Global analysis of gene expression patterns during disuse atrophy in rat skeletal muscle

doi:10.1113/jphysiol.2003.044701

Global analysis of gene expression patterns during disuse atrophy in rat skeletal muscle

Eric J. Stevenson, Paul G. Giresi, Alan Koncarevic and Susan C. Kandarian

Department of Health Sciences, Boston University, 635 Commonwealth Avenue, Boston, MA 02215, USA

ABSTRACT

Top
Abstract
Introduction
Methods
Results
Discussion
References

Muscular inactivity leads to atrophy, weakness, and decreased fatigue resistance. In order to provide a window into the dynamic processes that underlie muscle atrophy, we performed global gene expression analysis of rat soleus muscles using Affymetrix GeneChips at 1, 4, 7 and 14 days of hindlimb unloading. Expression of 309 known genes was significantly changed by at least 2-fold (212 upregulated, 97 downregulated). K-means clustering was used to divide these genes into co-regulated clusters based on the similarity of temporal expression patterns. This allowed the development of a timeline of the atrophy process with respect to the behaviour of genes in a broad array of functional categories. Regulatory genes were often upregulated early, in either a transient or sustained manner, but they also populated clusters with later patterns of activation, suggesting different phases of molecular adaptations. Other early events were the activation of ubiquitination genes and downregulation of protein chaperones. In comparison, clusters representing slightly delayed activation patterns included genes involved in fast contraction, glycolysis, translational inhibition, oxidative stress, protein degradation, and amino acid catabolism. Downregulated genes exhibited fewer unique expression patterns and included structural and regulatory genes of the extracellular matrix and cytoskeleton, and genes that define a slow-oxidative phenotype. Other novel findings include the tight co-activation of proteasome subunit and ubiquitination genes, differential regulation of serine proteases and serine protease inhibitors, and the identification of transcriptional, signalling, growth and cell cycle genes that probably play a role in the atrophy process. The present work has uncovered temporal patterns of gene expression that highlight the molecular processes that underlie muscle atrophy and provide new avenues for study.

(Resubmitted 7 April 2003; accepted after revision 30 May 2003; first published online 4 July 2003)
Corresponding author S. C. Kandarian: Department of Health Sciences, Boston University, 635 Commonwealth Avenue, Boston, MA 02215, USA. Email: skandar{at}bu.edu

INTRODUCTION

Top
Abstract
Introduction
Methods
Results
Discussion
References

Decreased physical activity associated with a sedentary lifestyle, space flight or prolonged periods of bed rest leads to a profound reduction in skeletal muscle mass and functional capacity (Booth & Baldwin, 1996). The physiological features of skeletal muscle atrophy include a marked tendency of slow muscle fibres to take on fast characteristics, increased fatiguability and myofibrillar protein loss due to decreased synthesis and increased degradation. Given the extent of remodelling that occurs, it is clear that atrophy is not simply a degradative process. For example, genes defining fast-twitch, glycolytic phenotypes (Booth & Baldwin, 1996) and those involved in proteolytic processes (Taillandier et al. 1996; Ikemoto et al. 2001) are actually upregulated in the face of this global protein loss. So, while certain genes are downregulated and preferentially degraded, others are spared from destruction or even transcriptionally upregulated.

The loss of muscle protein during disuse has been directly correlated with the activation of three proteolytic systems: the calcium-dependent calpains, lysosomal cathepsins and ubiquitin-proteasome system (Taillandier et al. 1996; Ikemoto et al. 2001). Recent work has shown that the ubiquitin-proteasome system is not only responsible for the bulk of protein degradation during atrophy (Jagoe & Goldberg, 2001), but it is also involved in defining much of its specificity (Bodine et al. 2001a; Gomes et al. 2001). Other evidence suggests, however, that the ubiquitin- proteasome system is not capable of initiating myofibrillar disassembly (Jagoe & Goldberg, 2001) and that additional proteolytic systems may be necessary (Tischler et al. 1990). Therefore, questions remain as to what proteolytic systems are at work and how they coordinate protein degradation. Furthermore, the regulatory triggers mediating accelerated proteolysis and decreased protein synthesis are not well understood. Several signalling pathways involving either Akt (Bodine et al. 2001b) or NF- $kappa$ B (Hunter et al. 2002) have been implicated in the regulation of disuse muscle atrophy but no direct linkages have been made between these pathways and gene targets, or even whether these components are necessary for disuse atrophy. It has also been shown that myogenic E-box-dependent mechanisms are responsible for the transactivation of several fast genes in response to inactivity (Swoap, 1998; Mitchell-Felton et al. 2000), but the upstream pathways regulating changes in muscle phenotype remain elusive. Defining signalling pathways and their protein targets in vivo remains one of the biggest challenges in the study of atrophy. This is complicated by the likelihood that several different pathways are working in parallel to mediate the atrophy process. Such complexity illustrates the need to apply a more global approach to analysing the molecular changes that occur during atrophy.

In recent years, several groups have used various approaches to studying expression of multiple genes during inactivity at one time point. These include serial analysis of gene expression after 12 days of immobilization (St-Amand et al. 2001), Affymetrix GeneChip analysis after 12 h (Bey et al. 2003), or 21 days (Stein et al. 2002) of hindlimb unloading, subtractive hybridization after 14 days of unloading (Cros et al. 2001), and cDNA array analysis after 35 days of unloading (Wittwer et al. 2002). These studies have uncovered the possibility that atrophy is regulated by changes in mRNA levels of genes involved in protein synthesis, proteolysis, oxidative stress, growth and cell cycle regulation and structural genes of the extracellular matrix and cytoskeleton. The current study reconfirms and significantly expands upon these findings by using Affymetrix GeneChips to monitor differential gene expression after 1, 4, 7 and 14 days of hindlimb unloading in rat soleus muscle. This approach not only identifies genes that are differentially expressed after unloading, but also provides a unique view of their expression patterns in relation to each other during the course of disuse. The temporal aspect of this analysis therefore provides a window into the dynamic molecular alterations that occur during disuse muscle atrophy.

The goals of this paper are to: (1) develop a general timeline of atrophy based on the temporal expression patterns of genes involved in contraction, metabolism, cytoskeleton, extracellular matrix (ECM), protein synthesis, oxidative stress, protein processing and degradation and regulatory genes (growth, proliferation, signalling and transcription); (2) determine the extent of coordinated expression of genes that share similar function in skeletal muscle (contraction, metabolism, oxidative stress and protein turnover); (3) identify the variety of distinct expression patterns among genes involved in regulating growth, proliferation, signalling and transcription; and (4) identify the expression patterns of genes not previously shown as being differentially regulated during disuse-related atrophy.

METHODS

Top
Abstract
Introduction
Methods
Results
Discussion
References

Hindlimb unloading protocol and experimental design

Female Wistar rats (6 weeks old) were hindlimb unloaded (HU) for 1, 4, 7 and 14 days using a standard elastic tail cast method (Mitchell-Felton et al. 2000). At each time point eight rats were randomly divided into control (ambulatory, n = 4) and HU (n = 4) groups. After each period of unloading, control and HU rats were anaesthetized with pentobarbitol sodium (60 mg kg^-1, I.P.), and soleus muscles were extracted. The rats were killed with an anaesthetic overdose after muscle extraction. The right and left soleus muscles from each rat were pooled, total RNA was isolated, labelled cRNA prepared and hybridized to a single GeneChip. Thus, at each time point eight chips were analysed (4 control and 4 HU) so a total of 32 GeneChips were used over the entire time course (see Supplementary material, Fig. 1). All procedures used conformed to National Institutes of Health Guide for the Care and Use of Laboratory Animals and Boston University's Institutional Animal Care and Use Committee.

View larger version
[in this window]
[in a new window]

Figure 1. Results from K-means cluster analysis
Genes were divided into 13 clusters using a K-means cluster analysis algorithm, which distinguishes sets of genes with similar expression profiles. Each data point represents the fold change calculated from the average intensity of control (n = 4) and unloaded (n = 4) GeneChips for each probe after 1, 4, 7 and 14 days (total of 32 chips). Below each cluster is a list of the gene functional categories most represented in that cluster. Detailed lists of the genes in each cluster and their associated functional categories are available online.

Total RNA isolation and cRNA synthesis

Total RNA was isolated from rat soleus muscles using a Trizol-based protocol recommended by Affymetrix (Affymetrix, Santa Clara, CA, USA). A second 'clean-up' step was used to improve the quality of total RNA using columns from Qiagen. RNA quality was judged based on the ratio of absorbance at 260 nm and 280 nm. Further, all RNA samples were analysed using agarose gel electrophoresis and stained to check for integrity of 18 and 28S RNA. Reverse transcription, second strand synthesis, labelled cRNA preparation, and hybridization to the rat U34A GeneChip were all performed at Partners Healthcare Gene Array Technology Center (Brigham and Women's Hospital, Boston, MA, USA) using the protocol recommended by Affymetrix. Hybridization conditions have been detailed elsewhere (Lockhart et al. 1996).

Expression profiling

Affymetrix rat U34A GeneChips were used in this study. The Affymetrix rat U34A GeneChip contains a total of 8799 probe sets representing mRNA transcripts and expressed sequence tag clusters from the Unigene database (Build 34). GeneChip expression patterns were compared between four unloaded rats and four control rats (i.e. 8 independent GeneChips) at each time point (1, 4, 7, and 14 days), so that a total of 32 GeneChips were analysed. The detail of the chip design is described elsewhere (Lockhart et al. 1996). Briefly, each gene probe is represented by ~16-20 'perfect match' (PM) 25-mer oligonucleotides paired with an equal number of 'mismatched' (MM) oligonucleotides that have a single nucleotide difference in a central position of the 25-mer. Comparison of the hybridization patterns of the PM and MM pairs allows for the elimination of non-specific hybridization signals. Each of the 16-20 probe pairs is designed to span sequences in the 3' region of the gene and is complementary to the cRNA targets prepared from tissue samples.

Image analysis was performed using Affymetrix Microarray Suite 5.0 (MAS 5.0) as previously described (Liu et al. 2002). Briefly, the MAS 5.0 statistical algorithm provides an intensity value (transcript abundance) which indicates whether a transcript is detected above the background level for each gene based on the hybridization performance of the PM and MM. A one-sided Wilcoxon's signed rank test was used to determine the detection call (present (P < 0.04), marginal (0.04 P < 0.06) or absent (P 0.06)) and a one-step Tukey's biweight estimate was used to calculate overall signal intensity.

Data analysis and statistical filtering

Step 1 - normalization. Signal intensities on each chip were normalized with respect to each other using a linear scaling method in which signal intensities for each probe set are multiplied by a normalization term (Butte et al. 2001). Normalization of signal intensities across all chips is necessary in order to eliminate the effects of chip-to-chip variations in target preparation, hybridization and scanning.

Step 2 - noise reduction. Probe sets that receive absent calls are usually associated with low intensity expression values and/or a high level of intra-probe set variability. Low intensity levels may represent the specific binding of a transcript expressed at low levels, but may also be the result of spurious binding and other factors that influence background and noise. In order to reduce the contribution of noise-based error in subsequent statistical analysis, each probe set that did not receive at least one present call across all 32 chips was removed. This reduced the data set from 8799 to 4208 probe sets (i.e. gene products). However, rather than depending entirely on the Affymetrix present/absent algorithm, the issue of noise and variability associated with 'absent' probe sets was dealt with primarily in subsequent steps in our statistical filter by testing for variability (and thus significance) by applying the 2-way analysis of variance (ANOVA) at P < 0.01 and using a stringent 2-fold cut-off.

