ISSN 00268933, Molecular Biology, 2013, Vol. 47, No. 6, pp. 836851. Pleiades Publishing, Inc., 2013.

Published in Russian in Molekulyarnaya Biologiya, 2013, Vol. 47, No. 6, pp. 959975.

GENOMICS.
TRANSCRIPTOMICS
UDC 577.214.6

Selection of Reference Genes for Quantitative RealTime PCR

in Six OilTea Camellia Based on RNAseq1
C. F. Zhou*, P. Lin*, X. H. Yao, K. L. Wang, J. Chang, and X. J. Han
Research Institute of Subtropical Forestry of the Chinese Academy of Forestry, Fuyang, Hangzhou, China 311400;
email: yaoxh168@163.com; zhouchangfu555@126.com
Received February 28, 2013; in final form, May 28, 2013

AbstractqRTPCR is becoming a routine tool in molecular biology to study gene expression. It is necessary
to find stable reference genes when performing qRTPCR. The expression of genes cloned in oiltea camellia
currently cannot be accurately analyzed due to a lack of suitable reference genes. We collected different tissues
(including roots, stems, leaves, flowers and seeds) from six oiltea camellia species to determine stable reference
genes. Five novel and ten traditional reference gene sequences were selected from the RNAseq database of
Camellia oleifera Abel seeds and specific PCR Primers were designed for each. Cycle threshold (Ct) data were
obtained from each reaction for all samples. Three different software tools, geNorm, NormFinder and Best
Keeper were applied to calculate the expression stability of the candidate reference genes according to the Ct
values. The results were similar between the three software packages, and indicated that the traditional genes
TUB3, ACT7 and the novel gene CESA were relatively stable in all species and tissues. However, no genes
were sufficiently stable across all species and tissues, thus the optimal number of reference genes required for
accurate normalization varied from 2 to 6. Finally, the relative expression of squalene synthase (SQS) and
squalene epoxidase (SQE) genes related to important ingredients squalene and tea saponin in oiltea camellia
seeds were compared by using stable to less stable reference genes. The comparison results validated the selec
tion of reference genes in the current study. In summary, for the different tissues of six oiltea camellia species
different optimal numbers of suitable reference genes were found.
DOI: 10.1134/S0026893313060198
Keywords: reference genes, realtime PCR, oiltea camellia

INTRODUCTION
The analysis of gene expression has become
increasingly important in biological research. Many
different research methods are now used to study gene
expression, such as northern blotting, serial analysis of
gene expression (SAGE), semiquantitative PCR
(semiPCR), gene microarrays, quantitative realtime
PCR (qRTPCR), RNase protection analysis (RPA);
and RNAseq [13]. qRTPCR is becoming a routine
tool in molecular biology to study gene expression,
and is a PCRbased technique for analysis of mRNA
expression in various biological samples. Compared
with other gene expression analysis methods, it is
highly sensitive and useful even for genes with low
expression levels or those which show small expression
changes [4]. At the same time, it is easy to perform and
enables rapid quantification. However, normalization
is required for accurate interpretation of results. Sev
eral strategies have been proposed for normalizing
qRTPCR data, including choosing samples of a sim
ilar size, quantifying RNA relative to genomic DNA,
and using an internal housekeeping or reference gene.
1 The article is published in the original.
* Zhou C.F. and Lin P. made an equal contribution to the study,
and should be regarded as joint first authors.

Normalizing to a reference gene is a simple and popu

lar method for internally controlling for error in qRT
PCR [5]. Ideal reference genes are expected to have a
stable expression level across various experimental
conditions, such as plant developmental stages, tissue
types and external stimuli. The most commonly used
reference genes include actin (ACTB), 18S ribosomal
RNA, translation elongation factor1 (TEF1) and glyc
eraldehyde3phosphate dehydrogenase (GAPDH)
[612]. However, the level of transcript expressed
from such genes is not always stable under all experi
mental conditions, which may lead to the misinterpre
tation of results.
Species of the genus Camellia whose seeds can pro
duce oil are referred to as oiltea camellia. These spe
cies are very important woody oil trees that have been
cultivated for a long time and are widely distributed in
southern China [13]. The oil extract from oiltea
camellia seeds is named tea oil and is extensively used
throughout China. Tea oil is regarded as one of the
healthiest edible vegetable oils in the world, with nutri
tional composition superior to olive oil. The content of
unsaturated fatty acids, such as oleic and linoleic acid
is above 90%. It is also very rich in other nutrients such
as squalene, sterols, vitamins and tocopherol.
Recently, improvement of high oil producing varieties

