ABSTRACT: Type of experiment: Comparison of liver samples between sham (control) and 20% TBSA (total burn surface area) burn rats collected at 1h, 4h, 8h, and 24h. Experimental factors: Burn injury and time; Experiment design: All experimental liver samples collected after 1h, 4h, 8h, and 24h after burn injury were compared to a common reference (sham control treated as 0h time point); Bio source: Sprague-Drawley male rats weighing 150-200 g supplied through Charles river laboratories, MA (www.criver.com); Bio material manipulations: Rats (n=3 for each scald-burn and sham-burn time point) were individually housed in a temperature-controlled (25oC) and light-controlled room (12h light-dark cycle) and allowed to adjust to their new surroundings for at least 5 days prior to the experiment. Water and rat chow were provided ad libitum to the animals. On the day of the treatment, the animals were randomly divided into two groups, burned and sham-burned. The burn injury consisted of a full-skin-thickness scald burn of the dorsum, calculated to be ~ 20% of the ratâs total body surface area (TBSA), induced by immersing the designated area in boiling water for 10 s [Yamaguchi, 1997 #416]. Rats were resuscitated with an intra-peritoneal injection of sterile saline solution (1.5 mL/Kg body weight/% TBSA) immediately after burn. At each time point, three animals belonging to each group was sacrificed, the serum collected, and stored at â80oC. Hybridization extract: Total RNA was isolated from approximately 50 mg of pooled liver tissue using the Nucleospin II RNA isolation kit from Clontech (Palo Alto, CA), as per the manufacturerâs instructions. Labeling protocol: Labeling was performed as instructed by Affymetrix, Inc (www.affymetrix.com). In brief, 10 μg of total RNA was used for the double-stranded cDNA synthesis using the T7-oligo (dT) promoter primer kit (Affymetrix Inc.) and superscript RT II enzyme (Invitrogen). 1 μg of cDNA was labeled through invitro transcription reaction using T7 DNA polymerase in the presence of biotin labeled ribonucleotides. The cRNA was cleaned up using the Qiagen RNeasy columns following manufactures instructions. The cRNA (20 μg) was fragmented using the fragmentation buffer supplied as part of the Gene Chip Sample Cleanup Module (Affymetrix). Image analysis and raw data description: The images were analyzed using MAS V.0 software provided by Affymetrix, Inc. The raw data output consists of several metrices: Signal, Detection, Detection p-value, Stat pairs, Stat pairs used. Signal is a measure of the abundance of the transcript. Detection is call that indicates whether the transcript is detected (P, present), undetected (A, absent), or at the limit of detection (M, marginal). Detection p-value is a measure of the significance of the detection call. Stat pairs, is the number of probe pairs for a particular probe set on an array. Stat pairs used, is the number of probe pairs per probe set used in the analysis. Normalized and summarized data: All chips were scaled to a target intensity of 500 to account for differences between the replicate chips and their hybridization efficiencies using Affymetrix GENECHIP MAS V.0 software. The entire dataset is available at gene expression omnibus (http://www.ncbi.nlm.nih.gov/geo/) with the accession number GSE802. Three filters were serially used to obtain the list of annotated genes that demonstrated differential expression in intensity between sham and burn injured animals. In the first filter, only genes that were flagged as âpresentâ by the MAS V.0 analysis software in at least one of the time points in all replicate chips were considered for further analysis. This step eliminated all genes for which the expression data was not reproducible between the replicate chips. The genes that passed this criterion were subjected to a second filter where ANOVA was performed to test each gene independently for a statistical difference in expression between groups (0, 1, 4, 8 and 24 h). The output of the analysis is the probability (p value) that a difference in expression can be observed by chance i.e., probability of getting a false positive. While the occurrence of false positives can be controlled by choosing a higher significance level (0.01 or 0.001) it also concomitantly increases the false negatives. Therefore, in this study we have chosen to work with ANOVA p-value cut-off 0.05. To reduce the occurrence of the false positives, we have used the false discovery rate (FDR) method (33) which adjusts the ANOVA p-value (cut-off 0.05) while taking the dependencies between the genes into consideration. The 695 genes that demonstrated a significant ANOVA p-value were then selected and their expression values averaged. The third filter selected 339 genes that were either up-regulated or down-regulated by at least 2-fold between any one time point (1, 4, 8 or 24h) and the 0h time point (Sham control). To determine the temporal profiles in the data, gene expression values were hierarchically clustered using the dChip software (www.dChip.org). The expression values for a gene across all samples were standardized by setting the mean to 0 and standard deviation to 1. The software then builds a hierarchical cluster tree based on centroid-linkage method using Pearson correlation coefficient as the distance metric. A set of genes were assigned to cluster groups empirically based on visual inspection of their temporal expression patterns.