Step 3 - significance testing. Significant changes in gene expression over the time course of disuse were assessed using a 2-way ANOVA. The 2-way ANOVA compares measured expression intensities and is independent of the present/absent algorithm. The 2-way ANOVA was used to test for the main effects of unloading and the interaction of unloading and time, in order to determine whether a probe set (or gene product) was significantly different between groups (P < 0.01). This reduced the data set from 4208 to 1273 probe sets (i.e. gene products).

Step 4 - multiple test correction. When using standard statistical methods to analyse microarray data, it is important to consider the effect of the number of tests being performed on type-I error (or false-positives). In this data set, the 2-way ANOVA was performed on 4208 probe sets. Testing multiple hypotheses in these situations can effectively increase type-I errors so it has become common to use correction methods to reduce this type of error. Therefore we have applied the false discovery rate (FDR) method of Benjamini & Hochberg (1995; Reiner et al. 2003) to each test, which calculates a new rank adjusted P value threshold (BH correction). The observed P values (P_obs) from the original test statistic were ranked based on significance, with the most significant (smallest P value) result being first and the least significant (largest P value) being last. A new set of thresholds (P_adj) were then determined based on rank:

P_adj = (P_original times rank)/N,

where P_original was the original threshold of 0.05 and N is the total number of comparisons (in this case N = 4208). So the most significant probe set would be evaluated against a P_adj = 1.19 times 10^-5 (0.05 times 1/4208 = 1.19 times 10^-5). These new rank-adjusted thresholds (P_adj) were then compared to the ranked P_obs and all probe sets where P_adj P_obs were considered significant.

The new rank-adjusted P value thresholds were compared to the ranked observed P values for the experimental effect (unloading) and the interaction effect. The resulting false positive rate (FPR) was P_sum = 0.016, which is the sum of the largest significant P value (P_sum) from the main effect and the interaction effect.

This also allowed us to calculate a false discovery rate (FDR), which is the ratio of the number of false discoveries (those that would be expected by chance, N_FD) divided by the number of total discoveries (N_TD).

FDR = N_FD/N_TD.

So, in our case, the total number of false discoveries (N_FD) was:

N_FD = N(P_sum),

where P_sum is multiplied by the number of comparisons (N = 4208). The number of probe sets found to be significantly different using these methods (N_TD) was 1273. This resulted in a FDR of 0.0538 (5.38 %). (Supplementary Fig. 1 available in Supplementary material online only describes the number of probe sets that pass each step and the FPR and FDR after the 2-way ANOVA and BH adjustment.)

Step 5 - fold change cut-off. The next step in filtering the data set was to eliminate those probe sets that did not have a minimum fold change of ±2, for at least one time point. Fold change was calculated using the following formula, if X- Y-:

where X_i-- represents the sample mean for the experimental group at the ith time point and $sigma$ ²_Xi is the variance term or S.D.², for the experimental group at the ith time point. The Y terms correspond to the same values except they are calculated for the control group at each time point. Since probe sets with lower intensities tend to have more intra-group variation, this calculation incorporates an estimate of the variance into the overall fold-change estimate (Mariani et al. 2003). When comparing probe sets with high variability, fold-change estimates will be smaller or more conservative. Therefore, only those comparisons where the overall differences are greater than the inherent noise would achieve sufficient fold changes to make the cut-off. After applying the ±2-fold cut-off, the data set was reduced from 1273 to 426 probe sets. The 2-fold cut-off was selected because for this paper we were interested in analysis of only the largest changes in gene expression, and because this is the most stringent fold cut-off suggested by the GeneChip manufacturer to reduce false positives (Lockhart et al. 1996). Of the 426 probe sets, 309 correspond to unique gene products or ESTs when redundant probes are considered.

K-means clustering and annotations

K-means clustering was performed using GeneSpring (Silicon Genetics, CA, USA) on the significantly changed genes that passed all filters (n = 426). K-means clustering has been described elsewhere (Dougherty et al. 2002). In brief, the K-means clustering algorithm initially divides genes into a number of equal sized groups based on a user-defined number (K). Centroids are first created representing the average location of each of these groups in 'expression space'. Each gene is then reassigned to the centroids that best match its expression pattern over time. The centroids are then recalculated and the process is repeated until the maximum number of iterations has been reached. The results are presented as multiple clusters representing sets of genes that share a distinct pattern of gene expression with respect to time.

The functional categories used were derived from Silicon Genetics and National Center for Biotechnology Information annotation records. Annotations were obtained for each gene from LocusLink and Unigene based on the GenBank sequences from which probes were derived. In the cases where rat functional information was incomplete, function was inferred based on curated homologues in the mouse and/or human (HomoloGene). Complete annotations and expression information for each probe set (n = 426) are available online (http://www.bu.edu/sargent/people/faculty/kandarian_susan.html).

DISCUSSION

Top
Abstract
Introduction
Methods
Results
Discussion
References

Data variability

The step-by-step results of statistical filtering are summarized in Supplementary Fig. 1 (see Supplementary material online). Of the 8799 probe sets on the U34A GeneChip, 426 were significantly changed with hindlimb unloading. Of these, 309 correspond to unique gene products or ESTs when redundant probes are considered. Pairwise intra-group comparisons (control vs. control or HU vs. HU) of signal intensities of all 8799 probe sets showed a mean correlation coefficient of 0.97 (± 0.002 S.E.M.) indicating low genetic variability. The correlation coefficients for all possible pairwise comparisons between chips including intra- and inter-group comparisons are available online (see Supplementary Fig. 2 online). Plots of the intra-group comparisons with the lowest (0.91) and highest (0.99) correlation coefficients are also available online (see Supplementary Fig. 3 online).

View larger version
[in this window]
[in a new window]

Figure 2. Differential expression of genes defining fast and slow contractile phenotype
Each data point represents the fold change calculated from the average intensity of control (n = 4) and unloaded (n = 4) GeneChips for each probe after 1, 4, 7 and 14 days (total of 32 chips). Novel findings are indicated by *.

View larger version
[in this window]
[in a new window]

Figure 3. Activation of genes involved in the utilization and transport of glycogen or glucose
Each data point represents the fold change calculated from the average intensity of control (n = 4) and unloaded (n = 4) GeneChips for each probe after 1, 4, 7 and 14 days (total of 32 chips). Novel findings are indicated by *.

K-means clustering used to develop a general timeline of atrophy

Atrophy of rat soleus muscle in response to unloading is marked by several well-established alterations in gene expression (Booth & Baldwin, 1996). These changes and the loss of muscle protein that accompanies them occur quickly after the removal of weight bearing providing the rationale for the time points studied here. Previous work has shown that functionally related genes tend to cluster together (Eisen et al. 1998). Therefore, identifying sets of co-regulated genes can help to identify functional relationships between genes that might not be immediately apparent. In order to visualize and interpret the time course data, K-means clustering was used to divide genes into groups based on similarity in expression patterns with respect to time. Each cluster represents a set of genes that share a relatively distinct pattern of differential expression in the context of the other clusters. The K-means algorithm emphasizes the shape and pattern of expression over time rather than magnitude of change. So, while in each cluster there may be significant differences in the magnitude of gene expression change, the response pattern is the distinguishing characteristic. We used this approach to develop a general timeline of atrophy, in the sense that K-means is used to identify sets of genes that are activated or deactivated early or late and in a transient or sustained manner. Figure 1 shows the results of the K-means analysis using the 426 differentially expressed genes in 13 clusters. Below each cluster is a list of functional categories represented in that cluster. Detailed annotations for genes in each cluster are available online (http://www.bu.edu/sargent/ people/faculty/kandarian_susan.html).

As expected, early events were marked by the differential expression of genes regulating proliferation, growth, signalling and transcription (Fig. 1A, B, C and H). Transcription and signalling factors were evenly distributed among clusters representing both early and late response patterns (all clusters except E and J), and in some cases, the changes were sustained over the entire time course (Fig. 1B, C, D, F, G, J and K). In addition, there were some clusters that exhibited a distinct peak or nadir at 4 or 7 days (Fig. 1D, E, F, I and J). These patterns suggest that even in the later stages of atrophy there are active regulatory processes occurring. This supports the idea that disuse is marked by several phases in which regulatory factors are sequentially activated or deactivated. For instance, specific early triggers may activate the genes involved in protein degradation but the maintenance of degradative processes may involve the later activation of different regulatory genes. It is possible that the genes within each cluster play a related role in the atrophy process or in the acquisition of the new faster phenotype. Another interesting aspect of these data involves the presence of several proliferative factors involved in regulation of the cell cycle in several of the clusters (Fig. 1A, D, F and I). These genes cluster in a manner similar to the regulatory genes mentioned above, indicating that they too may be involved in the initial switching mechanism that sets in motion the atrophic changes that occur as a result of inactivity as well as maintenance at later stages.

Initial regulatory signals may influence the expression patterns of genes that display response patterns that are slightly delayed. For example, the majority of genes that characterize a fast phenotype (i.e. fast contractile, calcium handling and glycolytic) are upregulated in clusters where most genes are activated by 4 days (rather than 1 day) (Fig. 1D, E and G). Genes that define a slow contractile and metabolic phenotype are downregulated in clusters representing similar delayed downregulated patterns (Fig. 1I and J). These results are consistent with the slow-to-fast shift in the expression of genes governing contractile properties found to occur with muscular inactivity (Booth & Baldwin, 1996). The data in Fig. 1 also reveal the activation patterns of different proteolytic systems that have been shown to mediate the loss of skeletal muscle protein during atrophy. These include ubiquitination genes (Fig. 1B), proteasome subunits (Fig. 1D) and lysosomal cathepsins (Fig. 1C and F). Several gene products that can inhibit the process of protein synthesis (translation inhibitors) are also upregulated (Fig. 1F). Genes known for their chemoprotective role against oxidative stress are distributed among clusters with both an early and a delayed pattern of activation. (Fig. 1C, D and G). Protein chaperones that regulate protein stability and refolding following damage are downregulated early during the course of atrophy (Fig. 1H and I). Transcripts encoding structural elements of both the cytoskeleton and extracellular matrix are downregulated gradually during the course of atrophy (Fig. 1K).

In the following sections, the detailed temporal expression patterns of individual genes in various functional categories will be presented. That is, the expression patterns of genes that share similar function are plotted on the same graph (independent of K-means clustering).

