Expression characteristics of the reference genes
We showed the full gene names and corresponding GeneBank accession numbers of the 21 candidate housekeeping genes in Table 1. The table 2 demonstrated the RNA concentrations at each stage of study, which ranged from 61.08 - 83.38 ng/μl. No significant between-group differences were apparent in RNA content. The extracted RNA quality is shown as 260 nm /280 nm ratio and 260 nm /230 nm ratio assessed by Nanodrop 2000 Spectrophotometer. Both ratios were close to 2.0, indicating the high quality of the RNA samples.
For the primer sequences and the amplified product length also see Table 1. Analysis of the melting curve assays, which showed the single peak, indicated the RT-qPCR amplification had good specificity (Figure 2A). The Ct (cycle threshold) -values obtained by RT-qPCR were used to quantify the gene expression levels, with a low Ct value indicating a high gene expression level and vice versa. Figure 2B illustrated the expression in each candidate reference gene abundance across the prenatal to postnatal periods of heart development. The candidate reference genes studied presented different expression levels with the mean Ct value ranged from 16.16 (Gapdh) to 25.19 (Gusb). The expression of the most frequently used internal reference gene in RT-qPCR experiments, Actb, revealed the unexpectedly significant variations at 7 different developmental stages of mice hearts. We found that some genes hold a more stable expression (such as Ubc, B2m, Vcp), whereas others were discrete, e.g., Ppia, Tfrc, Actb. These data illustrated the need for selection of appropriate reference genes for gene expression studies in mice hearts from different developmental stages.
For more efficient analysis of the expression pattern of those candidate reference genes at different development or maturation phases, we used the hierarchical cluster analysis (HCA) and orthogonal projections to latent structures - discriminant analysis (OPLS-DA) to globally visualize the expression classifications. Three phases can be distinguished in the gene expression features, that are (A) embryo stage, (B) first 7 days after birth, (C) 1 to 9 months after birth (Figure 3).
Stability of candidate reference genes
To assess the stability of gene expression, four different tools were used for each dataset tested in the two developmental stages addressed: GeNorm, NormFinder, Delta-Ct and BestKeeper. Table 3 demonstrated the results of the different statistical methods for analyzing the temporality expression variabilities of 21 housekeeping genes not only at earlier stages (embryonic stages). Although the scores and rank orders are not exactly the same as those reported in each of the calculation methods, the ranges of the dynamics are roughly similar in GeNorm, NormFinder, and Delta-Ct (Table 3).
GeNorm, NormFinder, and Delta-Ct, working on similar principles, held different computational approaches with BestKeeper. The stability values derived from BestKeeper based on the CV and SD values, while the former methods by comparing the average variation of a gene to all other. So, a consensual statistical analysis of the variabilities of the housekeeping genes was needed. Based on previous studies, we employed the RefFinder analysis to obtain a comprehensive evaluation of candidate reference genes by integrating all four analysis programs of the above-mentioned. The overall rank order of the most stable reference genes is shown in Table 3 for different comparisons across different developmental stages.
We summarized these results of the overall gene stability in each dataset with different condition using the heatmap or dot plot graph to illustrate the most suitable 5 reference genes for calculating the gene expression (Figure 4). Concerning the embryo development of LV, the Ppia, Rplp0, B2m, Vcp, and Gapdh were selected as the optimal reference genes when comparing the embryonic LV at 14-16 (E14-16) with LV at E17-20. In contrast, the Ppia and Gapdh were no more reliable choices for comparison of embryonic LV VS. hearts within 7 days of birth (E14-20 VS. D1-7). And the best selections of reference genes in this condition were the Vcp, Rplp0, Ywhaz, B2m, and Hprt1 (Table 3, Figure 4). Regarding the gene expression stabilities in LV on embryonic (E14-20) and postnatal maturation stages (M1-9), the results indicated the Rplp0, Gapdh, Tbp, Vcp, Gusb should be the appropriate reference genes for RT-qPCR experiments. To make the results of RT-qPCR as accurate as possible related to the early postnatal (D1-7) and postnatal maturation stages (M1-9), Reep5, Rplp0, Polr2a, Pgk1, Rpl5 represent the best set of reference genes (Table 3, Figure 4). Finally, among the 21 reference genes evaluated for their expression variabilities in comparing all LV samples under different phases of developmental maturity, the optimal set of reference genes should be Rplp0, Tbp, Vcp, Gusb, and Rpl5. On different conditions, the appropriate set of reference genes were not the same across groups. Our results showed that following genes should not be used for normalization in RT-qPCR experiments in mice hearts: 18S, Hmbs, Ubc, Psmb4, Tfrc and Actb. And the Rplp0 appeared to represent a good choice as reference gene when performing all the comparisons in the mice LV samples.
Gene expression levels normalized by different reference gene
We further verified the results by using the different internal reference genes to normalize the RT-qPCR data sets and to calculate the expression differences among groups. The most and least stable reference gene (Rplp0 and 18S) were included for RT-qPCR data analysis with the the same samples. As the results showed in Figure 5, the selection of proper internal reference gene induced very profound effects on gene expression levels. There was a large and significant difference in Vcp gene expression profile among the groups when Rplp0 as reference gene for normalization (Figure 5A), however, no significance when 18S used as reference gene (Figure 5B). As for the expression of Pgk1 in mice LVs with different stages, it is evident from Figure 5 that the statistical significance was obviously increased by the selecting of reference gene as Rplp0 rather than 18S.