Identification of reference genes for RT-qPCR data normalisation in aging studies

Abstract

Aging is associated with changes in gene expression levels that affect cellular functions and predispose to age-related diseases. The use of candidate genes whose expression remains stable during aging is required to correctly address the age-associated variations in expression levels. Reverse transcription quantitative-polymerase chain reaction (RT-qPCR) has become a powerful approach for sensitive gene expression analysis. Reliable RT-qPCR assays rely on the normalisation of the results to stable reference genes. Taken these data together, here we evaluated the expression stability of eight frequently used reference genes in three aging models: oncogene-induced senescence (OIS), in vitro and in vivo aging. Using NormFinder and geNorm algorithms, we identified that the most stable reference gene pairs were PUM1 and TBP in OIS, GUSB and PUM1 for in vitro aging and GUSB and OAZ1 for in vivo aging. To validate these candidates, we used them to normalise the expression data of CDKN1A, APOD and TFRC genes, whose expression is known to be affected during OIS, in vitro and in vivo aging. This study demonstrates that accurate normalisation of RT-qPCR data is crucial in aging research and provides a specific subset of stable reference genes for future aging studies.

Figure 1

Evaluation of RT-qPCR primer efficiencies. The amplification efficiency for each primer threshold cycle (Ct) and the logarithm of the initial cDNA concentrations were plotted to calculate the slope (S) of each primer pair. Standard curves were generated from at least four dilution points for each primer pair. RT-qPCR reactions for each sample were run in duplicate, with standard deviations <0.85.

Figure 2

Ct values of eight candidate reference genes in different aging models and the optimal number of reference genes needed for accurate normalisation calculated by geNorm. Box and whisker plots show the raw Ct values of the candidate reference genes during (a) oncogene-induced senescence in BJ fibroblasts, (b) in vitro aging in HDFs and (c) in vivo aging in HMECs from young and aged donors. (d) Pairwise variation (V_n/n+1) of candidate reference genes was obtained by geNorm to determine the required number of reference genes for accurate normalisation in each aging model. A discontinuous line indicates a pairwise variation (V) of 0.15, the cut-off value defined by geNorm. Abbreviations: 4-OHT: 4-Hydroxytamoxifen; Ct: cycle threshold; HDFs: Human Dermal Fibroblasts; HMECs: Human Mammary Epithelial Cells; EP: early passage (<10); LP: late passage (>20); YDs: young donors; ADs: aged donors. Notes: the boxes include values from the 25th to the 75th percentiles, the line across the box indicates the median, and whiskers show the minimum and maximum values for each reference gene. One-way ANOVA test was conducted. In those cases were p-values were < 0.05, a Bonferroni post-hoc test was conducted. Bonferroni corrected p-values are shown (*p < 0.05, **p < 0.01, ***p < 0.001).

Figure 3

Expression levels of target genes normalised to different reference genes. Gene expression levels were normalised to the indicated reference gene or gene pair in the X axis. (a) Relative expression levels of CDKN1A in control and in senescent BJ fibroblasts. (b) Relative expression of APOD and TFRC in EP and LP HDFs and (c) in HMECs from a young donor and an aged donor (YD 48R & AD 112R, respectively). Relative fold changes in gene expression were obtained according to the 2^−ΔΔCt method. Abbreviations: 4-OHT: 4-Hydroxytamoxifen; HDF: Human Dermal Fibroblasts; EP: early passage (<10); LP: late passage (>20); HMECs: Human Mammary Epithelial Cells; YDs: young donors; ADs: aged donors; OIS: oncogene-induced senescence. Data are presented as mean and SD. P-values obtained from pairwise comparisons using Welch’s t test are shown (*p < 0.05, **p < 0.01, ***p < 0.001).