Validation of slow-to-fast fibre type switch during disuse atrophy

The expression changes characterizing the slow-to-fast shift in phenotype (Booth & Baldwin, 1996) were reconfirmed in this study highlighting the reliability of this data set. Fast contractile genes were globally upregulated (Fig. 2, left column). Myosin heavy chains (MHC) 2B and 2D/X are upregulated most dramatically by 190- and 60-fold respectively (Fig. 2, top left). While these large fold changes are likely because these transcripts are not expressed at levels within the detection limits of GeneChips in control soleus samples, this does highlight the strong transcriptional component involved in the phenotypic alterations that occur during atrophy. The remaining fast isoforms of contractile genes follow a tightly co-regulated pattern that increases dramatically between 7 and 14 days of unloading (Fig. 2, bottom left). These include Serca1, a DHP receptor subunit, and fast isoforms of myosin light chain (MLC), tropomyosin (Tpm) and troponin (Tpn). Slow isoforms of MHC, MLC, and Tpn were on the other hand downregulated genes (Fig. 2, right column). Of these MLC2, slow transcripts were lost most drastically, being downregulated 40-fold by day 7. These results support work previously done showing the phenotypic shift that occurs during disuse atrophy (Booth & Baldwin, 1996) but are novel in the sense that they display the coordinated expression of these genes with respect to each other.

Decreased fatigue resistance is another hallmark of atrophy and may be related to alterations in abundance or specific activities of mitochondrial enzymes. This is reflected in the data, which show a generalized decrease in genes regulating carbohydrate and fatty acid oxidation in the mitochondria (http://www.bu.edu/sargent/people/faculty/kandarian_susan.html). This is accompanied by a coordinated increase in genes that favour the breakdown, transport and utilization of glycogen and glucose (Fig. 3). These results agree with previous studies that demonstrate disuse atrophy is characterized by a fibre-type shift from oxidative type I to glycolytic type II fibres in soleus muscle (Booth & Baldwin, 1996). Glycolytic gene activation is not tightly coordinated as evidenced by their presence in multiple clusters. Generally, however, it can be said that glycolytic genes are gradually upregulated over the course of atrophy and that for the most part they are not significantly increased until day 4 and beyond.

Protein turnover

Upregulation of translational inhibitors. Consistent with the idea that decreased protein synthesis is involved in the loss of muscle protein, two translational regulators with the capacity to inhibit translation are upregulated. Eukaryotic translation initiation factor 4E binding protein 1 (PHAS-I) is upregulated by 1 day and is maintained at 14 days (Fig. 4A). When unphosphorylated, PHAS-I is a translational repressor that inhibits eukaryotic translation initiation factor 4E (eIF4E)-dependent mRNA translation. An early (12 h) increase in PHAS-I was also shown by Bey et al. (2003) but the present work shows this is sustained after 14 days of unloading. PHAS-I/eIF4E complexes are enriched in rat medial gastrocnemius extracts after 14 days of hindlimb unloading suggesting a role in decreased protein synthesis (Bodine et al. 2001b). Eukaryotic elongation factor 2 kinase (eEF2K), another inhibitor of protein synthesis, was upregulated at 4 days and had a peak increase of 5-fold by 7 days. Phosphorylation of eukaryotic elongation factor 2 (eEF2) by eEF2K results in a global reduction in ribosomal capacity (Wang et al. 2001). Therefore, an increase in eEF2K expression if reflected at the activity level, could lead to a decrease in the protein synthetic capacity in unloaded muscle.

View larger version
[in this window]
[in a new window]

Figure 4. Regulation of protein balance
A, regulators of protein synthesis; B, protein chaperones; C, subunits of glutathione S-transferase (GST) and glutathione peroxidase 1; D, other antioxidant genes; E, ubiquitination; F, proteasome subunits; G, lysosomal proteases and cystatin C; H, inhibitors of calcium-dependent proteolysis; I, other protease systems; J, markers of amino acid metabolism and transport. Each data point represents the fold change calculated from the average intensity of control (n = 4) and unloaded (n = 4) GeneChips for each probe after 1, 4, 7 and 14 days (total of 32 chips). Novel findings are indicated by *.

Decreased expression of protein chaperones. Chaperones play several important roles in the maintenance of proteins present in a cell including: (1) protein stabilization and folding during synthesis; (2) refolding of proteins damaged by metabolic, environmental and oxidative stress; and (3) protection of proteins from degradation. Molecular chaperones were downregulated as shown in Fig. 4B. The most dramatic decreases occurring were glucose-regulated protein 78 (Grp78) and heat shock protein 70 (Hsp70) (Fig. 4B). Hsp70 in particular is known to associate with nascent polypeptides as they exit the ribosome, facilitating translation (Ku et al. 1995). It has been suggested that the decreases in polysomal Hsp70 that occur during muscle atrophy could slow polypeptide elongation rate (Ku et al. 1995). While other studies also show decreases in Hsp70 with disuse (Ku et al. 1995) the current study demonstrates the dynamics of this change over the time course of atrophy as well as the concomitant decrease in other protein chaperones. Interestingly, none of the chaperones share a distinct downregulated pattern, which is reflected in their presence in two downregulated K-means clusters (Fig. 1H and I). This indicates they are regulated by different triggers and may be involved in a variety of different processes. Overall, a global reduction in the expression of protein chaperones could affect decreased efficiency of translation or ability to handle stress-related damage.

Activation of oxidative stress markers. Previous studies have shown that both unloading and immobilization are marked by changes that are consistent with an oxidative stress response (Kondo et al. 1993; Koesterer et al. 2002). These studies have shown increased accumulation of reactive oxygen species and trace metals - both of which could lead to increased oxidative damage to cellular proteins and structures. Antioxidant and chemoprotective enzymes are globally activated during the course of unloading atrophy (Fig. 4C and D). The most striking changes in expression of genes involved in oxidative stress were the increase in the various isoforms of glutathione-S-transferase (GST), particularly the m and mu subunits (Fig. 4C). GST is a chemoprotective enzyme that could be upregulated as a compensatory response to increased production of reactive oxygen species (ROS) or oxidative stress (Owuor & Kong, 2002). While glutathione peroxidase was downregulated, the glutathione peroxidase precursor was upregulated throughout the period of unloading. The antioxidant defence protein selenoprotein P was upregulated 2-fold by 4 days (Fig. 4D). Other antioxidant defence genes that were upregulated by at least 1.5-fold in one time point and passed statistical tests, but fell below the 2-fold cut-off, include peroxiredoxin 5, selenoprotein W and superoxide dismutase 1 (data not shown). Thioredoxin interacting factor, a gene that inhibits the antioxidant capabilities of thioredoxin (Junn et al. 2000) is upregulated, indicating this chemoprotective story is more complicated than expected.

Along these lines, several genes were upregulated that are consistent with a response to increased accumulation of iron or heavy metals (Fig. 4D). These include feroxidase, transferrin receptor and rhodenase. Metallothionein, another gene normally upregulated in response to trace heavy metal accumulation particularly copper and iron (Coyle et al. 2002), was upregulated more than 2-fold, but did not pass the 2-way ANOVA. Metallothionein has been shown to be upregulated in skeletal muscle in response to starvation-induced atrophy (Jagoe et al. 2002). Iron accumulation has been shown to play an important role in both inactivity and age-related atrophy processes (Kondo et al. 1992; Cook & Yu, 1998). While the role of ROS and trace metal accumulation during atrophy is not well understood, investigators believe that ROS production in muscle wasting conditions is a trigger for the activation of proteolytic systems that promote muscle protein degradation (Reid, 2001). Overall, the complex patterns of change in the oxidative stress genes suggest they are regulated by multiple parallel or sequential cellular processes or switches initiated by increased oxidative stress.

Upregulation of proteolytic machinery. The majority of muscle protein loss after 3-4 days of inactivity is primarily mediated by acceleration in proteolytic processes (Thomason & Booth, 1990). At least three proteolytic systems are known to be involved in muscle protein degradation. These include the ubiquitin-proteasome system, lysosomal cathepsins and Ca²⁺-dependent calpains. Several studies have shown members of each of these pathways are upregulated during disuse (Taillandier et al. 1996; Ikemoto et al. 2001); however, the time course and parallel regulation of these genes has not previously been reported.

Increased proteolysis by the ubiquitin-proteasome system. The ubiquitin-proteasome system is responsible for the majority of protein breakdown that occurs during disuse atrophy (Jagoe & Goldberg, 2001). Our data support this, and uncover additional components of this system not previously reported. Proteins are marked for degradation by ubiquitin through the combined action of ubiquitin activating enzymes (E1s), ubiquitin-conjugating enzymes (E2s) and ubiquitin protein ligases (E3s). E3s are involved in the specific targeting of proteins for degradation by the proteasome (Jagoe & Goldberg, 2001). Atrogin1/MAFbx and MuRF1 are two muscle-specific E3s recently identified as being important to the regulation of protein loss during atrophy (Bodine et al. 2001a; Gomes et al. 2001). Since E3s are responsible for the degradation of specific proteins during atrophy, the identification of E3s involved in atrophy is important in developing therapeutic targets to mitigate protein loss. Our data confirm the upregulation of Atrogin1 but also show an early upregulation during unloading (Fig. 4E). MuRF1 was also upregulated in our data set, but fell below the 2-fold cutoff. Nedd4 is another E3 that is upregulated to a similar extent to Atrogin1. Nedd4 has been implicated in targeting membrane proteins for degradation by the proteasome (Snyder et al. 2001) but has not previously been shown to be upregulated during disuse atrophy. Ubiquitin C was upregulated in a similar manner to Atrogin1 and Nedd4 (Fig. 4E). Several ubiquitin-conjugating enzymes (E2s) were also upregulated, but did not pass our fold-change cut-off. These data are consistent with other findings showing an overall activation of genes that regulate protein ubiquitination (Taillandier et al. 1996; Ikemoto et al. 2001; Jagoe & Goldberg, 2001). One of the most interesting novel findings with regard to protein turnover involves the striking co-activation of eight different gene products comprising structural and regulatory components of the 26S proteasome (Fig. 4F). Previous work has shown the activation of C2, C3, C8 and C9 subunits with atrophy (Taillandier et al. 1996; Ikemoto et al. 2001; Bey et al. 2003), but the co-regulation of these genes was not studied. In the present work each proteasome subunit reaches a peak by day 4 and remains elevated to a smaller degree at days 7 and 14. This pattern of co-regulation suggests that each subunit may share common cis-regulatory elements. This delayed pattern of proteasome activation may be one reason ubiquitin-mediated proteolysis does not play a significant role in muscle protein degradation until days 3 or 4.

Activation of lysosomal proteases. Cellular proteins can also be targeted by lysosomal enzymes, the cathepsins. While some of the cathepsins (cathepsins C, D and L) are upregulated to similar levels as components of the ubiquitin proteasome system by day 4, they are increased to even greater levels at later time points (Fig. 4G). This pattern of activation indicates that the cathepsins may play a greater role in muscle protein degradation during the later stages of atrophy. The activation of cathepsins B, C and L has been shown with atrophy (Taillandier et al. 1996; Ikemoto et al. 2001; Bey et al. 2003) but the time course of this activation with respect to other proteolytic genes has not been demonstrated. Cystatin C, a major inhibitor of cathepsin B (Pavlova et al. 2000), is also activated with a pattern similar to the cathepsins. Cystatin C activation is a novel finding, that raises the possibility that preferential inhibition of cathepsin B may play a role in the later stages of atrophy.

