Identification and Evaluation of Reference Genes for Accurate Transcription Normalization in Safflower under Different Experimental Conditions
Identification and Evaluation of Reference Genes for Accurate Transcription Normalization in Safflower under Different Experimental Conditions
Dandan Li 0 1 2
Bo Hu 0 1 2
Qing Wang 0 1 2
Hongchang Liu 0 1 2
Feng Pan 0 1 2
Wei Wu 0 1 2
0 Agronomy College, Sichuan Agricultural University , Wenjiang 611130, Chengdu, Sichuan , P. R. China
1 Funding: This work was financially supported by the National Natural Science Foundation of China (
2 Editor: Obul Reddy Bandapalli, University of Heidelberg , GERMANY
Safflower (Carthamus tinctorius L.) has received a significant amount of attention as a medicinal plant and oilseed crop. Gene expression studies provide a theoretical molecular biology foundation for improving new traits and developing new cultivars. Real-time quantitative PCR (RT-qPCR) has become a crucial approach for gene expression analysis. In addition, appropriate reference genes (RGs) are essential for accurate and rapid relative quantification analysis of gene expression. In this study, fifteen candidate RGs involved in multiple metabolic pathways of plants were finally selected and validated under different experimental treatments, at different seed development stages and in different cultivars and tissues for real-time PCR experiments. These genes were ABCS, 60SRPL10, RANBP1, UBCL, MFC, UBCE2, EIF5A, COA, EF1-β, EF1, GAPDH, ATPS, MBF1, GTPB and GST. The suitability evaluation was executed by the geNorm and NormFinder programs. Overall, EF1, UBCE2, EIF5A, ATPS and 60SRPL10 were the most stable genes, and MBF1, as well as MFC, were the most unstable genes by geNorm and NormFinder software in all experimental samples. To verify the validation of RGs selected by the two programs, the expression analysis of 7 CtFAD2 genes in safflower seeds at different developmental stages under cold stress was executed using different RGs in RT-qPCR experiments for normalization. The results showed similar expression patterns when the most stable RGs selected by geNorm or NormFinder software were used. However, the differences were detected using the most unstable reference genes. The most stable combination of genes selected in this study will help to achieve more accurate and reliable results in a wide variety of samples in safflower.
Competing Interests: The authors have declared
that no competing interests exist.
Safflower (Carthamus tinctorius L.) belongs to the Asteraceae family is a multipurpose plant.
Safflower is used in traditional Chinese medicine (TCM) . Most of the medicinal effects of
the extract come from the water soluble yellow pigment ‘carthamidin’ . Furthermore,
safflower is an ancient oilseed crop widely grown for its high-quality esculent oil . Two major
fatty acids—oleic acid (C18:1Δ9) and linoleic acid (C18:2Δ9,12)—are found in safflower seed oil,
which together account for approximately 90% of the total fatty acids in this plant. Safflower
oil is well known for its relatively high level of linoleic acid content of more than 70% of total
fatty acids (FA) [4–7]. With the rapid development of molecular biology techniques, there has
been a tendency to focus on the expression patterns of key genes of interest in safflower, such
as genes that regulate the synthesis of flavonoid compounds, oleic acid and linoleic acid, for
developing new and exceptional cultivars from the perspective of genetic engineering and plant
Transcriptome analysis techniques, including RNA-seq and microarrays, can offer a
representative snapshot of the transcriptome that can only be applied to the limited treatments or
tissues. Real-time PCR has become an important method for in-depth analysis of gene
expression due to its wider range of quantification, higher accuracy, and lower cost [10–13]. However,
the accuracy of the results of relative quantification of gene expression is affected by many
factors, such as primer design, experimental operation, and choice of optimal annealing
temperature, but the choice of optimal RGs has become the most important factor . An erroneous
quantification result may emerge when using unstable RGs for normalization in gene
expression analysis .
Ideal RGs should be expressed at a constant level and not be affected by various
experimental elements. The theoretical range of the RG expression levels should be similar to the target
genes [16–18]. β-actin, glyceraldehyde-3-phosphate hydrogenase (GAPDH), ubiquitin (UBQ),
elongation factor-1α (EF1-α) and 18s rRNA are common RGs that are often used in
quantification experiments [19–24]. However, it is reported that there are no universal RGs under
variable experimental conditions. Some common RGs are not constantly expressed under different
experimental conditions in different tissues, especially in different species [25–27]. Therefore,
it is necessary to systematically evaluate the stability of candidate RGs for diverse experimental
conditions prior to their use in RT-qPCR normalization .
During the past few years, there have been a great number of reports related to the
identification and evaluation of appropriate RGs emerged for plants, such as papaya , citrus ,
switchgrass , strawberry , coffee , pea , peanut , tobacco , tomato ,
soybean  and arabidopsis . However, so far, there are no reports on the identification
and evaluation of RGs in safflower under various experimental conditions. Based on the results
of transcriptome analysis in safflower in our research group (unpublished data), some genes
with relative stable expression levels at different seed developmental stages were selected as
candidate RGs. These genes are the ABC superfamily (ABCS), 60s ribosomal protein L10
(60sRPL10), ran-binding protein RANBP1 and the related RanBD domain proteins (RanBP1),
ubiquitin-protein ligase (UBQL), multifunctional chaperone (MFC), ubiquitin-conjugating
enzyme E2 (UBCE2), translation initiation factor 5A (EIF5A), acetyl-CoA acetyltransferase
(COA), elongation factor 1 beta/delta chain (EF1β), elongation factor 1 alpha (EF1),
glyceraldehyde 3-phosphate dehydrogenase (GAPDH), F0F1-type ATP synthase, beta subunit (ATPS),
transcription factor MBF1 (MBF1), GTP-binding ADP-ribosylation factor Arf1 (GTPB), and
glutathione S-transferase (GST). Multiple abiotic stresses, such as gibberellin and paclobutrazol
spraying with different concentrations, cold stress, and salt stress, were applied in this study.
