Total liver RNA was extracted applying the Trizol-RNA extraction procedure (Gibco BRL, Life Technologies, Merelbeke, Belgium). The extracted RNA was purified using Dnase I and RNase inhibitor (Fermentas, St. Leon-Rot, Germany) followed by a phenol/chloroform extraction. RNA integrity was checked by denaturing formaldehyde-agarose gel electrophoresis. RNA quality was measured using spectrophotometry (NanoDrop, NanoDrop Technologies, Rockland, DE, USA): absorbance at 260 nm (RNA), 280 nm (proteins) and 230 nm (organic contaminants and solvents) was determined and all RNA samples had 260/280 and 260/230 ratios >1.8.
Label
Cy5
Label protocol
Fluorescently labelled cRNA was constructed using the Quick Amp labelling kit (Agilent Technologies, Diegem, Belgium) according to Agilent’s two-color microarray-based gene expression analysis protocol (version 5.7, http://www.agilent.com). Briefly, total RNA was reverse transcribed into first and second strand cDNA. Using T7 RNA polymerase, the second strand cDNA was converted in cRNA while either Cy3-CTP or Cy5-CTP was incorporated. The labelled cRNA was purified using the Qiagen RNeasy mini spin column kit (Qiagen, Venlo, The Netherlands) and labelling efficiency was determined by spectrophotometry at 550nm (Cy3) and 650 nm (Cy5) (NanoDrop, NanoDrop Technologies, Rockland, USA).
Total liver RNA was extracted applying the Trizol-RNA extraction procedure (Gibco BRL, Life Technologies, Merelbeke, Belgium). The extracted RNA was purified using Dnase I and RNase inhibitor (Fermentas, St. Leon-Rot, Germany) followed by a phenol/chloroform extraction. RNA integrity was checked by denaturing formaldehyde-agarose gel electrophoresis. RNA quality was measured using spectrophotometry (NanoDrop, NanoDrop Technologies, Rockland, DE, USA): absorbance at 260 nm (RNA), 280 nm (proteins) and 230 nm (organic contaminants and solvents) was determined and all RNA samples had 260/280 and 260/230 ratios >1.8.
Label
Cy3
Label protocol
Fluorescently labelled cRNA was constructed using the Quick Amp labelling kit (Agilent Technologies, Diegem, Belgium) according to Agilent’s two-color microarray-based gene expression analysis protocol (version 5.7, http://www.agilent.com). Briefly, total RNA was reverse transcribed into first and second strand cDNA. Using T7 RNA polymerase, the second strand cDNA was converted in cRNA while either Cy3-CTP or Cy5-CTP was incorporated. The labelled cRNA was purified using the Qiagen RNeasy mini spin column kit (Qiagen, Venlo, The Netherlands) and labelling efficiency was determined by spectrophotometry at 550nm (Cy3) and 650 nm (Cy5) (NanoDrop, NanoDrop Technologies, Rockland, USA).
Hybridization protocol
The hybridization was performed according to Agilent’s two-color microarray-based gene expression analysis protocol (version 5.7, http://www.agilent.com). Two probe volumes (Cy3 vs. Cy5) corresponding with 300 ng cRNA were mixed and applied onto every microarray. The hybridization design comprised one separate n+2 A-optimal design for each of the Forafac® mixtures (Knapen et al., 2009). The microarray slides were incubated at 65°C for 17 hours in a hybridization chamber which was placed in a rotating (10 rpm) Agilent hybridization oven. The slides were washed with Agilent wash buffers and acetonitrile and finally submersed in Agilent stabilization and drying solution to protect against ozone-induced degradation of the cyanine dyes.
Scan protocol
Microarrays were scanned at 532 and 635 nm using the Genepix Personal 4100A confocal scanner (Axon instruments, Union City, USA) at a resolution of 5 µm. The photomultiplier tube voltage (PMT) was adjusted for each slide for the separate wavelengths to obtain an overall red/green ratio of one.
Data processing
The resulting images were analyzed using the Genepix Pro software 4.1 (Axon Instruments, Union City, USA) for spot identification and for quantification of the fluorescent signal intensities. Statistical analysis was performed in the open source statistical programming environment R, using the Bioconductor package limma (Linear Models for Microarray Data; Smyth, 2005). Spots for which the mean foreground signal was lower than the mean local BG+2SD (BG: background signal, SD: standard deviation of the background) on every array in the design were removed before analysis, requiring that the foreground intensities lie outside the 97.5% confidence interval of the background intensities (Allemeersch, 2006). Median intensity data were background corrected using a normal-exponential convolution model (Ritchie et al., 2007) and normalized using the lowess method (Smyth and Speed, 2003). An empirical Bayes method was used to estimate contrasts of interest from the normalized and linear model-fitted data (Smyth, 2004). The different exposure concentrations were contrasted against the control. Gene transcripts were considered significantly different if both the log2 ratio of a gene in a contrast was ≤ -0.75 or ≥ +0.75 and the p-value was <0.05.
