Mdx mice show a milder phenotype than Duchenne patients despite bearing an analogous genetic defect. Our aim was to sort out genes, differentially expressed during the evolution of skeletal muscle mdx mouse disease, to elucidate the mechanisms by which these animals overcome the lack of dystrophin. Genome-wide microarray-based gene expression analysis was carried out at 3 wk and 1.5 and 3 mo of life. Candidate genes were selected by comparing: 1) mdx vs. controls at each point in time, and 2) mdx mice and 3) control mice among the three points in time. The first analysis showed a strong upregulation (96%) of inflammation-related genes and in >75% of genes related to cell adhesion, muscle structure/regeneration, and extracellular matrix remodeling during mdx disease evolution. Lgals3, Postn, Ctss, and Sln genes showed the strongest variations. The analysis performed among points in time demonstrated significant changes in Ecm1, Spon1, Thbs1, Csrp3, Myo10, Pde4b, and Adamts-5 exclusively during mdx mice lifespan. RT-PCR analysis of Postn, Sln, Ctss, Thbs1, Ecm1, and Adamts-5 expression from 3 wk to 9 mo, confirmed microarray data and demonstrated variations beyond 3 mo of age. A high-confidence functional network analysis demonstrated a strong relationship between them and showed two main subnetworks, having Dmd-Utrn-Myo10 and Adamts5-Thbs1-Spon1-Postn as principal nodes, which are functionally linked to Abca1, Actn4, Crebbp, Csrp3, Lama1, Lama3, Mical2, Mical3, Myf6, Pxn, and Sparc genes. Candidate genes may participate in the decline of muscle necrosis in mdx mice and could be considered potential therapeutic targets for Duchenne patients.
- skeletal muscle
- temporal gene expression analysis
- gene functional networks
mdx (C57BL/10ScSn-Dmdmdx) mice, the most widely used animal model in the study of Duchenne muscular dystrophy (DMD), bear a nonsense mutation in exon 23 of the dystrophin gene that causes a lack of this protein (57). In the skeletal muscle of mdx mice, fiber necrosis is absent before the 2nd wk of life, but becomes very active between the 3rd and 8th wk, diminishing progressively to remain at lower intensity levels throughout the rest of the mdx lifespan (41, 52, 61). In the last few years, we have analyzed the gastrocnemius muscle group pathological changes throughout the mdx mouse lifespan and have identified three main phases in its disease: 1) premuscle necrosis phase from birth to 2–3 wk of life, 2) maximal necrosis phase expanding from 3 wk to 1.5 mo of life, and 3) steady phase, after 3 mo of life, when muscle necrosis begins to decrease progressively and remains lastingly stable (52). It is well known that the reappearance of utrophin in the muscle cell membrane participates in the reduction of muscle fiber necrosis (23, 27, 53). However, compared with DMD patients, other factors in addition to utrophin must act in the mdx mice to explain the differences of disease intensity shown by these animals.
The global gene expression profile of dystrophic muscle, based on DNA microarray analysis, has been previously studied both in patients with DMD (10, 19, 26, 29, 30) and in mdx mice (8, 13, 44, 46–48, 54, 60, 62, 64). In particular, Porter et al. (45) applied for the first time the high-density oligonucleotide arrays from Affymetrix platform to study gene expression profiles of three different muscle types, such as limb, masticatory, and extraocular muscles, in adult wild-type mice. Afterward, they used the microarrays technology to compare the skeletal muscle gene expression profiles between mdx and wild-type mice and described the strong expression of the genes involved in the inflammatory response and their probable contribution to the pathogenesis of the dystrophin-deficient muscle (43). However, these authors, as well as other groups (13, 60, 62), studied gene expression patterns at single points in time of mdx mouse muscle disease evolution. Subsequent studies were directed to analyze gene expression profiles at diverse moments of the mdx mouse lifespan (47, 48, 54, 64), though only the comparisons between mdx and control mice at each point in time were considered. For this reason, no data were available with regard to the changes that occurred throughout the different points in time during mdx mouse or wild-type life evolution. Other studies have shown that the response to the lack of dystrophin varies in different muscle groups of mdx mice. Thus, while the diaphragm appears to be the most severely affected muscle (24, 39, 47, 48, 60), hindlimb muscles change their fate after a period of high-grade necrosis to manage a nonsevere pathology (24, 48, 60), and extraocular muscles appear to be unaffected by the dystrophin deficiency (49, 50). Some of these studies used DNA microarray-based analysis to compare the gene expression profiles between hindlimb/diaphragm and hindlimb/extraocular muscles (45–48, 54, 62, 64) and showed an interesting upregulation of some genes involved in cell adhesion and elasticity, such as Periostin, Elastin, and Cathepsin S, in extraocular muscles compared with hindlimb muscle. In the spared extraocular muscles, this would suggest a possible protective role for some of these genes against dystrophinopathy (46).
However, all the aforementioned studies, except the latest performed by Baban and Davies (8), were unable to survey the complete mouse genome, since they were performed using the Affymetrix MG-U74A gene chips, which contain probes for 12,422 transcripts that covers only 25–35% of the mouse genome (13, 45–48, 62), or they used manufactured chips that included only 1,082 (54, 60) or 7,776 transcripts (64). Therefore, a significant number of genes that could play a crucial role in the pathogenesis or in the amelioration of mdx mouse dystrophinopathy may have gone undetected. To overcome this limitation, in the present study we used the novel Affymetrix MOE 430 2.0 gene chips, which include probe sets covering the entire mouse genome [∼45,000 cDNAs/expressed sequence tags (ESTs)].
Based on our knowledge of the longitudinal evolution of mdx mouse disease (52), we selected three key points in mdx muscle pathology to determine the temporal gene expression variations in both mdx and wild-type mouse strains, seeking to find out which genes were differentially expressed throughout the evolution of mdx mouse disease.