Calcium-dependent proteases. While isolated proteasomes are able to degrade soluble actin and myosin they are not able to degrade intact myofibrils (Jagoe & Goldberg, 2001). This suggests that myofibrillar proteins must be released from the sarcomere before they can be degraded by the proteasome. It has been suggested that calcium-activated calpains mediate this disassembly (Huang & Forsberg, 1998). Previous studies have shown calpain activation during disuse, but results have been mixed and controversial (Tischler et al. 1990; Taillandier et al. 1996; Ikemoto et al. 2001; Tidball & Spencer, 2002). Taillandier et al. (1996) have reported that calpain 2 (m-calpain) is upregulated 1.8-fold after 9 days of hindlimb unloading in rats, while Ikemoto et al. (2001) and Tidball & Spencer (2002) have shown no change with disuse. On the other hand, pharmacological inhibition of calcium-dependent proteases attenuates protein degradation during unloading (Tischler et al. 1990). Calpain 2 did not significantly change in our data set. Calpain 3 is a muscle-specific isoform specifically bound to titin in the z-disk (Sorimachi et al. 2000). Interestingly, calpain 3 was significantly downregulated by 1.5- and 1.6-fold by days 7 and 14 respectively, but it did not pass the 2-fold cut-off (data not shown). Another unexpected result involves the upregulation of calpastatin, a specific inhibitor of calpain 1 and calpain 2, which is upregulated 2.8-fold by day 1 and reaches 4-fold upregulation by day 14 (Fig. 4H). Moreover, the calpain activator, diazepam-binding inhibitor (Dbi) (Melloni et al. 2000) was downregulated by 2.2-fold at day 7. We are aware of no studies that examine protein levels or activation state of calpain 3, calpastatin or Dbi during disuse atrophy. However, overexpression of a calpastatin transgene has been shown to attenuate unloading atrophy in rodents (Tidball & Spencer, 2002). These perplexing results highlight the need for further studies to determine the role of the calpain/calpastatin activity during skeletal muscle atrophy.

Other protease systems. Since the role of calpains in myofibrillar disassembly is far from definitive, we were also interested in examining the expression of other protease systems beyond those described above. One novel aspect of these data is the differential expression of several serine proteases and serine protease inhibitors (serpins). One of the earliest and most striking changes in gene expression involved the activation of serpin 2B (spin2B) (Fig. 4I). Spin2B is upregulated 4-fold at day 1, reaching its peak at 10-fold at days 4 and 7. Another serpin, serpin clade H (originally identified as a collagen-binding protein) was downregulated. Several serine proteases are gradually upregulated over the time course (kallikrein 1, substilin-like Ca²⁺-dependent serine protease and mannose-binding protein associated serine protease). The coordinated actions of serine proteases and their inhibitors are involved in a variety of cellular processes and have been well studied in other cell types (Sangorrin et al. 2002). While increases in protease inhibitors seem counterintuitive in the face of increased degradation, protease inhibitors have been shown to regulate the availability of other proteases or substrates necessary for degradation (Sangorrin et al. 2002). The activity and expression of these genes have not previously been characterized with respect to atrophy. Recent work, however, has identified a serine protease and its endogenous inhibitor that play a role in the disassembly of normal myofibrillar protein turnover in mice (Sangorrin et al. 2002). It is possible therefore, that serine protease cascades are the missing key to myofibrillar disassembly during disuse muscle wasting.

Many of these proteases have also been characterized based on their involvement in complement activation, or have been localized to the ECM of cells (Kaplan et al. 1999; Sim & Laich, 2000). Therefore, it is also possible that they exist in the extracellular space and mediate matrix remodelling of muscle or endothelial cells. It has been demonstrated that the composition of the ECM in soleus muscle is altered after unloading (Miller et al. 2001). Our results show that tissue inhibitor of metalloproteinase 1 (Timp1) was markedly downregulated while neurolysin was upregulated (Fig. 4I). Wittwer et al. also showed upregulation of metalloproteinase transcripts and their inhibitors after 5 weeks of unloading (Wittwer et al. 2002). Beyond altering the elastic properties of muscle tissue, remodelling of the ECM can have significant effects on muscle cell phenotype. Several studies have shown that alterations in ECM composition can lead to changes in gene expression and proliferative state (Danen & Yamada, 2001; Kovanen, 2002). In addition, the action of proteases on the ECM can release sequestered growth factors freeing them to bind to muscle cell receptors and initiating signalling cascades (Eliceiri, 2001). Further work localizing the expression and activity of these proteins in muscle is necessary before conclusions can be made as to their role in ECM remodelling or sarcomeric disassembly.

Amino acid transport and catabolism. Two markers of amino acid catabolism are upregulated with unloading (Fig. 4J). Glutamine synthetase (GS) is upregulated 7-fold by day 7, while cysteine dioxygenase 1 (cdo1), involved in degradation of cysteine to pyruvate, is also upregulated at this point by 3-fold. GS is also induced by corticosteroid administration, which alone elicits muscle weakness and atrophy (Labow et al. 1999). GS is the major enzyme that synthesizes glutamine from glutamate thereby allowing the transport of amino acids out of the musculature and into circulation (Sun et al. 1999). Cdo1 is involved in degradation of cysteine to pyruvate (Ensunsa et al. 1993). Both pyruvate and glutamate can be used as gluconeogenic precursors by the liver. Several other genes involved in the catabolism and transport of amino acids and their derivatives are also upregulated during the course of atrophy (Fig. 4J). This response has also been shown during muscle wasting cause by starvation and glucocorticoid treatment (Labow et al. 1999). It is possible that these genes are activated in an attempt by the musculature to utilize and dispose of the increasing levels of peptide fragments and amino acids released during muscle protein degradation.

Regulatory genes

Figure 5 represents another view of the K-means clusters presented in Fig. 1. This is the same set of clusters; however in this view only genes that are classified as being involved with growth, proliferation, transcription, or signalling are displayed. While transcriptional profiling cannot completely delineate the signalling pathways at work, it can provide insight into the complex regulatory networks that underlie the atrophy process by highlighting certain genes for further study. As seen with ubiquitination and proteasome genes, functionally related genes may cluster together with respect to expression pattern. Therefore the presence of several regulatory genes in the same co-regulated cluster may suggest they share similar functions during atrophy. Full annotations and functional information for each of these genes are available online (http://www.bu.edu/sargent/people/faculty/kandarian_susan.html).

View larger version
[in this window]
[in a new window]

Figure 5. Recapitulation of the K-means analysis from Fig. 1 with only regulatory genes displayed
Regulatory genes include those classified as regulating growth (G), cellular proliferation (P), signalling (S) or transcription (T). Each gene name is followed by a label with the appropriate classification. Cluster E contained no regulatory genes, so no data is displayed. Each data point represents the average fold change for each gene after each time point of unloading. Each data point represents the fold change calculated from the average intensity of control (n = 4) and unloaded (n = 4) GeneChips for each probe after 1, 4, 7 and 14 days (total of 32 chips).

MyoD and muscle-specific gene regulation. One of the earliest responses seen during the course of atrophy is the dramatic upregulation of the muscle-specific transcription factor MyoD (Fig. 5B). MyoD upregulation is usually thought of in the context of initiating myoblast differentiation during skeletal muscle development (McKinsey et al. 2001). While its role in adult muscle is not well understood, it appears to be active in modulating the transcriptional response of muscle genes to different activity paradigms (Booth & Baldwin, 1996). MyoD activates muscle gene expression by recruiting CBP/p300 and other coactivators with histone acetyltransferase (HATs) activity to E-boxes in muscle gene control regions (McKinsey et al. 2001). This effectively loosens the hold of histones on local regions of DNA, and opens up these control regions for binding by other regulatory factors. E-box-dependent transcriptional activation has been implicated in the regulation of MHC IIB and SERCA1 (both fast genes) in response to unloading in rat soleus muscle (Swoap, 1998; Mitchell-Felton et al. 2000). Our MyoD results agree with those published during the initial stages of atrophy (Wheeler et al. 1999; Mitchell-Felton et al. 2000). MyoD upregulation, however, is also seen with increased mechanical load (Adams et al. 1999), where there is a downregulation of fast genes, indicating that it may play a permissive role in the changes that occur in muscle gene expression during adaptations to different activity paradigms. As seen in Fig. 2, fast-type genes are upregulated while genes favouring a slow phenotype are downregulated. Since the upregulation of several of these genes involves E-box-dependent regulatory regions, the question arises as to how expression specificity is accomplished under different situations involving MyoD activation. The answer is likely to lie in the dependence of MyoD activity in association with the co-activators or inhibitors present as well as the number and spatial distribution of E-boxes and other regulatory cis-elements.

Early during the process of atrophy, Cbp/p300-interacting transactivator 2 (Cited2 or Mrg1) is highly upregulated and remains elevated at day 14. Cited2 is a transcriptional co-activator that physically associates with Cbp/p300 enhancing its ability to transactivate a variety of genes (Braganca et al. 2003). Cbp/p300 activates gene expression by acting as a bridge between activators and the general transcriptional machinery and may regulate transcription by enhancing chromatin remodelling (McKinsey et al. 2001). Surprisingly, the expression pattern of MyoD and Cited2 are tightly co-regulated (r² = 0.96) in the same cluster. Given that both MyoD and Cited2 can modulate CBP/p300-based gene transcription and that both of theses genes fall in the same expression cluster it is postulated that these genes somehow interact to control muscle-specific gene expression in unloaded muscle.

Inhibitor of DNA binding 1 (Id1), a transcriptional repressor in muscle and other tissues, is downregulated with unloading (Fig. 5I). Id1 is able to sequester E proteins (myogenic bHLH proteins) and thus inhibit the action of muscle regulatory factors (MRFs) such as MyoD (Jen et al. 1992). The activity of Id1 is thought to regulate muscle fibre size at the transcriptional level (Gundersen & Merlie, 1994). Increases in Id1 have been shown with ageing (Alway et al. 2002) and after muscle disuse prior to 4 days of denervation or nerve impulse block (Gundersen & Merlie, 1994). These contradictory results are perplexing; however, upon closer inspection we found the sequence for the Affymetrix probe was derived from a special splice variant of Id1 called Id1.25 (L23148). In contrast to Id1, the carboxy-terminus of the protein encoded by Id1.25 can form homodimers, which have been suggested to regulate the ability of Id1.25 to bind and inactivate MyoD or E2A gene products (Springhorn et al. 1994). Decreases in either form of Id1 would allow for the increased transcription of certain muscle-specific genes by MyoD or other MRFs. Therefore, specificity of transcriptional activation may involve alterations in ID1/MRF dynamics.

Notch signalling. Transducin-like enhancer of split 4 (Tle4) falls into a cluster of genes activated with a peak at 7 days (Fig. 5F). Tle4 is a repressor that interacts with hairy/enhancer of split proteins (notch signalling) and is expressed during development (Kageyama et al. 2000). Interestingly in the 4-day transient cluster D, there is a member of the hairy-related and enhancer of split family (Hes1), which is significantly upregulated. The activation of myogenic notch signalling pathways during disuse atrophy has not been previously reported. The upregulation of both Hes1 and Tle4 raises the possibility that an interaction between the two is somehow governing transcriptional regulation during unloading. Notch signalling has also been shown to interact with myogenic E proteins as well as to regulate myogenesis.

