Human muscle gene expression responses to endurance training provide a novel perspective on Duchenne muscular dystrophy

The FASEB Journal • Research Communication

Human muscle gene expression responses to endurance training provide a novel perspective on Duchenne muscular dystrophy James A Timmons,*,‡,1 Ola Larsson,‡ Eva Jansson,† Helene Fischer,† Thomas Gustafsson,† Paul L Greenhaff,§ John Ridden,* Jonathan Rachman,* Myriam Peyrard-Janvid,* Claes Wahlestedt,‡ and Carl Johan Sundberg* *Department of Physiology and Pharmacology, †Department of Laboratory Medicine, Division of Clinical Physiology, ‡Centre for Genomics and Bioinformatics; Karolinska Institutet, Stockholm, Sweden; and §Centre for Integrated Systems Biology and Medicine, University Medical School, Nottingham, UK Global gene expression profiling is used to generate novel insight into a variety of disease states. Such studies yield a bewildering number of data points, making it a challenge to validate which genes specifically contribute to a disease phenotype. Aerobic exercise training represents a plausible model for identification of molecular mechanisms that cause metabolicrelated changes in human skeletal muscle. We carried out the first transcriptome-wide characterization of human skeletal muscle responses to 6 wk of supervised aerobic exercise training in 8 sedentary volunteers. Biopsy samples before and after training allowed us to identify ⬃470 differentially regulated genes using the Affymetrix U95 platform (80 individual hybridization steps). Gene ontology analysis indicated that extracellular matrix and calcium binding gene families were most up-regulated after training. An electronic reanalysis of a Duchenne muscular dystrophy (DMD) transcript expression dataset allowed us to identify ⬃90 genes modulated in a nearly identical fashion to that observed in the endurance exercise dataset. Trophoblast noncoding RNA, an interfering RNA species, was the singular exception— being up-regulated by exercise and down-regulated in DMD. The common overlap between gene expression datasets may be explained by enhanced ␣7␤1 integrin signaling, and specific genes in this signaling pathway were up-regulated in both datasets. In contrast to these common features, OXPHOS gene expression is subdued in DMD yet elevated by exercise, indicating that more than one major mechanism must exist in human skeletal muscle to sense activity and therefore regulate gene expression. Exercise training modulated diabetes-related genes, suggesting our dataset may contain additional and novel gene expression changes relevant for the anti-diabetic properties of exercise. In conclusion, gene expression profiling after endurance exercise training identified a range of processes responsible for the physiological remodeling of human skeletal muscle tissue, many of which were similarly regulated in DMD. Furthermore, our analysis demonstrates that numerous genes previ-

ABSTRACT

750

ously suggested as being important for the DMD disease phenotype may principally reflect compensatory integrin signaling.—Timmons, J. A., Larsson, O., Jansson, E., Fischer, H., Gustafsson, T., Greenhaff, P. L., Ridden, J., Rachman, J., Peyrard-Janvid, M., Wahlestedt, C., Johan, C. Sundberg Human muscle gene expression responses to endurance training provide a novel perspective on Duchenne muscular dystrophy. FASEB J. 19, 750 –760 (2005) Key Words: gene assay 䡠 diabetes transcriptome 䡠 aerobic training 䡠 DMD Skeletal muscle dysfunction is a major determinant of morbidity and mortality in humans. Physical inactivity is probably the key contributor to skeletal muscle dysfunction under many circumstances. In contrast, regular aerobic exercise combats aspects of diabetes, cardiovascular and “frailty” diseases, making it an effective treatment strategy relevant for more than half a billion patients worldwide (1–3). The potential importance of transcriptome-wide profiling in human skeletal muscle has been highlighted in recent studies involving diabetes patients (4, 5) and children suffering from muscular dystrophy (6 –9). This work holds promise that the underlying molecular mechanisms of disease states with a skeletal muscle component can be identified. Diabetes and muscular dystrophy are characterized by a reduction in oxidative gene expression (e.g., genes coding for mitochondrial proteins) (4 – 6). However, physical inactivity is also a common feature of both conditions, suggesting that down-regulation of these oxidative gene families (referred to as OXPHOS genes) may reflect the physical activity status of the patient rather than any specific disease process. 1

Correspondence: Centre for Genomics and Bioinformatics, Karolinska Institutet, Stockholm, SE171 77, Sweden. E-mail: [email protected] doi: 10.1096/fj.04-1980com 0892-6638/05/0019-0750 © FASEB

Little is known of the mechanisms that connect muscle activity to gene expression in humans. In disease states that affect muscle phenotype, it is difficult to determine when altered gene expression reflects a compensatory (or beneficial) response or whether altered expression directly reflects the underlying disease process. Analysis of biopsy material from patients suffering from Duchenne muscular dystrophy (DMD) identified numerous genes that differ from control tissue (7, 8). Increased expression of embryonic and insulin-like growth factor (IGF) -related genes was attributed to precipitating the disease phenotype (6, 7). An alternative interpretation could be that many genes identified in the DMD transcriptome reflect physiological remodeling of the muscle tissue in response to compensatory signaling (10, 11). We hypothesized that a more informative interpretation of human muscle disease gene expression profiles would be possible after genome-wide transcriptome profiling of skeletal muscle subjected to endurance exercise. Dysfunctional dystrophin glycoprotein complex signaling results in compensatory increases in ␣7␤1 integrin signaling (10, 11). Thus, if the ␣7␤1 integrin signaling pathway represents a mechanosensor for the molecular responses to exercise (12), one might expect a common subset of gene expression changes to be observed in response to exercise and muscular dystrophy. We therefore set out to provide the first robust transcriptome-wide profile of human skeletal muscle gene expression after a period of aerobic training. We then examined claims that gene expression changes associated with DMD (6, 7) were obligatorily related to degeneration of human skeletal muscle function.

ter four times a week (45 min) at 75% of their pretraining peak aerobic capacity (peak VO2) for 6 wk. They were fully supervised during all training sessions. Post-training biopsies were taken 24 h after the last training session. Physiological measurements (heart rate, peak VO2, exercise capacity, and other demographic data) and muscle biopsies were performed as described previously (13–15) and obtained during the same week. Before any molecular analysis, two groups were identified from the original 24 subjects: a high (n⫽8) and a low-responder group (n⫽8). The physiological characteristics of these two groups can be found in Table 1. Subjects were assigned to a group on the basis of their relative improvement in three main physiological parameters (maximal aerobic power, total work performed in a 15 min cycle trial, and reduction in submaximal heart rate during fixed intensity submaximal cycling). Once this ranking process had taken place, muscle tissue from the eight high-responder subjects (Fig. 1) was processed for Affymetrix microarray analysis. The eight high responders were chosen for the array analysis as an attempt to minimize intersubject heterogeneity, ensuring that large intersubject variability did not obviate the use of a microarray approach. Microarray RNA isolation, target preparation, and hybridization Total RNA was extracted from frozen muscle using the acid guanidium thiocyanate-phenol-chloroform extraction method. RNA concentration was determined using a spectrophotometer and quality was controlled by agarose gel electrophoresis. Ten micrograms of RNA per sample was reverse transcribed and processed according to the protocol provided by Affymetrix Inc. In vitro transcription (IVT) was carried out using the Bioarray high-yield RNA transcript labeling kit (P/N900182, Affymetrix). Unincorporated nucleotides from the IVT reaction were removed using the RNeasy column (QIAGEN Inc., Chatsworth, CA, USA). The U95v1 A-E array, consisting of five GeneChip威 arrays, contains ⬃63,000 probe sets. Each sample was hybridized to the Human Genome HG-U95 A-to-E chips resulting in a total of 80 hybridizations.