MATERIALS AND METHODS
Breeding pairs of experimental C57BL/10ScSn-mdx, as well as pairs of genetically matched C57BL/10ScSn control mice, were initially obtained from Jackson Laboratories. Colonies of mdx and C57BL/10ScSn mice were kept at a constant temperature with natural night-day cycles and were fed with pellets and water “ad libitum.” All experimental procedures were reviewed and approved by the Hospital Universitari Vall d'Hebron Animal Experimentation Ethical Committee and fulfilled the requirements established by the Spanish Government and the European Community (Real Decreto 223/1988 and BOE 256, 10/25/90).
In this study, to minimize the genetic variability of muscle samples, we used two approaches. First, we analyzed the gene expression profile of a single skeletal muscle, the medial gastrocnemius (MG). We did not base our analysis on a pool of different muscles, as it could have yielded different expression profiles. Second, we performed temporal gene expression profiling by extracting the MG muscles of the same individual from both legs at two different times, to minimize the interindividual genetic variability. The MG muscle represented an excellent candidate for biopsy, due to its easy accessibility by surgery. It was also ideal, since its biopsy is well tolerated by the animals, allowing us to perform another, later biopsy in the other leg to obtain two MG samples from the same individual at two different times. Moreover, MG muscle is composed of ∼20–30% type I (red) fibers and 70–80% type II (white) fibers (7, 68) and, therefore, is more representative of skeletal muscle tissue, in general, than muscle composed exclusively of red or white fibers.
The transcript expression profiles in MG muscles from mdx and wild-type mice were analyzed at 3 wk, 1.5 mo, and 3 mo of life by using the MOE 430 2.0 gene chips from Affymetrix (n = 3 for each condition). The differentially expressed transcripts, which showed differences ≥1.5-fold, were obtained by performing three different comparisons: 1) genes differentially expressed in mdx compared with controls at each point in time (Supplemental File S1);1 2) temporal analysis of the genes differentially expressed in mdx mice among the three points in time, also compared with the variations in control mice (Supplemental File S2); and, 3) temporal analysis of the genes differentially expressed in control mice among the three points in time, also compared with the variations in mdx mice (Supplemental File S3). The first comparison that we performed, comparing the gene expression between mdx and control mice at every point in time, was similar to that performed in previous longitudinal studies (48, 54, 64). However, the other two comparisons were directed to elucidate the genes that varied throughout the points in time analyzed in every mouse strain. Therefore, we obtained, on the one hand, the genes that varied in mdx mice but not in wild type, and, on the other hand, the genes that varied in control animals but remain unchanged in mdx mice among the times analyzed. To present the results in a more comprehensive form, all of the genes were classified in seven different categories: cell adhesion & extracellular matrix; proteolysis; muscle structure & regeneration; inflammation & immune response; cell signaling & cell communication; metabolism; and others/unknown. The resulting genes from our study were classified in their functional categories using information from Affymetrix (www.affymetrix.com) and from the Gene Ontology database accessible on the Jackson Laboratory Mouse Genome Informatics website (www.informatics.jax.org).
Muscle biopsy surgical procedure.
Only male littermate animals were studied to avoid intersexual and genetic variations. Five mice from each strain were used for biopsies: two mice at 3 wk and at 3 mo, one at 3 wk and at 1.5 mo, one at 1.5 mo and at 3 mo, and one was only biopsied at 1.5 mo to complete the MG triplicates. All male animals were anesthetized by an intraperitoneal injection of a mixture of 90 mg/kg ketamine and 10 mg/kg xylazine before tissue collection. MG muscles were excised and immediately frozen in liquid nitrogen and stored at −80°C until used. MG biopsies were performed with the mouse in prone position, by the fixation of tail and extremities with adhesive strips to a cork surface to expose the ventral side of the right crus. External surface disinfection with chlorhexidine was applied to the surgical area, and a sterile surgical field was prepared. A vertical midline skin incision at this level was performed for dissection of posterior muscles with microsurgical instruments and under microscopical (M651 Surgical Microscope, Leica Microsystems) view support. The medial head of the gastrocnemius muscle (gastrocnemius caput mediale) was extracted by sectioning the MG tendon before its insertion in the common calcanean tendon (tendo calcaneus communis). We tried to respect the vasculo-nervous elements (saphenus medialis) that run underneath them to avoid bleeding. Skin closure was performed with a running reabsorbable suture (B. Braun surgical, Spain) and postsurgery analgesia (1.5 mg/kg meloxicam) was provided in all cases. Functional recovery of the extremities was observed in all of the operated animals.
RNA purification and hybridization to arrays.
Frozen MG muscles were finely powdered under liquid nitrogen in a mortar. Briefly, total RNA was extracted using TRIzol (GIBCO-BRL Life Technologies) and purified on RNeasy Micro kit columns (Qiagen). Extracted total RNA was used to synthesize double stranded cDNA using the One Cycle cDNA Synthesis Kit (Affymetrix), which incorporates a T7 RNA polymerase promoter. Biotin-labeled antisense cRNA was obtained starting with 5 μg of total RNAs and the oligo dT primer 5′-GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-(dT)24. cRNas were purified using the columns from the GeneChip Sample Cleanup Module (Affymetrix), and then 20 μg of cRNAs was fragmented at 94°C for 30 min in 40 μl of 40 mM Tris-acetate, pH 8.1, 100 mM KOAc, 30 mM Mg(OAc)2. The fragmented samples were checked using the Bioanalyzer 2100 (Agilent Technologies) to verify their quality and then added to a hybridization cocktail containing Control oligonucleotide B2 (50 pM) and Eukaryotic Hybridization controls (BioB, BioC, BioD, cre) at 1.5, 5, 25, and 100 pM final concentration, respectively, from the GeneChip Eukaryotic Hybridization Control Kit (Affymetrix), herring sperm DNA (0.1 mg/ml), and acetylated BSA (0.5 mg/ml).
