I/II type fiber percentage was carried out as described to determine the fiber type distribution for the HS-R1 treatment (32). Control experiments demonstrated that fiber type distribution as determined via ATPase or immunohistochemistry was in good correspondence (r2 = 0.97). Additional experiments with monoclonal antibody F1-652 against embryonic myosin heavy chain (1:20, Developmental Studies Hybridoma Bank; gift of F. Kadi, Örebro University, Örebro, Sweden) were carried out basically as described for the adult myosin isoforms (32). The percentages of endomysial tenascin-C-positive fibers and centrally nucleated fibers, respectively, per total fiber number were determined as described previously (32).
Total RNA was extracted from pooled 25-μm cryosections from the belly portion of the m. solei and subjected to reverse-transcription and hybridization on Atlas Rat 1.2 cDNA microarrays (no. 7854; Clontech Laboratories, Ozyme, France) as described (64). Six microarray experiments were run with samples from different muscles of the C14, HS14, and HS-R1 treatments, and five filters were hybridized with cDNA derived from other muscles of the HS-R5 protocol (see Supplemental Online Fig. 1). 1 Filter sets from two different lots were used. For repetitive use, the membranes were stripped, thereby removing on average >90% of the signal. The raw signal and local background of each probe spot on the membrane, given as the sum of pixels, were estimated with AIDA Array Easy software (Raytest Schweiz, Urdorf, Switzerland). Data sets have been deposited at Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/projects/geo) under sample codes GSM10045—GSM10062, and GSM17357–GSM17361, which link to the series GSE667, GSE668, GSE669, and GSE1084, respectively. For details on minimal information on gene expression see Supplemental Online Fig. 1 and Ref. 9.
Statistical analysis of microarray data.
Differentially expressed genes were identified with two statistical approaches depending on their level of detection. 382 cDNAs were detected 30% above background on all filters from the HS14, HS-R1, or C14 treatment. Statistical significance of level changes for these abundant transcripts was tested as described with permutation analyses on L1 regression models of logarithmized raw signals in scatterplots for the comparison of two treatments (Statistica software package 6.1 and Ref. 64). Residues were exported into Microsoft Excel 2000 and used to test the significance of a trend under application of the sign test (64). The expression ratio for each transcript was estimated from the median of potentiated residues (to the base of 10) from the individual scatterplots. To correct for multiple tests, the False Discovery Rate method introduced by Benjamini et al. (6) was applied at a two-tailed P = 0.05.
Additionally, transcripts that were detected 30% above background in more than four experiments with targets from one treatment but that were detected fewer than two times on the filters probed with cDNA from the compared treatment were identified. This led to 26 additionally detected mRNAs, e.g., aquaporin 4, being analyzed. Signal values were background-corrected and normalized to the sum of signals of 25 “stable” transcript signals from the respective microarray filter scan (see Supplemental Online Fig. 1). Differences between two treatments for these cDNA signals were analyzed from the normalized expression signals with one-way ANOVA using Fisher's post hoc test for least significant difference under multiplicity-error correction using the false discovery rate method (Statistica software package 6.1 and Ref. 6).
Hierarchical cluster analysis.
Supervised hierarchical cluster analysis was carried out with publicly available Cluster software (27) for the normalized signal values of the 355 transcripts whose level was significantly altered upon reloading. Data were log-transformed and centered for mean of genes and arrays. After complete linkage, hierarchical clustering for centered correlations of genes and arrays was carried out. Results were visualized in TreeView and exported into Corel Draw version 10 (Corel) for figure assembly in Microsoft PowerPoint 2000 (Microsoft, Kildare, Ireland). Additionally, graphs created in SigmaPlot 2002 version 8.0 (SPSS) were included in the presentations (see Fig. 2).
Finally, the gene ontologies were identified for which the majority of transcripts were clustered to the same nodes. For each node the ratio of transcript number (per ontology) to the total number of transcripts per ontology was calculated. This value was subjected to centered hierarchical clustering analysis against the node identification and visualized using TreeView software (see Fig. 3 and Ref. 27).
