Reference Gene Selection for Gene Expression Analyses in Mouse Models of Acute Lung Injury

qRT-PCR still remains the most widely used method for quantifying gene expression levels, although newer technologies such as next generation sequencing are becoming increasingly popular. A critical, yet often underappreciated, problem when analysing qRT-PCR data is the selection of suitable reference genes. This problem is compounded in situations where up to 25% of all genes may change (e.g., due to leukocyte invasion), as is typically the case in ARDS. Here, we examined 11 widely used reference genes for their suitability in commonly used models of acute lung injury (ALI): ventilator-induced lung injury (VILI), in vivo and ex vivo, lipopolysaccharide plus mechanical ventilation (MV), and hydrochloric acid plus MV. The stability of reference gene expression was determined using the NormFinder, BestKeeper, and geNorm algorithms. We then proceeded with the geNorm results because this is the only algorithm that provides the number of reference genes required to achieve normalisation. We chose interleukin-6 (Il-6) and C-X-C motif ligand 1 (Cxcl-1) as the genes of interest to analyse and demonstrate the impact of inappropriate normalisation. Reference gene stability differed between the ALI models and even within the subgroup of VILI models, no common reference gene index (RGI) could be determined. NormFinder, BestKeeper, and geNorm produced slightly different, but comparable results. Inappropriate normalisation of Il-6 and Cxcl1 gene expression resulted in significant misinterpretation in all four ALI settings. In conclusion, choosing an inappropriate normalisation strategy can introduce different kinds of bias such as gain or loss as well as under- or overestimation of effects, affecting the interpretation of gene expression data.

Automated summary

Selecting suitable reference genes is crucial in qRT-PCR data analysis. The stability of reference genes varies in acute lung injury models, and inappropriate normalization can lead to significant misinterpretation of gene expression data.