836

SELECTION OF REFERENCE GENES FOR QUANTITATIVE REALTIME PCR

has become one of the most important areas of camel

lia scientific research. Selection and hybridization are
the most popular and ageold methods for breeding,
but, the progress require more time and resources.
Therefore, improving the production and quality of
tea oil using molecular approaches should become
especially necessary in the future. So far only the evo
lutionary relationships of the Camellia species have
been studied using molecular markers [14], also an
EST library has been constructed for camellia seed
[15] and transcriptome sequencing of different seed
developmental stages was finished [16]. Many func
tional genes have been cloned, such as the Dehydrin
like Protein [17], metallothionein gene [18], FAD2
gene [19], SAD gene [20], calmodulin genes [21], and
an aldoketo reductase [22]. However, analyses of
gene expression in different species, tissues or growth
stages could not be performed because no suitable ref
erence genes for genus Camellia were found. There
fore, the identification of reliable reference genes in
Camellia has become a crucial factor to allow accurate
functional gene expression analysis. In this study, we
aimed at identifying potential reference genes suitable
for transcript normalization in different tissues of six
camellia species. These reference genes will allow
more accurate and reliable qRTPCR normalization
for gene expression studies in the genus Camellia.
EXPERIMENTAL
Plant materials and biological samples. Six eco
nomically important and widely cultivated oiltea
camellia species were selected from the Camellia Ger
mplasm Collection Park at the Subtropical Forestry
Research Institute of the Chinese Academy of For
estry. These six species were Zhejiang (C. chekiango
leosa Hu), Guangning (C. semiserrata C.W. Chi),
Xiaoguo (C. meiocarpa Hu), Wantian (C. polyodonta
F.C. How), Youxian (C. yuhsienensis Hu) and Putong
(C. oleifera C. Abel). Five different tissues, including
roots, stems, leaves, flowers and seeds, were collected
from six adult trees of each species. Ten samples of
each tissue were picked from those 36 trees. Vegetative
tissue samples, such as roots, leaves and stems were
taken from young tissues in March. Flowers were har
vested at full bloom; the flowers of Putong and
Xiaoguo were harvested in November, and the flowers
of the other species were harvested in March the fol
lowing year. Fruits at the fast growth stage were col
lected in July and peeled to extract the seeds immedi
ately. All samples were harvested at about 10:00 in the
morning and cut into small pieces. For each species
tissue samples from six trees were mixed together.
Plant materials were frozen in liquid nitrogen immedi
ately after being harvested and stored in a freezer at
80C until the total RNA was isolated.
RNA extraction and cDNA synthesis. The mixed
five tissue samples from each species were taken from
the 80C freezer and ground into fine powder in a
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

837

mortar with liquid nitrogen. About 100 mg of the pow

der was used for RNA extraction. Total RNA from all
samples was prepared using the RN38 EASYspin plus
Plant RNA kit (Aidlab Biotech, Beijing, China)
according to the manufacturers instructions. A
Nanodrop 2000 microvolume spectrophotometer was
used to detect the RNA concentration and quality
(Table 1). Only RNA samples with a 260/280 wave
length ratio of 1.82.1 and a 260/230 wavelength ratio
between 0.1 and 1 were used for cDNA synthesis, sev
eral samples with the 260/230 values greater than
1 were diluted. cDNA synthesis was performed with
38 L total RNA (the final content of RNA in the
reaction mixture was adjusted to about 1 g for all
samples) according to the protocols of the Super
Script FirstStrand Synthesis System (Invitrogen,
Carlsbad, CA. USA) in a total volume of 20 L.
Finally, the cDNA was diluted 1 : 15 with nuclease
free water for qRTPCR [23].
Selection of candidate reference genes based on
RNASeq and primer design. RNASeq of Putong
oiltea camellia seeds from July to October were
sequenced by using solexa technology. Total number of
reads, Total Nucleotides, Q20 percentage, gap per
centage, GC percentage, number of contig, Scaffold
and Unigenes in each month are listed in Table 2.
There were 8310777 reads in average of RNASeq
from July to October. Only 0.04% of the gaps were not
sequenced. Transcriptome assembly was carried out by
short reads assembly software SOAP denovo [24],
detailed assembling steps are presented in Fig. 1. The
assembling results showed 43461, 47932, 30022 and
39251 unigenes respectively from July to October.
After the repeated unigenes were removed, a total of
80310 unigenes were detected according to this RNA
Seq of Putong oiltea camellia seeds. 21789 unigenes
with protein function annotations were generated.
The unigenes with function annotations can be classi
fied into twentyfour functionalcategories and
account for 27.13% of all unigenes. All unigenes were
queried against the KEGG pathway database, and
42638 unigenes were given the pathway annotations
related to 265 pathways, including metabolism,
genetic information processing, cell metabolism path
ways and so on.
RPKM values (Reads per kb per Million reads)
[25] of unigenes were used to analyze genes expression
in RNASeq databases. The least fluctuation in
RPKM values among different growth periods or tis
sues may represent the most stably expressed genes.
Coefficient variations of the RPKM values from July
to October were used to determine gene expression
fluctuations. The coefficient variation was calculated
as the Mean divided by the Standard deviation. All
unigenes of the two dozen traditional reference genes
and several reported novel reference genes were taken
from the RNASeq database. Fifteen unigenes with
the least RPKM values variations among the four seed
development stages were selected (Table 3), and

