Selection of candidate reference genes using the RefGenes tool
RefGenes is an in silico method enabling the identification of genes with high expression stability within microarray libraries of wheat subjected to drought. Using this tool to examine normalized and well-annotated microarray experiments, we found 20 candidate RGs. Among these genes, we selected five candidates with stable expression levels under drought conditions. The candidate RGs obtained in this analysis were used for validation in qPCR (Table 1).
Table 1. Primers sequences and amplicons characteristics of candidate RGs.
Gene name
|
GenBank Accession number
|
Primer sequence (5'→3')
|
Amplicon length (bp)
|
Reference
|
Triticum aestivum alpha-tubulin mRNA
(TUB)
|
U76558
|
F: CCCTGAGGTTTGATGGTGCT
|
156
|
Rampino et al. [2006]
|
R: TGGTGATCTCAGCAACGGAC
|
Triticum aestivum mRNA for actin
(ACT)
|
AB181991
|
F: GGAGAAGCTCGCTTACGTG
|
136
|
Wei et al. [2015]
|
R: GGGCACCTGAACCTTTCTGA
|
Triticum aestivum glyceraldehyde-3-phosphate dehydrogenase (GAPC) mRNA
(GAPDH)
|
EF592180
|
F: AACGACCCCTTCATCACCAC
|
150
|
Wei et al. [2015]
|
R: GTTCCTGCAGCCAAACACAG
|
Triticum aestivum ubiquitin (WUB1) mRNA
(UBI)
|
AY297059
|
F: GGAGTCCACCCTTCACTTGG
|
130
|
Li et al. [2012]
|
R: GACACAGGCACCATTCGAG
|
Wheat translation elongation factor 1 alpha-subunit (TEF1) mRNA
(TEF1)
|
M90077.1
|
F: AGGCTGACTGTGCTGTTCTC
|
106
|
Liu et al. [2012]
|
R: AGAGTGAAAGCAAGGA
|
EST BJ254354
|
BJ254354
|
F: TGTTGAGGAGACAGTTGCCC
|
101
|
This study
|
R: GTTTGTCGGGCAATGCAGAG
|
EST wpa1c.pk012.d13
|
CA596223
|
F: AGAACTTGGCGTACAGGCTC
|
109
|
This study
|
R: GGCAGAGACTCGTACATCGG
|
EST wdi1c.pk002.n12
|
CA728440
|
F: CCCATCCAGCTCACACTGAC
|
134
|
This study
|
R: CGTGTCCGGCTTAAAACGAG
|
EST CJ705892
|
CJ705892
|
F: GCCTCAGTGGTAGGAGCATT
|
116
|
This study
|
R: TTCAGCAAATGCGGTGGTTG
|
EST wre1n.pk0067.d7
|
CA644093
|
F: CAGTCTGCACTGTGGCACTA
|
113
|
This study
|
R: CCAGCCGCCTAAACTTCTGA
|
Expression levels of the reference genes
To identify the most stable housekeeping genes, cDNA of all tested lines (stress imposed and control) was used in qPCR. The specificity of the primers was estimated by qPCR melting curve analysis. A single peak of the melting curve was observed for most of the tested primers (8 of 10 primer pairs), confirming the specificity of the amplicons. Only for two primer pairs (for the TEF1 and CA596223.1 genes) unspecific products of expression were observed, and because of that fact, they were excluded from further analysis. Moreover, no signal was detected in the NTC samples. We used the standard curve method with a pool of all the cDNAs to determine the PCR efficiency (E) and the correlation coefficient (R2) for each primer pair. The obtained results were analysed according to Bustin et al. [10], and the results indicated that the acceptable range of efficiency was from 80 to 120%. According to Tyburski et al. [30], a slope equal to -3.32 is evidence of high reaction efficiency, and R²=1 indicates that the same expression level was observed in the calibrator and tested sample. We obtained E values varying between 83.01% and 112.75% and R2 from 0.83 to 1 (Table 2). The raw quantification cycle (Cq) values were estimated for determination of the gene expression levels. The Cq values for analysed samples ranged between 20.16 and 37.60 (Fig. 1).
Table 2. Slope, efficiency and R2 values for analyzed candidate RGs.
Gene
|
Slope
|
Efficiency [%]
|
R2
|
ACT
|
-3.32
|
100.00
|
1
|
GAPDH
|
-3.36
|
98.44
|
1
|
TUB
|
-3.41
|
96.45
|
1
|
UBI
|
-3.23
|
103.98
|
0.99
|
BJ254354
|
-3.53
|
91.99
|
0.93
|
CA728440
|
-3.81
|
83.01
|
0.94
|
CJ705892
|
-3.25
|
103.09
|
0.98
|
CA644093
|
-3.05
|
112.75
|
0.83
|
Expression stability of the reference genes
The expression stability of selected RGs was estimated using five different algorithms: geNorm, NormFinder, BestKeeper, RefFinder and the delta Ct method. For each platform, eight RGs were ranked from the most stable to the least stable. In the results generated by the software packages, differences were observed. The ranking of RGs using geNorm was mostly in agreement with that of NormFinder. We found that the first three genes with the most stable expression and the gene with the least stable expression were the same for these two platforms. RefFinder and delta Ct analysis gave the same rankings among all 8 RGs. However, the results of BestKeeper and RefFinder showed different rankings for the most and least stable candidate genes (Table 3).
