I. Experimental Design: a. Type of experiment: Time course b. Experimental factors: 3 month old mice subjected to lengthening contractions of the extensor digitorum longus muscle (EDL). Samples of EDL collected at 6 and 72 h and compared to non-treated control. c. How many hybridizations in exp: 9 d. Common reference used for all hyb: no e. Quality control steps: All arrays used from same lot. Triplicate hybridizations performed for each time point.
II. Samples used, extracts, preparation and labeling: Animals. Nine male C57BL/6 mice (3-4 mo of age, 27.8 ± 3.3 g body mass, Harlan Sprague-Dawley, Indianapolis, IN). Muscle Injury. The extensor digitorum longus (EDL) muscle of six mice were exposed to lengthening contractions. Mice were anesthetized with 2% avertin (0.015 ml/kg) and supplemental doses were administered if the mouse responded to a toe pinch. Animals were placed on a plexiglass platform that was maintained at 37°C. The distal femur of the right hindlimb was fixed between screws and the foot was taped to the platform. The distal tendons of the EDL were exposed by incision and tied to the lever arm of a servomotor (Aurora Scientific, Richmond Hill ON, Canada) with 4-0 silk suture. Needle electrodes were placed adjacent to the peroneal nerve to stimulate dorsiflexor contraction (Grass Instruments, West Warwick, RI). Maximal isometric force (Po) was determined by stimulating the dorsiflexors maximally at optimum length (Lo) for force development. To induce muscle injury, 75 lengthening contractions were performed at 0.25 Hz for 5 min. For each lengthening contraction, the muscle was stimulated at 150 Hz and stretched 100 ms after the initiation of stimulation. Muscle stretch involved 20 % strain relative to Lo. Force deficit was evaluated 10 min after the 75 contractions and then the incision was closed with 7-0 silk suture. Animals were returned to their cage to recover until the time of sacrifice. Prior to sacrifice by cervical dislocation, animals were anesthetized with avertin and the EDL muscles were excised and flash frozen in liquid nitrogen. Muscles were stored in liquid nitrogen until RNA isolation. RNA Isolation. Total RNA was isolated from EDL muscles from control mice (n=3) and mice sacrificed at 6 h (n=3) and 72 h (n=3) after lengthening contractions according to the modified protocol of Chomzynski et al.. Frozen muscles were added to preweighed tubes containing TriReagent (Sigma-Aldrich, St. Louis, MO). Tissue was homogenized in TriReagent using a Tissue Tearor (Fischer, Pittsburgh, PA) at 30,000 rpm for 1 minute. Homogenates were transferred to sterile, RNAase free microcentrifuge tubes and incubated for 10 min at room temperature. Chloroform (0.2 ml; Sigma-Aldrich, St. Louis, MO) was added and after incubation, the samples were centrifuged for 15 min at 12,000 g and 4°C. Total RNA was precipitated from the aqueous phase by the addition of isopropanol (Sigma-Aldrich), and pelleted by centrifugation. Pellets were washed in ethanol (75 and 95 %) and air dried before resuspension in diethylpyrocarbonate (DEPC) treated water. RNA concentration was determined by optical density at 260 nm.
III. Hybridization procedures and parameters: Microarray Hybridization and Analysis. To reduce measurement error, all membranes (Atlas mouse 1.2; Clontech, Palo Alto, CA) were prepared from the same lot. Additionally, each membrane was hybridized only once to avoid errors associated with stripping and multiple hybridizations. In order to minimize individual biological variability, total RNA was pooled from three animals for hybridization of arrays. Three arrays were used for the pooled sample at each time point. First strand cDNA probe generation from total RNA and microarray hybridization were performed according to manufacturer’s instructions (Atlas cDNA Expression Arrays User Manual, Clontech). 33P labeled cDNA was generated from sample RNA using [33P]-dATP (3000 Ci/mmol; ICN, Costa Mesa, CA) and Maloney murine leukemia virus (MMLV) reverse transcriptase (Clontech). Labeled probes were purified using nucleospin columns (Clontech). Array membranes were prehybridized in Express Hyb containing sheared Salmon testes DNA (Gibco BRL, Rockville, MD) for 30 min at 71°C in rotating hybridization tubes. Radiolabeled probe (3 x 106 cpm) was incubated in denaturing solution (Clontech) for 20 min at 71°C, after which Cot1 DNA/neutralizing solution was added and incubation continued for 10 min. The probe mixture was added to the hybridization tube and incubated with rotation overnight. Following hybridization, membranes were washed four times for 30 min in 2X saturated sodium citrate (SSC) and 1 % sodium dodecyl sulphate (SDS) at 71°C, once for 30 min in 0.1X SSC and 0.5% SDS at 71°C, and once for 5 min in 2X SSC at room temperature. Membranes were removed and immediately wrapped in plastic and exposed to a phosphor storage screen (Molecular Dynamics, Sunnyvale, CA) for four days.
IV. Measurement data and specifications: Data Analysis. The membrane image was acquired using a Storm phosporimager (Molecular Dynamics) and Image Quant software. Array images were analyzed using Atlas Image 2.0 (Clontech) software to determine raw intensity levels of expression. Raw intensity data was imported into GeneSpring (Silicon Genetics, Redwood City, CA) expression analysis software. Intensity levels were normalized to the median expressed intensity on each of the arrays and averaged for the three arrays at each time point. Normalized expression levels for control arrays were established as a value of 1 and data from 6 and 72 h were expressed as fold changes relative to control. In order to simplify interpretation of the resulting dataset, a k-means cluster analysis was employed to group genes based on similarities in patterns of expression. To reduce the likelihood of false positives, the data was filtered using a criterion 2 fold change in expression prior to clustering. Miller et al. (61) have calculated that using an array of 10,000 features with a coefficient of variation (CV) of 20 percent, a 2 fold criterion for altered gene expression would result in zero false positives. The mean and median CV (15.8 and 14.3%, respectively) for all arrays in the current study fell within these guidelines for the elimination of false positives.