838

ZHOU et al.

Table 1. RNA concentration and quality detected by Nanodrop 2000

Species

Tissues

Nucleic acid conc., ng/L

A260

A280

260/280

260/230

Zhejiang

Roots
Stems
Leaves
Flowers
Seeds

208.60
308.30
266.50
326.90
496.90

5.21
7.70
3.40
8.10
12.42

2.91
3.56
1.60
3.95
5.87

1.79
2.16
2.13
2.12
2.12

0.73
1.46
0.43
0.62
1.14

Guangning

Roots
Stems
Leaves
Flowers
Seeds

107.70
225.60
647.60
267.30
319.90

2.64
5.64
11.36
9.60
8.00

1.34
2.83
5.78
4.53
3.82

1.45
2.00
1.97
2.12
2.09

0.47
1.26
1.24
0.75
1.12

Wantian

Roots
Stems
Leaves
Flowers
Seeds

170.60
122.80
289.90
384.10
711.20

4.25
3.07
4.30
9.60
17.78

2.20
1.51
2.06
4.53
8.25

1.93
2.03
1.11
2.12
2.15

1.09
0.74
0.41
1.06
1.59

Youxian

Roots
Stems
Leaves
Flowers
Seeds

136.80
242.50
334.90
174.20
590.40

3.42
6.06
8.06
4.17
14.77

1.77
2.86
3.91
2.16
7.49

1.94
2.12
2.06
1.93
1.97

1.08
1.4
1.31
0.59
1.21

Putong

Roots
Stems
Leaves
Flowers
Seeds

448.10
446.80
254.30
158.60
493.00

9.76
12.17
5.73
3.97
12.33

4.99
6.28
2.84
1.99
5.82

1.96
1.78
2.02
1.99
2.12

0.67
0.82
1.02
0.31
0.81

Xiaoguo

Roots
Stems
Leaves
Flowers
Seeds

378.90
410.20
401.50
132.60
276.40

8.69
11.63
10.24
3.57
5.89

4.18
5.77
5.38
1.84
3.07

2.08
2.02
1.90
1.94
1.92

1.01
1.21
0.73
0.68
0.92

included nine traditional references and two reported

novel references. The nine traditional genes were
Actin 7 (7), Cyclophilin B (CYPB), Tublin 3
(TUB3), TATA box banding protein component of
TFIID and TFIIIB (TBP), CyclinD42 (CYCD42),
Translation elongation factor 1 (TEF1), phospho
glycerate kinase (PGK), Glyceraldehyde3phosphate
dehydrogenase (GAPDH) and RNA polymerase II asses
sory factor (RNAPII). The two reported novel refer
ence genes were clathrin coat associated protein AP2
complex subunit (CAC), Coatomer protein subunit 1
(COP1). They were also chosen to be novel stable
reference genes in Arabidopsis, tomato and Cupres
saceae [26, 27]. Besides all these reported reference
genes, four unigenes with a minimal Coefficient vari

ation among all unigenes were chosen to be candi

date references in this paper. They were cellulose
synthetase A (CESA), Glyoxylate reductase (GARY),
nuclear protein localization protein 4 (NPL4), vacuolar
sorting receptor 3 (VSR3). Primers were designed with
Primer3 (v0.4.0; http://frodo.wi.mit.edu/primer3/)
according to the conserved sequences with the follow
ing parameters: optimal length 2022 nucleotides,
melting temperature 6065C, and product size range
from 160 to 220 base pairs. We then used OligoAnaly
ser to make sure no hairpins or dimmers could be
formed.
qRTPCR conditions. qRTPCR reactions were
performed in fast optical 0.1 mL, 96well plates using an
ABI 7300 Real Time PCR System (Applied Biosystems,
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