MATERIALS AND METHODS Array statistical analysis Endurance exercise training protocol Twenty-four sedentary males of Scandinavian origin took part in the aerobic training portion of the study. The study was approved by the ethics committee of the Karolinska Institutet, Stockholm, Sweden, and informed consent was obtained from each volunteer. Subjects abstained from strenuous exercise for 3 wk prior to obtaining pretraining muscle biopsies from the vastus lateralis. Subjects trained on a cycle ergome-

The data presented in the paper were quality controlled using the Microarray Suite software (MAS 5.0). Microarray data were subject to normalization using the Robust Multi-Array Average (RMA) expression measure (16, 17) and the MAS5.0 algorithm separately. Normalized data were then analyzed using the Significance Analysis of Microarray (SAM) method as an Addin within Microsoft Excel (version 1.21) (18). SAM provides a fold change parameter and an estimate of the false

TABLE 1. Baseline demographic and physiological parametersa

LRG HRG

Height (cm)

Age (year)

180 ⫾ 3 183 ⫾ 3 P ⫽ 0.53

23 ⫾ 1 24 ⫾ 1 P ⫽ 0.54

Mass (kg)

Resting mean BP (mmHg)

HR (BPM)

Submaximal HR (BPM)

RER

Maximal peak VO2 (L/min)

15 min work (KJ)

77 ⫾ 3 77 ⫾ 6 P ⫽ 0.97

92 ⫾ 1 88 ⫾ 4 P ⫽ 0.25

71 ⫾ 5 70 ⫾ 6 P ⫽ 0.94

170 ⫾ 5 171 ⫾ 5 P ⫽ 0.85

1.0 ⫾ 0.0 1.0 ⫾ 0.0 P ⫽ 0.45

3.7 ⫾ 0.1 3.5 ⫾ 0.3 P ⫽ 0.48

220 ⫾ 9 204 ⫾ 16 P ⫽ 0.37

a Values are mean ⫾ SE taken before training. HRG is the responder group (top 8 subjects from 24) and LRG is the low responder group (bottom 8 from the 24). This was based on group membership assigned after measuring training improvements following 6 wk of aerobic training and prior to any molecular analysis. Heart rate (HR) is in beat per minute (BPM). Blood pressure (BP) is mmHg. Submaximal heart rate was measured during 10 min constant load cycling at 75% peak VO2 and is beats BPM. RER is the respiratory exchange ratio. 15 min work (KJ) is the total work done in 15 min of self-paced cycling. Peak VO2 is the maximal oxygen uptake measured during an incremental maximal exercise protocol (liters/min). There were no differences between the two groups for any baseline values (unpaired t test P values presented).

SKELETAL MUSCLE GENE EXPRESSION

751

Figure 1. Average changes in physiological variables used to assign group membership after endurance training. Values are mean (⫾se) changes derived from the individual percentage change in each parameter after 6 wk aerobic training (n⫽8 for each group). Submax-HR is in beat per minute and is the reduction in submaximal heart rate after 6 wk training, during 15 min constant load cycling. Peak VO2 is the increase in maximal oxygen uptake measured during an incremental maximal exercise protocol (liters per minute). Max-work is the increase in total work done in 15 min of cycling (KJ). Percentage adaptation for each subject was calculated from the sum of the percentage changes for each of the above three variables. discovery rate (FDR). The method for generating significant gene list from SAM is complex. First, the fold change (FC) criteria used during the operation of the SAM algorithm within Excel determines the number of significant genes yielded for a fixed statistical criterion. To yield the highest number of genes with a 1.5 FC and a FDR of ⬍ 5%, the FC criterion during the operation of SAM varied between 1.4 and 1.5 depending on the dataset characteristics. EASE (19, http://david.niaid.nih.gov/david/ease.htm) was used for gene ontology analysis. Genespring 6.1 was used for probe set calculations, preliminary cluster analysis and dataset comparisons (http://www.silicongenetics.com). All probe sets were updated using the Affymetrix Web site during August 2004. The DMD Affymetrix U95av2 dataset (8) was normalized using RMA. We used the 12 DMD quadriceps samples (average age 3 years; range 0.8 –7) and 9 healthy control quadriceps samples (average age 6.4 years; range 1–17) originating from the Kunkel laboratory (8). We did not include two samples from middle-aged control subjects or one intercostal muscle sample as we thought these were not relevant control samples. SAM analysis was carried out to the same fold change criteria as the original study, where ⬎ 2 FC has been directly identified as a reliable change by Real Time PCR in this bank of DMD tissue (8) Quantitative RT-PCR for validating Affymetrix dataset Total RNA was prepared using the TRIzol method (Invitrogen, San Diego, CA, USA) and quantified using a spectrophotometer (260 nm). RNA (2 ␮g) was reverse transcribed by Superscript reverse transcriptase (Life Technologies, Gothenburg, Sweden) using random hexamer primers (Roche Diagnostics GmbH, Mannheim, Germany) in a total volume of 20 ␮L. Detection of mRNA was performed using a ABI-PRISM威 752

Vol. 19

May 2005

7700 Sequence Detector (Perkin-Elmer Applied Biosystems Inc., Foster City, CA, USA). Oligonucleotide primers and TaqMan probes were designed using Primer Express version 1.5 (Perkin-Elmer Applied Biosystems) and synthesized by Cybergene (Stockholm, Sweden) or ordered as a “gene assay by demand” product (Perkin-Elmer Applied Biosystems) (for sequences or gene assay by demand numbers, see appendix files: RTPCR.xls). Probes were designed to cover exon-exon boundaries to avoid amplification of genomic DNA. 18S rRNA was selected as an endogenous control to correct for potential variations in RNA loading (there are no established low-abundance housekeeping genes for this experimental paradigm). The ⌬⌬Ct method (20) was used to calculate relative changes in mRNA abundance. The threshold cycle (Ct) for 18 s is subtracted from the Ct for the target gene to adjust for variations in mRNA/cDNA generation efficacy. This is carried out for pre- and post-training samples. The pre-exercise value reflects baseline gene expression levels and is subtracted from the postexercise value to calculate the increase or decrease in mRNA abundance. For paired analysis of ⌬Ct, a paired t test was used and precise P values presented. For comparison between high- and low-responder RTPCR data, an unpaired t test was used to compare respective fold changes.

