Dna microarrays

Global gene expression was determined during log-phase growth of B. breve UCC2003 in mMRS supplemented with either LNT, LNnT, LNB, lactosamine-HCl or lactose. The obtained transcriptome was compared to that determined for log-phase B. breve UCC2003 cells when grown in mMRS supplemented with ribose. DNA microarrays containing oligonucleotide primers representing each of the 1864 identified open reading frames on the genome of B. breve UCC2003 were designed and obtained from Agilent Technologies (Palo Alto, Ca., USA). Methods for cell disruption, RNA isolation, RNA quality control, complementary DNA synthesis and labelling were performed as described previously68 (link). Labelled cDNA was hybridized using the Agilent Gene Expression hybridization kit (part number 5188-5242) as described in the Agilent Two-Colour Microarray-Based Gene Expression Analysis v4.0 manual (publication number G4140-90050). Following hybridization, microarrays were washed in accordance with Agilent’s standard procedures and scanned using an Agilent DNA microarray scanner (model G2565A). Generated scans were converted to data files with Agilent’s Feature Extraction software (Version 9.5). DNA-microarray data were processed as previously described69 (link)70 (link)71 (link). Differential expression tests were performed with the Cyber-T implementation of a variant of the t-test72 (link).

200 ng of IP or input RNA were labelled using Agilent’s QuickAmp labelling kit and hybridized to DNA microarrays (Agilent). Initial HA-Tat and FLAG-Tat IPs were hybridized to a custom microarray containing multiple probes to all RefSeq mRNAs (includes standard Agilent probes)
[70 (link)]. For these custom arrays, RefSeq mRNAs with at least one probe with an IP/input ratio above 2-fold and p-value < 0.001 and at least one other with an IP/input ratio above 1.5-fold and p-value < 0.001 were judged to be enriched by the IP. Subsequent dye-swap and control IPs were analysed using catalogue Agilent 44K human gene expression arrays (AMADID 014850). For these arrays, the ratio and p-values were averaged for RefSeq mRNAs with multiple probes and then those mRNAs with an IP/input ratio above 2 and p-value < 0.001 were judged to be enriched by the IP. Log2 IP/input ratios were filtered to remove those with insignificant p-values and the remaining data clustered using average-linkage hierarchical clustering. A total of 3 Tat RNA IP microarray experiments were performed.

We collected four public available datasets from patients with primary breast cancer profiled using Affymetrix or Agilent DNA microarrays: the Netherlands Cancer Institute NKI2 dataset from van de Vijver et al. (256 patients), the KJX64/KJ125 datasets from Sotiriou et al. (189 patients), Uppsala dataset from Ivshina et al. (249 patients) and the Bordet Institute TRANSBIG dataset from Desmedt et al. (198 patients). Some samples of the NKI2 cohorts were excluded from our study due to missing or biased data. Redundant patients (74 samples) present in KJX64/KJ125 and Uppsala datasets were removed from the validation tests so they were only considered once. Gene expression and clinical data of public series were retrieved from Gene Expression Omnibus (GEO) public database http://www.ncbi.nlm.nih.gov/geo, author’s website and publications [6 (link),13 (link),14 (link),16 (link)]. All datasets were retrospective. They are described in Additional file 9: Table S7.