The probe array was equilibrated to room temperature and prehybridized with 1× hybridization buffer (100 mM MES, 1 M [Na+], 20 mM EDTA, 0.01% Tween 20) at 45°C for 10 min with rotation. The hybridization cocktail was heated to 99°C for 5 min in a heat block, transferred to 45°C heat block for 5 min, and spun at maximum speed in a microfuge for 5 min. We used 200 μl of the hybridization mixture to fill the MOE 430 2.0 GeneChip Affymetrix cartridge (Affymetrix). After removal of the hybridization buffer and the hybridization of the arrays at 45°C for 16 h with rotation in the Affymetrix GeneChip Hyb Oven 640, GeneChips were washed and marked with streptavidin phycoerythrin in the Fluidics Station 450 (Affymetrix) using the protocol EukGE-WS2-v5 provided by Affymetrix. Once washed and streptavidin phycoerythrin-marked, the GeneChips were scanned in an Agilent G3000 GeneArray Scanner (Agilent Technologies).
Microarray data analysis.
CEL files were imported into the affy package (34) in Bioconductor (http://www.bioconductor.org/) and preprocessed using the robust multiarray analysis algorithm with the default parameters (14). Genes were filtered according to the following criteria: signal ≥log (100), mean absolute fold-change ≥1.5. Genes complying with these criteria were then processed using the package limma (a linear model for microarray analysis by Gordon Smyth, Natalie Thorne, and James Wettenhall at The Walter and Eliza Hall Institute of Medical Research), and false discovery rate was used as the method for multitest correction (Benjamini and Hochberg's step-up method). A complete data set of our microarray analysis has been submitted to the EMBL-EBI ArrayExpress repository. The accession number assigned for our GEO-deposited microarray data is E-MEXP-1623.
Analysis of gene expression by semiquantitative RT-PCR.
The RNA used for the RT-PCR was the same as that used for the microarray analysis for the time points of 3 wk, 1.5 mo, and 3 mo. The RNA samples corresponding to 6 and 9 mo were obtained from freshly biopsied MG muscle, which, immediately after surgery, was frozen and finely powdered under liquid nitrogen in a mortar. Total RNA from all the samples was extracted using TRIzol (GIBCO BRL Life Technologies) and purified on RNeasy Micro kit columns (Qiagen). Reverse-transcriptase (RT) reaction was performed using up to 5 μg of total RNA by using the Superscript II RT kit (Invitrogen, Carlsbad, CA) following the manufacturer's instructions and adding, to each RT reaction, 100 ng of Random primers (Promega, Southampton, UK), and 20 units of SUPERase-in RNase inhibitor (Ambion, Huntingdon, UK) to prevent RNA degradation. The gene expression levels of Sarcolipin, Periostin, Cathepsin S, Thbs1, Ecm1, and Adamts-5, were analyzed by semiquantitative PCR, and each data point was normalized by the mRNA abundance of the ribosomal protein L19. As we have described in previous studies, L19 gene expression shows no significant changes throughout the first 9 mo of life in mdx mice (53). All the oligonucleotide primer pairs used for each of the genes in this study were designed according to the sequences derived from GenBank (accession numbers given in parenthesis) and corresponded to the following nucleotides: Sarcolipin (NM_025540), 483 bp PCR product: Sln-Ups: 5′-TGAAGACAAGCCTTGGTGTG-3′, Sln-Down: 5′-ACAGTGAGTTGGGGGAGGTA-3′; Periostin (NM_015784), 291 bp PCR product: Post-Ups:5′-CCGATGTCTCGAAGCTGAGAG-3′, Post-Down:5′-CATGGATGACTCGAGCAC-3′; Cathepsin S (NM_021281), 348 bp PCR product: Ctss-Ups: 5′-CAACTGCAGAGAGACCCTAC-3′, Ctss-Down: 5′-GTATTTCACCTCAGTGACGCAG-3′; Thbs1 (NM_011580), 330 bp PCR product: Thbs1-Ups: 5′-CCTGTTGCATATGTGTGGAAGC-3′, Thbs1-Down: 5′-GAGACCACACTGAAGATCTG-3′; Ecm1 (NM_007899), 509 bp PCR product: Ecm1-Ups: 5′-CAGACCAGCGAGAGATGAC-3′, Ecm1-Down: 5′-CTTCAGATTGTCTGGAGAAG-3′; Adamts-5 (NM_011782), 404 bp PCR product: Adts5-Ups: 5′-GGCATCATTCATGTGACACC-3′, Adts5-Down: 5′-CGAGTACTCAGGCCCAAATG-3′; ribosomal protein L19 (M62952), 287 bp PCR product: L19-Ups: 5′-CATTCCCGGGCTCGTTGC-3′, L19-Down: 5′-TTCTTGCGGGCCTTGTCTGC-3′. Complementary DNA (15 ng) from the RT reaction was used as a template for each PCR in a 20-μl reaction volume including 0.4 μM of each primer, 0.2 mM dNTPs (Invitrogen), 1.5 mM MgCl2, Taq buffer, and 1 unit of EcoTaq DNA polymerase (Ecogen, Barcelona, Spain). The thermal cycling conditions included an initial activation step at 94° C for 3 min, followed by denaturation at 94°C for 1 min, annealing for 1 min, and extension at 72°C for 1 min, including a final extension step at 72°C for 10 min. The number of cycles and annealing temperature for semiquantitative PCR were optimized for each gene and were the following: Periostin (32 cycles/59°C); Sarcolipin (32 cycles/60°C); Cathepsin S (32 cycles/60°C); Thbs1 (32 cycles/59°C); Ecm1 (35 cycles/58°C); Adamts-5 (32 cycles/58°C). Five microliters of each PCR product were resolved on a 1.5% agarose gel, and the different bands were quantified using a Bio-Rad Gel Doc 2000 Documentation System (Bio-Rad, Hercules, CA).
