RESUMO
Normalisation to standard reference gene(s) is essential for quantitative real-time polymerase chain reaction (RT-qPCR) to obtain reproducible and comparable results of a gene of interest (GOI) between subjects and under varying experimental conditions. There is limited evidence to support selection of the commonly used reference genes in rat ischaemic and toxicological kidney models. Employing these models, we determined the most stable reference genes by comparing 4 standard methods (NormFinder, qBase+, BestKeeper and comparative ΔCq) and developed a new 3-way linear mixed-effects model for evaluation of reference gene stability. This new technique utilises the intra-class correlation coefficient as the stability measure for multiple continuous and categorical covariates when determining the optimum normalisation factor. The model also determines confidence intervals for each candidate normalisation gene to facilitate selection and allow sample size calculation for designing experiments to identify reference genes. Of the 10 candidate reference genes tested, the geometric mean of polyadenylate-binding nuclear protein 1 (PABPN1) and beta-actin (ACTB) was the most stable reference combination. In contrast, commonly used ribosomal 18S and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were the most unstable. We compared the use of PABPN1×ACTB and 2 commonly used genes 18S and GAPDH on the expression of 4 genes of interest know to vary after renal injury and expressed by different kidney cell types (KIM-1, HIF1α, TGFß1 and PECAM1). The less stable reference genes gave varying patterns of GOI expression in contrast to the use of the least unstable reference PABPN1×ACTB combination; this improved detection of differences in gene expression between experimental groups. Reduced within-group variation of the now more accurately normalised GOI may allow for reduced experimental group size particularly for comparison between various models. This objective selection of stable reference genes increased the reliability of comparisons within and between experimental groups.