Three safflower cultivars that have different contents of linoleic acid were used, and various
tissues were selected for this study. To evaluate the stability of candidate RGs more appropriately,
both geNorm3.5  and NormFinder  were used for the analyses. Furthermore, in order
to validate the newly identified RGs, a detailed expression analysis of CtFAD2 genes that
control the formation of linoleic acid were performed using both the most stable and unstable
reference genes for normalization. Although we could not identify any single gene expressed
constantly even in a particular species, one or two appropriate and stable RGs in specific given
conditions used in RT-qPCR experiments could be selected. The selection and validation of
candidate RGs in RT-qPCR experiments under different experimental conditions should be
carried out before they are utilized for the normalization of RT-qPCR data. In this work, the
referential recommendations for using these candidate genes are offered to confirm an accurate
normalization of transcript level under a particular condition in gene expression studies in
safflower by RT-qPCR.
Materials and Methods
Safflower (Carthamus tinctorius L.) accessions PI401477 (81.84/11.13, % oleic/linoleic),
PI470942 (76.7/14.3) and PI544021 (16.47/79.1) was provided by Dr. Bradley of American
Germplasm Resource Information Network (GRIN). Safflower cultivar Chuanhong No.1 was
provided by Ms. Tang Li from Yaan San Jiu Medicine Co. Ltd. Young leaves were taken from
1-month-old seedlings, and for the reproductive organ, seeds were harvested at 0, 10, and 20
days after flowering (DAF). All plant materials were immediately frozen in liquid nitrogen
after harvesting and then stored at -80°C until the total RNA was isolated.
Samples were collected under various stress conditions. A summary of the stress assays is
shown in Table 1. In our previous experiment, safflower seeds were planted in the field in
September and were spayed by different concentrations gibberellin acid (GA3) and paclobutrazol
(PP333) at the bud stage in the next year. The results show that 0.09 mM GA3 and 0.51 mM
PP333 can significantly improve the linoleic acid and oleic acid contents of seed oil, respectively,
in safflower accession PI544021, so 0.09 mM GA3 and 0.51 mM PP333 were selected and the
seeds at 0 and 10 DAF were used for this experiment.
Safflower accessions PI401477, PI470942 and PI544021 were subjected to cold treatments.
The plants was grown in the field and at the bud stage were transferred to 12°C for 48 h in an
illumination incubator (25°C, 16 h lighting/day). After 48 h, the plants were continued to grow
at room temperature, and the seeds were collected at 10 and 20 DAF.
Safflower cultivar Chuanhong No.1 was subjected to a salt stress treatment and abscisic acid
(ABA) treatment. For the salt stress treatment, a progressive increase in the salt concentration
in the solution (0, 100, 150 mM NaCl) was applied for the 1-month-old seedlings grown in
illumination incubator (25°C, 16 h lighting/day). Leaves were taken at 0, 6, and 12 h after reaching
the last salt solution concentration. For ABA treatment, 1-month-old seedlings (the growth
conditions were the same as above) were sprayed with 100 μM ABA, and the leaves were
collected at 0, 2, and 5 h after ABA spraying.
For different seed development stage samples, safflower accession PI544021 was planted
under normal conditions, and seeds were selected at 0, 10, and 20 DAF.
Different tissue samples were selected from safflower accession PI544021 under normal
development conditions. Seeds and leaves were selected at 10 DAF, and the flowers were
selected at 0 DAF.
All experiments were performed with biological triplicates. Details of the above-mentioned
experimental conditions are summarized in Table 1.
GA3 concentration: 0.09 mM
PP333 concentration: 0.51 mM
12°C for 48 h
Biological stages / Time points
0 DAF in bud stage
Note: DAF: days after flowering. GA3: gibberellin acid. ABA: abscisic acid. PP333: paclobutrazol. The same abbreviations are used in the following tables.
Total RNA isolation and cDNA synthesis
A total of 10 mg of plant tissue was used for RNA isolation. Total RNA was extracted using the
Trizol method (Trizol was purchased from TaKaRa, Dalian) and then treated with DNAse I for
digestion using the RNAse-free kit (TaKaRa, Dalian) to decrease potential DNA
contamination. The quality of the RNA samples was verified by 1.0% agarose gel electrophoresis (AGE).
The RNA concentration and purity were determined with a NanoDrop ND-2000
spectrophotometer. RNA samples with a 260/230 ratio higher than 2.0 and 260/280 ratio between 1.8 and
2.1, as well as 18S and 28S ribosomal RNA bands with a concentration ratio of approximately
1:2, were selected and used in the subsequent experiment. First-strand cDNA was synthesized
using the PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa, Dalian) according to the
manufacturer’s instructions. The final cDNA products were diluted for RT-qPCR experiments.