Supervised analysis for transcript-structure relationships of the 355 reloading-affected transcripts was carried out with combined K-means cluster algorithm for 20 nodes and hierarchical cluster analysis for centered correlations. Normalized signal values for those samples for which concurrent expression and fiber (muscle) data were available (32) were log-transformed and mean-centered for genes and arrays before clustering analysis. This led to the inclusion of measures from three to six independent muscle samples per treatment. A relationship of a transcript with a structural parameter (% endomysial tenascin-C-positive fibers, % centrally nucleated fibers, % type II or hybrid-type fibers, soleus-to-body weight ratio) was called significant when it was being grouped into the same K-means node and when the correlation coefficient r from hierarchical clustering was >0.70.
The transcripts were grouped into ontologies based upon the information available through BD Biosciences Clontech-AtlasInfo 3.2 (http://atlasinfo.clontech.com), medline (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi), or EXPASY SWISS-PROT and TrEMBL (http://www.expasy.org/cgi-bin/sprot-search-de). The cataloging into one of these groups was, for some mRNAs, not ultimate as they are involved in multiple pathways. Expression ratios of altered mRNAs were related to the level seen in age controls (set to 1) with the use of Microsoft Excel 2000 and exported in SigmaPlot 2002 version 8.0 (SPSS) for preparation of graphical presentations.
Aliquots corresponding to 600 ng of total RNA were reverse-transcribed using random hexamer primers with the Omniscript Reverse Transcriptase kit (Qiagen, Basel, Switzerland). Real-time PCR amplification reactions were carried out in triplicate on 30-μl aliquots in 96-well plates on an ABI Prism 5700 Sequence Detection System with probe detection via SYBR Green (PE Biosystems, Rotkreuz, Switzerland). Primers for target mRNAs, EF-2, RPL19, CTSL, FABP3, and COX4 were designed with the Primer Express software (PE Biosystems) as described in Wittwer et al. (64). The major 28S rRNA species was used as a reference (standard) in parallel PCR reactions. Volumina corresponding to 6 and 0.6 ng of reverse-transcribed total RNA, respectively, were used for each amplification reaction for target and 28S rRNA. The amount of target mRNA relative to total RNA was estimated as previously described by relating the respective target values to the corresponding 28S reference values (32). Significance of a transcript level alteration as determined by RT-PCR was verified with the appropriate one-tailed statistical test based upon the hypothesis of a codirectional correspondence to the level change as identified with microarray analysis. For transcripts in which the hypothesis of a normal distribution was rejected based upon the Kolmogorov-Smirnov-Lillefors test (Statistica 6.1), a Kruskal-Wallis ANOVA was applied; otherwise a one-way ANOVA statistics test was used.
Muscle mass, fiber transformations, and fiber damage with unloading and reloading.
The muscle-to-body weight ratio of the m. solei used for microarray analysis was significantly reduced by 52% after HS14 (P < 0.05). After 1 day of reloading, the soleus muscle-to-body ratio was not significantly altered, but after 5 days of reloading, it was increased and was not different from control (HS-R5, Fig. 1A). The percentage of endomysial tenascin-C-positive fibers in the muscles studied was nonsignificantly altered after HS14 but was fourfold increased after 1 and 5 days of reloading (HS-R1 and HS-R5, Fig. 1A). Centrally nucleated fibers were rare in control soleus muscles and were not significantly altered with 14 days of unloading. After 5 days of reloading the percentage of centrally nucleated fibers was 10-fold increased vs. control (C14). The proportion of hybrid type I/II fibers in the muscles was increased ninefold after HS14 at the expense of pure type I fibers. None of the determined fiber type percentages in 1-day reloaded soleus muscle was different from the values in atro-phied muscle. Conversely, the hybrid fiber percentage dropped again after 5 days of reloading when type I fiber percentage had almost returned to control levels (see Fig. 1B). A distinctive staining of small muscle fibers for embryonic myosin heavy chain was occasionally noted after 1 and 5 days of reloading, but staining was absent in the suspended and control animals (data not shown).
Global expression profiles in control vs. unloaded and reloaded rat soleus muscles.