Functional networks analysis.
The top 10 candidates and two genes associated with the disease (Dmd and Utrn) were included in the query gene list of the web-interface MouseNet v.1 (www.functionalnet.org) (35). This software reports all connected MouseNet neighbors to your query genes in a high-confidence functional network [score >0.5, ∼1.7 million linkage, ∼15,500 genes (∼72% of the total mouse genome)]. In our analysis we considered the top 200 predictions, which were exported to Cytoscape 2.6.0 for graphical representation of the functional network (56).
The information obtained from our study is quite extensive and, due to space constraints, cannot be presented in great detail. Additional complete and detailed files of the different comparisons performed in our study (mdx mice vs. age-matched controls at each time point and mdx or wild-type mouse temporal gene expression variations among the selected time-points) are available online (Supplemental Files S1, S2, and S3, respectively). The most relevant results are presented following the main lines of our experimental strategies described in materials and methods and are summarized in Tables 1–3. It is worth noting that genes showing differential expression in the temporal analysis, but which varied in both mouse strains at the same time, were not included in Tables 2 and 3 due to the fact that their variations could have been attributed to the natural life evolution of this mouse strain. To present the main results in a more comprehensive format and allow for comparisons, we have followed the approach of previous authors (44, 48, 54, 60, 64). All of the analyzed genes have been grouped in seven categories: cell adhesion & extracellular matrix, proteolysis, muscle structure & regeneration, inflammation & immune response, cell signaling & cell communication, metabolism, and others/unknown.
Differential expression between mdx mice and age-matched controls.
The comparison between mdx and age-matched control mice revealed 206 genes/ESTs differentially expressed in at least one of the analyzed time-points (Supplemental File S1). The most representative genes for each gene category are summarized in Table 1. Most of these transcripts showed statistically significant variations at only one point in time, although 43% of the total genes showed variations at more than one time-point, especially at 1.5 and 3 mo of age.
Mdx muscle showed a marked induction of genes belonging to the inflammation & immune response, cell adhesion & extracellular matrix, proteolysis and muscle structure & regeneration families (96, 75, 83, and 80% of upregulated genes, respectively). However, the genes belonging to the cell signaling & cell communication, metabolism, and others/unknown families showed an opposite tendency (63, 64, and 73% of downregulated genes, respectively).
As displayed in Table 1, the inflammation-related Lgals3 gene showed the strongest upregulation (23.01-fold at 3 mo), followed by the striking upregulation, especially at 3 mo of life, of Sarcolipin, Cathepsin S, and Periostin (21.55-, 12.68-, and 11.12-fold, respectively). A large number of genes related to inflammation/immune response, such as several antigens (Cd163, Cd52, Cd53, Ly86, Maged2, and some H2-types), chemokines (Ccl6, Ccl9, Cxcl14), cytokines (Ii, Ifi27), and cell receptors were found upregulated in mdx muscle samples, with the exception of the Pla2g7 gene, which was the only inflammation-related gene found downregulated. Sarcolipin, Cathepsin S, and Periostin were the genes related to cell adhesion, muscle structure/regeneration, and extracellular matrix remodeling that showed the strongest variations. Sarcolipin, which reached similar upregulation values to that of Lgals3 at 3 mo, showed a time-dependent upregulation in its expression levels (8.08-, 13.88-, and 21.55-fold, at 3 wk and 1.5 and 3 mo, respectively). Cathepsin S, like Sarcolipin, was found upregulated at the three time points analyzed, showing 9.37-, 11.23-, and 12.68-fold higher than control mice at 3 wk, 1.5 mo, and 3 mo, respectively. Periostin (also known as osteoblast-specific factor-2, Osf-2) showed no statistically significant differences at 3 wk, but its expression levels rose up to 4.11- and 11.12-fold at 1.5 and 3 mo, respectively. Furthermore, our data showed a clear upregulation (ranging from two- to ninefold) of many other genes such as the well-known myogenic markers Myogenin, β6-tubulin, and Igf-2, and also Elastin, Hs6st2, Lox, Mfap2, Col14a1, Prg4, Hp, Mest, Pi16, Acta2, Casq2, Ckap4, Coro1a, Fath, and Myom3.
Differential expression among selected time points in mdx mice.
The analysis of the temporal gene expression throughout the mdx mouse lifespan, revealed a total of 121 differentially expressed genes/ESTs (≥1.5 fold) in at least one of the three comparisons performed among the time points investigated (Supplemental File S2). Only the genes whose expression varied in mdx but not in wild-type animals are presented in Table 2. We detected a total of 71 genes/ESTs that varied throughout the mdx mouse lifespan, 43 up- and 28 downregulated. Approximately one-third of these genes/ESTs (35.2%) were differentially expressed in more than one comparison, especially between 3 wk and 1.5 mo and between 3 wk and 3 mo. This indicates that their expression levels were promptly increased from 3 wk to 1.5 mo and remained elevated until 3 mo of the mdx lifespan.