RESULTS Physiological parameters from the Stockholm training study Twenty-four male subjects completed 6 wk of aerobic cycling training at 75% of their pretraining peak aerobic power. Table 1 presents baseline physiological characteristics of the high- and low-responder groups. No differences in baseline demographics or physiological characteristics existed between the low- and highresponder groups. The highest eight responders represent the group used for Affymetrix analysis. The mean change in selected physiological characteristics after 6 wk aerobic training can be found in Fig. 1 (n⫽8/ group). The cumulative training response was significantly greater in the high-responder group (P⬍0.001), as was improvement in each individual physiological parameter. There was no relationship between baseline physiological characteristics and the magnitude of improvement measured for any parameter (n⫽24). Results of the microarray analysis procedures Intelligent use of microarray technology should consider the functional (physiological) characteristics of the patients from where each tissue sample being analyzed was obtained. We decided to carry out microarray analysis using only the eight subjects that demonstrated a large and measurable physiological adaptation. To our knowledge this is the first human muscle gene expression training study to use tissue from subjects proven to have physiologically responded to the training intervention. We normalized the datasets using two different approaches. The first used RMA expression (16) and the second approach used MAS5.0 (see ref 17). The merits of these two approaches

The FASEB Journal

TIMMONS ET AL.

have been discussed extensively and it appears that RMA is emerging as the more acceptable method, since it is better at assigning low variation values to low expression genes (17). RMA normalization identified ⬃4-fold the number of differentially expressed genes compared with MAS-5.0. SAM analysis was carried out using RMA normalized data from all subjects selected for microarray analysis (n⫽8) identifying 236 significant probe sets. Often more than one probe set detected a similar change in gene expression, providing additional validation of the expression data. Moreover, the most homogeneous group of subjects, identified using correlation analysis, enhanced the sensitivity of our analysis. As a consequence, we detected a total of 503 significant probe sets when we analyzed the four most similar subjects. All five RMA normalized chip sets contributed to the analysis in the following rank order A⬎⬎B⬎C⬎D⫽E. The MAS5.0 normalized dataset yielded significant data from the A chip only, producing a list of 80 significant probe sets, of which 32 were unique to the MAS5.0 analysis. In total, we found 387 known genes to be modulated (327 upregulated ⬎1.5-fold and 60 down-regulated ⬎1.45fold). We found 85 genes of unknown identity (64 up-regulated ⬎1.5-fold and 23 down-regulated ⬎1.45-fold); these are presented according to their unique probe set values. A list of all unique genes is presented in the “full gene list.xls” document with a common FDR criteria of ⬍5%.

Results from gene ontology and functional group enrichment analysis We used EASE (19) to determine which gene ontology (GO) groups and functional groups were over-represented in the significant gene list (full gene list.xls). EASE calculates a score (EASE Score) based on the principles of jackknifing (19). In short, one gene is removed from each category cluster to ascertain the reliability of the reported statistical significance value. Such an approach avoids a spuriously high enrichment score for a single gene from a small GO group. We generated a list of GO or functional groups with a significant EASE score and an FDR of ⬍ 5%. One of the limitations of this approach is that the same gene contributes to the enrichment score in many related gene ontology groups. It can be seen that genes related to extracellular matrix (ECM) reorganization and calcium ion binding were up-regulated by endurance exercise training (Table 2). After endurance training, there were fewer significant down-regulated functional groups, which probably reflected the small number of genes we detected (functional groups.xls). The validity of these down-regulated groups (Table 2) may be less robust, as each parameter is based on a fewer number of observations. A comprehensive list of all significant functional groups can be found in the “functional groups.xls” file, where Web links are provided to further online analysis possibilities. The individual genes within each GO or function group are listed in this

TABLE 2. Gene ontology clustering using EASE demonstrates close link between DMD and exercise data sets

GO group

Exercise up-regulated processes GO Molecular function GO Molecular function GO Molecular function GO Molecular function GO Molecular function DMD up-regulated processes GO Molecular function GO Molecular function GO Molecular function GO Molecular function GO Molecualr function Exercise down-regulated processes GO Biological process GO Molecular function GO Molecular function DMD down-regulated processes GO Biological process GO Biological process GO Biological process

List hits

List total

Population hits

Population total

Structural molecule activity Extracellular matrix structural constituent Cell adhesion molecule activity Actin binding Calcium ion binding

50 18

293 293

733 91

12318 12318

3.02E-11 ⬍0.1% 3.51E-11 ⬍0.1%

32 23 35

293 293 293

345 237 577

12318 12318 12318

1.63E-10 ⬍0.1% 4.68E-08 ⬍0.1% 8.65E-07 ⬍0.1%

Structural molecule activity Cell adhesion molecule activity Extracellular matrix structural constituent Calcium ion binding Glycosaminoglycan binding

30 21 12

123 123 123

533 281 73

8173 8173 8173

6.16E-10 ⬍0.1% 4.87E-09 ⬍0.1% 8.48E-09 ⬍0.1%

20 8

123 123

395 70

8173 8173

5.37E-06 ⬍0.1% 7.65E-05 ⬍0.1%

Muscle contraction Protein binding Structural constituent of muscle

8 17 4

46 48 48

139 1629 49

12011 12318 12318

7.51E-07 ⬍0.1% 2.15E-04 ⬍0.1% 8.48E-04 ⬍0.1%

Carbohydrate metabolism Energy pathways Glycogen metabolism

7 5 3

32 32 32

303 184 28

8110 8110 8110

8.64E-04 ⬍0.1% 5.00E-03 ⬍0.1% 5.02E-03 ⬍0.1%

Gene category

EASE score

Global FDR

a A list of all significant genes (n ⫽ 472) within the exercise transcriptome was used (along with the entire Affymetrix U95 transcriptome) to generate information on the representation of significant gene ontology (GO) groups within the significant dataset using EASE. Examination of the significant GO or functional groups indicated that gene transcripts for ECM-related and calcium binding processes represented the most enriched sets of genes for up-regulated processes. Selected, highly significant data are presented above.

SKELETAL MUSCLE GENE EXPRESSION

753

appendix file (represented by Affymetrix probe set numbers) allowing for further independent analysis. Validation of the exercise training Affymetrix expression data We took several approaches to validating the transcript expression changes. First, biopsy material from the eight high-responder subjects was used for RT-PCR of 12 genes (RTPCR.xls). Second, we identified an additional ⬃20 genes from the literature that were modulated in a similar manner to our dataset (see RTPCR.xls). Third, given that metabolic genes may only be modulated in a modest fashion at a transcriptional level, we created a list of OXPHOS genes (see OXPHOS.xls) similar to the idea presented by Mootha et al. (4). This contained 67 genes, mostly up-regulated, and the majority did not meet the study statistical criteria (⬍5% FDR). However, the directional changes are entirely consistent with the known mitochondrial biogenesis after aerobic training (see ref 21). Finally, we examined the expression of three genes (FABP4, IGF2, and IGFBP4) in muscle biopsies from the 8 subjects that demonstrated the least adaptation to aerobic training (lowest 8 responders from the original 24 subjects). This allowed us to establish that altered gene expression related to the extent of physiological adaptation for these genes (Fig. 2).