Expression of all detected transcripts in m. soleus during the HS14, HS-R1, HS-R5, and C14 treatments was compared. Supervised hierarchical cluster analysis demonstrated that arrays of the same treatment were well clustered (see Fig. 2).
Transcriptome adjustments with unloading-induced muscular atrophy.
To elucidate the specific adaptations in mRNA levels with unloading-induced atrophy of m. soleus, gene expression data were analyzed for statistical differences between control and 14 days of suspension. From the 408 analyzed mRNAs, the level of 91 was increased, whereas 80 were reduced by 14 days of unloading (P < 0.05). On the basis of post hoc estimates, the majority of alterations (83%) were <1.5-fold, 7% being above twofold (see Table 1). mRNA changes for genes belonging to several gene ontologies ran parallel (see Supplemental Online Table 1).
Major themes of expressional alterations with reloading.
Upon 1 day of reloading of atrophied m. soleus, 117 mRNAs were increased, while 115 were reduced with reloading (P < 0.05, Table 1). A greater proportion of mRNA level changes (i.e., 15%) was above twofold than with unloading. Continued reloading for 5 days affected the concentration of multiple transcripts, whereby 14% of changes were greater than twofold. The level of 355 transcripts was altered relative to 14 days of unloading with either 1 or 5 days of reloading. Supervised hierarchical cluster analysis revealed distinctive patterns of expressional alterations with reloading of atrophied muscle (Fig. 2).
Subsequent analysis of gene ontologies revealed that level changes of multiple transcripts belonging to the same gene ontology were coregulated (Fig. 3). Multiple gene messages for factors involved in synthetic, folding, and proteolytic aspects of protein turnover were twofold induced with 1 or 5 days of reloading (pattern B in Fig. 2, Supplemental Online Table 2). From these gene ontologies, however, only members of the messages involved in protein synthesis were prominently coregulated (Fig. 3).
On the contrary, there was a transient, correlated downregulation of transcript levels of fatty acid transporters, respiratory chain constituents, voltage-gated cation channels, and some glycolytic factors with 1 day of reloading relative to unloaded muscle (pattern D in Fig. 2, Fig. 3). After 5 days of reloading, the levels of these former mRNAs were induced again (Supplemental Online Table 2).
A prominent reduction of expression of cell regulatory factors was noted after 5 days of reloading (pattern E in Fig. 2). The most notable ones of these adjustments included a co-regulated drop of mRNA levels of five factors involved in complex lipid metabolism and six ion channels (Fig. 3, Supplemental Online Table 2).
The specificity of microarray results for the same filter setup has been validated before (64). Additional RT-PCR experiments for selected transcripts (EF-2, RPL19, CTSL, FABP3, COX4) showed a good correspondence and high correlation (r2 ≥ 0.79 for EF-2, RPL19, CTSL, FABP3, COX-4) between expression signals from microarray and RT-PCR values with unloading and reloading (Fig. 4).
Association of expression changes to structural alterations.
The relationships between gene transcript levels and fiber characteristics were analyzed by K-means clustering combined with hierarchical clustering. This revealed an association of 40 and 50 mRNAs with the percentage of endomysial tenascin-C-positive and centrally nucleated muscle fibers, respectively. Generally, these transcripts corresponded to a few distinctive expression patterns (Table 2) as identified from the correlation of transcripts (see Fig. 2). From the transcripts being associated with endomysial tenascin-C-positive fibers, the majority of transcripts were involved in protein turnover, i.e., nine protein synthetic and 10 proteolytic factors (see Table 2). Additionally, four transcription factor mRNAs and gene messages for seven other signal transducers were correlated with the percentage of these fibers being surrounded by tenascin-C (Table 2). The mRNAs that were related to the percentage of centrally nucleated muscle fibers (see Table 3) included several transcripts involved in oxidative metabolism, i.e., five fatty acid transporter, seven mitochondrial respiratory chain constituents, and two factors of beta-oxidation mRNAs. Additionally, three voltage-gated cation channel mRNAs, three redox factors, and 16 signal transducers were associated with fibers showing central nuclei. Concerning the alterations in muscle mass, an association of soleus muscle-to-body weight with the mRNA levels of 33 genes was detected. These associated transcripts largely overlapped those being correlated with the percentage of centrally nucleated muscle fibers (underlined, Table 3). They included encoded main components of various aspects of oxidative metabolism such as fatty acid transport, beta-oxidation, mitochondrial respiratory chain, and redox regulation as well as voltage-gated ion channels. Under the employed criteria, the level alterations of not even a single mRNA were associated with the percentage of fast (slow) and hybrid-type muscle fibers.
