Candidate reference genes and PCR amplification
Eighteen candidate reference genes were selected from the transcriptome of S. superba and ColGAPDH was cited from Yang [22]. The presence of a single PCR product expected size (Figure S1) and single peak in the melting curve (Figure S2) confirmed the specific amplification. The amplification efficiency (E) of all PCR reactions ranged from 93.47 % for SsuTUA1 to 109.03 % for SsuUBCJ2 (Table 1), suggested that these genes were suitable for further gene expression analysis. Meanwhile, the standard curves showed good linear relationships, with correlation coefficients (R2) above 0.99 (Table 1).
Ct values of candidate reference genes
To assess the expression stability of 19 candidate reference genes in different tissues, the transcript abundances were presented as their Ct values. The Ct values varied in different tissues, which varied from 16.752 (SsuMet2) to 33.379 (SsuCas), while the mean Ct values varied from 18.032 (SsuMet2) to 25.556 (SsuMDH) (Table 2). The Ct range > 4 of SsuCas, SsuUBC17, SsuUBC2 and SsuUDP in different tissues, suggested these genes varied greatly and were unstable in different tissues.
Analysis of reference gene stability using three bioinformatic programs
In order to reduce the analysis error, the candidate gene stability ranking in different tissues was determined separately using geNorm, NormFinder and BestKeeper. So as to screen out the reference genes suitable for the experimental treatment, and provide beneficial reference for subsequent research.
The geNorm program is used to rank the genes expression stability by calculating the average expression stability values (M) based on the 2−∆Ct value [25]. The smaller M value of reference gene, the more stably expressed. Meanwhile, if M value > 1.5, it is not suitable as a reference gene [25]. The M values of tested genes evaluated by geNorm were shown in Fig.1. SsuTUA1 and SsuRIB were ranked as the two most stable genes in different tissues, while SsuCas and SsuUDP were the two least stable genes.
The pairwise variation value (Vn/Vn+1) calculated by the geNorm determines the optimal number of reference genes. When Vn/Vn+1< 0.15, the optimal number of reference genes is n, otherwise, the number is n+1 [25]. In this study, except for V18/V19, the other value of Vn/Vn+1 < 0.15 (Fig.2), indicating that two reference genes would be sufficient for gene normalization and an increase do not improve sensitivity.
NormFinder ranks the expression stability of reference genes by calculating the average pairwise variation in one relative to other candidate genes and the smaller stability value, the more suitable as reference gene [26]. For different tissues, the most stable gene was SsuACT, followed by SsuRIB, while the least stable gene was SsuCas, which were not altogether with the genes selected by the geNorm (Fig.3).
The expression stability is represent by the standard deviation (SD), coefficient of variance (CV) and correlation coefficient (r) of Ct values in the BestKeeper program, and the most stable reference genes are identified as those with the lowest SD and CV and the most r [27]. In this study, SsuACT and SsuUBCJ2 were identified as the most stable genes for different tissues, while SsuUBC17, SsuTUB, SsuUBC2, SsuUDP, SsuCas are unstable because of the SD > 1(Table 3).
The rankings of the nineteen tested genes were not perfectly consistent by geNorm, NormFinder and BestKeeper (Table 4). To provide a comprehensive evaluation of candidate reference genes, further analysis was thus carried out using geometric mean, which integrates geNorm, Normfinder and BestKeeper. The comprehensive ranking recommended by the geometric mean method was shown in Table 4, and SsuACT was the most stable genes for different tissues.
The best combination of reference genes was determined based on the optimal number calculated by geNorm and the ranking list obtained using geometric mean method. So SsuACT + SsuRIB was found to be the best combination of reference genes for different tissues.
Validation of the identified reference genes
In order to examine the reliability of the candidate reference genes for normalization, SsuSND1 expression profiles in different tissues were normalized using the two most stable candidate reference genes (SsuACT and SsuRIB), combination of stable genes (SsuACT + SsuRIB), the least stable reference gene (SsuCas) well as ColGAPDH (Fig.4). When SsuACT, SsuRIB, SsuACT + SsuRIB and ColGAPDH were used for normalization, the expression patterns of SsuSND1 were similar, and the relative expression of xylem, leaf and fruit were higher than others, and SsuSND1 was hardly expressed in bud and root, but the expression was most appropriate by SsuACT and SsuACT + SsuRIB. However it was analyzed by SsuCas, the expression pattern was not compatible and the expression level were too high in fruit and too low in bud. It was suggested that the selected reference genes were reliable.