The strongest variations throughout the mdx mouse life evolution were detected from 3 wk to 3 mo (Table 2), in particular the downregulation of the Ramp2 gene (−17.47-fold) followed by C3, Aqp4, Pfkfb3, Cxcl14, Mcf2l, and Csrp3 (6.80-, −5.64-, 4.61-, 4.22-, −4.17-, and −4.08-fold, respectively). We focused our attention on the temporal changes of the genes related to cell adhesion, muscle structure/regeneration, and extracellular matrix remodeling, which were found markedly upregulated in mdx mice compared with age-matched controls (Supplemental File S1). Only seven genes belonging to cell adhesion & extracellular matrix, proteolysis, and muscle structure & regeneration were found to be differentially expressed during the mdx lifespan, namely Ecm1, Spon1, Thbs1, Adamts-5, Csrp3, Myo10, and Pde4b. Both Ecm1 and Spon1 were rapidly upregulated from 3 wk to 1.5 mo (2.32- and 3.85-fold, respectively), whereas Thbs1 only showed statistically significant changes (>2-fold) by decreasing from 3 wk to 3 mo. Adamts-5 gene expression levels increased 2.76-fold from 3 wk to 1.5 mo remaining unaltered up to 3 mo of life. On the other hand, both Csrp3 and Pde4b decreased their expression levels from 3 wk to 3 mo (−4.08- and −2.52-fold, respectively), whereas the Myo10 gene almost doubled its mRNA expression.
Differential expression among selected time points in control mice.
The analysis of the temporal gene expression during the lifespan of the control mice revealed a total of 255 differentially expressed genes/ESTs (≥1.5 fold) in at least one of the three possible comparisons performed among the time points investigated (Supplemental File S3). Among them, only 211 genes/ESTs varied in wild-type animals but not in the mdx (Table 3). Like the mdx mice, the wild-type animals showed differentially expressed genes/ESTs in approximately one-third (31.7%) in more than one comparison, particularly between 3 wk and 1.5 mo and between 3 wk and 3 mo. However, in contrast to that observed in mdx mice, the majority (56% of the total) of the genes that varied, were downregulated exclusively during the wild-type mouse lifespan.
The strongest changes observed throughout the wild-type mouse lifespan were the downregulation, from 3 wk to 3 mo, of Igf-2, Tceal7, Periostin, Peg3, and Mybph (−8.96, −8.06, −7.30, −7.21, and −7.06-fold, respectively) and the upregulation of the Zmynd17 gene (8.56-fold), as shown in Table 3. Numerous genes that showed a decrease in their mRNA expression throughout the natural evolution of wild-type mice (Supplemental File S3) were also found upregulated in mdx mice compared with age-matched controls (Supplemental File S1). This trend was especially observable for the genes belonging to the muscle structure & regeneration and cell adhesion & extracellular matrix families, with 60% (9 of 15) and 46% (6 of 13) of these genes following this tendency. Many genes involved in cell adhesion, extracellular matrix development, and muscle structure/regeneration showed a decrease in their expression levels throughout the natural life span of the wild-type animals, but they remained unchanged at higher levels of expression in mdx mice, such as Hs6st2, Lox, Mfap2, Periostin, Col14a1, Acta2, Casq2, Ckap4, Fath, Igf2, Myogenin, Myom3, Sarcolipin, and Tubb6, or Elastin, which experienced an attenuated decrease.
Corroboration of microarray data by semiquantitative RT-PCR analysis at 3 wk and 1.5, 3, 6, and 9 mo.
To corroborate microarray data, we analyzed several genes involved in cell adhesion, extracellular matrix development, and muscle structure. The expression levels of Sarcolipin, Periostin, Cathepsin S, Thrombospondin-1, Ecm1, and Adamts-5 were analyzed by semiquantitative RT-PCR at 3 wk and 1.5, 3, 6, and 9 mo of life (Figs. 1 and 2). Our findings confirmed the results obtained by microarray analysis and demonstrated that these changes persisted, at least up to 9 mo of life.
Sarcolipin mRNA expression (Fig. 1, A and B) was strongly upregulated (8.15-fold) among mdx and control muscle at 3 wk but showed a decrease (almost twofold) in its expression in mdx mice at 1.5 mo. This was followed by a progressive increase up to its maximum levels at 9 mo of life (10.98-fold higher in mdx compared with age-matched control muscle). Periostin mRNA levels were found upregulated more than twofold in mdx mice compared with control mice throughout the entire period analyzed (Fig. 1, C and D). Periostin expression increased from 3 wk to 1.5 mo and then decreased progressively throughout the mdx lifespan, whereas a continuous decrease was observed in its mRNA expression in wild-type mice. The strongest differences between mdx and control mice were detected at 1.5 and 3 mo of life (2.98- and 2.91-fold, respectively) and remained elevated up to 9 mo of life (2.05-fold). Cathepsin S, which showed 5.11-fold higher expression levels in mdx than in control animals at 3 wk (Fig. 1, E and F), increased its mRNA expression levels at 1.5 mo to decrease progressively up to 9 mo, where it showed the lowest differences between mdx and control muscle (3.04-fold).
As shown in Fig. 2, Thbs1 expression levels declined in mdx and control mice (Fig. 2, A and B) throughout their lifespans, although this decrease was more pronounced in dystrophic animals (a condition that reduced its mRNA expression almost twofold from 3 wk to 6 mo). Ecm1, which showed no differences between mdx and wild-type mice at 3 wk (Fig. 2, C and D), promptly increased its expression in mdx mice (which almost doubled its mRNA levels from 3 wk to 3 mo), followed by a slight decline in its mRNA levels up to 9 mo of life. Adamts-5 mRNA expression showed no differences between mdx and wild-type mice at 3 wk (Fig. 2, E and F) but increased 1.91-fold from 3 wk to 1.5 mo in mdx mice, corroborating the upregulation detected by microarrays. The upregulation of Adamt-5 was maintained up to 9 mo of life in mdx mouse muscle, whereas no significant changes were observed during the wild-type mouse lifespan.
Functional network association analysis for the induced genes in mdx mouse skeletal muscle.