Rebuilding of atrophic skeletal muscle is a complex process that takes weeks to be completed (13, 23, 41, 54). In rat soleus muscle after HS, these necessary steps involve initial fiber damage and regeneration before recovery of fiber mass and fiber contractile and metabolic characteristics become apparent (23, 32, 41, 45, 56). The molecular mechanisms underlying the different phases of load-dependent muscle remodeling are poorly understood. The current report demonstrates that extensive transcriptional reprogramming of distinct gene ontologies is involved in the early phase of reloading of atrophic rat m. soleus. Clustered expression patterns indicated that the transcriptional response of individual transcripts from diverse gene ontologies is coregulated and biphasic during the first 5 days of reloading (see Figs. 2 and 3, Table 2). Multiple reloading-induced mRNA levels of genes involved in protein turnover (i.e., protein synthesis and proteasomal degradation) and gene transcription were associated with the percentage of endomysial tenascin-C-positive fibers. Conversely, several transcripts involved in myocellular oxidative metabolism (i.e., fatty acid transport, mitochondrial respiration), redox regulation, and voltage-gated ion transport were transiently reduced with reloading and related to the percentage of centrally nucleated fibers and correlated with muscle weight. Our study thus demonstrates that a posteriori statistical testing of transcript signals can allow the identification of the gene responses that are associated with mechanical damage and rebuilding of atrophic skeletal muscle fibers.
M. soleus is composed of multiple cell types that undergo pronounced changes in composition and abundance with atrophy and reloading (17, 21, 43–45). The problem of standardization therefore is paramount. Only relative estimates for expression changes are possible. To deal adequately with this problem, we applied permutation analyses on L1 regression models of scatterplots of log-transformed raw cDNA signals for all array pairs from two experimental conditions (treatments). Differentially expressed genes were identified as described with a nonparametric sign test from the difference relative to the regression line and multiplicity error correction (63, 64). The justification for the application of regression analysis in our study is given based on the basic assumption that altered mRNAs show up as outliers in (linear) comparisons of expression signals between two experimental conditions (47). This regression approach is useful to compare large sets of data from two conditions because it is robust to outliers, does not depend on normal distributions, and does not involve background correction or normalization. This circumvents a major type of bias. To control the multiplicity error, the false discovery rate rather than the conservative Bonferroni correction was applied. Categorically expressed genes, i.e., genes that were not detected in all the treatment groups, were identified with an ANOVA approach because their analysis with the L1 regression approach would have introduced a systematic bias.
The power of this regression-based approach for the identification of differentially expressed genes between the experimental conditions of this study is highlighted by the agreement with our RT-PCR data and the literature. Initially, we had validated the accuracy of microarray results for the same filter setup with RT-PCR (64). Additionally, for this study a good correspondence was obtained for the expressional changes as identified by mRNA-related microarray and total RNA-related RT-PCR analysis (Fig. 4). The lowering of this relationship is expected due to differences between the two techniques and possibly also reflects eventual shifts in the mRNA-to-rRNA ratio due to increases in total RNA content in suspended soleus muscle after several days of reloading (15). Accordingly, the identified mRNA alterations (see Tables 2 and 3, Supplemental Online Tables 1 and 2) confirm previously reported transcriptional adaptations (36, 58–60).
Expressional adaptations related to the “stable” muscle atrophy phenotype.