Jak-Stat signalling. The transcription factors klf9, c-jun, stat5b, and nur77 also have expression patterns represented by cluster F that is transiently upregulated at 7 days (Fig. 5F). Each of these transcriptional regulators may be involved in parallel pathways that regulate different target genes. Janus kinase 2 (Jak2) is also in this cluster raising the possibility that Jak-Stat pathways are involved in muscle wasting. Jak-Stat pathways are most commonly known for modulating cytokine responses. Suppressor of cytokine signalling 2 (Socs2) is a known target of Stat5b (Sadowski et al. 2001) and deletion of the Socs2 gene in mice leads to significant overgrowth (Metcalf et al. 2000). Therefore, Jak-Stat activation of growth inhibitors such as Socs2 may play a role in muscle atrophy with disuse. The significance of such a late and tightly co-regulated response in the context of unloading remains to be determined and is an area ripe for investigation.

Synaptic vesicle remodelling. One of the earliest, growth-related genes to be upregulated was synaptic vesicle glycoprotein 2b (Sv2B). Sv2B was immediately and markedly upregulated by 9.7-fold after 1 day of unloading in cluster A (Fig. 5A). Synaptic vesicle glycoproteins are specialized transporters involved in the reuptake of neurotransmitters into synaptic vesicles after exocytosis and synaptic transmission. While Sv2B is not required for transmission or structural integrity of synaptic vesicles, knockouts have shown that it is necessary for organism growth and long-term survival (Xu & Bajjalieh, 2001). Sc2 is another synaptic vesicle glycoprotein, but is downregulated in cluster J, a cluster that is populated by other genes that have ECM or membrane-based proliferation functions. The relevance of these changes in the context of atrophy is not clear. They may, however, serve as markers of synaptic remodelling or even modulate vesicle trafficking in muscle cells.

Insulin-like growth factor binding proteins. Insulin-like growth factor binding protein 5 (Igfbp5) appears in the same upregulated cluster as MyoD and Cited2 (Fig. 5B). Igfbp5 transcripts were elevated 4-fold by 1 day, 7-fold by 7 days, and remained elevated at 14 days. Other studies have shown that Igfbp5 activation is loading- and activity-dependent (Spangenburg et al. 2002) but have not shown how early it is upregulated and the degree of coordination with other regulatory factors. It is thought that Igfbp5 may inhibit the growth effects of insulin growth factor (IGF)-I via its ability to sequester IGF-I in the ECM (Schneider et al. 2002). Igfbp5 has also been shown to modulate cell growth through IGF-independent mechanisms depending on the localization of its expression (Schneider et al. 2002). Igfbp5 has a nuclear localization sequence and has been shown to interact with nuclear proteins to effect gene transcription (Amaar et al. 2002).

Tgf- $beta$ signalling. Many studies have demonstrated the powerful role that Tgf- $beta$ growth factors such as myostatin play in regulating muscle size (Lee & McPherron, 2001). Myostatin is a negative regulator of muscle growth (Lee & McPherron, 2001). Mutations or knockouts of myostatin lead to the development of a double muscled phenotype characterized by hyperplasia and hypertrophy (McPherron & Lee, 1997; Lee & McPherron, 2001). Studies examining expression mRNA levels of myostatin have been mixed, showing modest increases in plantaris after HU (Wehling et al. 2000) but no change in soleus muscles (Carlson et al. 1999; Kawada et al. 2001). Probes for myostatin do exist on the GeneChip, but levels of expression in controls lie within the low end of detectable limits leading to a high degree of variability. For this reason, the probe for myostatin does not pass our statistical filtering even though samples from our unloaded muscles do show increased myostatin expression over time. While the role that differential regulation of myostatin plays during atrophy is ambiguous, we report here the novel finding that activin IIB (ActIIB), the myostatin receptor, is upregulated 2-fold by day 1 and is further activated 4- and 4.8-fold by days 7 and 14, respectively (Fig. 5B). Activation of the ActIIB receptor could possibly increase the sensitivity of muscle cells to myostatin and lead to decreased growth. While no experimental evidence exists to support this, it has been shown that expression of a dominant negative form of ActIIB receptor leads to the same double muscled phenotype in myostatin knockout mice mentioned above (Lee & McPherron, 2001). Another novel result involves the transient upregulation of follistatin by 3.7-fold at day 7 (Fig. 5F). Follistatin physically interacts with myostatin to inhibit its binding and growth-preventing activity (Hill et al. 2002). Overexpression of follistatin in mice results in hypertrophy similar to that seen with myostatin knockout and dominant negative ActIIB expression (Lee & McPherron, 2001). This delayed activation of a myostatin inhibitor could be an attempt by the cell to counteract early increases in myostatin expression and sensitivity. Another possibility is that this shift, favouring myostatin inhibition during the later time points, may be a molecular switch mechanism involved in the maintenance of the atrophied phenotype.

Cellular proliferation. Muscle fibres, which are post-mitotic and multinucleated, are not lost during periods of inactivity and atrophy but rather decrease in size due to protein loss. It has been demonstrated, however, that myonuclei are lost with atrophy (Darr & Schultz, 1989; Schultz et al. 1994). The opposite seems to be true during hypertrophy, when myonuclei numbers increase and may be a result of satellite cell proliferation (Barton-Davis et al. 1999). However, changes in the number of non-muscle cells during muscle growth or atrophy may be major contributors to these processes and cannot be distinguished from muscle cells in whole tissue gene expression studies.

In cluster A, G0/G1 switch protein 2 (G0S2) is rapidly and transiently upregulated at day 1 (Fig. 5A). G0S2 is the rat homologue for a mouse and human protein that is required for commitment to the G1 phase of the cell cycle (Russell & Forsdyke, 1991). The immediate response of G0S2 suggests it may be involved in a switching mechanism that initiates the cascade of expression changes setting in motion the atrophic changes that occur as a result of inactivity. Several cell cycle regulators are upregulated at later time points as well. Gadd45 and Gas5, two genes that also promote cell cycle arrest, are upregulated in the 4-day transient cluster (Fig. 5D). Cyclin-dependent kinase inhibitor 1 (p21) is upregulated in cluster F by 9.7-fold (Fig. 5F). At high levels, p21 plays a cooperative role with MyoD in inducing growth arrest and cell cycle withdrawal. Several other cell cycle and proliferative genes are differentially regulated during unloading; full annotations are available online (http://www.bu.edu/sargent/people/ faculty/kandarian_susan.html). The exact role of cell cycle regulators in differentiated post-mitotic skeletal muscle tissue is not well understood. However, increases in genes favouring cell cycle arrest would be consistent with the loss of cell types with proliferative capacity such as satellite, fibroblast or endothelial cells (Desplanches et al. 1987; Schmalbruch & Lewis, 2000).

Conclusions

In this work we have identified the overall timeline for multiple biological events as reflected by gene expression patterns during the development of unloading-induced muscle atrophy. There were 309 genes significantly changed by at least 2-fold (214 upregulated and 95 downregulated) at any one time point during the 1, 4, 7, and 14 days of unloading. K-means clustering showed that genes encoding transcription factors, signalling proteins, and growth regulators have complex and varied activation and deactivation patterns and these were either transient or sustained. Other genes in functional categories that increased expression such as glycolysis, fast contraction, proteolysis, amino acid metabolism, or categories of genes with decreased expression such as chaperones, slow contraction, oxidative metabolism, extracellular matrix, and the cytoskeleton, tended to have less varied patterns of activation. This supports the dogma that regulatory genes are upstream of, and moderate the expression of, structural genes. Genes encoding proteins involved in oxidative stress were less tightly co-regulated, reflecting the complexity of this process and its likely involvement in atrophy. Each of the various proteolytic systems had a unique pattern of upregulation with atrophy including amino acid catabolism. The relationships among various gene expression patterns shown here using temporal transcriptional profiling will allow pursuit of a variety of avenues for further study of the regulation of disuse atrophy.

REFERENCES

Top
Abstract
Introduction
Methods
Results
Discussion
References

Adams GR, Haddad F & Baldwin KM (1999). Time course of changes in markers of myogenesis in overloaded rat skeletal muscles. J Appl Physiol 87, 1705-1712 [Abstract/Full Text]

Alway SE, Degens H, Lowe DA & Krishnamurthy G (2002). Increased myogenic repressor Id mRNA and protein levels in hindlimb muscles of aged rats. Am J Physiol Regul Integr Comp Physiol 282, R411-422 [Abstract/Full Text]

Amaar YG, Thompson GR, Linkhart TA, Chen ST, Baylink DJ & Mohan S (2002). Insulin-like growth factor-binding protein 5 (IGFBP-5) interacts with a four and a half LIM protein 2 (FHL2). J Biol Chem 277, 12053-12060 [Abstract/Full Text]

Barton-Davis ER, Shoturma DI & Sweeney HL (1999). Contribution of satellite cells to IGF-I induced hypertrophy of skeletal muscle. Acta Physiol Scand 167, 301-305 [Medline]

Benjamini Y , & Hochberg Y (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc B 57, 289-300

Bey L, Akunuri N, Zhao P, Hoffman EP, Hamilton DG & Hamilton MT (2003). Patterns of global gene expression in rat skeletal muscle during unloading and low-intensity ambulatory activity. Physiol Genomics 13, 157-167 [Abstract/Full Text]

Bodine SC, Latres E, Baumhueter S, Lai VK, Nunez L, Clarke BA, Poueymirou WT, Panaro FJ, Na E, Dharmarajan K, Pan ZQ, Valenzuela DM, Dechiara TM, Stitt TN, Yancopoulos GD & Glass DJ (2001a). Identification of ubiquitin ligases required for skeletal muscle atrophy. Science 294, 1704-1708

Bodine SC, Stitt TN, Gonzalez M, Kline WO, Stover GL, Bauerlein R, Zlotchenko E, Scrimgeour A, Lawrence JC, Glass DJ & Yancopoulos GD (2001b). Akt/mTOR pathway is a crucial regulator of skeletal muscle hypertrophy and can prevent muscle atrophy in vivo. Nat Cell Biol 3, 1014-1019 [Medline]

Booth F , & Baldwin K (1996). Muscle plasticity: energy demand/supply processes. In Handbook of Physiology, section 12, Exercise: Regulation and Integration of Multiple Systems, ed. Rowell L & Sheperd J, chap. 24, pp. 1075-1123. American Physiological Society, Bethesda, MD, USA

Braganca J, Eloranta JJ, Bamforth SD, Ibbitt JC, Hurst HC & Bhattacharya S (2003). Physical and functional interactions between AP-2 transcription factors, p300/CBP, and CITED2. J Biol Chem 278, 16021-16029 [Abstract/Full Text]

Butte AJ, Ye J, Haring HU, Stumvoll M, White MF & Kohane IS (2001). Determining significant fold differences in gene expression analysis. Pac Symp Biocomput 6, 6-17

Carlson CJ, Booth FW & Gordon SE (1999). Skeletal muscle myostatin mRNA expression is fiber-type specific and increases during hindlimb unloading. Am J Physiol 277, R601-606 [Medline]

Cook CI , & Yu BP (1998). Iron accumulation in aging: modulation by dietary restriction. Mech Ageing Dev 102, 1-13 [Medline]

Coyle P, Philcox JC, Carey LC & Rofe AM (2002). Metallothionein: the multipurpose protein. Cell Mol Life Sci 59, 627-647 [Medline]