With a view to exploring the functional connections between the most relevant genes detected in this study, we used the web-interface MouseNet v.1, the first version of a probabilistic functional gene network of Mus musculus genes (35). In our functional genomics approach, we selected 10 genes belonging to cell adhesion & extracellular matrix, proteolysis, and muscle structure & regeneration families, named Sln, Ctss, and Postn, which showed the strongest upregulation between mdx and control mice at every point in time, and Ecm1, Spon1, Thbs1, Csrp3, Myo10, Pde4b, and Adamts-5, whose expression varied exclusively throughout mdx mouse lifespan, not in control animals, in the analysis performed between the different points in time. These 10 genes were analyzed in the web-interface MouseNet v.1 with high-confidence setting. Sln and Ecm1 were valid mouse gene names but are not present in Mousenet; Ctss and Pde4b are disconnected seed genes in MouseNet. Therefore, the functional network analysis was performed with only six seed genes. In the resulting network we found a strong functional relation among Spon1, Thbs1, and Adamts5, which made sense since these three genes encode for proteins containing thrombospondin type 1 repeats. Interestingly, if we consider the top 200 functional predictions for these six seed genes we observe that Spon1, Thbs1, and Adamts5 are component of a bigger functional network containing Postn and Csrp3. We decided to expand further our functional network model by including Dystrophin (Dmd) and Utrophin (Utrn) in our analysis. Interestingly, these two genes and Myo10 presented a high level of functional association. Thus, the functional map resulting from considering the top 200 functional predictions for these eight (6+2) seed genes shows two main subnetworks having Dmd-Utrn-Myo10 and Adamts5-Thbs1-Spon1-Postn as principal nodes (Supplemental File S4). Importantly, these two subnetworks are functionally linked by the seed gene Csrp3 and 10 other genes, namely Actn4, Abca1, Lama1, Lama3, Crebbp, Sparc, Mical2, Mical3, Myf6, and Pxn (Fig. 3). Following a “guild”-by-association strategy, these 10 genes may be considered as a guild that confers low skeletal muscle fiber necrosis to mdx mouse.
In this study, we directed our research to sort out genes differentially expressed during the evolution of mdx mouse disease that could explain the milder phenotype of mdx mice with respect to human DMD disease. Although a similar approach has been employed previously by other authors (47, 48, 54, 64), we introduced several changes that add extra value to the results obtained in our study. 1) We addressed our analysis to explore the gene expression profiles throughout three crucial time points of the disease evolution of mdx mice: a) 3 wk of age, when muscle necrosis begins, b) 1.5 mo, when maximal necrosis is reached, and c) 3 mo, when muscle necrosis begins to decrease. 2) We analyzed the complete mouse genome expression profile in mdx and control mouse skeletal muscle samples by using the MOE 430 2.0 Gene chips from Affymetrix (∼45,000 genes/ESTs). 3) We analyzed the gene expression profile of a single muscle, the MG, and, to minimize genetic background differences, we analyzed two different muscle samples obtained from the same animal at two different time points. Finally, 4) we performed a functional network association analysis for the most relevant genes detected in this study, by using the new web-interface MouseNet v.1, the first version of a probabilistic functional gene network of M. musculus genes (35).
The results obtained comparing mdx and age-matched control mice at each established time point (Table 1 and Supplemental File S1) were similar to those reported in previous microarray-based studies (13, 44, 46–48, 54, 60, 62, 64). Our findings corroborated the strong inflammatory response observed throughout the mdx mouse lifespan and, in particular, that of the inflammation-related Lgals3 gene (44). We also found a clear-cut upregulation in >75% of the genes belonging to the muscle structure/regeneration, cell adhesion, and extracellular matrix remodeling families. Well-known myogenic markers, such as Myogenin, β6-tubulin, and Igf-2, and also Elastin, Hs6st2, Lox, Mfap2, Col14a1, Prg4, Hp, Mest, Pi16, Acta2, Casq2, Ckap4, Coro1a, Fath, and Myom3, showed between two- and ninefold upregulation in dystrophic animals. Following Lgals3, the strongest variations were detected for the Sarcolipin, Periostin, and Cathepsin S genes, whose upregulation in mdx skeletal muscle has been previously reported in earlier studies (44, 54, 60, 62). Sarcolipin was found to be upregulated in mdx mice from 3 wk to 9 mo of life (Fig. 1, A and B). This protein has been recently described as an inhibitor of the sarcoplasmic reticulum Ca2+ ATPase pump activity (9), which provokes a decrease in calcium transport and muscle contraction both in heart (9) and skeletal muscle (63). Sarcolipin upregulation in mdx mice could act as a response mechanism to diminish or attenuate muscle contraction peak force amplitudes and rates of contraction and relaxation. In mdx mice, Periostin mRNA expression first increased at 1.5 mo (being 2.98-fold higher than in controls) to decrease progressively throughout the time points analyzed (Fig. 1, C and D). However, the expression of Periostin showed a continuous and more pronounced decrease during the lifespan of wild-type mice than that of mdx mice. The bone-related protein Periostin interacts with integrins, mediating cell adhesion and matrix remodeling (17, 25) and also increasing endothelial cell survival and angiogenesis in certain types of cancer (12). The higher Periostin mRNA levels in mdx mouse skeletal muscle could be due to the strong inflammatory response and the associated release of several cytokines, such as transforming growth factor (TGF)-β, which has been described as an inductor of Periostin expression both at mRNA and protein level (21, 32). Recent data obtained in our laboratory (M. Marotta, unpublished observations) demonstrate that TGF-β1 increases Periostin mRNA expression levels both in healthy (2 mo old) and DMD (5 yr old) human primary cultured muscle cells and that this increase is more pronounced in DMD than in healthy TGF-β-treated myoblasts (4.92- and 2.39-fold, respectively). Cathepsin S, like Sarcolipin, was upregulated in mdx mice at all time points analyzed both by microarray and RT-PCR analysis (Fig. 1, E and F). Cathepsin S, a protease that plays a key role in extracellular matrix remodeling (59, 66), has also a potent elastolytic activity (18). The fact that Cathepsin S induction was accompanied by the upregulation (around twofold) of the three components of the elastic system, Elastin, Lox, and Mfap2 (Supplemental File S1), suggests that it plays a possible role in the modulation of Elastin production and its deposition in mdx skeletal muscle. The upregulation of these genes in dystrophic mice indicates a potential response mechanism, which could reinforce the link between the extracellular matrix and the cytoskeleton, modulating muscle contraction rate or increasing fiber elasticity, and thus, protecting mdx skeletal muscle fibers against contraction-induced damage. This protective mechanism has already been proposed by Porter et al. (46), who established that the natural sparing of mdx extraocular muscles and the compensation/regeneration of pathological mdx hindlimb muscles are due to constitutive properties, which provide protection against contraction-induced injury, rather than molecular adaptations.
