The Hup A content analyses
HPLC–UV was performed to detect the HupA content in H. serrata. Typical chromatograms from HupA standard and three tested samples are shown in Figure S1, indicating that HupA has good peak shape and is well separated from different tissues. A linear relationship exists between the peak area (measured at 308 nm) and the concentration of HupA in the sample injected into the HPLC. The results showed that there was obvious difference for the HupA content in different tissues. The highest HupA content (72 μg/g) was found in the leaves of H. serrata. The lowest content (19 μg/g) of HupA was found in root tissues of H. serrata (Figure 1).
The screening of candidate RGs
In consideration of the varied difference of Hup A concentrations in different tissues, the root, stem and leaf samples were collected and proceeded the full-length transcriptome sequencing by Nanopore. After assembly, 43,443 unigenes were retrieved. CPM(counts per million)is the index for measuring the expression of unigenes. Based on the CPM value and reported literatures, ten traditional RGs and three new RGs candidate were choosed. The three new RGs candidate had stable expression in full-length transcriptome sequencing. They were annotated hypothetical or uncharacterized proteins by NCBI Nr database, furthermore, they were not used as RGs before. The three new RGs candidate as following: ONT.10684 represented the high expression level (CPM over 100), EVM0022608 was the middle level (CPM 29-34), and EVM0017784 was the low level (CPM less than 5). The detail information of total thirteen candidate RGs was showed in Table 1.
Verification of the primer specificity and RT-qPCR amplification efficiency
The primer information of thirteen candidate RGs was given in Table 2. Each primer pair was designed except the conserved domains to ensure the specificity (Figure S2). Initially, the agarose gel electrophoresis yielded a specific fragment of expected size (Figure S3A, S3B and S3C). Further, the melting curve analysis in the RT-qPCR reaction showed the single peak for each primer pair indicating an absence of non-specific product amplification (Figure S4). For all primer pairs, the amplification efficiencies were spanning from 90.4% to 103.6%, and the correlation coefficient (R2) were greater than 0.990 (Table 2). Taken together, these results indicated each primer pair was specificity and the RT-qPCR assays were highly efficient.
Expression profiles of candidate RGs
The expression profiles of RT-qPCR products for all experimental samples are shown in Figure2. The results illustrated that the mean Ct values of all RGs ranged from 24.04 to 29.43. Lower Ct value indicates the higher expression abundance, conversely means the lower expression profiles. EF1dt and UBQ1 were highly expressed with mean Ct values between 24.04 and 24.08 while EFTS was the least expressed gene on account of its highest mean Ct value (29.43). All candidate genes showed expression variability in different samples as evident from a wide range of Ct values. Genes such as GAPDHB and EFTS showed relatively smaller variation (< 2 cycles), while others like UBQ11 had the highest expression variation (3.07 cycles). The results indicated that there was still variable expression even for relative stable housekeeping genes.
geNorm analysis
To identify the most stable RG, geNorm algorithm calculated the average expression stability values (M values) of each RG. As Figure3 shown, each M value was less than 1.5, which suggested the appropriateness of all RGs for normalization consideration in different tissues of H. serrata. Concretely analyzing, EF1dt, HisH2A and GAPDHB were the most stable genes in each H. serrata samples, while HisH3.3 and EFTS were the least stable genes in each H. serrata samples.
NormFinder analysis
NormFinder evaluates each RG according to the stability value. Lower stability value indicates more stable gene expression, and vice versa. As shown in Table 3, GAPDHB and HisH2A were obviously stable in all samples, and EFTS (highest stability value = 0.210) was the least stable gene. For the root samples, HisH2A and a-tub3 were most stable, and the EFTS (stability value = 0.506) still was the least stable gene. Whereas GAPDHB and HisH2A were the most stable gene and HisH3.3 (stability value = 0.358) was the least one in stem. In leaf tissues, the most stable RGs were EF1dt, HisH2A and, GAPDHB, meanwhile, the least stable RG was Actin7 (stability value = 0.364). Overall, with NormFinder analysis, GAPDHB, HisH2A and EF1dt were the most stable genes, while EFTS and HisH3.3 were the least stable genes in different tissues of H. serrata.
BestKeeper analysis
The stability standard deviation (SD) and its relationship to the BestKeeper index were considered as two important evaluation criteria in BestKeeper analysis [23]. The results showed that each RG had a SD value < 1.0, which indicated that the candidate RGs were relatively stable for RT-qPCR normalization. In present, GAPDHB, EF1dt and HisH2A were the top three ranked genes with lowest CV ± SD values in all samples, stem and leaf tissues (Table 4). In root samples, the top three ranked genes were HisH2A, UBQ11 and a-tublin. Rather, HisH3.3 was deemed to the least RG with the highest CV ± SD value (27.35 ±0.21 and 28.66 ±0.20) in all samples and stem tissues, while EFTS (in root tissues) and Actin7 (in leaf samples) showed the least stable expression. Taken together, with BestKeeper analysis, GAPDHB, EF1dt and HisH2A were the most stable genes, while HisH3.3 and EFTS were the least stable genes in different tissues of H. serrata.
RefFinder analysis
Athough the results (geNorm analysis, NormFinder and BestKeeper analysis) were similar, it was not strictly consistent. Therefore, we performed overall evaluate using RefFinder to recommend a comprehensive ranking of the most stable genes in diverse tissues (Table 5). In root tissues, the final ranking calculations based on the RefFinder found HisH2A (GM = 1.67), GAPDHB (3.33) and a-tub3 (3.33) were the best genes. For stem samples, the top three stable RGs were GAPDHB (1.67), EF1dt (2.00) and HisH2A (3.67), while EF1dt (1.33), HisH2A (2.00) and GAPDHB (2.33) in leaf samples. Across all samples, the top three stable RGs were GAPDHB (1.67), HisH2A (1.67) and EF1dt (2.33). On the other hand, HisH3.3 and EFTS were ranked as the two least stable genes (Table 5).
Optimal Number of RGs for Normalization
Though a single and stable RG is sufficient for quantifying gene expression, the use of more than one RG for effective normalization of gene expression data is suggested [29]. Based on the geNorm software, the optimal number of RGs needed for normalization was determined by pairwise variation (Vn/n+1). In our data, the all pairwise variation of V2/3 values were lower than 0.15 (Figure 3), which suggested that the combination of the two most stable RGs was optimum for normalization.
Together with RefFinder analysis, GAPDHB and HisH2A were the best combination for normalization in different tissues of H. serrata.
RG Validation
To demonstrate the utility of identified stable RGs, four HupA-biosynthetic related genes LDC, MET CYP and CAO were selected in H. serrata. For the purpose of comparison, expression values of target genes were normalized with respect to the most stable gene pair (GAPDHB and HisH2A) and the least stable gene pair (EFTS and HisH3.3) in H. serrata different tissues. When normalized using the most stable genes, the transcription levels of LDC, MET, CYP and CAO were (over 2 –fold) in the tissues of leaf, stem and root were compared, and the expression trend was consistent with that of transcriptome sequencing dates (Figure 4). By contrast, when normalized using the least stable genes, the transcription level of MET and CYP were not up-regulated (less than 2 –fold) in stem and leaf tissues. The transcription level of LDC was down-regulated (0.77 –fold) in stem tissues, and the CAO was down-regulated (0.67 –fold) in leaf tissues. This expression trend was not consistent with that of transcriptome sequencing dates. In all, the expression of the most stable gene pair was more reliable than the least stable gene pair.