All the horses were genotyped for the GYS1 c.926G>A mutation 21 (wild type GG, heterozygous GA or homozygous AA).

A biopsy of the gluteus medius muscle was performed in 13 Norman adult horses: 4 females and 9 males, 4–17 years old. This muscle was chosen because of (i) its important propulsion function in horse locomotion, (ii) its early and major histological involvement in the course of the disease 1 and (iii) the good standardization of biopsy sampling 22. The horses were sedated and received a local anaesthesia before making a small skin incision according to a standardized protocol used in routine for muscle disease diagnostic. The biopsy of the dorsal compartment of the gluteus medius was performed at the first third distance between the sacral and coxae tuber. A biopsy needle with automatic sampling was vertically inserted until its extremity just under the fascia corresponding to a 3.5 cm depth for the gluteus medius. Each muscle biopsy was divided in two parts, and the first part was frozen in isopentane cooled with liquid nitrogen for histological analysis and the second part was collected for genomic analysis. The blood was removed by absorbing it with a compress and immediately frozen in the RNA later^® at -80°C. Informed consent was obtained from the horse owner, the French National Studs (Les Haras Nationaux), and the study was approved by the animal care committee of the National Research Institute of Agriculture (INRA).

Briefly, the muscle samples were serially sectioned (8 mm thick) and each section was stained with either hematoxylin-eosin-safranine (HES), or periodic acid-Schiff (PAS) with or without incubation for 20 minutes at 37°C with α-amylase 1. Intracytoplasmic presence of amylase-resistant material was used as diagnosis criteria for PSSM: seven were PSSM horses and six were control horses 2. Morphometric analysis was performed additionally for histologic quantitative evaluation using a digital camera (Nikon DXM 1200, Badhoevedo, Netherlands) combined with image-analyzing software (Lucia imaging software, Laboratory Imaging Inc., Prague, Czech Republic). Percentages of aggregate-containing and vacuolated fibers were calculated as described previously 1 from each PAS-amylase-stained and HES-stained sections respectively.

Additional biopsies from one PSSM horse (#49) was designated for routine electron microscopy and subsequently fixed in 2.5% buffered glutaraldehyde, postfixed in osmium tetroxide, embedded in epoxy resin, cut into 1-mm-thick sections, stained with uranyl acetate and lead citrate, and examined by transmission electron microscopy.

Gene expression analysis was performed using a homemade equine long oligonucleotide microarray which included 384 equine transcripts: 50 probes of the mitochondrial genome and 334 probes of the nuclear genome [see Additional file 1 and GEO Platform accession number GPL8349: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GPL8349]. This equine oligonucleotide microarray was designed according to the derived method used for open access human and mouse long oligonucleotide microarrays 23. Briefly, the list of human annotated genes (available in the NCBI public data base) involved in muscle structure, contraction, energetic metabolism and inflammation processes was used to identify the equine candidate genes that could be interesting to put on the microarray. Then, this list of candidates was crossed with the orthologous equine sequences available in 2006 on the NCBI data base using a blast method 24. In addition, 50 other probes of the equine mitochondrial genome were designed to measure the expression of the 13 genes (2 different probes per protein gene) coding for the protein units of the respiratory chain as well as 22 tRNA and 2 rRNA. Finally, 384 probes of 50 nucleotides long were calculated and synthesized with a 5' terminal – NH2 modification in order to attach the probe on the slides. The microarrays were spotted at the microarray service (LEFG, CEA, Evry, France) by printing the equine probes, suspended in a spotting buffer composed of 50% dimethyl sulfoxide (DMSO) and TE on hydrogel slides (Schott Nexterion), with a Microgrid-II-robot. The equine microarray can be spotted alone on a slide or jointly with the 25 K mouse oligonucleotide microarray 23.

This equine microarray has been validated by a reproducibility test, competition test, biological test for equine transcript analysis extracted from different muscles of different equine breeds and health status [unpublished data]. Fourteen microarrays were used for analysis of each PSSM muscle sample against reference control muscles using two fluorochromes in a dye-swap design (duplicate measurements).

Total RNA was extracted from tissues by a phenol-chloroform method (TRI

ZOL^®

The synthesis of RNA with a polyA tail was performed using in vitro transcription with aminoallyl UTP. A total RNA (1 μg) was used to start the synthesis using the in vitro transcription kit according to the manufacturer's protocol (Aminoallyl MessageAmp^® II aRNA). This method equally increased the number of mRNA copies of each transcript without changing the proportion of the initial population of transcripts. The integrity and concentration of aminoallyl RNA (aaRNA) were checked again by micro channel electrophoresis (RNA 6000 Nano LabChip^®, Bioanalyzer^®) and a spectrophotometer (Nonodrop^®). The aaRNA were stored at -80°C.