It is worth noting that the data obtained from the temporal gene expression analysis among the different time points, both in mdx and wild-type mice (Tables 2 and 3 and Supplemental Files S2 and S3, respectively), demonstrate that the higher expression detected for many genes in the mdx mice compared with age-matched controls (Table and Supplemental File S1) is due to their sustained expression throughout mdx mouse lifespan. This tendency was especially observable for the genes belonging to the cell adhesion & extracellular matrix, proteolysis, and muscle structure & regeneration categories, such as Sarcolipin, Periostin, Myogenin, β6-tubulin, Igf-2, Elastin, Hs6st2, Lox, Mfap2, Col14a1, Mest, Pi16, Acta2, Casq2, Ckap4, Fath, and Myom3. This sustained gene expression in mdx muscle, probably caused by the necrosis and regeneration processes that are continuously activated during mdx lifespan, is in agreement with the data observed by Yan and colleagues (67) in a microarray-based gene expression study of the skeletal muscle regeneration process. These authors demonstrated an increased expression of many genes related to muscle structure/regeneration, cell adhesion, and extracellular matrix remodeling, such as Myogenin, Igf-2, collagens, Matrix metalloproteinases, Troponin, and Cathepsins.
The analysis performed among the different time points, both in the mdx and control mice (Tables 2 and 3 and Supplemental Files S2 and S3), demonstrated significant gene expression variations exclusively during the mdx mouse lifespan, but not in controls, in only seven genes related to cell adhesion, muscle structure/regeneration, and extracellular matrix remodeling, namely Ecm1, Spon1, Thbs1, Csrp3, Myo10, Pde4b, and Adamts-5. Remarkably, three of these genes, Spon1, Thbs1, and Adamts-5, encode for proteins that contain thrombospondin type 1 repeats, which have been reported to bind to several ECM components such as laminin, fibronectin, heparin, sulfatide, heparan sulfate proteoglycan, and CD36 (15, 22, 38, 51). Thbs1 expression showed a pronounced decline throughout the mdx mouse lifetime (Fig. 2, A and B). Thrombospondin-1 is a multifunctional protein involved in cell adhesion and spreading that is expressed during the inflammatory response, wound healing, and angiogenesis (3). It could act as a regulator of the organization of the cortical actin cytoskeleton by modulating F-actin polymerization and organization (1, 2). This effect could be relevant, since F-actin plays an important role in the linkage of the cytoskeleton to cell membrane by interacting with either dystrophin (55) or integrin-associated proteins, such as Talin or Vinculin (33, 40). Interestingly, a similar pattern of expression at the protein level has been recently described by Cohn and collaborators (21) for thrombospondin-1 and periostin, two of the genes that showed a strong decrease in the mRNA levels during mdx lifespan. These authors demonstrated increased protein levels of periostin and thrombospondin-1 in 2-mo-old mdx muscle, and a remarkable decrease of their levels after the administration of a TGF-β-neutralizing antibody or the angiotensin II type 1 receptor blocker losartan. The inhibition of TGF-β1 pathway normalized muscle architecture and function in mdx mice and decreased periostin and thrombospondin-1 protein levels, which coincided with the improvement of mdx disease. These data agree with our observations in mdx mouse skeletal muscle from 1.5 mo onward, given that the decrease of periostin and thrombospondin-1 mRNA levels coincided with the amelioration of mdx mouse muscular dystrophy. On the other hand, Spondin-1 (or F-spondin), seems to play an important role in neuronal development by promoting neural cell adhesion and neurite extension (36). Although its function in skeletal muscle is unknown, its possible role in cell adhesion is supported by the fact that M-spondin, a protein with a high homology with F-spondin, is localized at the muscle attachment sites in the Drosophila embryo (65). On the other hand, both genes Csrp3 and Pde4b, which are downregulated throughout the first 3 mo of the mdx mouse lifespan, have been related to muscle regeneration. The Csrp3 gene, which encodes for the muscle LIM protein (6), acts as a positive regulator of myogenesis and myogenic differentiation (5) and in the maintenance of normal muscle function in response to injury (11). Csrp3 is also essential for the structural organization of the cardiac sarcomere and the sensor for cardiomyocytes stretching (16), indicating that its downregulation in mdx skeletal muscle could be related to muscle contractility adaptations in response to contraction-induced injury. Moreover, Pde4b downregulation seems to have a beneficial effect in mdx muscle, since the inhibition of phosphodiesterase 4, the major cAMP-modifying phosphodiesterase found in skeletal muscle, has recently been demonstrated to increase muscle mass and force in atrophic skeletal muscle in rodents (31). The mRNA expression of Ecm1 was raised almost twofold in mdx mice (Fig. 2, B and C) and was maintained practically unchanged up to 9 mo of life. Ecm1 increase in mdx skeletal muscle could play a role in the modulation of extracellular matrix remodeling or cell adhesion properties, since Ecm1 mutations are characteristic of the rare, inherited disease lipoid proteinosis, that causes excessive basement membrane deposition in skin and mucosa (28). The Myo10 gene encodes for the Myo10 protein, a myosin that belongs to the group of MyTH-FERM myosins, which can act as a link to integrins and microtubules (58). At the end of their tail, these myosins contain a FERM domain that shares ∼28% of its protein sequence identity with the FERM domain of Talin (58), a cytoskeletal protein involved in the anchorage of integrins to actin filaments (20). Myo10 participates in the relocalization of integrins to form adhesive structures and to promote filopodial extension, and could act as a link between integrins and the cytoskeleton (69), which suggests that Myo10 upregulation could respond to a mechanism of mdx skeletal muscle to reinforce the interactions between the myofiber cytoskeleton and the plasma membrane.
