Background: Epstein-Barr virus-transformed lymphoblastoid cell lines (EBV-LCLs) and skin fibroblasts are often used to establish or confirm the effect of mutations on gene expression levels. Relative quantification of gene expression is usually achieved by comparison of the expression level of a gene of interest with that of a suitable reference gene. Hence, the choice of reference gene is critical in all experiments that require normalization of expression data.
Methods: For normalization of gene expression levels in both EBV-LCLs and skin fibroblasts, we compared six common reference genes (ACTB, GAPDH, GUSB, HPRT1, PPIB, and TFRC) with six alternative reference genes (CLK2, COPS5, PCYT1A, RAD51L1, RNF10, and RNF111). The alternative genes were selected on the basis of their stability in expression levels in EBV-LCLs according to microarray data. The 12 genes were ranked according to stability in expression levels based on the standard deviation (SD) of the cycle threshold (Ct), geNorm, NormFinder, and the SD of the comparative cycle threshold (DeltaDeltaCt). In addition, we predicted the validity of an observed difference in expression level of a gene of interest between two samples when a specific reference gene is used.
Results and conclusions: In our dataset, GUSB and CLK2 were the best choices as reference genes for EBV-LCLs and fibroblasts, respectively. For almost all reference genes, expression level differences of <2-fold between two samples were most likely not significant (p > 0.05). GAPDH for EBV-LCLs and GAPDH and HPRT1 for fibroblasts should not be used for normalization in these cell lines because of their variability in expression levels.