Background: Studies of postmortem human brain are important for investigating underlying pathogenic molecular mechanisms of neuropsychiatric disorders. They are, however, confounded by pre- and postmortem factors. The purpose of this study was to identify sources of variation that will enable a better design of gene expression studies and higher reliability of gene expression data.
Methods: We assessed the contribution of multiple variables to messenger RNA (mRNA) expression of reference (housekeeping) genes measured by reverse transcriptase-polymerase chain reaction (RT-PCR) by multiple regression analysis in a large number (N = 143) of autopsy samples from the hippocampus and white and grey matter of the dorsolateral prefrontal cortex (DLPFC) of patients with schizophrenia and normal control subjects.
Results: The strongest predictor of gene expression was total RNA quality. Other significant factors included pH, postmortem interval, age and the duration of the agonal state, but the importance of these factors depended on transcript measured, brain region analyzed, and diagnosis. The quality of RNA obtained from the DLPFC white matter was also adversely affected by smoking.
Conclusions: Our results show that normalization of expression data of target genes with a geometric mean of multiple housekeeping genes should be used to control for differences in RNA quality between samples. The results also suggest that accurate assessment of other confounding factors and their inclusion as regressors in the analysis is critical for obtaining reliable and accurate quantification of mRNA expression.