Characteristics of the DMD gene expression dataset One of the main aims of our study was to facilitate a more rational interpretation of human muscle disease gene expression profiles (4 –9). There are few robust global transcriptome datasets available for direct comparison. One of the available studies (8) provided us with access to all raw data that could be normalized and analyzed in a manner consistent with the exercise training dataset. Two studies did not allow for direct comparison using SAM analysis (6, 7). Using the original study criteria, SAM analysis of the DMD study (8) yielded 187 significant probe sets (DMD.xls). Remarkably, 81 of these probe sets demonstrated nearly identical modulation when compared with the exercise transcription profiling dataset. Not only were there a large number of genes in common, but GO/functional group analysis indicated that the most common biological processes associated with up-regulated gene expression were common to both exercise and the DMD datasets (Table 2). Only one gene, TncRNA, was present in both datasets and demonstrated modulation in the opposite direction (see Fig. 3). TncRNA was down-regulated 2-fold in DMD while being up-regulated 2-fold by exercise training. If we extend our comparison to other published analysis (6, 7), ⬃30 additional gene expression responses were shared between the endurance exercise training and DMD. In contrast, an obvious difference between the two conditions was that OXPHOS genes tend to be up-regulated by exercise (OXPHOS.xls) and down-regulated in DMD (Table 2) (6). This provides evidence that these two clusters of genes (“common genes” and OXPHOS genes) may respond to distinct signals in human skeletal muscle. The overlap between the DMD mRNA expression profile and the endurance training profile is substantially greater than the overlap seen when an analogous type of comparison is made between two cell-based systems modeling a simple biological process (22). This strongly indicates it would not be possible to observe such an overlap by chance alone.

DISCUSSION Figure 2. Selected changes in transcript expression in the high- (n⫽8) vs. low- (n⫽8) responder groups. Values are average fold changes ⫾ se in human skeletal muscle gene expression after 6 wk of aerobic training, determined using TaqMan Real Time PCR. After 6 wk training (n⫽24), the highest and the lowest responders to exercise training were identified from % improvement in maximal aerobic capacity, the % reduction in submaximal heart rate (10 min submaximal cycle) and the % improvement in work done during a 15 min maximal cycling test. This ranking was carried out before any genomic analysis was carried out and therefore it was blinded to the results of the study. Low responders were significantly different from high-responder group using un-paired t tests and the criteria of *P ⫽ 0.07, #P ⬍0.02 754

Vol. 19

May 2005

One of the primary objectives of our study was to create a dataset that allowed for a more insightful analysis of human skeletal muscle disease. To achieve this we generated a transcriptome-wide skeletal muscle mRNA expression profile after a period of endurance exercise training. This has afforded us several novel and important observations. In particular, it has allowed for a new interpretation of the transcriptome profile of human DMD. We have generated information that pertains to diabetes-induced alterations in muscle gene expression and activity sensing mechanisms in human skeletal muscle. Inspection of our data indicates that a large number of ECM-related genes (23–26) were found to be up-

The FASEB Journal

TIMMONS ET AL.

Figure 3. Individual genes common to EXERCISE and Duchenne muscular dystrophy transcript expression profiles. Data were generated using the Affymetrix U95 oligonucleotide array, was normalized using RMA and fold change and significance values were generated using SAM analysis. Genes that demonstrated a significant change in expression (compared with their respective control samples) and were common to the 6 wk of aerobic exercise (EXERCISE) and in Duchenne muscular dystrophy (DMD, raw image files originated from ref 8). Changes in gene expression between baseline (control) and post-training (or disease) were plotted against each other. A total of 76 genes were expressed in the common group whereas 75 from this total were expressed in a similar fashion. ACTC is gene coding for cardiac ␣ actin, DMD is the dystrophin gene and TncRNA is trophoblast noncoding RNA.

regulated by endurance exercise training. Electronic clustering of functionally related genes (Table 2) supports this conclusion by ranking ECM genes as one of the highest ranked class of up-regulated genes in the molecular function GO classification. The highest ranked grouping, structural molecular activity contains many ECM genes, such as Bgn, which codes for biglycan, important for matrix development (27). Genes coding for components of collagen V (COL5A2) and collagen VI (COL6A1 and COL6A2) play an important regulatory role in ECM maturation, as does the basement membrane collagen (COL4), where reduced expression promotes apoptosis, mitochondrial dysfunction and muscle degeneration (28). FN1, ITGB1, and ITGB5 are genes that code for integrin signaling molecules while increased expression of the Laminin gene family (LAMA4, LAMB1, LAMC1) was also evident. Increased expression appears therefore to reflect greater motor activity, presumably through a specific sensor of muscle activation. In support of this idea, some endurance training up-regulated genes have been shown to be down-regulated in rodent models of disuse atrophy (26, 29) supporting the idea that the present transcript expression changes reflect increased muscle activity. In support of the automated gene clustering by EASE, manual annotation provided a list of ⬃45 ECM genes from the total list of 387 genes (⬎12%), further supporting the conclusion that regulation of human skeletal muscle ECM remodeling after endurance training occurs at the level of gene expression (ECMrelated.xls). Exercise-induced gene expression and new insight into muscular dystrophy Duchenne muscular dystrophy is a devastating disease affecting 1 in 3500 male births. There is obvious muscle weakness by 3 years of age and death typically within the second decade of life (30). DMD is initiated by mutaSKELETAL MUSCLE GENE EXPRESSION

tions in the dystrophin gene (31), resulting in an absence of the dystrophin protein. Dystrophin is responsible for communicating mechanical signals from the ECM via the intracellular cytoskeleton, thus regulating muscle gene expression. Reduced dystrophin expression can be partly compensated for by an increase in its homologue utrophin (30). Lack of dystrophin is known to be associated with an up-regulation of ␣7␤1 integrin expression (10), which, as an alternative mechanotransducer, appears to be able to reestablish some degree of functional signaling with the laminins (11). The sequence of events that lead to the degeneration of muscle in DMD remains unclear. Novel approaches for the treatment of DMD are being developed in response to a greater understanding of the molecular basis of the disease process (32). Thus, it is important that the most relevant therapeutic targets are identified from human DMD gene expression studies, as the disease profile in many murine models does not closely reflect the human disease. The present study indicates that some proposed molecular targets deemed to be responsible for muscle degeneration in DMD (6) may not be responsible for precipitating muscle degeneration. There are several similarities between the DMD gene expression profile and our new endurance training dataset. Exercise modestly reduced dystrophin gene expression and increased UTRN and components of integrin (FN1, ITGB1, ITGB1BP3, and ITGB5) and laminin (LAMA4, LAMB1, LAMC1) signaling systems. DMD was associated with up-regulation of the laminin system (LAMA4, LAMB1), an increase in FN1 and ITGA7, and a substantial down-regulation of the dystrophin gene (DMD.xls). Genes responsible for regulating integrin signaling processes such as SPARC or SPP1 were elevated in the DMD and endurance training datasets. SPARC functions to regulate tissue remodeling through growth factor binding and mediates cell-matrix interactions (33). Together, these observations 755