Cros N, Tkatchenko AV, Pisani DF, Leclerc L, Leger JJ, Marini JF & Dechesne CA (2001). Analysis of altered gene expression in rat soleus muscle atrophied by disuse. J Cell Biochem 83, 508-519 [Medline]

Danen EH , & Yamada KM (2001). Fibronectin, integrins, and growth control. J Cell Physiol 189, 1-13 [Medline]

Darr KC , & Schultz E (1989). Hindlimb suspension suppresses muscle growth and satellite cell proliferation. J Appl Physiol 67, 1827-1834 [Abstract]

Desplanches D, Mayet MH, Sempore B & Flandrois R (1987). Structural and functional responses to prolonged hindlimb suspension in rat muscle. J Appl Physiol 63, 558-563 [Abstract]

Dougherty ER, Barrera J, Brun M, Kim S, Cesar RM, Chen Y, Bittner M & Trent JM (2002). Inference from clustering with application to gene-expression microarrays. J Comput Biol 9, 105-126 [Medline]

Eisen MB, Spellman PT, Brown PO & Botstein D (1998). Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A 95, 14863-14868 [Abstract/Full Text]

Eliceiri BP, (2001). Integrin and growth factor receptor crosstalk. Circ Res 89, 1104-1110 [Abstract/Full Text]

Ensunsa JL, Hirschberger LL & Stipanuk MH (1993). Catabolism of cysteine, cystine, cysteinesulfinate, and OTC by isolated perfused rat hindquarter. Am J Physiol 264, E782-789 [Medline]

Gomes MD, Lecker SH, Jagoe RT, Navon A & Goldberg AL (2001). Atrogin-1, a muscle-specific F-box protein highly expressed during muscle atrophy. Proc Natl Acad Sci U S A 98, 14440-14445 [Abstract/Full Text]

Gundersen K , & Merlie JP (1994). Id-1 as a possible transcriptional mediator of muscle disuse atrophy. Proc Natl Acad Sci U S A 91, 3647-3651 [Abstract]

Hill JJ, Davies MV, Pearson AA, Wang JH, Hewick RM, Wolfman NM & Qiu Y (2002). The myostatin propeptide and the follistatin-related gene are inhibitory binding proteins of myostatin in normal serum. J Biol Chem 277, 40735-40741 [Abstract/Full Text]

Huang J , & Forsberg NE (1998). Role of calpain in skeletal-muscle protein degradation. Proc Natl Acad Sci U S A 95, 12100-12105 [Abstract/Full Text]

Hunter RB, Stevenson E, Koncarevic A, Mitchell-Felton H, Essig DA & Kandarian SC (2002). Activation of an alternative NF-kappaB pathway in skeletal muscle during disuse atrophy. FASEB J 16, 529-538 [Abstract/Full Text]

Ikemoto M, Nikawa T, Takeda S, Watanabe C, Kitano T, Baldwin KM, Izumi R, Nonaka I, Towatari T, Teshima S, Rokutan K & Kishi K (2001). Space shuttle flight (STS-90) enhances degradation of rat myosin heavy chain in association with activation of ubiquitin-proteasome pathway. FASEB J 15, 1279-1281 [Full Text]

Jagoe RT , & Goldberg AL (2001). What do we really know about the ubiquitin-proteasome pathway in muscle atrophy? Curr Opin Clin Nutr Metab Care 4, 183-190 [Medline]

Jen Y, Weintraub H & Benezra R (1992). Overexpression of Id protein inhibits the muscle differentiation program: in vivo association of Id with E2A proteins. Genes Dev 6, 1466-1479 [Abstract]

Junn E, Han SH, Im JY, Yang Y, Cho EW, Um HD, Kim DK, Lee KW, Han PL, Rhee SG & Choi I (2000). Vitamin D3 up-regulated protein 1 mediates oxidative stress via suppressing the thioredoxin function. J Immunol 164, 6287-6295 [Abstract/Full Text]

Kageyama R, Ohtsuka T & Tomita K (2000). The bHLH gene Hes1 regulates differentiation of multiple cell types. Mol Cells 10, 1-7 [Medline]

Kaplan F, Ledoux P, Kassamali FQ, Gagnon S, Post M, Koehler D, Deimling J & Sweezey NB (1999). A novel developmentally regulated gene in lung mesenchyme: homology to a tumor-derived trypsin inhibitor. Am J Physiol 276, L1027-1036 [Medline]

Kawada S, Tachi C & Ishii N (2001). Content and localization of myostatin in mouse skeletal muscles during aging, mechanical unloading and reloading. J Muscle Res Cell Motil 22, 627-633 [Medline]

Koesterer TJ, Dodd SL & Powers S (2002). Increased antioxidant capacity does not attenuate muscle atrophy caused by unweighting. J Appl Physiol 93, 1959-1965 [Abstract/Full Text]

Kondo H, Miura M, Nakagaki I, Sasaki S & Itokawa Y (1992). Trace element movement and oxidative stress in skeletal muscle atrophied by immobilization. Am J Physiol 262, E583-590 [Medline]

Kondo H, Nakagaki I, Sasaki S, Hori S & Itokawa Y (1993). Mechanism of oxidative stress in skeletal muscle atrophied by immobilization. Am J Physiol 265, E839-844 [Medline]

Kovanen V, (2002). Intramuscular extracellular matrix: complex environment of muscle cells. Exerc Sport Sci Rev 30, 20-25 [Medline]

Ku Z, Yang J, Menon V & Thomason DB (1995). Decreased polysomal HSP-70 may slow polypeptide elongation during skeletal muscle atrophy. Am J Physiol 268, C1369-1374 [Medline]

Labow BI, Souba WW & Abcouwer SF (1999). Glutamine synthetase expression in muscle is regulated by transcriptional and posttranscriptional mechanisms. Am J Physiol 276, E1136-1145 [Medline]

Lee SJ , & McPherron AC (2001). Regulation of myostatin activity and muscle growth. Proc Natl Acad Sci U S A 98, 9306-9311 [Abstract/Full Text]

Liu WM, Mei R, Di X, Ryder TB, Hubbell E, Dee S, Webster TA, Harrington CA, Ho MH, Baid J & Smeekens SP (2002). Analysis of high density expression microarrays with signed-rank call algorithms. Bioinformatics 18, 1593-1599 [Abstract]

Lockhart DJ, Dong H, Byrne MC, Follettie MT, Gallo MV, Chee MS, Mittmann M, Wang C, Kobayashi M, Horton H & Brown EL (1996). Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 14, 1675-1680 [Medline]

McKinsey TA, Zhang CL & Olson EN (2001). Control of muscle development by dueling HATs and HDACs. Curr Opin Genet Dev 11, 497-504 [Medline]

McPherron AC , & Lee SJ (1997). Double muscling in cattle due to mutations in the myostatin gene. Proc Natl Acad Sci U S A 94, 12457-12461 [Abstract/Full Text]

Mariani TJ, Budhraja V, Mecham BH, Gu CC, Watson MA & Sadovsky Y (2003). A variable fold change threshold determines significance for expression microarrays. FASEB J 17, 321-323 [Full Text]

Melloni E, Averna M, Salamino F, Sparatore B, Minafra R & Pontremoli S (2000). Acyl-CoA-binding protein is a potent m-calpain activator. J Biol Chem 275, 82-86 [Abstract/Full Text]

Metcalf D, Greenhalgh CJ, Viney E, Willson TA, Starr R, Nicola NA, Hilton DJ & Alexander WS (2000). Gigantism in mice lacking suppressor of cytokine signalling-2. Nature 405, 1069-1073 [Medline]

Miller TA, Lesniewski LA, Muller-Delp JM, Majors AK, Scalise D & Delp MD (2001). Hindlimb unloading induces a collagen isoform shift in the soleus muscle of the rat. Am J Physiol Regul Integr Comp Physiol 281, R1710-1717 [Abstract/Full Text]

Mitchell-Felton H, Hunter RB, Stevenson EJ & Kandarian SC (2000). Identification of weight-bearing-responsive elements in the skeletal muscle sarco(endo)plasmic reticulum Ca²⁺ ATPase (SERCA1) gene. J Biol Chem 275, 23005-23011 [Abstract/Full Text]

Owuor ED , & Kong AN (2002). Antioxidants and oxidants regulated signal transduction pathways. Biochem Pharmacol 64, 765-770 [Medline]

Pavlova A, Krupa JC, Mort JS, Abrahamson M & Bjork I (2000). Cystatin inhibition of cathepsin B requires dislocation of the proteinase occluding loop. Demonstration by release of loop anchoring through mutation of his110. FEBS Lett 487, 156-160 [Medline]

Reid MB, (2001). Invited review: redox modulation of skeletal muscle contraction: what we know and what we don't. J Appl Physiol 90, 724-731 [Abstract/Full Text]

Reiner A, Yekutieli D & Benjamini Y (2003). Identifying differentially expressed genes using false discovery rate controlling procedures. Bioinformatics 19, 368-375 [Abstract]

Russell L , & Forsdyke DR (1991). A human putative lymphocyte G0/G1 switch gene containing a CpG-rich island encodes a small basic protein with the potential to be phosphorylated. DNA Cell Biol 10, 581-591 [Medline]

Sadowski CL, Choi TS, Le M, Wheeler TT, Wang LH & Sadowski HB (2001). Insulin Induction of SOCS-2 and SOCS-3 mRNA expression in C2C12 skeletal muscle cells is mediated by Stat5B. J Biol Chem 276, 20703-20710 [Abstract/Full Text]

St-Amand J, Okamura K, Matsumoto K, Shimizu S & Sogawa Y (2001). Characterization of control and immobilized skeletal muscle: an overview from genetic engineering. FASEB J 15, 684-692 [Abstract/Full Text]

Sangorrin MP, Martone CB & Sanchez JJ (2002). Myofibril-bound serine protease and its endogenous inhibitor in mouse: extraction, partial characterization and effect on myofibrils. Comp Biochem Physiol B Biochem Mol Biol 131, 713-723 [Medline]

Schmalbruch H , & Lewis DM (2000). Dynamics of nuclei of muscle fibers and connective tissue cells in normal and denervated rat muscles. Muscle Nerve 23, 617-626 [Medline]

Schneider MR, Wolf E, Hoeflich A & Lahm H (2002). IGF-binding protein-5: flexible player in the IGF system and effector on its own. J Endocrinol 172, 423-440 [Medline]

Schultz E, Darr KC & Macius A (1994). Acute effects of hindlimb unweighting on satellite cells of growing skeletal muscle. J Appl Physiol 76, 266-270 [Abstract]

Sim RB , & Laich A (2000). Serine proteases of the complement system. Biochem Soc Trans 28, 545-550 [Medline]

Snyder PM, Olson DR, Mcdonald FJ & Bucher DB (2001). Multiple WW domains, but not the C2 domain, are required for inhibition of the epithelial Na⁺ channel by human Nedd4. J Biol Chem 276, 28321-28326 [Abstract/Full Text]

Sorimachi H, Ono Y & Suzuki K (2000). Skeletal muscle-specific calpain, p94, and connectin/titin: their physiological functions and relationship to limb-girdle muscular dystrophy type 2A. Adv Exp Med Biol 481, 383-395; discussion 395-387 [Medline]