Selection of Potential RGs and Primer Design in Safflower
On the basis of the results of transcriptome sequencing in safflower with the different seed
development stages in our research group (data unpublished), combined with commonly used
RGs, some of the stably expressed genes according to the transcriptome analysis results were
selected as candidate RGs to seek better RGs with wider adaptation in various experimental
Primer 3 (v.0.4.0) (http://bioifo.ut.ee/primer3-0.4.0/) online version was used for specific
primer design for RT-qPCR using the default parameters for a qPCR experiment. The primers
were selected with no hairpins or dimer, and target specificity was detected by Blastn against
the nt/nr databank. The primer lengths were between 20–22 bp, the amplicon lengths varied
from 90 to 150 bp, and the melt temperatures were between 58 and 63°C. The names,
transcriptome numbers, primer and amplicon lengths, melt temperature and primer sequences of
the fifteen candidate RGs are shown in Table 2 and Table 3. All primer amplification
efficiencies were between 90 and 110% from real-time PCR results. A single peak in the melting curve
of the amplification products of each candidate gene and a single band of AGE verified
indicates that a single PCR product was amplified effectively and that the specificity of each primer
Ran-binding protein RANBP1 and
related RanBD domain proteins
Multifunctional chaperone (14-3-3
Ubiquitin-conjugating enzyme E2
Elongation factor 1 beta/delta chain
elongation factor 1 alpha
60s ribosomal protein L10
was good. The single peak in the melting curves and the result of the agarose gel
electrophoretogram are shown in Figs 1 and 2.
RT-qPCR analysis was performed using the Bio-Rad CFX96 real-time system. Each PCR
reaction contained 5 μl SsoFast EvaGreen supermix (Bio-Rad), 1 μl of the diluted cDNA reaction
sequence of forward (F) and reverse
Transcription factor MBF1
sequence of forward (F) and reverse (R)
Fig 1. The melting curves of the 15 candidate RGs. A single peak is shown (each including 3 technical
replicates of the cDNA pool of the total samples used in this study).
mixture (The cDNA diluted to 200 ng/μl was used for the qPCR assay), 1 μl specific primer
with a concentration of 200 nM and 3 μl ddH2O in a total reaction volume of 10 μl. Reactions
were performed at 95°C for 1 min, 40 cycles of 95°C for 10 s, and 58°C for 30 s in 96-well
reaction plates. The specificity of PCR reactions was verified by a melting curve analysis of each
amplified sample product. Each real-time PCR reaction was performed in triplicate.
Stability analysis of reference gene expression
The fluorescence raw data were produced and preserved by Bio-Rad CFX96 Manager 3.0
software. The rapid calculation of threshold cycles and amplification efficiency for every gene was
performed. Only genes with amplification efficiency between 90 and 110%, a single peak in
their melting curves and a single band in the agarose gel electrophoretogram were selected for
the following further analysis.
Because no RGs were used in this experiment, relative quantitative data were acquired by
the delta-Cq method for the following geNorm and NormFinder analysis. Delta-Cq data were
calculated in Microsoft Excel 2010 using the lowest Cq value (the highest expression value) as
the calibrator. The relative quantities were imported to the gene expression stability analysis
program geNorm3.5  and NormFinder .
The geNorm VBA applet for Microsoft Excel can calculate a gene expression normalization
factor (NF) for each sample on the basis of the geometric mean of the minimal number of RGs,
Fig 2. The results of agarose gel (1.0%) electrophoresis of RGs. The amplification of a single product of
the expected size for each gene tested in this study is shown.
estimates of an expression stability value (M) for a reference gene as the average pairwise for
that a particular gene with all other tested RGs, and an estimate of the pairwise variation
(Vvalue), Vn/Vn+1, reflecting the effect of additional new genes on the normalization factor. The
lower the M value is, the greater the stability of the RG set. Stepwise removal of the gene with
the lowest expression stability (the highest M value) allows ranking of the tested genes based
on their expression stability. Based on the principle of the lowest number of genes, Vn/Vn+1
with a pairwise variation (V value) lower than 0.15 was accepted as the optimum number of
NormFinder software was also used to identify the optimal normalization gene among a set
of candidate genes. NormFinder uses a solid statistical framework to estimate both inter- and
intra-variations in a given sample set or given experimental design. The best combination of
two RGs was given, the stability value for each gene was provided, and the systematic error
introduced by using the gene for normalization were evaluated by NormFinder . In most
cases, the use of two RGs was more accurate than only one most stable gene. Those candidate
genes with the lowest variation both within and between the groups were considered the most
Determination of the CtFAD2 expression profile
In order to validate whether the most stable genes selected by the geNorm and NormFinder
programs were the most suitable for quantification normalization, different RGs were used for
normalization in the CtFAD2 expression profile. The enzyme encoded by CtFAD2 is primarily
responsible for the synthesis of linoleic acid from oleic acid in seed storage lipids. An usually
large CtFAD2 gene family with 11 members from safflower has been isolated according to a
previous report . Eight of these genes, CtFAD2-1, CtFAD2-2, CtFAD2-3, CtFAD2-4,
CtFAD2-5, CtFAD2-6, CtFAD2-7 and CtFAD2-8, have been cloned in our research group by
Ling-Liang Guan . Seven of the genes were selected for expression analysis in this study.
The primer information and accession numbers of these genes in NCBI are shown in Table 4.
For comparing the differences of the expression profile of seven CtFAD2 genes, the seeds
under cold stress and normal temperature conditions were collected at 10 and 20 DAF,
respectively. The cold stress was the same as the above description.