Comparison of the expressional alterations in 14 days suspended rat m. soleus in this study to those identified in rat m. soleus with 35 days of HS (64) revealed broadly similar expressional adaptations (see Supplemental Online Table 1). This similarity concerned important metabolic, homeostatic, and regulatory myocellular processes. Much of the multifaceted nuclear program, which is characteristic of prolonged atrophy state of rat m. soleus (64), was thus established after HS14.
Some differences in mRNA concentrations relate to the earlier induction of gene expression of proteasomal factors and transcription factors in rat m. solei within the first week of suspension (58) and thus reflect the enhanced apoptotic cell death after HS14 (1, 49). The observed expressional dissimilarities of mitochondrial, proteasomal, cytoskeletal, and cell cycle transcripts point to ongoing remodeling of some muscular structures between 2 and 5 wk of muscle unloading. These latter processes include the loss of total mitochondrial mass with HS (21), the ongoing removal of myofibrillar structures between 3 and 30 days of atrophy (39, 60, 61), and the drop in satellite cell proliferation between 3 and 30 days of HS (17). The data hence imply that the atrophied rat soleus muscles analyzed were, with the exception to degradation remodeling events, in a “quasi-steady state”.
Coregulated transcriptome changes induced by mechanical loading of atrophic muscle.
Reloading had a pronounced effect on the degree and magnitude of transcript level changes compared with unloading (Fig. 2, Table 1). For several gene ontologies, supervised hierarchical cluster analysis revealed an association with a distinct expression pattern (patterns A–C in Figs. 2 and 3, Tables 2 and 3). In particular, the correlated increase in protein synthetic factor mRNAs was a main observation after 1 day of reloading.
Similarly, the correlated, sometimes transient, overshoot of mRNA levels of several transcription factors, signal transducers, cytoskeletal and trafficking factors after 1 day of reloading (see Table 2, Supplemental Online Table 2) points to the involvement of certain cell regulatory events in the early phase of muscle loading. In contrast to the possible and expected inflammatory response that occurs during early reloading (16, 45, 56), our data provided little evidence for transcription level alterations of factors related to this aspect of the damage response (Supplemental Online Table 2). This contention is supported by the lack of alteration in centrally nucleated fibers after 1 day of reloading, as internal nuclei at early time points (i.e., 2 days) of reloading may also indicate transient macrophage infiltration (56). This is in contrast to the heavy inductions of these transcripts with muscle growth as induced by muscle overloading (12), or pharmacologically induced muscle regeneration (37). These differences presumably reflect the rather normal, nonpathological range within which the rat suspension-reloading model permits us to investigate skeletal muscle plasticity.
The conclusion that important, biphasic transcriptional alterations occur with reloading of atrophic muscle was further corroborated by the observations of a correlated pattern of transiently reduced transcripts after 1 day of reloading. These transcripts were overshooting after 5 days of reloading (see pattern D in Fig. 2). Among these gene ontologies we note genes related to several aspects of oxidative metabolism, such as fatty acid transport, mitochondrial respiratory chain, and redox regulation, as well as voltage-gated cation channels (see pattern D in Fig. 3, Supplemental Online Table 2). The latter association underlines the concept that expressional adaptations of voltage-gated cation channels constitute an important part of adaptations of muscle fibers (64). These transcriptional adaptations indicate a retooling of the T- and SR (sarcoplasmic reticulum)-tubular system with muscle fiber transformations (19, 53). Additionally, with the clustered mRNA level drop in voltage-gated cation channels a related drop in transcript levels of two glycolytic factors, i.e., phosphofruktokinase, aldolase A, was observed (Fig. 3, Supplemental Online Table 2). The latter adaptations of glycolytic mRNAs can be seen in context with the described initial decrease of protein level for the glycolytic enzymes lactate dehydrogenase B and β-enolase, between 2 and 7 days after reweighing of 3-wk suspended rats (40).
Surprisingly, the levels of several transcripts for ion channels and factors of complex lipid metabolism were coincidentally reduced with 5 days of reloading (Fig. 3, Supplemental Online Table 2). This relates to the typically lower resting membrane chloride conductance of slow-twitch than fast-type muscle fibers (reviewed in Ref. 10). This lowering of membrane channel mRNAs with muscle reloading therefore may be related to a fast-to-slow transformation of muscle fiber types.