All the samples of PSSM muscles were hybridized against the reference control muscles. This reference was made by pooling together all the mRNA extracted after in vitro transcription from the 6 control muscles of the sound horses. Briefly, the hybridization protocol 24 included the following steps. After purification of the amino-modified RNA, monofunctional forms of the Cyanine 3 (Cy3) or Cyanine 5 (Cy5) fluorochrome were coupled to 5 μg of the studied sample and control sample by an indirect method: the ester groups linked to the fluorochromes were bound on the amino-allyl groups incorporated on the RNA during the in vitro transcription. The marked targets were then mixed and hybridized on the same chip for immediate comparison of the two nucleic acid populations. For each sample, a dye-swap (duplicate measurement) was performed to avoid asymmetry of marker affinity (Cy3 and Cy5). The two results were averaged before statistical results.

The 5 μg of aaRNA were completely vacuum-dried and resuspended in 9 μL of coupling buffer and 11 μL of dye, the tubes were mixed by vortexing and incubated for 30 minutes in the dark. The reaction was quenched by addition of 0.73 M Hydroxylamine and incubation for 15 minutes in the dark. Excess dye was removed from the labeled aaRNA by purification on a column. Purified labeled aaRNA were subsequently mixed with 10 μg poly (A) RNA, 10 μg yeast tRNA, 10 μg Cot 1 mouse RNA, then precipitated with 0.5 volumes of 7.5 M ammonium acetate and 2.5 volumes of absolute ethanol. After centrifugation at 16,000 × g for 45 minutes to 1 hour at 4°C, the supernatant was removed and the pellet was washed once with 700 μL of 70% ethanol. The samples were centrifuged at 16000 g for 15 minutes at 4°C. The aaRNA pellet was vacuum-dried for 1 minute and then dissolved in 50 μL of the hybridization mix preheated at 50°C: 2 × SSC, 0.1% SDS and 0.1% Salmon sperm DNA. It was mixed by vortexing and incubated in a water bath at 50°C to dissolve the mRNA pellet. Each slide was rehydrated in a humidified chamber for 1 hour at room temperature then dried in a dessicator for 1 hour. The slides were prehybridized in 0.3% Ethanolamine and 0.05 M Na Borate pH9. The slides were plunged 1 minute in water and dried by centrifugation for 7 minutes at 100 × g. The probe was heated for 2 minutes at 100°C and placed between the slide and a cover slip. The arrays were incubated overnight in a water bath at 50°C in a humidified slide chamber and washed in a first solution of 0.1 × SSC with 0.1% SDS for 5 minutes followed by two washing steps in 0.1× SSC for 5 minutes each at room temperature, under agitation and safe from the light. The arrays were dried by centrifugation at 100 × g for 7 minutes. The image of fluorescence intensities of Cy3 (532 nm) and Cy5 (635 nm) were scanned separately with the laser scanner (GenePix^® 4000B).

The tiff images (10 μm/pixel) were analyzed using a specific software for microarray image analysis (GenePix^® Pro 6.0 software) which allowed locating the spots automatically with manual control and quantifying the fluorescence intensity of each spot. For each microarray, the data output were the fluorescence intensities of all the spots corresponding to each gene. These data were imported in a genomic data analysis software (Gene Spring^® software) to normalize the data before filtering and statistical analysis. For each sample (horse), a duplicate analysis (dye-swap) was made and the average fluorescence results were used. Then, all the results were normalized per spot and per microarray by the Lowess regression method in order to remove systematic variations. Finally, for each sample and each gene an expression ratio was obtained:

Ratio = Expression of each PSSM muscle/expression in the reference control muscle

For statistical analysis, a log transformation ratio was used to normalize the distribution. A list of the significant genes were obtained by filtering the genes with a cut-off t-test p-value p < 0.05 with Benjamini and Hochberg multiple testing correction method to avoid a false discovery rate 25. The Student t-test detected the mean expression ratio of replicates which are significantly different from one (expression ratio = 1, means no gene expression change).