The mean value of quantified gene expression was calculated from three biological and
three technical repeats per sample. The normalized CtFAD2 gene expression data were
calculated by dividing the target gene’s raw expression value for each sample by the appropriate
In a qPCR experiment, the cycle at which the fluorescent signal is significantly different from
the background was considered the threshold cycle (Ct) value. The Ct value also reflects the
difference in the transcription level in some ways. The mean Ct values for the 15 RGs in the 29
samples under study were used to compare the expression rates within genes and the set of
samples. As shown in Fig 3, the data exhibited a wide range of expression levels from 16.65 to
31.22 in all samples, and the majority of the selected candidate genes exhibited Ct values
ranging from 18.83 to 25.83. Within all samples, the expression variation of a single gene was lower
in EF1, EF1β, UBCE2, and EIF5A and higher in MFC, COA, MBF1 and GAPDH. The MFC
with the variation value of 8.62 and the EIF5A with the variation of 2.24 were the highest and
lowest variation gene, respectively. The genes with lower expression levels were MFC,
RANBP1, and COA with mean Ct values of 27.82, 25.92, and 25.53, respectively. The genes
Nucleotide sequence (50-30)
with higher expression levels were EIF5A, GAPDH, and EF1 with mean Ct values 20.25, 19.96,
and 18.30, respectively. Box-plot analysis only provides an approximate estimate of the stability
of candidate RGs, so it is necessary to perform a further analysis in order to identify the most
suitable reference gene or the best combination of genes for qPCR experiments under different
Stability analysis of expression
geNorm software was used for stability analysis of gene expression under different sets of
samples: 1) ABA stress samples, 2) salt stress samples, 3) cold stress samples, 4) GA3 stress samples,
5) PP333 stress samples, 6) different cultivars, 7) different seed development stages, 8) different
tissues, and 9) total samples. Based on the mean expression stability value (M), 15 candidate
genes were ranked under different experimental conditions and are shown in Fig 4. High
Fig 3. Threshold cycle (Ct) values of the candidate RGs across the experimental samples. A box-plot
graph of Ct values shows the median values as lines across the box. Upper and lower boxes indicate the 75th
percentile to the 25th percentile. Whiskers represent the maximum and minimum values.
expression stability values with an M value lower than the cut-off (M 0.15) established by
Vandesompele et al  for 15 genes are presented. When all of the samples were taken
together, the average expression stability value (M) for 60SRPL10 + EF1 was the lowest, and
that of the MFC was the highest. This indicated that 60SRPL10 + EF1 had the most stable
expression and MFC was the most variably expressed. In addition, 60SRPL10 + EF1 had the
lowest average expression stability value, and MBF1 had the highest M value both under the
cold stress and PP333 stress. For salt stress samples, UBCE2 + MBF1 present the lowest
expression stability value, and COA had the highest. When only different cultivates were considered,
the UBCE2 + EF1β were the most stable genes and GAPDH was the most unstable gene in
stability analysis. For GA3 stress samples, ATPS + GST was the most stably expressed among the
15 RGs, and MFC was the most variably expressed. For ABA stress samples, the most stable
combination was EF1 + ATPS, and MFC was the most unstable gene. For different seed
development stage samples, 60SRPL10 + ATPS were considered the most stable combination of
genes and MBF1 was the most unstable gene according to the geNorm analysis results. Finally,
when different tissues were considered for stability analysis, ABCS + 60SRPL10 was the most
stable combination of RGs, and UBCS was the most unstable gene.
The optimal number of RGs was also calculated by geNorm automatic analysis for accurate
normalization. Under different experimental conditions, the optimal number of RGs may be
different than the results of pairwise variation analysis (Fig 5). Except for the total samples and
the different tissue samples, the V2/3 values of the other sets of samples were all lower than the
cut-off of 0.15, indicating that it is sufficient to use two RGs for accurate normalization.
However, when pairwise variation was evaluated in all samples, the V2/3 value was 0.158, the V3/4
value was 0.205, and the V4/5 value was 0.182. The V5/6 value was 0.15, so the five most stable
RGs, including 60SRPL10, EF1, UBCE2, EIF5A and ATPS, should be used for more accurate
normalization. For different tissues samples, the V3/4 value was 0.172 and the V 4/5 value was
0.135, so it was better to use four RGs for normalization.
NormFinder was also used to analyze the expression stability and to rank the stability value
from high to low; the ranking of all candidate RGs in different set of samples is shown in
Table 5 and Table 6. When all 29 samples sets were considered, EIF5A was the most stable
gene with a stability value of 0.060, and the best combination of two genes was EF1 + EIF5A
with a stability value of 0.050. For ABA stress samples, the best gene was UBCE2 with a stability
value of 0.180, and the best combination of two genes was 60SRPL10 + UBCE2 with a stability
value of 0.141. For GA3 and PP333 stress samples, the best combinations of two genes were GST
Fig 4. Gene expression stability values (M) of the candidate RGs calculated by geNorm. The figure
shows the ranking of the gene expression stability performed in ABA stress, GA3 stress, paclobutrazol stress,
cold stress, different cultivars, different development stages, and total samples. The most stable genes are
on the right, and the least stable genes are on the left. GA3: gibberellin acid. ABA: abscisic acid.
Fig 5. Gene expression pairwise variation (V) of the candidate RGs calculated by geNorm. Pairwise
variation (Vn/Vn+1) was analyzed between the normalization factors NFn and NFn+1. Asterisk indicates the
optimal number of RGs required for normalization. ABA (abscisic acid stress), GA3 (gibberellin stress), LT
(low temperature), Paclobutrazol (paclobutrazol stress), Total (total samples), DC (different cultivates), DDS
(different development stages), Salt (salt stress), DT (different tissues).
+ ATPS and ATPS + UBCE2 with a stability value of 0.014 and 0.060, respectively. ATPS
represented the best stability in the previous two sets of samples. For salt stress, the best combination
of genes was EF1 + GTPB with a stability value of 0.093, and the best gene was EF1 with a
stability value of 0.132. When only low temperature stress samples were considered, NormFinder
identified that the most stable expression gene was EIF5A with a stability value of 0.079, and
the best combination of two genes was EIF5A + 60SRPL10 with a stability value of 0.082.
