Statistical differences resulting from selection of stable reference genes after hypoxia and hypothermia in the neonatal rat brain.

Martín Bustelo,D L A Van Den Hove,Martín A Bruno,Antonio W D Gavilanes,César F Loidl,Harry W M Steinbusch,Manuel Rey-Funes

doi:10.1371/journal.pone.0233387

Abstract

Real-time reverse transcription PCR (qPCR) normalized to an internal reference gene (RG), is a frequently used method for quantifying gene expression changes in neuroscience. Although RG expression is assumed to be constant independent of physiological or experimental conditions, several studies have shown that commonly used RGs are not expressed stably. The use of unstable RGs has a profound effect on the conclusions drawn from studies on gene expression, and almost universally results in spurious estimation of target gene expression. Approaches aimed at selecting and validating RGs often make use of different statistical methods, which may lead to conflicting results. Based on published RG validation studies involving hypoxia the present study evaluates the expression of 5 candidate RGs (Actb, Pgk1, Sdha, Gapdh, Rnu6b) as a function of hypoxia exposure and hypothermic treatment in the neonatal rat cerebral cortex–in order to identify RGs that are stably expressed under these experimental conditions–using several statistical approaches that have been proposed to validate RGs. In doing so, we first analyzed RG ranking stability proposed by several widely used statistical methods and related tools, i.e. the Coefficient of Variation (CV) analysis, GeNorm, NormFinder, BestKeeper, and the ΔCt method. Using the Geometric mean rank, Pgk1 was identified as the most stable gene. Subsequently, we compared RG expression patterns between the various experimental groups. We found that these statistical methods, next to producing different rankings per se, all ranked RGs displaying significant differences in expression levels between groups as the most stable RG. As a consequence, when assessing the impact of RG selection on target gene expression quantification, substantial differences in target gene expression profiles were observed. Altogether, by assessing mRNA expression profiles within the neonatal rat brain cortex in hypoxia and hypothermia as a showcase, this study underlines the importance of further validating RGs for each individual experimental paradigm, considering the limitations of the statistical methods used for this aim.

Highlights

In qPCR analysis, reference genes (RGs) with stable expression levels are essential internal controls for relative quantification of mRNA expression
The selection of RGs in qPCR experiments has an enormous impact on the reliability and interpretation of results in gene expression studies making it a crucial, yet often understated, process
It is recognized that normalization of qPCR results against a single RG is likely to be inadequate and that normalization against a panel of RGs containing at least three stable RGs is preferred

Summary

Introduction

In qPCR analysis, reference genes (RGs) with stable expression levels are essential internal controls for relative quantification of mRNA expression. RGs normalize variations of candidate gene expression under different conditions [1, 2]. The ideal RG should be expressed at constant levels regardless of e.g. experimental conditions, developmental stages or treatments [3, 4], and should have expression levels comparable to that of the target gene [5]. Increasing evidence suggests that the expression of commonly used RGs often varies considerably under different experimental conditions, as reviewed previously [6, 7]. The choice of unstable RGs for the normalization of qPCR data may give rise to inaccurate results, concomitant with potential expression changes in genes of interest being missed or overemphasized. The identification of stable RGs is a prerequisite for reliable qPCR experiments [8,9,10]

Methods

Results

Conclusion