strongly support the idea that, like DMD (10), exercise training in human skeletal muscle activates the ␣7␤1 integrin system. There are compelling reasons to believe that changes in gene expression relate to functional adaptation in response to muscle activity. SPARC expression is reduced during skeletal muscle atrophy and sarcopenia (29). The skeletal muscle-specific protease calpain p94 was suppressed by training, consistent with recent observations in humans (34) and genes involved with translation (EIF4A1, EEFIA1, and RPL3) being elevated, again contrasting with muscle atrophy (29). The endurance training dataset contained nine up-regulated ECM genes that are down-regulated during muscle atrophy (29), providing support that our observations reflect physiological adaptation in response to increased physical activity levels. During development, integrins form a conduit for communication between extracellular and intracellular domains and play a critical role in muscle assembly and tissue angiogenesis (35). Integrin signaling has been proposed as a physiological activity sensor in skeletal muscle (12). To date there is no direct support that this signaling process operates to translate physical activity levels into changes in gene expression in adult human skeletal muscle. There is evidence for increased ␣7␤1 integrin signaling in DMD (10) and the aforementioned tissue remodeling process are clearly important for the known adaptation to endurance exercise training. Reflecting on the present observations, it would appear that a significant proportion of the human DMD gene expression profile mimics that of the response to endurance training (see Table 2 and Fig. 3), and it is reasonable to suggest that the common factor may be enhanced ␣7␤1 integrin signaling. Table 2 clearly indicates that the most up-regulated processes after endurance training are common to the changes observed in DMD muscle tissue. Figure 3 strengthens this observation by demonstrating this involved an identical subset of genes. In a recent DMD transcript expression study by the Kunkel laboratory (9), 80% of genes from the group of most up-regulated genes were up-regulated in the endurance training dataset. The study by Chen et al. (6) demonstrated 12 ECM genes and 4 signaling molecules to be up-regulated in DMD and ␣-sarcoglycan deficient dystrophy patients, all of which is consistent with our endurance exercise training study (full gene list.xls). These common features are suggestive of some parallel between DMD disease progression and muscle degeneration after extreme overtraining. DMD patients demonstrate evidence for muscle satellite cell activation, and hence myogenesis (30). Many factors operate within the setting of the ECM environment to influence the process of myogenesis (36 –38). We detected increased MYADM expression, a phenotypic marker of pluripotent stem cell differentiation, and MARCKS (39), which codes for a signaling molecule that regulates myoblast fusion. The tetraspanin CD81 was up-regulated and is thought to contribute to myoblast fusion in response to integrin signaling. In756

Vol. 19

May 2005

creased expression of ACTC is thought to be indicative of the appearance of an embryonic-like disease phenotype (6, 7) yet was a characteristic of exercise traininginduced muscle remodeling. In contrast to previous assertions (6, 7) this embryonic phenotype or cardiac lineage is unlikely to be a unique marker for the de-regulation of muscle phenotype but rather more likely reflects a simple increase in tissue remodeling. Bakay et al. (7) suggested that dysregulation of the muscle IGF axis may precipitate the degeneration of skeletal muscle in DMD patients. We have demonstrated that endurance training up-regulates IGF1 and IGF2 mRNA as well as the IGF regulatory proteins IGFBP2, IGFBP4, IGFBP7, and PRSS11. The functions of the IGF binding proteins involve activation and inhibition of IGF-related signaling events (40). In addition to the six IGF-related genes common to our study and that of Bakay et al. (7), we demonstrated a decrease in IGFBP5 expression in our analysis of exercise and DMD datasets (see supplementary data). IGF binding protein 5 (IGFBP5) has been shown to directly promote IGF-1 activity (40). However, like other IGF binding proteins, IGFBP5 has IGF-independent activities (40, 41). The importance of decreased IGFBP5 expression for IGFrelated signaling therefore is unclear. To support our contention that alterations in IGF-related gene expression reflect the positive process of increased physiological capacity; we measured two examples of the IGF axis (IGF2 and IGFBP4) in two distinct groups of subjects (Fig. 2). One group demonstrated substantial increases in muscle performance (high responders); the other group (low responders) failed to demonstrate robust improvements in exercise capacity (Fig. 1). All subjects were part of the original training group. As can be seen from Fig. 2, increases in IGF2 and IGFB4 expression were greater in subjects that demonstrated the greater physiological adaptation, establishing a link between our IGF-related gene expression changes and positive adaptation after regular endurance exercise training. If disruption of IGF-related signaling in DMD is not supported by analysis of gene expression profiles, then other potential scenarios must be explored. It is known, for example, that the IGF signaling pathway interacts with transforming growth factor ␤ (TGF-␤) signaling, influencing muscle cell differentiation and proliferation in a complex manner (42). In DMD TGFB and TGFB3 are up-regulated (8) whereas these genes did not appear in the exercise training analysis. This would support the theory that TGFB signaling plays a role in the degeneration of skeletal muscle in DMD (43) partly through the disruption of IGF-related signaling. This may explain why, in animal models of DMD, IGF-1 administration can attenuate the decline in skeletal muscle function (44). In contrast, the argument that DMD phenotype is caused by the up-regulation of several IGF binding proteins blocking the normal physiological role of the IGF pathway (6) is not particularly convincing. Our electronic reanalysis of the Haslett et al. (8) tissue samples generated ⬎ 2-fold change values for 162

The FASEB Journal

TIMMONS ET AL.

transcripts, differentially expressed from control muscle tissue. In total, 45% of these genes were modulated by aerobic exercise training. Strikingly, 75 of the 76 genes in common (81 probe sets) were regulated in the same direction and by a similar magnitude (DMD.xls and Fig. 3). This is too great an overlap to be merely a chance observation (22). Intriguingly, the exception in our analysis was TncRNA, a trophoblast derived noncoding RNA. This noncoding RNA directly regulates gene expression without coding for a protein. TncRNA has been linked with suppression of major histocompatibility complex (MHC) antigens (45). This interfering RNA inhibits all class II transactivator (CIIAT) -mediated transcriptional activity (45), potentially influencing the regulation of numerous signaling and proliferation genes (46). Indeed, disruption of mRNA processing contributes to muscle dystrophies (47), suggesting that interfering RNA species could have relevance for idiopathic muscular dystrophies. It is important to emphasize that there were numerous genes that were not found to be common to the DMD and exercise gene expression profiles. The unique genes (DMD.xls) may in fact better define the DMD disease phenotype. We noted potential senescence genes (22) exclusive to the DMD gene list (DMD.xls). On the other hand gene ontology/functional group analysis indicated that many of the novel genes within the DMD up-regulated gene list belonged to functional processes identified within the exercise training response analysis (Table 2 and functional groups.xls). This suggests that the number of individually modulated genes, common to both datasets (Fig. 3) actually underestimates the degree of commonality between the two conditions. Overall, our analysis suggests that despite the clinical features of DMD involving impaired physical activity, muscle gene expression in the quadriceps of young DMD patients reflects a response that has great similarity to that observed after endurance exercise training. This is in stark contrast with the down-regulation of metabolic genes in DMD (6, Table 2) as exercise typically elevates transcript levels of OXPHOS genes, to some extent (see OXPHOS.xls). Overall, our analysis indicates that many of the genes established to be up-regulated during DMD may reflect increased muscle remodeling (turnover) rather than a specific disease process. Emerging evidence suggests that exercise therapy may accelerate muscle degeneration in models of DMD (48) indicating that enhanced ␣7␤1 integrin signaling due to both lack of dystrophin (10) and exercise-induced activation, may ultimately result in an overtraining-like response within the muscle. Our direct comparison made between DMD and the gene expression changes observed after endurance training provides insight into the potential regulators of skeletal muscle remodeling. From the analysis presented in Table 2, ECM and calcium binding-genes are clear examples of commonly regulated processes in DMD and endurance training. The candidate signaling pathways responsible for the regulation of these comSKELETAL MUSCLE GENE EXPRESSION

