After removal of RNA, 2 μg of cDNA was fragmented with DNase and end-labeled (GeneChip®
WT Terminal Labeling Kit; Affymetrix). Size distribution of the fragmented and end-labeled cDNA, was assessed using an Agilent 2100 Bioanalyzer. 2 μg of end-labeled fragmented cDNA was used in a 200-μl hybridization cocktail containing added hybridization controls and hybridized on arrays for 16 hours at 48°C. Standard Caspase cleavage post hybridization wash and double-stain protocols (FS450_0001; Selleckchem CT99021 GeneChip HWS kit, Affymetrix) were used on an Affymetrix GeneChip Fluidics Station 450. Arrays were scanned on an Affymetrix GeneChip scanner 3000 7G. Microarray analysis Scanned arrays were first analyzed using Affymetrix Expression Console software to obtain Absent/Present
calls and assure that all quality parameters were in the recommended range. Subsequent analysis was carried out with DNA-Chip Analyzer 2008. First a digital mask was applied, leaving for analysis only the 8305 probe sets on the array representing Sinorhizobium meliloti transcripts. Then the 6 arrays were normalized to a baseline array with median CEL intensity by applying an Invariant Set Normalization Method . Normalized CEL intensities of the arrays were used to obtain model-based gene expression indices based on a PM (Perfect Match)-only model . Replicate data (triplicates) for each of the wild-type and tolC mutant strains were weighted gene-wise by using inverse squared standard error as weights.
Genes compared were considered to be differentially expressed if the 90% lower confidence bound of the fold change between experiment and baseline was PD0332991 nmr above 1.2, resulting in 3155 differentially expressed transcripts with a median False Discovery Rate (FDR) of 0.4%. The lower confidence bound criterion means that we can be 90% confident that the fold change is a value between the lower confidence bound and a variable upper confidence bound. Li and Wong  have shown that the lower confidence bound is a conservative estimate of the fold change and therefore more reliable as a ranking statistic for changes CYTH4 in gene expression. For a second analysis Partek Genomics Suite 6.4 was used. Here the 6 arrays were normalized and modeled using Robust Multichip Averaging (RMA). After RMA, probe sets analyzing expression of transcripts of Medicago truncatula and Medicago sativa, were filtered out. For the remaining S. meliloti probe sets differential expression was determined using 1-way Analysis of Variance (ANOVA). FDR analysis with a cut-off of 5% determined 2842 transcripts as differentially expressed, corresponding to an ANOVA p-value cut-off of <0.017. A set of 2067 differentially expressed transcripts was identified in the two independent analyses performed. All further analyses focused on this core set. Fold change values presented in Tables 1 and 2 and in the additional files 1 and 2 were obtained using Partek Genomics Suite 6.4.