Spangenburg EE, Abraha T, Childs TE, Pattison JS & Booth FW (2002). Skeletal muscle IGF-binding protein 3 and 5 expressions are age, muscle, and load dependent. Am J Physiol Endocrinol Metab 284, E340-350 [Abstract/Full Text]

Springhorn JP, Singh K, Kelly RA & Smith TW (1994). Posttranscriptional regulation of Id1 activity in cardiac muscle. Alternative splicing of novel Id1 transcript permits homodimerization. J Biol Chem 269, 5132-5136 [Abstract]

Stein T, Schluter M, Galante A, Soteropoulos P, Tolias P, Grindeland R, Moran M, Wang T, Polansky M & Wade C (2002). Energy metabolism pathways in rat muscle under conditions of simulated microgravity. J Nutr Biochem 13, 471 [Medline]

Sun YN, Mckay LI, Dubois DC, Jusko WJ & Almon RR (1999). Pharmacokinetic/Pharmacodynamic models for corticosteroid receptor down-regulation and glutamine synthetase induction in rat skeletal muscle by a receptor/gene-mediated mechanism. J Pharmacol Exp Ther 288, 720-728 [Abstract/Full Text]

Swoap SJ, (1998). In vivo analysis of the myosin heavy chain IIB promoter region. Am J Physiol 274, C681-687 [Medline]

Taillandier D, Aurousseau E, Meynial-Denis D, Bechet D, Ferrara M, Cottin P, Ducastaing A, Bigard X, Guezennec CY, Schmid HP & Attaix D (1996). Coordinate activation of lysosomal, Ca2+-activated and ATP-ubiquitin-dependent proteinases in the unweighted rat soleus muscle. Biochem J 316, 65-72 [Medline]

Thomason DB , & Booth FW (1990). Atrophy of the soleus muscle by hindlimb umweighting. J Appl Physiol 68, 1-12 [Abstract]

Tidball JG , & Spencer MJ (2002). Expression of a calpastatin transgene slows muscle wasting and obviates changes in myosin isoform expression during murine muscle disuse. J Physiol 545, 819-828 [Abstract/Full Text]

Tischler ME, Rosenberg S, Satarug S, Henriksen EJ, Kirby CR, Tome M & Chase P (1990). Different mechanisms of increased proteolysis in atrophy induced by denervation or unweighting of rat soleus muscle. Metabolism 39, 756-763 [Medline]

Wang X, Li W, Williams M, Terada N, Alessi DR & Proud CG (2001). Regulation of elongation factor 2 kinase by p90(RSK1) and p70 S6 kinase. EMBO J 20, 4370-4379 [Abstract/Full Text]

Wehling M, Cai B & Tidball JG (2000). Modulation of myostatin expression during modified muscle use. FASEB J 14, 103-110 [Abstract/Full Text]

Wheeler MT, Snyder EC, Patterson MN & Swoap SJ (1999). An E-box within the MHC IIB gene is bound by MyoD and is required for gene expression in fast muscle. Am J Physiol 276, C1069-1078 [Medline]

Wittwer M, Fluck M, Hoppeler H, Muller S, Desplanches D & Billeter R (2002). Prolonged unloading of rat soleus muscle causes distinct adaptations of the gene profile. FASEB J 16, 884-886 [Full Text]

Xu T , & Bajjalieh SM (2001). SV2 modulates the size of the readily releasable pool of secretory vesicles. Nat Cell Biol 3, 691-698 [Medline]

Acknowledgements

The authors thank Joachim Theilhaber of Aventis pharmaceuticals for statistical analysis consultation. This work was supported by the National Space Biomedical Research Institute (MA00207) and by a NASA Graduate Student Researchers Program fellowship to E.J.S. (NGT5-50307).

Supplementary material

The online version of this paper can be found at:

DOI: 10.1113/jphysiol.2003.044701

and contains material entitled:

Experimental design and data analysis, data variability and intra-group comparisons.

This article has been cited by other articles:

D. J. Mazzatti, M. A. Smith, R. C. Oita, F.-L. Lim, A. J. White, and M. B. Reid
Muscle unloading-induced metabolic remodeling is associated with acute alterations in PPAR{delta} and UCP-3 expression
Physiol Genomics, July 9, 2008; 34(2): 149 - 161.
[Abstract] [Full Text] [PDF]