mon processes, as well as the opposing OXPHOS response, deserves further consideration. The calcineurin pathway is an established model of a calcium activated signaling pathway that can control skeletal muscle gene expression (49). Two endogenous inhibitors of the calcineurin signaling pathway, PPIB (50) and LNK (51) were up-regulated in our training study whereas a third was down-regulated (FKBP5(1)). Moreover, a key effecter of the calcineurin pathway, CAMKII was up-regulated by 2-fold. An interesting observation was the increased expression of a number of genes coding for calcium binding proteins (e.g., S100A4, S100A10, S100A11, and S100A13) in the DMD and the endurance exercise training groups. A chronic elevation in cytosolic calcium is proposed to contribute to the dysregulation of skeletal muscle in DMD (see ref 30). It therefore seems plausible that both situations involve a chronic elevation in cytosolic calcium and that activation of the calcineurin pathway explains the overlap in gene expression profiles. This suggestion can be challenged, however, as calcineurin dependent remodeling of skeletal muscle typically involves the up-regulation of OXPHOS genes (49) whereas such genes are down-regulated in DMD. Our interpretation, relying on a large number of data points, indicates that remodeling of human skeletal muscle must involve at least two distinctly regulated signaling pathways, since DMD can mimic exercise training with respect to a large number of genes (Table 2) while opposing the OXPHOS response. We would like to suggest, based on the discussion above, that enhanced signaling through the integrin pathways provides the most likely explanation for the common features of within each gene expression dataset. The reduced physical development of children suffering from DMD (and hence low physical activity levels) provides a clue that physical inactivity may be responsible for the reduction in OXPHOS gene expression. Specific inhibition of factors regulating OXPHOS gene expression cannot, however, be ruled out and merits further investigation. Exercise and insights into the diabetes transcriptome Insulin resistance is another important skeletal disease phenotype (4, 5) resulting in disruption of glucose homeostasis. Endurance exercise can reverse insulin resistance, making the muscle more responsive to circulating insulin. In the present endurance training study we did not characterize changes in insulin sensitivity, but rather categorized subjects of the basis of improvement in aerobic capacity. This parameter is, however, highly related to insulin sensitivity (52), so it appears reasonable to discuss exercise-induced gene expression changes in the context of muscle insulin sensitivity. Indeed, some genes involved with metabolism were substantially regulated at the mRNA level after 6 wk aerobic training. FABP4, the gene encoding aP2, was up-regulated to a greater extent in those subjects that improved their 757

aerobic capacity the most (Fig. 2), emphasizing that our dataset contains expression changes that directly relate to the magnitude of physiological adaptation. FABP4 is thought to regulate cell fatty acid availability (53) and hence influence muscle substrate selection by the mitochondria. Rodent models of insulin resistance have demonstrated that reducing FABP4 expression actually enhances insulin sensitivity. It is therefore intriguing that in humans, aerobic exercise results in greatly enhanced FABP4 mRNA expression. It is plausible that this increase in gene expression did not result in increased protein expression. This would, however, make the greater response in the high-responder group (Fig. 2) more difficult to rationalize. We observed a ⬎ 1.5-fold increase in FABP5, COX6, CYBRD1, LPL, and other enzymes involved with oxidative metabolism during a period when muscle insulin sensitivity should have been enhanced. Many other OXPHOS genes were up-regulated by a modest amount (OXPHOS.xls). Several genes previously associated with human diabetes were significantly modulated by exercise. RRAD (54), UCP3 (55), and PDHK4 (56) have been shown to be elevated in diabetes and all were, logically, suppressed by aerobic training. Indeed, PDHK4 was one of the most down-regulated genes, and this may reflect concurrent repression of its transcriptional activator FOXO3A (56). PEA15 expression is increased in diabetic muscle (57) and overexpression increases basal glucose transport (58). Aerobic exercise training increased PEA15 expression, suggesting that in human diabetic muscle tissue, the PEA15 response reflects a compensatory adaptation in the face of insulin resistance. Recently, Mootha et al. (4) and Patti et al. (5) claimed that diabetes was associated with a specific reduction in OXPHOS genes. Chen et al. (6) demonstrated downregulation of OXPHOS genes in muscular dystrophy using a microarray approach. We think it plausible that inactivity per se is a likely common factor explaining all of these observations, suggesting that a thorough characterization of patient physical activity levels must be carried out before specific disease-related pathways are proposed from the outcome of gene array profiling (4, 5).

endurance exercise gene expression profile contained diabetes-related genes, suggesting our dataset may contain genes that could represent future anti-diabetic drug targets. The authors would like to thank Tim Buchanan (PGRD, UK) for handling all aspects of ICH Good Clinical Practice implementation and his excellent management of the exercise project data. We also thank Claire M. Johnson and Frank Burslem (PGRD, UK) for their initial advice on the Affymetrix platform and the assistance of the Affymetrix Core facility (Karolinska Institutet, NOVUM). We would like to acknowledge the support of CIF (Sweden), Swedish Heart and Lung Foundation, and the Thurings Foundation.

REFERENCES 1.

2.

3.

4.

5.

6.

7. 8.

CONCLUSIONS We have identified ⬃500 of the most modulated genes in human skeletal muscle after aerobic training. A total of ⬃50 of these genes were confirmed by a variety of methodologies as being modulated by aerobic training. Strikingly, at least 100 genes responsive to endurance exercise training are regulated in a similar manner in DMD. Genes once considered to be involved in the pathogenesis of DMD should be reevaluated in light of the present study. In addition, we have provided evidence that human skeletal muscle gene expression is regulated by at least two distinct molecular pathways, which can operate in opposing directions, under certain circumstances (such as in DMD). Finally, the 758

Vol. 19

May 2005

9.

10.

11.

12. 13.

