Approval from the Animal Experiment Committee of the University of Kuopio and the Provincial Government CX-5461 nmr of Eastern Finland was obtained for all animal study plans. Following euthanasia, liver tissues were excised, sliced, and snap-frozen. The tissues were later homogenized and total RNA was extracted using Qiagen RNeasy kits according to the manufacturer’s instructions (Qiagen, Mississauga, Canada) as previously described (Boutros et al., 2011). The isolated RNA was assayed on Affymetrix RAT230-2 (Wis and F344; performed with six biological replicates each) or RAE230-A (L-E, H/W, LnA, and LnC; performed with four biological replicates each) arrays at The Centre for Applied Genomics at The

Hospital for

Sick Children (Toronto, Canada). The two platforms RAT230-2 and RAE230-A differ by the number of probe sets contained on the array. The platform RAE230-A is a subset of RAT230-2 and hence shares many of the same genes as RAT230-2. Our statistical comparisons were performed within the same platform; thus any variability is balanced and no bias is introduced. We rigorously assessed the technical quality of each array and none were excluded from subsequent data analyses. Animal handling and reporting comply with ARRIVE guidelines (Kilkenny et al., 2010). Raw quantitated array data (CEL files) were loaded into the R Y-27632 in vivo statistical environment (v2.12.2) using the affy package (v1.28.0) of the BioConductor library (Gentleman et al., 2004). Data were screened for spatial and distributional homogeneity and none were excluded from this study. Data were pre-processed with a sequence-specific version of RMA algorithm – GCRMA – as implemented in R (gcrma package v2.22.0). Probes were remapped to Entrez Gene IDs using rat2302rnentrezgcdf (v13.0.0) and rae230arnentrezgcdf Digestive enzyme (v13.0.0) R packages (Dai et al., 2005). Entrez Gene annotation was downloaded from NCBI on 2011-02-22. Individual strains were treated as separate cohorts and animals within a cohort were pre-processed together to avoid confounding effects from co-normalization of diverse strains. Raw and

pre-processed microarray data are available in the GEO repository under accession GSE31411. Following pre-processing, we employed general linear-modeling to identify genes affected by TCDD treatment relative to the vehicle control. The expression profiles across all animals within a cohort were determined using a per-gene linear model that assesses both basal levels and TCDD-induced effects. Coefficients were fit to terms representing each effect and the standard errors of the coefficients were adjusted using an empirical Bayes moderation of the standard error (Smyth, 2004). To test if each coefficient was statistically different from zero, we applied model-based t-tests, followed by a false-discovery rate adjustment for multiple-testing (Storey and Tibshirani, 2003).