Adams GR, Haddad F & Baldwin KM (1999). Time course of changes in markers of myogenesis in overloaded rat skeletal muscles. J Appl Physiol 87, 1705-1712	[Abstract/Full Text]
Alway SE, Degens H, Lowe DA & Krishnamurthy G (2002). Increased myogenic repressor Id mRNA and protein levels in hindlimb muscles of aged rats. Am J Physiol Regul Integr Comp Physiol 282, R411-422	[Abstract/Full Text]
Amaar YG, Thompson GR, Linkhart TA, Chen ST, Baylink DJ & Mohan S (2002). Insulin-like growth factor-binding protein 5 (IGFBP-5) interacts with a four and a half LIM protein 2 (FHL2). J Biol Chem 277, 12053-12060	[Abstract/Full Text]
Barton-Davis ER, Shoturma DI & Sweeney HL (1999). Contribution of satellite cells to IGF-I induced hypertrophy of skeletal muscle. Acta Physiol Scand 167, 301-305	[Medline]
Benjamini Y , & Hochberg Y (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc B 57, 289-300
Bey L, Akunuri N, Zhao P, Hoffman EP, Hamilton DG & Hamilton MT (2003). Patterns of global gene expression in rat skeletal muscle during unloading and low-intensity ambulatory activity. Physiol Genomics 13, 157-167	[Abstract/Full Text]
Bodine SC, Latres E, Baumhueter S, Lai VK, Nunez L, Clarke BA, Poueymirou WT, Panaro FJ, Na E, Dharmarajan K, Pan ZQ, Valenzuela DM, Dechiara TM, Stitt TN, Yancopoulos GD & Glass DJ (2001a). Identification of ubiquitin ligases required for skeletal muscle atrophy. Science 294, 1704-1708
Bodine SC, Stitt TN, Gonzalez M, Kline WO, Stover GL, Bauerlein R, Zlotchenko E, Scrimgeour A, Lawrence JC, Glass DJ & Yancopoulos GD (2001b). Akt/mTOR pathway is a crucial regulator of skeletal muscle hypertrophy and can prevent muscle atrophy in vivo. Nat Cell Biol 3, 1014-1019	[Medline]
Booth F , & Baldwin K (1996). Muscle plasticity: energy demand/supply processes. In Handbook of Physiology, section 12, Exercise: Regulation and Integration of Multiple Systems, ed. Rowell L & Sheperd J, chap. 24, pp. 1075-1123. American Physiological Society, Bethesda, MD, USA
Braganca J, Eloranta JJ, Bamforth SD, Ibbitt JC, Hurst HC & Bhattacharya S (2003). Physical and functional interactions between AP-2 transcription factors, p300/CBP, and CITED2. J Biol Chem 278, 16021-16029	[Abstract/Full Text]
Butte AJ, Ye J, Haring HU, Stumvoll M, White MF & Kohane IS (2001). Determining significant fold differences in gene expression analysis. Pac Symp Biocomput 6, 6-17
Carlson CJ, Booth FW & Gordon SE (1999). Skeletal muscle myostatin mRNA expression is fiber-type specific and increases during hindlimb unloading. Am J Physiol 277, R601-606	[Medline]
Cook CI , & Yu BP (1998). Iron accumulation in aging: modulation by dietary restriction. Mech Ageing Dev 102, 1-13	[Medline]
Coyle P, Philcox JC, Carey LC & Rofe AM (2002). Metallothionein: the multipurpose protein. Cell Mol Life Sci 59, 627-647	[Medline]
Cros N, Tkatchenko AV, Pisani DF, Leclerc L, Leger JJ, Marini JF & Dechesne CA (2001). Analysis of altered gene expression in rat soleus muscle atrophied by disuse. J Cell Biochem 83, 508-519	[Medline]
Danen EH , & Yamada KM (2001). Fibronectin, integrins, and growth control. J Cell Physiol 189, 1-13	[Medline]
Darr KC , & Schultz E (1989). Hindlimb suspension suppresses muscle growth and satellite cell proliferation. J Appl Physiol 67, 1827-1834	[Abstract]
Desplanches D, Mayet MH, Sempore B & Flandrois R (1987). Structural and functional responses to prolonged hindlimb suspension in rat muscle. J Appl Physiol 63, 558-563	[Abstract]
Dougherty ER, Barrera J, Brun M, Kim S, Cesar RM, Chen Y, Bittner M & Trent JM (2002). Inference from clustering with application to gene-expression microarrays. J Comput Biol 9, 105-126	[Medline]
Eisen MB, Spellman PT, Brown PO & Botstein D (1998). Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A 95, 14863-14868	[Abstract/Full Text]
Eliceiri BP, (2001). Integrin and growth factor receptor crosstalk. Circ Res 89, 1104-1110	[Abstract/Full Text]
Ensunsa JL, Hirschberger LL & Stipanuk MH (1993). Catabolism of cysteine, cystine, cysteinesulfinate, and OTC by isolated perfused rat hindquarter. Am J Physiol 264, E782-789	[Medline]
Gomes MD, Lecker SH, Jagoe RT, Navon A & Goldberg AL (2001). Atrogin-1, a muscle-specific F-box protein highly expressed during muscle atrophy. Proc Natl Acad Sci U S A 98, 14440-14445	[Abstract/Full Text]
Gundersen K , & Merlie JP (1994). Id-1 as a possible transcriptional mediator of muscle disuse atrophy. Proc Natl Acad Sci U S A 91, 3647-3651	[Abstract]
Hill JJ, Davies MV, Pearson AA, Wang JH, Hewick RM, Wolfman NM & Qiu Y (2002). The myostatin propeptide and the follistatin-related gene are inhibitory binding proteins of myostatin in normal serum. J Biol Chem 277, 40735-40741	[Abstract/Full Text]
Huang J , & Forsberg NE (1998). Role of calpain in skeletal-muscle protein degradation. Proc Natl Acad Sci U S A 95, 12100-12105	[Abstract/Full Text]
Hunter RB, Stevenson E, Koncarevic A, Mitchell-Felton H, Essig DA & Kandarian SC (2002). Activation of an alternative NF-kappaB pathway in skeletal muscle during disuse atrophy. FASEB J 16, 529-538	[Abstract/Full Text]
Ikemoto M, Nikawa T, Takeda S, Watanabe C, Kitano T, Baldwin KM, Izumi R, Nonaka I, Towatari T, Teshima S, Rokutan K & Kishi K (2001). Space shuttle flight (STS-90) enhances degradation of rat myosin heavy chain in association with activation of ubiquitin-proteasome pathway. FASEB J 15, 1279-1281	[Full Text]
Jagoe RT , & Goldberg AL (2001). What do we really know about the ubiquitin-proteasome pathway in muscle atrophy? Curr Opin Clin Nutr Metab Care 4, 183-190	[Medline]
Jen Y, Weintraub H & Benezra R (1992). Overexpression of Id protein inhibits the muscle differentiation program: in vivo association of Id with E2A proteins. Genes Dev 6, 1466-1479	[Abstract]
Junn E, Han SH, Im JY, Yang Y, Cho EW, Um HD, Kim DK, Lee KW, Han PL, Rhee SG & Choi I (2000). Vitamin D3 up-regulated protein 1 mediates oxidative stress via suppressing the thioredoxin function. J Immunol 164, 6287-6295	[Abstract/Full Text]
Kageyama R, Ohtsuka T & Tomita K (2000). The bHLH gene Hes1 regulates differentiation of multiple cell types. Mol Cells 10, 1-7	[Medline]
Kaplan F, Ledoux P, Kassamali FQ, Gagnon S, Post M, Koehler D, Deimling J & Sweezey NB (1999). A novel developmentally regulated gene in lung mesenchyme: homology to a tumor-derived trypsin inhibitor. Am J Physiol 276, L1027-1036	[Medline]
Kawada S, Tachi C & Ishii N (2001). Content and localization of myostatin in mouse skeletal muscles during aging, mechanical unloading and reloading. J Muscle Res Cell Motil 22, 627-633	[Medline]
Koesterer TJ, Dodd SL & Powers S (2002). Increased antioxidant capacity does not attenuate muscle atrophy caused by unweighting. J Appl Physiol 93, 1959-1965	[Abstract/Full Text]
Kondo H, Miura M, Nakagaki I, Sasaki S & Itokawa Y (1992). Trace element movement and oxidative stress in skeletal muscle atrophied by immobilization. Am J Physiol 262, E583-590	[Medline]
Kondo H, Nakagaki I, Sasaki S, Hori S & Itokawa Y (1993). Mechanism of oxidative stress in skeletal muscle atrophied by immobilization. Am J Physiol 265, E839-844	[Medline]
Kovanen V, (2002). Intramuscular extracellular matrix: complex environment of muscle cells. Exerc Sport Sci Rev 30, 20-25	[Medline]
Ku Z, Yang J, Menon V & Thomason DB (1995). Decreased polysomal HSP-70 may slow polypeptide elongation during skeletal muscle atrophy. Am J Physiol 268, C1369-1374	[Medline]
Labow BI, Souba WW & Abcouwer SF (1999). Glutamine synthetase expression in muscle is regulated by transcriptional and posttranscriptional mechanisms. Am J Physiol 276, E1136-1145	[Medline]
Lee SJ , & McPherron AC (2001). Regulation of myostatin activity and muscle growth. Proc Natl Acad Sci U S A 98, 9306-9311	[Abstract/Full Text]
Liu WM, Mei R, Di X, Ryder TB, Hubbell E, Dee S, Webster TA, Harrington CA, Ho MH, Baid J & Smeekens SP (2002). Analysis of high density expression microarrays with signed-rank call algorithms. Bioinformatics 18, 1593-1599	[Abstract]
Lockhart DJ, Dong H, Byrne MC, Follettie MT, Gallo MV, Chee MS, Mittmann M, Wang C, Kobayashi M, Horton H & Brown EL (1996). Expression monitoring by hybridization to high-density oligonucleotide arrays. Nat Biotechnol 14, 1675-1680	[Medline]
McKinsey TA, Zhang CL & Olson EN (2001). Control of muscle development by dueling HATs and HDACs. Curr Opin Genet Dev 11, 497-504	[Medline]
McPherron AC , & Lee SJ (1997). Double muscling in cattle due to mutations in the myostatin gene. Proc Natl Acad Sci U S A 94, 12457-12461	[Abstract/Full Text]
Mariani TJ, Budhraja V, Mecham BH, Gu CC, Watson MA & Sadovsky Y (2003). A variable fold change threshold determines significance for expression microarrays. FASEB J 17, 321-323	[Full Text]
Melloni E, Averna M, Salamino F, Sparatore B, Minafra R & Pontremoli S (2000). Acyl-CoA-binding protein is a potent m-calpain activator. J Biol Chem 275, 82-86	[Abstract/Full Text]
Metcalf D, Greenhalgh CJ, Viney E, Willson TA, Starr R, Nicola NA, Hilton DJ & Alexander WS (2000). Gigantism in mice lacking suppressor of cytokine signalling-2. Nature 405, 1069-1073	[Medline]
Miller TA, Lesniewski LA, Muller-Delp JM, Majors AK, Scalise D & Delp MD (2001). Hindlimb unloading induces a collagen isoform shift in the soleus muscle of the rat. Am J Physiol Regul Integr Comp Physiol 281, R1710-1717	[Abstract/Full Text]
Mitchell-Felton H, Hunter RB, Stevenson EJ & Kandarian SC (2000). Identification of weight-bearing-responsive elements in the skeletal muscle sarco(endo)plasmic reticulum Ca²⁺ ATPase (SERCA1) gene. J Biol Chem 275, 23005-23011	[Abstract/Full Text]
Owuor ED , & Kong AN (2002). Antioxidants and oxidants regulated signal transduction pathways. Biochem Pharmacol 64, 765-770	[Medline]
Pavlova A, Krupa JC, Mort JS, Abrahamson M & Bjork I (2000). Cystatin inhibition of cathepsin B requires dislocation of the proteinase occluding loop. Demonstration by release of loop anchoring through mutation of his110. FEBS Lett 487, 156-160	[Medline]
Reid MB, (2001). Invited review: redox modulation of skeletal muscle contraction: what we know and what we don't. J Appl Physiol 90, 724-731	[Abstract/Full Text]
Reiner A, Yekutieli D & Benjamini Y (2003). Identifying differentially expressed genes using false discovery rate controlling procedures. Bioinformatics 19, 368-375	[Abstract]
Russell L , & Forsdyke DR (1991). A human putative lymphocyte G0/G1 switch gene containing a CpG-rich island encodes a small basic protein with the potential to be phosphorylated. DNA Cell Biol 10, 581-591	[Medline]
Sadowski CL, Choi TS, Le M, Wheeler TT, Wang LH & Sadowski HB (2001). Insulin Induction of SOCS-2 and SOCS-3 mRNA expression in C2C12 skeletal muscle cells is mediated by Stat5B. J Biol Chem 276, 20703-20710	[Abstract/Full Text]
St-Amand J, Okamura K, Matsumoto K, Shimizu S & Sogawa Y (2001). Characterization of control and immobilized skeletal muscle: an overview from genetic engineering. FASEB J 15, 684-692	[Abstract/Full Text]
Sangorrin MP, Martone CB & Sanchez JJ (2002). Myofibril-bound serine protease and its endogenous inhibitor in mouse: extraction, partial characterization and effect on myofibrils. Comp Biochem Physiol B Biochem Mol Biol 131, 713-723	[Medline]
Schmalbruch H , & Lewis DM (2000). Dynamics of nuclei of muscle fibers and connective tissue cells in normal and denervated rat muscles. Muscle Nerve 23, 617-626	[Medline]
Schneider MR, Wolf E, Hoeflich A & Lahm H (2002). IGF-binding protein-5: flexible player in the IGF system and effector on its own. J Endocrinol 172, 423-440	[Medline]
Schultz E, Darr KC & Macius A (1994). Acute effects of hindlimb unweighting on satellite cells of growing skeletal muscle. J Appl Physiol 76, 266-270	[Abstract]
Sim RB , & Laich A (2000). Serine proteases of the complement system. Biochem Soc Trans 28, 545-550	[Medline]
Snyder PM, Olson DR, Mcdonald FJ & Bucher DB (2001). Multiple WW domains, but not the C2 domain, are required for inhibition of the epithelial Na⁺ channel by human Nedd4. J Biol Chem 276, 28321-28326	[Abstract/Full Text]
Sorimachi H, Ono Y & Suzuki K (2000). Skeletal muscle-specific calpain, p94, and connectin/titin: their physiological functions and relationship to limb-girdle muscular dystrophy type 2A. Adv Exp Med Biol 481, 383-395; discussion 395-387	[Medline]
Spangenburg EE, Abraha T, Childs TE, Pattison JS & Booth FW (2002). Skeletal muscle IGF-binding protein 3 and 5 expressions are age, muscle, and load dependent. Am J Physiol Endocrinol Metab 284, E340-350	[Abstract/Full Text]
Springhorn JP, Singh K, Kelly RA & Smith TW (1994). Posttranscriptional regulation of Id1 activity in cardiac muscle. Alternative splicing of novel Id1 transcript permits homodimerization. J Biol Chem 269, 5132-5136	[Abstract]
Stein T, Schluter M, Galante A, Soteropoulos P, Tolias P, Grindeland R, Moran M, Wang T, Polansky M & Wade C (2002). Energy metabolism pathways in rat muscle under conditions of simulated microgravity. J Nutr Biochem 13, 471	[Medline]
Sun YN, Mckay LI, Dubois DC, Jusko WJ & Almon RR (1999). Pharmacokinetic/Pharmacodynamic models for corticosteroid receptor down-regulation and glutamine synthetase induction in rat skeletal muscle by a receptor/gene-mediated mechanism. J Pharmacol Exp Ther 288, 720-728	[Abstract/Full Text]
Swoap SJ, (1998). In vivo analysis of the myosin heavy chain IIB promoter region. Am J Physiol 274, C681-687	[Medline]
Taillandier D, Aurousseau E, Meynial-Denis D, Bechet D, Ferrara M, Cottin P, Ducastaing A, Bigard X, Guezennec CY, Schmid HP & Attaix D (1996). Coordinate activation of lysosomal, Ca2+-activated and ATP-ubiquitin-dependent proteinases in the unweighted rat soleus muscle. Biochem J 316, 65-72	[Medline]
Thomason DB , & Booth FW (1990). Atrophy of the soleus muscle by hindlimb umweighting. J Appl Physiol 68, 1-12	[Abstract]
Tidball JG , & Spencer MJ (2002). Expression of a calpastatin transgene slows muscle wasting and obviates changes in myosin isoform expression during murine muscle disuse. J Physiol 545, 819-828	[Abstract/Full Text]
Tischler ME, Rosenberg S, Satarug S, Henriksen EJ, Kirby CR, Tome M & Chase P (1990). Different mechanisms of increased proteolysis in atrophy induced by denervation or unweighting of rat soleus muscle. Metabolism 39, 756-763	[Medline]
Wang X, Li W, Williams M, Terada N, Alessi DR & Proud CG (2001). Regulation of elongation factor 2 kinase by p90(RSK1) and p70 S6 kinase. EMBO J 20, 4370-4379	[Abstract/Full Text]
Wehling M, Cai B & Tidball JG (2000). Modulation of myostatin expression during modified muscle use. FASEB J 14, 103-110	[Abstract/Full Text]
Wheeler MT, Snyder EC, Patterson MN & Swoap SJ (1999). An E-box within the MHC IIB gene is bound by MyoD and is required for gene expression in fast muscle. Am J Physiol 276, C1069-1078	[Medline]
Wittwer M, Fluck M, Hoppeler H, Muller S, Desplanches D & Billeter R (2002). Prolonged unloading of rat soleus muscle causes distinct adaptations of the gene profile. FASEB J 16, 884-886	[Full Text]
Xu T , & Bajjalieh SM (2001). SV2 modulates the size of the readily releasable pool of secretory vesicles. Nat Cell Biol 3, 691-698	[Medline]