Ekelund, L. G., Haskell, W. L., Johnson, J. L., Whaley, F. S., Criqui, M. H., and Sheps, D. S. (1988) Physical fitness as a predictor of cardiovascular mortality in asymptomatic North American men. The Lipid Research Clinics Mortality Follow-up Study. N. Engl. J. Med. 319, 1379 –1384 Paffenbarger, R. S., Jr., Hyde, R. T., Wing, A. L., Lee, I. M., Jung, D. L., and Kampert, J. B. (1993) The association of changes in physical-activity level and other lifestyle characteristics with mortality among men. N. Engl. J. Med. 328, 538 –545 Manson, J. E., Greenland, P., LaCroix, A. Z., Stefanick, M. L., Mouton, C. P., Oberman, A., Perri, M. G., Sheps, D. S., Pettinger, M. B., and Siscovick, D. S. (2002) Walking compared with vigorous exercise for the prevention of cardiovascular events in women. N. Engl. J. Med. 347, 716 –725 Mootha, V. K., Lindgren, C. M., Eriksson, K. F., Subramanian, A., Sihag, S., Lehar, J., Puigserver, P., Carlsson, E., Ridderstrale, M., Laurila, E., et al. (2003) PCG-1␣-responsive genes involved in oxidative phosphorylation are co-coordinately down regulated in human diabetes. Nat. Genet. 34, 267–273 Patti, M. E., Butte, A. J., Crunkhorn, S., Cusi, K., Berria, R., Kashyap, S., Miyazaki, Y., Kohane, I., Costello, M., Saccone, R., et al. (2003) Coordinated reduction of genes of oxidative metabolism in humans with insulin resistance and diabetes: potential role of PGC1 and NRF1. Proc. Natl. Acad. Sci. USA 100, 8466 – 8471 Chen, Y. W., Zhao, P., Borup, R., and Hoffman, E. P. (2000) Expression profiling in the muscular dystrophies: identification of novel aspects of molecular pathophysiology. J. Cell Biol. 151, 1321–1336 Bakay, M., Zhao, P., Chen, J., and Hoffman, E. P. (2002) A web-accessible complete transcriptome of normal human and DMD muscle. Neuromuscul. Disord. S1, 125–141 Haslett, J. N., Sanoudou, D., Kho, A. T., Bennett, R. R., Greenberg, S. A., Kohane, I. S., Beggs, A. H., and Kunkel, L. M. (2002) Gene expression comparison of biopsies from Duchenne muscular dystrophy (DMD) and normal skeletal muscle. Proc. Natl. Acad. Sci. USA 99, 15000 –15005 Haslett, J. N., Sanoudou, D., Kho, A. T., Han, M., Bennett, R. R., Kohane, I. S., Beggs, A. H., and Kunkel, L. M. (2003) Gene expression profiling of Duchenne muscular dystrophy skeletal muscle. Neurogenetics 4, 163–171 Hodges, B. L., Hayashi, Y. K., Nonaka, I., Wang, W., Arahata, K., and Kaufman, S. J. (1997) Altered expression of the alpha7beta1 integrin in human and murine muscular dystrophies. J. Cell Sci. 110, 2873–2881 Burkin, D. J., Wallace, G. Q., Nicol, K. J., Kaufman, D. J., and Kaufman, S. J. (2001) Enhanced expression of the alpha7beta1 integrin reduces muscular dystrophy and restores viability in dystrophic mice. J. Cell Biol. 152, 1207–1218 Carson, J. A., and Wei, L. (2000) Integrin signaling's potential for mediating gene expression in hypertrophying skeletal muscle. J. Appl. Physiol. 88, 337–343 Norrbom, J., Sundberg, C. J., Ameln, H., Kraus, W. E., Jansson, E., and Gustafsson, T. (2004) PGC-1␣ mRNA expression is

The FASEB Journal

TIMMONS ET AL.

14.

15.

16.

17.

18. 19. 20.

21.

22.

23.

24.

25. 26.

27.

28.

29.

30. 31.

influenced by metabolic perturbation in exercising human skeletal muscle. J. Appl. Physiol. 96, 189 –194 Timmons, J. A., Gustafsson, T., Sundberg, C. J., Jansson, E., Hultman, E., Kaijser, L., Chwalbinska-Moneta, J., ConstantinTeodosiu, D., Macdonald, I. A., and Greenhaff, P. L. (1998) Acetyl group availability limits oxidative ATP regeneration at the onset of skeletal muscle contraction in humans. J. Clin. Invest. 101, 79 – 85 Tyni-Lenne, R., Dencker, K., Gordon, A., Jansson, E., and Sylven, C. (2001) Comprehensive local muscle training increases aerobic working capacity and quality of life and decreases neurohormonal activation in patients with chronic heart failure. Eur. J. Heart Fail. 3, 47–52 Irizarry, R. A., Hobbs, B., Collin, F., Beazer-Barclay, Y. D., Antonellis, K. J., Scherf, U., and Speed, T. P. (2003) Exploration, data normalization and summaries of high density oligonucleotide array probe level. Biostatistics 4, 249 –264 Bolstad, B. M., Irizarry, R. A., Astrand, M., and Speed, T. P. (2003) A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19, 185–193 Tusher, V. G., Tibshirani, R., and Chu, G. (2001) Significance analysis of microarrays applied to the ionizing radiation response. Proc. Natl. Acad. Sci. USA 98, 5116 –5121 Hosack, D. A., Dennis, G., Jr., Sherman, B. T., Lane, H. C., and Lempicki, R. (2003) Identifying biological themes within lists of genes with EASE. Genome Biol. 4, R70 Winer, J., Jung, C. K., Shackel, I., and Williams, P. M. (1999) Development and validation of real-time quantitative reverse transcriptase-polymerase chain reaction for monitoring gene expression in cardiac myocytes in vitro. Anal. Biochem. 270, 41– 49 Bengtsson, J., Gustafsson, T., Widegren, U., Jansson, E., and Sundberg, C. J. (2001) Mitochondrial transcription factor A and respiratory complex IV increase in response to exercise training in humans. Pflugers Arch. 443, 61– 66 Larsson, O., Scheele, C., Liang, Z., Moll, J., Karlsson, C., and Wahlestedt, C. (2004) Kinetics of senescence associated changes of gene expression in an epithelial temperature-sensitive SV40 large T antigen model. Cancer Res. 64, 482– 489 Koskinen, S. O., Kjaer, M., Mohr, T., Sorensen, F. B., Suuronen, T., and Takala, T. E. (2000) Type IV collagen and its degradation in paralyzed human muscle: effect of functional electrical stimulation. Muscle Nerve 23, 580 –589 Lluis, F., Roma, J., Suelves, M., Parra, M., Aniorte, G., Gallardo, E., Illa, I., Rodriguez, L., Hughes, S. M., Carmeliet, P., et al. (2001) Urokinase-dependent plasminogen activation is required for efficient skeletal muscle regeneration in vivo. Blood 97, 1703–1711 Mayer, U. (2003) Integrins: redundant or important players in skeletal muscle? J. Biol. Chem. 278, 14587–14590 Pattison, J. S., Folk, L. C., Madsen, R. W., Childs, T. E., and Booth, F. W. (2003) Transcriptional profiling identifies extensive down regulation of extracellular matrix gene expression in sarcopenic rat soleus muscle. Physiol. Genomics 15, 34 – 43 Xu, T., Bianco, P., Fisher, L. W., Longenecker, G., Smith, E., Goldstein, S., Bonadio, J., Boskey, A., Heegaard, A. M., Sommer, B., et al. (1998) Targeted disruption of the biglycan gene leads to an osteoporosis-like phenotype in mice. Nat. Genet. 20, 78 – 82 Irwin, W. A., Bergamin, N., Sabatelli, P., Reggiani, C., Megighian, A., Merlini, L., Braghetta, P., Columbaro, M., Volpin, D., Bressan, G. M., et al. (2003) Mitochondrial dysfunction and apoptosis in myopathic mice with collagen VI deficiency. Nat. Genet. 35, 367–371 Lecker, S. H., Jagoe, R. T., Gilbert, A., Gomes, M., Baracos, V., Bailey, J., Price, S. R., Mitch, W. E., and Goldberg, A. L. (2004) Multiple types of skeletal muscle atrophy involve a common program of changes in gene expression. FASEB J. 18, 39 –51 Blake, D. J., Weir, A., Newey, S. E., and Davies, K. E. (2002) Function and genetics of dystrophin and dystrophin-related proteins in muscle. Physiol. Rev. 2, 291–329 Monaco, A. P., Neve, R. L., Colletti-Feener, C., Bertelson, C. J., Kurnit, D. M., and Kunkel, L. M. (1986) Isolation of candidate cDNAs for portions of the Duchenne muscular dystrophy gene. Nature (London) 323, 646 – 650

SKELETAL MUSCLE GENE EXPRESSION

32. 33.

34.