In order to validate the equine microarray results, quantitative RT-PCR were conducted in triplicate for the top ten up and down-regulated gene transcripts both in control and PSSM horses. The primers were designed using Primer express software (Applied Biosystem) (Table 1). All the reactions were carried out on the real time PCR system 7500 (Applied Biosystems). Briefly, reverse transcription was performed on 20 ng of aaRNA obtained by aminoallyl amplification (RNA used for microarray hybidization and RT-PCR were issued from the same preparation). Random primers (100 ng) and 10 pmol of dATP, dCTP, dGTP and dTTP were added to RNA in a 13 μL final volume. The reaction tubes were incubated at 70°C for 5 min. The random annealing primer was performed by an incubation at room temperature (~25°C) for 10 min. Four microliters of 5× first strand buffer, 200 pmol of DTT and 200 U of Superscript II reverse transcriptase were added in the tube for a 20 μL final volume and incubated at 42°C for 1 hour. The enzyme was inactivated by incubating the mixture at 70°C for 15 min. The cDNA concentration was measured by optical density with a spectrometer that measured absorbance at 260 and 280 nm (Nanodrop^®). The equine cDNA solution was diluted in water to obtain a final concentration of 20 ng/μL and stored at -20°C. Complementary DNA (100 ng) obtained from aaRNA was amplified in a 20 μL PCR reaction with 10 μmol of each specific primer and 10 μL SYBR^® Green PCR Master Mix. Amplification steps were as follows: 10 min denaturation at 95°C, 40 cycles of denaturation at 95°C for 15 s and annealing extension at 60°C for 1 min. The comparative threshold cycle (CT) method was used for the calculation of amplification as specified by the manufacturer. The expression results of the quantitative RT-PCR of the control and PSSM horses were normalized by the mean CT of the reference gene 12S rRNA measured in all the muscle samples and the relative expression ratio of PSSM against the control group was calculated.

A biological interpretation of the data was performed using two types of analysis to identify the main biological functions and the main metabolic pathway disorders revealed by the expression profile of the PSSM muscles. The categorization of the gene functions was generated through the use of Ingenuity^® Pathway Analysis http://www.ingenuity.com. The analysis identified the biological functions that were the most significant to the gene list results. The significant genes associated with biological functions in the knowledge database (gene functions described in published papers) were considered for the analysis. The Fischer exact test was used to calculate a p-value determining the probability that each biological function assigned to that data set is due to chance alone.

The analysis of the main metabolic pathways and their dysfunctions was performed using a Java/Perl software, Predictsearch^® (Prediguard, http://prediguard.com/predictsearch.htm) which has been previously described in 26. This software allows the characterization of the pathways as well as the functional networks in which the selected genes were involved. Briefly, gene aliases corresponding to the significant modulated genes were submitted as queries to PubMed in order to collect titles and abstracts of all related publications. The results are presented in a 2D graph that shows the relationships between each gene and cell type, cellular compartment, and biological function.

The PSSM horse genotypes for the GYS1 c.926G>A mutation were 5 heterozygous (AG) and 2 homozygous (AA) (Table 2). The 6 control horses were wild type (GG). Heterozygosity seemed to be consistent with the dominant mode of inheritance.

Intracytoplasmic presence of amylase-resistant material, used as diagnosis criteria for PSSM, was observed in biopsies of the 7 out of the 13 horses (Figure 1). Two horses were graded as mildly affected (less than 3% of fibers containing PAS positive amylase resistant material) and five horses as severely affected (10 to 20% of affected fibers) (Table 2) using quantitative evaluation of aggregate-containing fibers as previously published 1. The two homozygous horses were histologically severely affected. Heterozygous horses could present the mild (n = 2) or the severe (n = 3) phenotype.

Some other myopathic changes were also identified in affected horses with marked individual variations. Numerous fibers underwent segmental necrosis that was associated with mononuclear cell infiltration foci (Figure 1). Some of these foci were centered on fragmented fibers with many macrophages phagocytizing cellular necrotic debris. Strikingly, fiber necrosis was more prominent in homozygous PSSM horses. In this marked inflammatory context, some endomysial fibrosis was also present. Additionally, in all PSSM muscles, we noted an increased number of central nuclei. This change corresponds to muscular regeneration. Some degree of anisocytosis of muscular fibers was always present in affected muscles and was more severe in the two homozygous horses. Fatty infiltration of muscular fascia and replacement of muscle fibers by adipose tissue was also sometimes observed, this change being particularly severe in one homozygous horse.

Ultrastructural evaluation of one severely affected horse (#49) was performed (Figure 2). Extensive accumulation of an abnormal polysaccharide appearing as dense particles, displaced and partially replaced mitochondria and myofibrils, a number of which was severely decreased. This accumulation was exclusively cytoplasmic, membrane unbound and particularly abundant around nuclei and in the subsarcolemmal region of the cytoplasm, where numerous mitochondria are normally present in sound animals. Strikingly, we also reported some mitochondrial ultrastructural changes, namely a decrease of the cristae number, formation of myelinic bodies and swelling.

Thirty-three percent of the probes of the micorarray were significantly hybridized with a high fluorescence value, especially for mitochondrial genes. All the data of the present study have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO series accession number GSE15572 http://www.ncbi.nlm.nih.gov/sites/entrez?db=gds&cmd=search&term=GSE15572.