Finally, because the different cultivars, different seed development stages samples and different
tissue samples had no control samples and NormFinder raw data cannot be divided into two
groups, NormFinder did not calculate the most stable combination of two genes for all
candidate RGs (Table 2 and Table 3). Thus, when different seed development stage samples and
different cultivars were considered, EF1β + UBCE2 were the most stable RGs with stability values
of 0.025 and 0.053, respectively. For different tissue samples, MBF1 + RAN were the most stable
genes and MFC was the least stable gene in ABA, GA3, and all the sets of samples. Under the
different seed development stages, cold and PP333 stress samples, MBF1 had the highest
stability value. COA + GAPDH were the least stable genes under salt stress and in different cultivate
samples, respectively. In addition, UBQL was considered the most unstable gene for different
tissue samples by NormFinder.
RG validation by quantification of the CtFAD2 expression profile with
different normalization factors
In order to validate whether the most stable genes selected by geNorm and NormFinder
programs were the most suitable for quantification normalization, different RGs were used for
CtFAD2 is the key gene that controls the formation of linoleic acid. Seven CtFAD2 genes
selected in this study had been identified, and their sequence had been submitted in NCBI for
safflower. Under cold stress, the content of accumulated polyunsaturated fatty acids (PUFA) in
plant tissues will usually increase significantly [47–48].
60SRPL10 + EF1 was selected as the most stable combination of genes by geNorm
algorithms, and 60SRPL10 + EIF5A were identified as the most stable genes by NormFinder and
were used in expression analysis of 7 CtFAD2 genes under cold stress for normalization (Fig
6A and 6B). MBF1, as the least stable gene identified by both geNorm and NormFinder, was
also used for normalization to further verify whether the use of unstable RGs can lead to an
inaccurate result for quantification expression analysis (Fig 6C). The results showed that the
fold changes of CtFAD2-1, CtFAD2-2, CtFAD2-3, CtFAD2-4, CtFAD2-5, and CtFAD2-6 were
all greater than 1 at 10 DAF, indicating that the expression of these genes under cold stress in
safflower seeds was higher than in the control (seeds under normal temperature). These results
were in accordance with the result that the relative percentage content of linoleic acid in
safflower seed had an obvious increase at 10 DAF under cold stress according to fatty acid analysis
by GC-MS. Fig 6C shows that, under cold stress, the expression of all genes was nearly zero at
10 DAF and that the fold change of all genes except for CtFAD2-1 was extremely high
compared with the control at 20 DAF. Overall, it was nearly similar to the expression pattern of 7
CtFAD2 gene under cold stress at 10 and 20 DAF regardless of whether 60SRPL10 + EF1 or
60SRPL10 + EIF5A was used for normalization (Fig 6A and 6B). However, when MBF1 was
used as the reference gene for normalization, the result was obviously different from those
mentioned previously. Unstable RGs can lead to the NF value becoming larger or smaller,
corresponding to the relative expression value of target genes becoming smaller or larger,
respectively. Therefore, it is very necessary to use an appropriate reference gene for relative gene
With the development of molecular biology techniques, such as the transcriptome, microarray,
digital gene expression profile and real-time quantitative PCR, investigation of gene expression
patterns at the transcript level has become feasible. RT-qPCR has become a frequently used
technology for specific gene expression analysis .
Note: DDS: different development stage, DFC: different cultivars, and DT: different tissues.
Although EF1α and UBQ were used as reference genes in safflower gene expression analysis
with the RT-qPCR technique, the validation of a suitable set of RGs is still of crucial
importance [48–50]. This safflower transcriptome analysis identified a group of stably expressed
genes under different seed development stages, and these data resources made it convenient for
us to select some relatively stable candidate RGs for normalization in order to further identify
and evaluate the most appropriate RGs for stable expression across a variety of experimental
conditions. This work is the first particular study aimed at the identification and validation of a
series of candidate reference genes for normalization of transcription in safflower. For this
study, some candidate RGs that had a relatively more stable and higher expression level based
on the transcriptome data were selected in safflower. After determining the AGE and solubility
curves, 15 candidate RGs were selected in order to validate their normalization potential
among 29 samples under different abiotic stresses and at different developmental stages.
In the present study, geNorm and NormFinder were used to select the best RGs for the
normalization of gene expression data in safflower. It is acceptable that there were differences
between analysis results regarding the ranking of the candidate RGs from the two programs
because of their differences in mathematical approaches. Although the differences in the
ranking of genes existed, the most and the least stable genes were nearly consistent under different
sets of samples from the two distinct program analysis results. MFC was always the least stable
gene, whether in ABA, GA3-sprayed samples or total samples from the two analysis methods.
In addition, MBF1 was the most unstable gene in cold stress, PP333 stress and different seed
development stages samples. Although GAPDH was commonly used as the RG for
normalization in gene expression analysis, in our study, it was the least stable gene in different varieties of
safflower, and COA was the least stable gene in salt stress samples. This result indicated that
the common RGs for normalization should be used carefully before an evaluation for every set
of samples under different experimental conditions. From a single gene perspective, combined
Fig 6. Relative quantification of 7 CtFAD2 genes using different combinations of RGs or single RGs
for normalization. a. Gene expression normalized with the most stable RGs selected by geNorm for the low
temperature stress samples. b. Gene expression normalized with the most stable RGs selected by
NormFinder for low temperature stress samples. c. Gene expression normalized with the most unstable
genes selected by both geNorm and NormFinder for low temperature stress samples.
with the results of two methods, EF1, EIF5A, ATPS, UBQE2 and 60SRPL10 were strongly
recommended as RGs for relative quantification of gene expression in safflower. In addition, EF1
had been selected as the best reference gene in switchgrass , strawberry , perennial
ryegrass , rice  lolium  and Brachiaria . However, there were some differences in
the best combination of genes for normalization from the analysis results of the two programs.
