Validating internal controls for quantitative plant gene expression studies

Abstract

Background: Real-time reverse transcription PCR (RT-PCR) has greatly improved the ease and sensitivity of quantitative gene expression studies. However, accurate measurement of gene expression with this method relies on the choice of a valid reference for data normalization. Studies rarely verify that gene expression levels for reference genes are adequately consistent among the samples used, nor compare alternative genes to assess which are most reliable for the experimental conditions analyzed.

Results: Using real-time RT-PCR to study the expression of 10 poplar (genus Populus) housekeeping genes, we demonstrate a simple method for determining the degree of stability of gene expression over a set of experimental conditions. Based on a traditional method for analyzing the stability of varieties in plant breeding, it defines measures of gene expression stability from analysis of variance (ANOVA) and linear regression. We found that the potential internal control genes differed widely in their expression stability over the different tissues, developmental stages and environmental conditions studied.

Conclusion: Our results support that quantitative comparisons of candidate reference genes are an important part of real-time RT-PCR studies that seek to precisely evaluate variation in gene expression. The method we demonstrated facilitates statistical and graphical evaluation of gene expression stability. Selection of the best reference gene for a given set of experimental conditions should enable detection of biologically significant changes in gene expression that are too small to be revealed by less precise methods, or when highly variable reference genes are unknowingly used in real-time RT-PCR experiments.

Figure 1

Scatterplot of residuals after regression of tissue means from each experiment on overall mean for all genes. Genes were ordered based on absolute value of mean residuals (increasing from left to right).

Figure 2

Regression lines for several genes showing predicted regression lines and actual means over both experiments. The most stable and consistent control genes would have the lowest slope and closest fit to the regression line. UBQ (second from bottom) had the highest and TUA (top) the second highest stability indices in this experiment. CYP (third from bottom) had the lowest stability index. See Table 1 for descriptions of tissue samples.