Evidence for an important remodeling of muscle fiber makeup in atrophied muscles was provided by the structural observations demonstrating the appearance of regenerating, i.e., centrally nucleated, fibers and alterations toward a reestablishment of normal muscle weight and fiber type distribution after 5 days of reloading (Fig. 1). Linking mechano-induced gene activity to recovery of form and function of atrophied rat soleus muscle, important interactions become apparent between gene ontologies and fiber morphological characteristics (see Tables 2 and 3). The increased mRNA levels of nine protein synthetic mRNAs (RPS11, RPS12, RPS19, RPS29, RPL11, RPL13, RPL19, RPL21; EF-2), six proteasomal subunits (PSMA3, PSMB1, PSMC1, PSMA1, PSMA7, PSMC3), four transcription factors (APEX, CSBP, STAT3, TCEB1), and some cell regulatory mRNAs were associated with the percentage of muscle fibers which demonstrated endomysial tenascin-C staining (Table 2). This extracellular matrix molecule is a marker of mechanical stress (14). Previously, we identified in the rat suspension-reloading model that the decoration of the muscle fiber periphery with tenascin-C precedes and is associated with the appearance of centrally nucleated fibers (32). Similarly, wing overloading caused the rapid ectopic expression of tenascin-C in fibroblasts of chicken anterior latissimus dorsi muscle (33). Therefore, the observed transcriptional adaptations are a likely consequence of unaccustomed mechanical stress of muscle fibers and associated to connective tissue with early reloading of the m. soleus weakened during atrophy. The higher levels of protein synthetic factors imply that an increased translational capacity contributes to the enhanced protein synthesis rates during the first week of reloading of suspended rat m. soleus (59). At the same time, the substantial increase in several proteasomal subunits (Table 2, Supplemental Online Table 2) relates to the increased protein breakdown due to nonlysosomal and Ca2+-independent proteolysis with the first day of reloading of suspended rat m. solei (59). Our finding of a partial clustering of proteasomal factors mRNA levels to distinct expression patterns A and B (Table 2) supports the notion that transcriptional regulation of the proteasomal system with reloading of atrophic rat m. solei is coordinated to some extent. This is in contrast to previous suggestions (59). Myosin heavy chains are major targets of the proteasomal pathway (reviewed in Ref. 55). Our fiber type data imply that the transcriptional upregulation of proteasomal subunits is possibly related to the removal of myofibrillar structures with the “dedifferentiation,” i.e., the drop, of hybrid-type fibers between 1 and 5 days of reloading (Fig. 1B). Finally, the tenascin-C-associated enhanced levels of the transcription factors involved in DNA repair (APEX), the acute-phase response (STAT3), general RNA polymerization (TCEB1), and metabolism of hRNAs (CSBP) indicate that distinct nuclear processes are a consequence of increased mechanical stress of muscle fibers.
With regard to muscle fiber damage and regeneration, the coordinated upregulation of multiple transcripts after 5 days of reloading involved in fatty acid transport, mitochondrial respiratory chain, and redox regulation as well as voltage-gated ion transport in association with the number of centrally nucleated fiber is of importance (see Table 3). Central nuclei are a sign of muscle regeneration and were increased in the reloaded muscles with a similar time course, i.e., after 5 days, as regenerating fibers were described to be apparent (45, 56). The affected fatty acid transporters (LPL, FABP3, CPTI-M, CPTII) and mitochondrial respiratory chain components (COX4, COX5B, COX6A2, COX7A2, COX8H, ATP5B, ATP5G1) constitute main factors of oxidative metabolism. Similarly, the affected voltage-gated ion channels are essentially involved in the de-/repolarization of the sarcolemmal potential (reviewed in Ref. 64). Muscle fibers are the major source of these oxidative enzymes and sarcolemmal channels (5, 19, 51, 53). The stimulation of mitochondrial biogenesis has recently also been observed during regeneration of bupivacaine-damaged tibialis anterior muscle (25). A similar concerted drop of mitochondrial mRNA levels has also been observed in skeletal muscle after ischemic damage in mice (50). The transiently altered transcript levels of oxidative metabolic factors and sarcolemmal channels therefore possibly reflect a general transcript response of damaged muscle and the initiation of a regeneration program to support the increased demand in synthesis of the set of myocellular proteins during the rebuilding of damaged muscle fibers (13).