60SRPl10 + EF1 was considered the best combination of candidate genes for normalization
under total samples, paclobutrazol stress samples, and cold stress samples by geNorm analysis.
However, in the results with NormFinder, EF1 + EIF5A was identified as the best combination
among total samples, EIF5A + 60SRPL10 was identified as the best combination in cold stress
samples and ATPS + UBCE2 was identified as the best combination in paclobutrazol stress
samples. Under ABA stress, EF1 + ATPS was identified as the best combination by geNorm
and 60SRPL10 + UBCE2 was identified as the best combination by NormFinder. UBCE2
+ MBF1 was the best combination by geNorm, and EF1 + GTPB was identified as the best
combination by NormFinder under salt stress. Finally, ATPS + GST was identified as the best
combination of genes for normalization by both geNorm and NormFinder programs.
To verify the effectiveness of selected RGs, the seven copies of the CtFAD2 gene as target
gene was used to evaluate RGs efficiency for normalization in cold stress samples. Similar
expression patterns appeared when two stable combinations of candidate genes analyzed were
used for normalization by geNorm and NormFinder. However, when the least stable gene was
used for normalization, the result was obviously different from the previous result. At 10 and
20 DAF, the expression of CtFAD2 in safflower seeds was lower and highly significant,
respectively. The standardized expression quantity of the target gene was acquired by dividing by the
normalization factor (NF) of reference gene, so the lower or higher NF can directly lead to an
exponential increase or decrease in the expression level. Therefore, the identification and
stability evaluation of RGs should be made prior to their application in transcription normalization
studies of gene expression. Our study may become a reference for high-efficiency identification
and evaluation of RGs based on transcriptome data for other researchers and with other
species. Meanwhile, these results also provide a convenient reference for investigations of the
expression patterns of key genes of interest in safflower.
This work was financially supported by the National Natural Science Foundation of China
Conceived and designed the experiments: DDL BH WW. Performed the experiments: DDL
BH QW. Analyzed the data: DDL HCL. Wrote the paper: DDL WW FP BH.
1. Li DJ , Zhou MD , Ramanatha Rao V ( 1993 ) Characterization and evaluation of safflower germplasm . Beijing:geological publishing house . 260 p.
2. Mirzajani F , Bernard F , Zeinali SM , Goodarzi R ( 2015 ) Identification of hydroxy-safflor yellow A, safflor yellow B, and precarthaminin safflower using LC/ESI-MSMS . Journal of Food Measurement and Characterization . doi: 10.1007/s11694- 015 - 9240 -9
3. Li DJ ( 1993 ) Progress of safflower (Carthamus tinctorius L.) research and production in China . In Third International Safflower Conference , Beijing, China, pp 35 - 46 .
4. Fernandez-Martinez J , Rio M and Haro A ( 1993 ) Survey of safflower (Carthamus tinctorius L.) germplasm for variants in fatty acid composition and other seed characters . Euphytica 69 : 115 - 122 .
5. Futehally S and Knowles PF ( 1981 ) Inheritance of very high levels of linoleic acid in an introduction of safflower (Carthamus tinctorius L.) from Portugal . In First International safflower conference, Davis, California, pp 56 - 61 .
6. Guan LL , Wu W , Zheng YL ( 2008 ) Seed oil contents and fatty acid compositions of seed oil from different safflower accessions introduced into Yaan and the correlation analysis with the agronomic traits and photosynthetic parameters . Philipp Agric Sci 91 ( 4 ): 383 - 388 .
7. Hamdan YAS , Perez-Vich B , Velasco L , Fernandez-Martinez JM ( 2008 ) Inheritance of high oleic acid content in safflower . Euphytica 168 ( 1 ): 61 - 69 .
8. Guan LL , Wang YB , Shen H , Hou K , Xu YW , Wu W. ( 2012 ) Molecular cloning and expression analysis of genes encoding microsomal oleate desaturases (FAD2) from safflower ( Carthamus tinctorius L.). Plant Mol Biol Rep 30 : 139 - 148 .
9. Guan LL , Xu YW , Wang YB , Chen L , Shao JF , Wu W ( 2012 ) Isolation and Characterization of a Temperature-Regulated Microsomal Oleate Desaturase Gene (CtFAD2-1) from Safflower (Carthamus tinctorius L.). Plant Mol Biol Rep 30 ( 2 ): 391 - 402 .
10. Bustin SA , Benes V , Nolan T , Pfaffl MW ( 2005 ) Quantitative real-time RT-PCR a perspective . J Mol Endocrinol 34 : 597 - 601 . PMID: 15956331
11. Czechowski T , Bari RP , Stitt M , Scheible W- R , Udvardi MK ( 2004 ) Real-time RT-PCR profiling of over 1400 Arabidopsis transcription factors: unprecedented sensitivity reveals novel root- and shoot-specific genes . Plant J 38 : 377 - 379 .
12. Gachon C , Mingam A , Charrier B ( 2004 ) Real-time PCR: what relevance to plant studies . J Exp Bot 55 : 1445 - 1454 . PMID: 15208338
13. Bustin SA ( 2002 ) Quantification of mRNA using real-time reverse transcription PCR(RT-PCR):trends and problems . J Mol Endocrinol 29 : 23 . PMID: 12200227
14. Ginzinger DG ( 2002 ) Gene quantification using real-time quantitative PCR: An emerging technology hits the mainstream . Exp Hematol 30 : 503 - 512 . PMID: 12063017
15. Silver N , Best S , Jiang J , Thein SL ( 2006 ) Selection of housekeeping genes for gene expression studies in human reticulocytes using real-time PCR . BMC Mol Biol 7 : 33 . PMID: 17026756
16. Schmittgen TD , Zakrajsek BA ( 2000 ) Effect of experimental treatment on housekeeping gene expression: validation by real-time, quantitative RT-PCR . J Biochem Bioph Meth 46 : 69 - 81 .