Table 3. geNorm M and stability values (SV) of the eight candidate reference genes obtained by geNorm, NormFinder, BestKeeper, RefFinder algorithm and delta Ct method.
Rank
|
geNorm
|
NormFinder
|
BestKeeper
|
RefFinder
|
delta Ct
|
Gene
|
geNorm M
|
Gene
|
SV
|
Gene
|
SV
|
Gene
|
SV
|
Gene
|
SV
|
|
CJ705892
|
0.554
|
CJ705892
|
0.072
|
ACT
|
0.498
|
ACT
|
1.00
|
ACT
|
0.87
|
|
ACT
|
0.573
|
ACT
|
0.084
|
CJ705892
|
0.526
|
UBI
|
1.86
|
UBI
|
0.89
|
|
UBI
|
0.596
|
UBI
|
0.131
|
UBI
|
0.564
|
CJ705892
|
2.71
|
CJ705892
|
0.89
|
|
GAPDH
|
0.676
|
TUB
|
0.139
|
TUB
|
0.732
|
GAPDH
|
4.43
|
GAPDH
|
0.99
|
|
TUB
|
0.729
|
BJ254354
|
0.152
|
CA644093
|
0.761
|
TUB
|
4.95
|
TUB
|
1.03
|
|
BJ254354
|
0.771
|
GAPDH
|
0.153
|
GAPDH
|
0.842
|
BJ254354
|
5.96
|
BJ254354
|
1.04
|
|
CA644093
|
0.839
|
CA644093
|
0.216
|
BJ254354
|
0.851
|
CA644093
|
6.44
|
CA644093
|
1.14
|
8. |
CA728440
|
0.975
|
CA728440
|
0.300
|
CA728440
|
1.032
|
CA728440
|
8.00
|
CA728440
|
1.46
|
geNorm analysis
geNorm analysis indicated that the stability of gene expression (M-value) varied between 0.550 for the most stable gene and 0.975 for the least stable gene (Table 3). According to this algorithm, genes with the lowest M-value were considered to be the most stable, whereas genes with the highest M-value were considered to be the least stable [31]. Based on geNorm software results, we identified a CJ705892 gene as the most stable in the tested wheat lines. Among a set of commonly used housekeeping genes, actin (0.575) and ubiquitin (0.600) were assessed as the most stable. The rest of the genes obtained from the Genevestigator database indicated a low level of expression stability. The least stable was CA728440 (0.975).
NormFinder analysis
The stability of the eight selected RGs was further analysed using the NormFinder platform. The NormFinder software analyses datasets and estimates stability based on intra-group and inter-group variation. The genes with lower stability values were considered to be the most stable RGs, whereas the genes with higher stability values were ranked as the least stable [32]. Based on the NormFinder algorithm, we found that the CJ705892 gene (stability value: 0.072) was the most stable gene, followed by ACT (0.084) and UBI (0.131). The least stable was CA728440 (0.300) (Table 3). We observed that the results of NormFinder and geNorm were slightly different. However, both algorithms indicated that the CJ705892 gene and CA728440 were the most and least stable genes, respectively. Therefore, based on the geNorm and NormFinder analysis and previous data concerning the BestKeeper and RefFinder platforms, we concluded that a novel gene, CJ705892, developed by the RefGenes tool, was found to be the most stable RG in tested wheat lines under short-term drought.
BestKeeper analysis
The obtained data were also analysed using the BestKeeper algorithm. BestKeeper software is usually employed by assessing the correlation coefficients of each individual gene with the geometric mean of all genes (the BestKeeper Index) [33]. These results were different from those of NormFinder and geNorm. According to BestKeeper, the most stable gene was ACT (0.468), followed by CJ705892 (0.526). The least stable was CA728440 (1.032) (Table 3).
RefFinder analysis
The RefFinder platform requires only raw Cq values without any option to include PCR efficiency. The ranking obtained by this algorithm is based on the standard deviations of the RG Cq values [33]. The results obtained by geNorm and NormFinder were not provided by the RefFinder output. RefFinder assessed ACT (1.00) and UBI (1.861) as the most stable genes. The gene CJ705892 (2.711) was in third place (Table 3).
Delta Ct method
The results obtained by the delta Ct method showed the same results as we observed using the RefFinder software. The most stable genes were ACT (0.868) and UBI (0.887), followed by CJ705892 (0.895) (Table 3). Analysis of all datasets suggests that the results obtained by the NormFinder, geNorm and BestKeeper methods indicated that the gene CJ705892 is on top of the RG rankings, with some slight differences in the rankings. All statistical algorithms showed CA728440 as the least stable gene.