The extraction of RNA from blood samples was performed with PAXgeneTM blood RNA kit (PreAnalitiX GmbH, Hombrechtikon, Switzerland) following the manufacturer’s instructions.
Label
Cy5
Label protocol
One-hundred ng of each pool were submitted to double round of amplification using Amino Allyl MessageAmpTM II aRNA Amplification Kit (Ambion, Inc.). At the end of amplification protocol aRNA was labeled with Cy5 dye
Hybridization protocol
Hybridizations were performed according to the standard protocol supplied with CustomArrayTM 90K (CombiMatrix Irvine, CA, USA). Probes were designed from 12194 UniGenes (NCBI - Build #13) by mean of OligoWiz 2.0 software (Wernersson and Nielsen, 2005). For each Unigene, 2 nonredundant 35-40mer probes were spotted in triplicate on the chip. Bacterial and plant probes were also spotted on the slide as negative control.
Scan protocol
The microarray was scanned using a Perkin Elmer laser scanner and data were extracted from the image using CombiMatrix Microarray Imager Software (CombiMatrix Irvine, CA, USA).
Data processing
A quantile normalization algorithm was applied as described in Bolstad (2003). This method is a generalization of standard median normalization techniques which match only the 50% quantile of each sample’s signal distribution. Afterwards an in-slide replicates analysis from Midas was used to reduce data redundancy. To evaluate the significant over and under expressed genes between experimental groups, the ratios of T3 to T0 and of T51 to T0 were first calculated for each gene within replicates and data were log(2) transformed to ensure variance homogeneity. Finally, a tab delimited file was imported to MeV software (v 4.6- http://www.tm4.org/mev/; Saeed et al., 2003) and used to perform one-way ANOVA with group as fixed factor. P-values were based on permutation test, with 1000 permutations, and Bonferroni correction was applied setting alpha (critical p-value) at 0.001.