17. Thellin O , Zorzi W , Lakaye B , De Borman B ( 1999 ) Housekeeping genes as internal standards: use and limits . J Biotechnol 75 : 291 - 295 . PMID: 10617337
18. Cappelli K , Felicetti M , Capomaccio S , Spinsanti G , Silvestrelli M , Supplizi AV. ( 2008 ) Exercise induced stress in horses: Selection of the most stable reference genes for quantitative RT-PCR normalization . BMC Mol Biol 9 : 49 ( 1 ). doi: 10.1186/1471-2199-9-49 PMID: 18489742
19. Goidin D , Mamessier A , Staquet M-J , Schmitt D , Berthier-Vergnes O ( 2001 ) Ribosomal 18S RNA Prevails over Glyceraldehyde-3-Phosphate Dehydrogenase and β-Actin Genes as Internal Standard for Quantitative Comparison of mRNA Levels in Invasive and Noninvasive Human Melanoma Cell Subpopulations . Anal Biochem 295 : 17 - 21 . PMID: 11476540
20. Kouzaki H , O'Grady SM , Lawrence CB , Kita H ( 2009 ) Proteases induce production of thymic stromal lymphopoietin by airway epithelial cells through protease-activated receptor-2 . J Immunol 183 : 1427 . doi: 10.4049/jimmunol.0900904 PMID: 19561109
21. Radonic A , Thulke S , Mackay IM , Landt O , Siegert W , Nitsche A. ( 2004 ) Guideline to reference gene selection for quantitative real-time PCR . Biochem Bioph Res Co 313 : 856 - 862 .
22. Tong Z , Gao Z , Wang F , Zhou J , Zhang Z ( 2009 ) Selection of reliable reference genes for gene expression studies in peach using real-time PCR . BMC Mol Biol 10 : 71 . doi: 10.1186/ 1471 - 2199 - 10 -71 PMID: 19619301
23. Guénin S , Mauriat M , Pelloux J , Van Wuytswinkel O , Bellini C , Gutierrez L. ( 2009 ) Normalization of qRT-PCR data: the necessity of adopting a systematic, experimental conditions-specific, validation of references . J Exp Bot 60 : 487 - 493 . doi: 10.1093/jxb/ern305 PMID: 19264760
24. Kim BR , Nam HY , Kim SU , Kim SI , Chang YJ ( 2003 ) Normalization of reverse transcription quantitative-PCR with housekeeping genes in rice . Biotechnol Lett 25 : 1869 - 1872 . PMID: 14677714
25. Gutierrez L , Mauriat M , Guénin S , Pelloux J , Lefebvre JF , Louvet R , et al. ( 2008 ) The lack of a systematic validation of reference genes: a serious pitfall undervalued in reverse transcription-polymerase chain reaction (RT-PCR) analysis in plants . Plant Biotechnol J 6 : 609 - 618 . doi: 10.1111/j.1467- 7652 . 2008 .00346.x PMID: 18433420
26. Thellin O , Zorzi W , Lakaye B , De Borman B , Coumans B , Hennen G , et al. ( 1999 ) Housekeeping genes as internal standards: use and limits . J Biotechnol 75 : 291 - 295 . PMID: 10617337
27. Artico S , Nardeli S , Brilhante O , Grossi-de-Sa M , Alves-Ferreira M ( 2010 ) Identification and evaluation of new reference genes in Gossypium hirsutum for accurate normalization of real-time quantitative RTPCR data . BMC Plant Biol 10 : 49 . doi: 10.1186/ 1471 - 2229 - 10 -49 PMID: 20302670
28. Nicot N , Hausman JF , Hoffmann L , Evers D ( 2005 ) Housekeeping gene selection for real-time PCR normalization in potato during biotic and abiotic stress . J Exp Botany 56 ( 421 ): 2907 - 2914 .
29. Zhu XY , Li XP , Chen WX , Chen JY , Lu WJ , Chen L , et al. ( 2012 ) Evaluation of New Reference genes in Papaya for Accurate Transcript Normalization under Different Experimental Conditions . PloS One 7 ( 8 ) : e44405 . doi: 10.1371/journal. pone.0044405 PMID: 22952972
30. Mafra V , Kubo KS , Alves-Ferreira M , Ribeiro-Alves M , Stuart RM , Boava LP , et al. ( 2012 ) Reference genes for Accurate Transcript Normalization in Citrus Genotypes under Different Experimental Conditions . PloS One 7: e31263. doi: 10.1371/journal.pone.0031263 PMID: 22347455
31. Huang L , Yan H , Jiang X , Zhang XQ , Zhang YW , Huang X , et al. ( 2014 ) Evaluation of Candidate Reference genes for Normalization of Quantitative RT-PCR in Switchgrass Under Various Abiotic Stress Conditions . BioEnerg Res 7 ( 4 ): 1201 - 1211 .