35. 36.

37. 38.

39.

40. 41. 42.

43.

44.

45.

46. 47. 48.

49. 50. 51.

52.

53.

Bogdanovich, S., Perkins, K. J., Krag, T. O., and Khurana, T. S. (2004) Therapeutics for Duchenne muscular dystrophy: current approaches and future directions. J. Mol. Med. 82, 102–115 Francki, A., Motamed, K., McClure, T. D., Kaya, M., Murri, C., Blake, D. J., Carbon, J. G., and Sage, E. H. (2003) SPARC regulates cell cycle progression in mesangial cells via its inhibition of IGF dependent signaling. J. Cell. Biochem. 88, 802– 811 Jones, S. W., Hill, R. J., Krasney, P. A., O'Conner, B., Pierce, N., and Greenhaff, P. L. (2004) Disuse atrophy and exercise rehabilitation in humans profoundly affects the expression of genes associated with the regulation of skeletal muscle mass. FASEB J. 18, 1025–1027 Gullberg, D. (2003) Cell biology: the molecules that make muscle. Nature (London) 424, 138 –140 Henriquez, J. P., Casar, J. C., Fuentealba, L., Carey, D. J., and Brandan, E. (2002) Extracellular matrix histone H1 binds to perlecan is present in regenerating skeletal muscle and stimulates myoblast proliferation. J. Cell Sci. 115, 2041–2051 Rudnicki, M. A. (2003) Marrow to muscle, fission versus fusion. Nat. Med. 9, 1461–1462 Casar, J. C., Cabello-Verrugio, C., Olguin, H., Aldunate, R., Inestrosa, N. C., and Brandan, E. (2004) Heparan sulfate proteoglycans are increased during skeletal muscle regeneration: requirement of syndecan-3 for successful fiber formation. J. Cell Sci. 117, 73– 84 Kim, S. S., Kim, J. H., Lee, S. H., Chung, S. S., Bang, O. S., Park, D., and Chung, C. H. (2002) Involvement of protein phosphatase-1-mediated MARCKS translocation in myogenic differentiation of embryonic muscle cells. J. Cell Sci. 115, 2465–2473 Schneider, M. R., Wolf, E., Hoeflich, A., and Lahm, H. (2002) IGF-binding protein-5: flexible player in the IGF system and effector on its own. J. Endocrinol. 172, 423– 440 Ricort, J. M. (2004) Insulin-like growth factor binding protein (IGFBP) signaling. Growth Horm. IGF Res. 14, 277–286 Kamanga-sollo, E., Pampusch, M. S., White, M. E., and Dayton, W. R. (2003) Role of insulin-like growth factor binding protein (IGFBP)-3 in TGF-beta- and GDF-8 (myostatin)-induced suppression of proliferation in porcine embryonic myogenic cell cultures. J. Cell. Physiol. 197, 225–231 Murakami, N., McLennan, I. S., Nonaka, I., Koishi, K., Baker, C., and Hammond-Tooke, G. (1999) Transforming growth factor-␤2 is elevated in skeletal muscle disorders. Muscle Nerve 22, 889 – 898 Gregorevic, P., Plant, D. R., Leeding, K. S., Bach, L. A., and Lynch, G. S. (2002) Improved contractile function of the mdx dystrophic mouse diaphragm muscle after insulin-like growth factor-I administration. Am. J. Pathol. 161, 2263–2272 Geirsson, A., Lynch, R. J., Paliwal, I., Bothwell, A. L., and Hammond, G. L. (2003) Human trophoblast noncoding RNA suppresses CIITA promoter III activity in murine B-lymphocytes. Biochem. Biophys. Res. Commun. 14, 718 –724 Nagarajan, U. M., Bushey, A., and Boss, J. M. (2002) Modulation of gene expression by the MHC class II transactivator. J. Immunol. 169, 5078 –5088 Watchko, J. F., O'Day, T. L., and Hoffman, E. P. (2002) Functional characteristics of dystrophic skeletal muscle: insights from animal models. J. Appl. Physiol. 93, 407– 417 De Luca, A., Pierno, S., Liantonio, A., Cetrone, M., Camerino, C., Fraysse, B., Mirabella, M., Servidei, S., Ruegg, U. T., and Conte Camerino, D. (2003) Enhanced dystrophic progression in mdx mice by exercise and beneficial effects of taurine and insulin-like growth factor-1. J. Pharmacol. Exp. Ther. 304, 453– 463 Olson, E. N., and Williams, R. S. (2000) Calcineurin signaling and muscle remodeling. Cell 101, 689 – 692 Carpentier, M., Allain, F., Haendler, B., Slomianny, M. C., and Spik, G. (2002) Delineation of the calcineurin interacting region of cyclophilin B. Protein Sci. 9, 2386 –2393 Velazquez, L., Cheng, A. M., Fleming, H. E., Furlonger, C., Vesely, S., Bernstein, A., Paige, C. J., and Pawson, T. (2002) Cytokine signaling and hematopoietic homoeostasis are disrupted in Lnk-deficient mice. J. Exp. Med. 195, 1599 –1611 Seibaek, M., Vestergaard, H., Burchardt, H., Sloth, C., TorpPedersen, C., Nielsen, S. L., Hildebrandt, P., and Pedersen, O. (2003) Insulin resistance and maximal oxygen uptake. Clin. Cardiol. 26, 515–520 Uysal, K. T., Scheja, L., Wiesbrock, S. M., Bonner-Weir, S., and Hotamisligil, G. S. (2000) Improved glucose and lipid metabo-

759

54. 55.

56.

57.

760

lism in genetically obese mice lacking aP2. Endocrinology 141, 3388 –3396 Reynet, C., and Kahn, C. R. (1993) Rad: a member of the Ras family overexpressed in muscle of type II diabetic humans. Science 262, 1441–1444 Bao, S., Kennedy, A., Wojciechowski, B., Wallace, P., Ganaway, E., and Garvey, W. T. (1998) Expression of mRNAs encoding uncoupling proteins in human skeletal muscle: effects of obesity and diabetes. Diabetes 47, 1935–1940 Kwon, H. S., Huang, B., Unterman, T. G., and Harris, R. A. (2004) Protein kinase B-alpha inhibits human pyruvate dehydrogenase kinase-4 gene induction by dexamethasone through inactivation of FOXO transcription factors. Diabetes 53, 899 –910 Condorelli, G., Vigliotta, G., Iavarone, C., Caruso, M., Tocchetti, C. G., Andreozzi, F., Cafieri, A., Tecce, M. F., Formisano, P.,

Vol. 19

May 2005

Beguinot, L., et al. (1998) PED/PEA-15 gene controls glucose transport and is overexpressed in type 2 diabetes mellitus. EMBO J. 17, 3858 –3866 58. Condorelli, G., Vigliotta, G., Trencia, A., Maitan, M. A., Caruso, M., Miele, C., Oriente, F., Santopietro, S., Formisano, P., and Beguinot, F. (2001) Protein kinase C (PKC)-alpha activation inhibits PKC-zeta and mediates the action of PED/PEA-15 on glucose transport in the L6 skeletal muscle cells. Diabetes 50, 1244 –1252

The FASEB Journal

Received for publication August 11, 2004. Accepted for publication December 14, 2004.

TIMMONS ET AL.