Concerning recovery of soleus muscle weight, the association of the soleus-to-body weight ratio with the alterations of multiple transcripts of oxidative metabolism, redox regulation, and voltage-gated ion transport was remarkable (see Table 3). Generally, after 5 days of reloading, the levels of these transcripts were overshooting above values seen in control soleus (see pattern D in Fig. 2, Supplemental Online Table 2). These observations strongly imply that transcriptional reprogramming of oxidative metabolism is essentially involved in the reestablishment of oxidative capacity and mitochondrial mass with recovery of atrophied rat m. soleus in this model (23).
Several of the transcriptional alterations possibly play a role in the adjustment of myocellular homeostasis during the reversible fiber type transformation that was manifested with unloading and reloading (Fig. 1B). However, no mRNA level was found to be associated with the percentage of fast (slow) or hybrid-type fibers when control, 14-day suspended and 1- or 5-day reloaded muscles were compared. This suggests that the transcriptional makeup underlying myosin fiber type-specific homeostasis is not established at the studied time points of muscle atrophy and regeneration.
Regulation of coordinated transcriptional adaptations.
The coregulated level changes of mRNAs encoding factors of fatty acid transport, mitochondrial respiratory chain, voltage-gated cation channels on one side, as well as protein synthetic factors on the other, were a surprising observation (Fig. 3, Tables 2 and 3). This association extends the previous observations on “enzyme groups of constant proportion” to the transcriptional level and to additional gene ontologies. Previously such relationships have been shown to exist in skeletal muscle for enzymatic activities of components of mitochondrial metabolism, such as beta-oxidation, citrate cycle, and the respiratory chain (reviewed in Ref. 5). The existence of such “synexpression groups” calls for a mechanism capable of coordinating expression of the multiple members of these gene ontologies. Several of the coregulated mRNAs within cellular pathways involved in protein synthesis (ribosomal factors), voltage-gated cation transport, fatty acid transport, and mitochondrial respiration localize to different chromosomes (for details see http://www.ncbi.nih.gov/entrez/query.fcgi?db=gene). This indicates that any putative regulatory process involves mechanisms capable of coordinating gene transcription over several chromosomes or else includes posttranscriptional processes that influence RNA stability via common nontranslated elements within the transcript (18). With regard to the pronounced drop of mRNAs for fatty acid transporters, mitochondrial respiratory chain constituents, and voltage-gated cation channels (pattern D in Fig. 2, Table 3), when ribosomal factor RNAs are twofold increased (pattern B in Fig. 2, Table 2), the specific RNA degradation of these former gene ontologies is a plausible possibility.
Mechanical factors exert important control over the skeletal muscle phenotype (4, 12, 31, 54, 57), of which the underlying gene expressional events are little understood. Our investigation involving clustering analysis of microarray and structural data points to the involvement of distinct gene ontologies in the adaptations of muscle fibers in the early phase of mechano-dependent recovery from muscular atrophy. The observed transcript-structure associations indicate that a statistical approach works to identify biologically relevant clusters of transcripts within a set of microarray data. This study exposed a number of mechanistically important pathways involved in fiber damage and repair and revealed that co-regulation of transcript levels with reloading of atrophic muscle occurs across chromosomes. This begs the question as to the regulatory events controlling synexpression groups of genes.
The study was supported by the French Secretary of State for Foreign Affairs.
We are especially thankful to Marie-Hélène Sornay Mayet for carrying out the animal experiments and Dr. Hans Howald for helpful comments on the manuscript. We thank Samuel Müller and Michael Vock of the Department of Mathematical Statistics and Actuarial Sciences of the University of Berne for statistical advice.
↵1 Supplemental data for this article may be found at http://ajpregu.physiology.org/cgi/content/full/00833.2004/DC1.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society