32. Amil-Ruiz F , Garrido-Gala J , Blanco-Portales R , Folta KM , Munoz-Blanco J , Caballero JL ,. ( 2013 ) Identification and Validation of Reference genes for Transcript Normalization in Strawberry (Fragaria × ananassa) Defense Responses . PLoS ONE 8 ( 8 ) : e70603 . doi: 10.1371/journal. pone.0070603 PMID: 23940602
33. Cruz F , Kalaoun S , Nobile P , Colombo C , Almeida J , Barros LMG , et al. ( 2009 ) Evaluation of coffee reference genes for relative expression studies by quantitative real-time RT-PCR . Mol Breeding 23 ( 4 ): 607 - 616 .
34. Die JV , Román B , Nadal S , González-Verdejo CI ( 2010 ) Evaluation of candidate reference genes for expression studies in Pisum sativum under different experimental conditions . Planta 232 ( 1 ): 145 - 153 . doi: 10.1007/s00425- 010 - 1158 - 1 PMID: 20379832
35. Condori J , Nopo-Olazabal C , Medrano G , Medina-Bolivar F ( 2011 ) Selection of reference genes for qPCR in hairy root cultures of peanut . BMC Res Notes 4: 392. doi: 10.1186/1756-0500-4-392 PMID: 21985172
36. Schmidt GW , Delaney SK ( 2010 ) Stable internal reference genes for normalization of real-time RTPCR in tobacco (Nicotiana tabacum) during development and abiotic stress . Mol Genet Genomics 283 : 233 - 241 . doi: 10.1007/s00438- 010 - 0511 - 1 PMID: 20098998
37. Exposito-Rodriguez M , Borges AA , Borges-Pérez A , Pérez JA ( 2008 ) Selection of internal control genes for quantitative real-time RT-PCR studies during tomato development process . BMC Plant Biol 8 : 131 . doi: 10.1186/ 1471 - 2229 -8-131 PMID: 19102748
38. Hu RB , Fan CM , Li HY , Zhang QZ , Fu YF. ( 2009 ) Evaluation of putative reference genes for gene expression normalization in soybean by quantitative real-time RT-PCR . BMC Mol Biol 10 : 93 . doi: 10. 1186/ 1471 - 2199 - 10 -93 PMID: 19785741
39. Czechowski T , Stitt M , Altmann T , Udvardi MK , Scheible WR ( 2005 ) Genome-Wide Identification and Testing of Superior Reference genes for Transcript Normalization in Arabidopsis . J Plant Physiol 139 : 5 - 17 .
40. Vandesompele J , De Preter K , Pattyn F , Poppe B , Van Roy N , De Paepe A , et al. ( 2002 ) Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes . Genome Biol 3 : 1 .
41. Andersen CL , Jensen JL , Rntoft TF ( 2004 ) Normalization of real-time quantitative reverse transcriptionPCR data: a model-based variance estimation approach to identify genes suited for normalization, applied to bladder and colon cancer data sets . Cancer Res 64 : 5245 - 5250 . PMID: 15289330
42. Huang XQ , Madan A ( 1999 ) CAP3: A DNA sequence assembly program . Genome Res 9 : 868 - 877 . PMID: 10508846
43. geNorm software website . Available: http://medgen.ugent.be/~jvdesomp/genorm/. Accessed 2011 Jan, 11 .
44. NormFinder software website . Available: http://www.mdl.dk/publicationsnormfinder.htm. Accessed 2011 Jan, 11 .
45. Cao SJ , Zhou XR , Wood CC , Green AG , Singh SP , Liu XL , et al. ( 2013 ) A large and functionally diverse family of Fad2 genes in safflower ( Carthamus tinctorius L.). BMC Plant Biol 13 : 5. doi: 10.1186/ 1471 - 2229 - 13 - 5 PMID: 23289946
46. Guan LL ( 2011 ) Accumulation pattern and regulatory mechanisms of fatty acid in different safflower ( Carthamus tinctorius L.) tissues (D) . Sichuan: Sichuan Agriculture University.
47. Sakamoto T , Murata N ( 2002 ) Regulation of the desaturation of fatty acids are required to maintain chloroplast function at low temperatures . Plant Physiol 124 : 1697 - 1705 .
48. Wu Q , Liu T , Liu H , Zheng G ( 2009 ) Unsaturated fatty acid: Metabolism, synthesis and gene regulation . African J Biotechnol 8 ( 9 ): 1782 - 1785 .
49. Bustin SA ( 2000 ) Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays . J Mol Endocrinol 25 : 169 - 193 . PMID: 11013345
50. Guan LL , Wu W , Hu B , Li D , Chen JW , Hou K , et al. ( 2014 ) Developmental and growth temperature regulation of omega-3 fatty acid desaturase genes in safflower ( Carthamus tinctorius L.). Genet Mol Res 13 ( 3 ): 6623 - 6637 . doi: 10.4238/ 2014 . August.28.7 PMID: 25177943
51. Lee JM , Roche JR , Donaghy DJ , Thrush A , Sathish P ( 2010 ) Validation of reference genes for quantitative RT-PCR studies of gene expression in perennial ryegrass ( Lolium perenne L.). BMC Mol Biol 11 : 14 .
52. Jain M , Nijhawan A , Tyagi AK , Khurana JP ( 2006 ) Validation of housekeeping genes as internal control for studying gene expression in rice by quantitative realtime PCR . Biochem Bioph Res Co 345 ( 2 ): 646 - 651 .
53. Dombrowski J , Martin R ( 2009 ) Evaluation of reference genes for quantitative RT-PCR in Lolium temulentum under abiotic stress . Plant Science 176 ( 3 ): 390 - 396 .
54. Silveira ED , Alves-Ferreira M , Guimaraes LA , da Silva FR , Carneiro VTD ( 2009 ) Selection of reference genes for quantitative real-time PCR expression studies in the apomictic and sexual grass Brachiaria brizantha . BMC Plant Biol 9 : 10 .