After filtering the normalized expression ratios with a cut-off at p < 0.05 (normalized ratio significantly different from 1), a total of 129 genes exhibited significant changes of expression levels between PSSM and control muscles (Table 3). This set contained 16 up-regulated genes (normalized ratio > 1.5), 37 down-regulated genes (normalized ratio < 0.66) and 76 moderately modulated genes (0.66 < normalized ratio < 0.9 and 1.07 < normalized ratio < 1.5). This list of significant genes was used for the rest of the study. The maximal up-regulation value was observed for interleukin 18 (IL18) 6.2 fold change, and the following genes were up-regulated up to 2 fold: CTSS, LUM, CD44, FN1, GST01. The maximum down regulation was observed for the mitochondrial tRNA genes (TRNQ, TRNC, TRNN, TRNY, TRNA) from -4.83 to -2.86 fold changes. The three most down-regulated genes coding for proteins were the following: SLC2A2 or GLUT2 (-2.84), PRKCα (-2.72), VEGFα (-2.68).

The relative expression ratio of the top ten genes up- or down-regulated between PSSM and the control group were compared to the results obtained by microarray analysis (Table 4). A significant correlation of r = 0.54 (p < 0.01) between the two methods was calculated. As in microarray results, the mitochondrial genes ND3 and ND5 were found down-regulated using quantitative RT-PCR analysis. The hypoxia markers were also in agreement: VEGFα was down-regulated and HIF1α was up-regulated. All inflammation markers (IL18, CTSS, LUM, CD44, GSTO1, VCAM1, ITGB1, CXCR4) were up-regulated as in microarray results except FN1 and ITGB1 which were found down-regulated. The gene markers of aerobic pathway SLC25a17 was found up-regulated by quantitative RT-PCR and SCL25a15 down-regulated as in microarray data.

Our list of modulated genes was classified by categories of cellular functions, canonical pathways and signaling pathways using Ingenuity Pathway Analysis^®. Protein synthesis, apoptosis, cellular movement, growth and proliferation are the main significant cellular functions (p < 0.05) that are associated to the modulated genes (Figure 3). The main significant (p < 0.05) canonical pathways concerned oxidative phosphorylation, ubiquinone biosythesis, aminosugar and purine metabolism. The main significant signaling pathways (p < 0.05) were related to G-protein couple receptor, hypoxia, NF-κB, VEGFα and IL6, IL10, apoptosis and leukocyte extravasation signaling. The main metabolic disorders significantly (p < 3.3 10^-4) detected were mitochondrial dysfunction, hypoxia and alteration of NF-κB activity.

The main tissue, organ and cellular context and the biological activities that were cited in relationship with the set of genes were further investigated using Predictsearch^® analysis. Using the PubMed database, it was found that a part of the significant modulated genes were strongly associated with the following terms: skeletal muscle, mitochondria, myoblast, endoplasmic reticulum. (Figure 4). In addition, the more frequent biological activities cited in relationship with the set of genes were glycolysis, respiratory chain, hypoxia, glycogen synthase, inflammatory myopathy (Figure 5). The main metabolic pathway disorders are described in the figures 6, 7 and 8 using these data (see Additional file 2 for the legends of the pathway drawings). In the mitochondria, all the genes were down regulated (Figure 6). The tRNA were severely down regulated (fold change between -4.83 and -2.27). The other mitochondrial genes ND2, ND3, ND5, ND6, COX2 and COX3 were moderately down regulated but with low expression values (fold changes between -1.31 and -1.85). Other nuclear genes involved in the aerobic metabolism of the mitochondria were down regulated as well: UCP2, GNAS, SLC25a15, SLC2a2 (GLUT2), ATP5L, ATP5J, ATP5D and ATP5H.

A chronic inflammation process of the muscle was revealed by many up-regulated genes like interleukin 18 (IL18), CD44, CD69, CTSS, FN1, HSPA5, ITGB1, ITGB2, VCAM, CXCR4 LTF, GSTO1 and LUM (Figure 7).

The up-regulation of the glycogen synthase kinase-3 (GSK3β) under its active form could be responsible for glycogen synthase inhibition, and for hypoxic transcription factor inhibition (HIF1α) by phosphorylation as indicated by data mining analysis (Figure 8). In addition, VEGFα was down-regulated and avoided capillarization and oxygen distribution of the new regenerated muscle fibers. These gene regulations increased muscle hypoxia.

Other interesting genes were slightly up-regulated: ENO1 (1.39) was involved in the end of glycolysis; IGF2 (1.34) was involved in the growth hormone activity in muscle differentiation; EPO (1.28) was involved in hematopoietic function. Some other genes were slightly down-regulated: PYGL (-1.11) was involved in glycogen metabolism; IL6ST (-1.20), IL7 (-1.19) were involved in hematoprotein signaling; MIF (-1.30) was involved in cell inflammatory regulation. GSK3β, PYGL, GLUT2 are the only genes closely related to glycogen metabolism.