The increase of Adamts-5 expression during the mdx lifespan, detected by microarrays and corroborated by RT-PCR analysis (Fig. 2, E and F), is remarkable. Adamts (a disintegrin-like and metalloprotease domain with thrombospondin type 1 motifs) family, a complex secreted enzymes involved in cell binding, communication and migration (reviewed in Ref. 4), have a high similarity to Adam proteins, preserving their metalloprotease and disintegrin-like activity. This is relevant since Kronqvist and colleagues (37) have recently demonstrated that Adam12 protein alleviates the muscle pathology when overexpressed in mdx skeletal muscle. Adam12 could prevent muscle cell necrosis in mdx mice by increasing the expression and redistribution of α7-integrin, utrophin, and associated glycoproteins (42). All these data support the possible role of Adamts-5 in the amelioration of mdx dystrophinopathy.
Finally, we analyzed the possible association in functional networks of 10 relevant genes detected in our study (Ecm1, Spon1, Thbs1, Csrp3, Myo10, Pde4b, Adamts-5, Sln, Postn, and Ctss) also including the DMD-related Dystrophin (Dmd) and Utrophin (Utrn) genes. The complexity of functional relationships among genes forces their association to networks instead of lineal pathways. In the last few years, a vast amount of mouse functional genomic data has been collected, and bioinformatics teams are assembling this information to generate predictions of functions (43). Recently, Marcotte's laboratory has integrated ∼20 million experimental observations (DNA microarrays, protein-protein interactions, genetic interactions, literature, and comparative genomics methods) among ∼95% of the total mouse genes to build MouseNet v.1, the first version of a probabilistic functional gene network of M. musculus genes (35). To explore the functional relationship among the genes pointed out in our experiments, we ran 10+2 genes, such as Ecm1, Spon1, Thbs1, Csrp3, Myo10, Pde4b, Adamts-5, Sln, Postn, and Ctss, and also Dmd and Utrn, in the web-interface MouseNet v.1. Although our analysis showed four disconnected genes, Sln and Ecm1 (valid mouse gene names but not present in Mousenet) and Ctss and Pde4b, we found a strong functional relation for the six genes (Spon1, Thbs1, Csrp3, Myo10, Adamts5, and Postn) not only among them, but also with Dmd and Utrn. Interestingly, we obtained a functional map showing two main subnetworks having Dmd-Utrn-Myo10 and Adamts5-Thbs1-Spon1-Postn as principal nodes (Fig. 3), and a large number of genes functionally linked to these networks (the top 200 functional predictions are summarized in the Supplemental File S4). It is worth noting that we detected Actn4, Abca1, Lama1, Lama3, Crebbp, Sparc, Mical2, Mical3, Myf6, and Pxn genes functionally linked to these two subnetworks. Moreover, seed gene Csrp3 showed functional links either to Dmd-Utrn-Myo10 or to Adamts5-Thbs1-Spon1-Postn. In summary, we present here a high-confidence functional network containing two main nodes and several functional links that may orientate future experimental assays to identify those genes functionally relevant to confer advantageous features to the mdx mouse for muscular dystrophy and ideally to guide putative therapies.
Taken as a whole our results indicate that 1) there is a marked induction of genes belonging to the cell adhesion & extracellular matrix, proteolysis, and muscle structure & regeneration categories between mdx and age-matched controls; 2) mdx mouse skeletal muscle maintains the expression of many genes unchanged, which are normally downregulated after birth and throughout the lifespan of wild-type mice; 3) the nature of the genes that are up- or downregulated in hindlimb mdx mouse muscle probably reveal a response mechanism addressed to reinforce the link between the extracellular matrix and the cytoskeleton, modulating muscle contraction rate or increasing fiber elasticity in an attempt to protect the mdx skeletal muscle fibers against contraction-induced damage; and 4) there is a strong functional association between some of the genes detected in this work, which resulted in two main subnetworks having Dmd-Utrn-Myo10 and Adamts5-Thbs1-Spon1-Postn as principal nodes. Although further experiments at the protein level are required to demonstrate their direct implication in the improvement of mdx dystrophinopathy, some of these genes could be considered as potential therapeutic targets to ameliorate the evolution of Duchenne and Becker muscular dystrophies.
C. Ruiz-Roig was the recipient of a fellowship from the Generalitat de Catalunya (Barcelona, Spain). This study was supported by the Fondo de Investigaciones Sanitarias (FIS 2004/030106 and FIS 2006/061164) of the Ministerio de Sanidad y Consumo (Madrid, Spain).
For valued technical assistance we thank Marta Rosal, Alex Rojo, and Marielle Esteves from the Animal Facility Unit and Ricardo Gonzalo and Francisca Gallego from the Gene Expression Unit, all of them belonging to the Unitat Científico-Técnica de Suport (UCTS), Institut de Recerca, Hospital Universitari Vall d'Hebron (Barcelona, Spain).
↵1 The online version of this article contains supplemental material.
Address for reprint requests and other correspondence: M. Marotta, Laboratori de Neurologia Infantil, Institut de Recerca Hospital Universitari Vall d'Hebron, Passeig Vall d'Hebron 119-129, 08035 Barcelona, Spain (e-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2009 the American Physiological Society