SELECTION OF REFERENCE GENES FOR QUANTITATIVE REALTIME PCR

Reads (sample 1)

839

Reads (sample 2)

Assemble
Contig
Map reads to contigs
Contig 1

Contig 2

The same pipeline as sample 1

Assemble contigs to scaffolds

Scaffold

NNNNNN

Gap filling

Unigen

NN
NN

Unigen

Long sequence clustering

NN

NN

Unigen

Fig. 1. Detailed assembling steps of Putong oilcamellia seeds RNAseq from July to October.

Foster City CA, USA). The qRTPCR reaction vol

umes were 20 L, each containing 10 L 2 SYBR
Premix Ex TaqTM (Takara, Tokyo, Japan), 0.4 L
50 RO reference dye, 2 L of 15fold diluted syn
thesized cDNA, 0.4 L 10 M forward primer, 0.4 L
10 M reverse primer, and 6.8 L sterile distilled
water. PCR was then performed following the opera
tion manual (30 s at 95C; 40 cycles of 95C for 5 s and
60C for 34 s). After thermal cycling, melting curve
analysis (6095C, fluorescence read once every
0.3C) was performed to verify the specificity of the
amplicons. A negative PCR control lacking the cDNA
template was run for each primer pair. The threshold
cycle (Ct) values were calculated from the mean of four
technical replicates for each sample (every candidate
reference gene primer pair).

Analysis of gene stability. qRTPCR was performed

for each primer pair in a series of 10fold dilutions (101,
102, 103, 104, 105) of the mixed cDNA template to
obtain t values. The corresponding PCR amplifica
tion efficiencies () were calculated according to the
equation E = (101/slope 1) 100 [28], the correla
tion coefficients (R) and slope values were calculated
from the standard curve of Ct values for each gene
using the Excel software.
To select suitable reference genes, the expression
stability of every candidate reference gene was ana
lyzed with three different Microsoft Excelbased soft
ware tools: geNorm, NormFinder, and BestKeeper.
The raw Ct values were transformed into the required
data input format for geNorm and NormFinder. The
maximum expression level (the lowest Ct value) of
each gene was set to a value of 1. Relative expression

Table 2. Description of Camellia oleifera C. Abel RNAseq analysis

Sample

Total
reads

Q20
Gap
GC
Total
percentage, percentage, percentage, Contig
nucleotides, nt
%
%
%

Scaffold

Unigenes

July

8320068

1248010200

97.02

0.03

47.55

43216

43410

43461

August

8397376

1259606400

97.11

0.02

46.91

47568

47798

47932

September

8385809

1257871350

94.52

0.06

49.03

33827

35543

35589

October

8139855

1220978250

91.20

0.04

48.76

28817

29954

30022

Average

8310777

1246616550

94.96

0.04

48.06

38357

39176

39251

MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

273
369
59
978
449

Unigene 10169_All TUB3

Unigene 13817_All TBP

Unigene 57574_All CYCD42

Unigene 36161_All TEF1

Unigene 9686_All

MOLECULAR BIOLOGY

750
317

Unigene 14532_All NPL4

Unigene 5330_All

Vol. 47

COP1

No. 6

Unigene 4773_All

SQE

Unigene 63559_All SQS

Unigene 5501_All

Unigene 12278_All CAC

405

373

1148

55

17

Unigene 24263_All GARY

VSR3

1604

CESA

Unigene 5695_All

784

RNAPII

Unigene 6103_All

Unigene 11231_All GAPDH

327

119

Unigene 20799_All CYPB

PGK

August

September

October

208.80

83.43

154.22

4.99

120.94

65.71

1.52

100.84

64.62

84.13

256.22

244.56

4.70

28.76

116.47

13.44

21.17

1532

629

937

42

325

734

16

2100

598

214

397

976

82

534

283

107

269

812.31

144.69

129.45

3.92

127.52

66.14

1.47

135.78

50.69

56.63

232.99

251.00

6.72

42.81

124.17

12.43

22.71

329

192

843

35

250

612

13

993

523

285

246

774

45

305

183

106

223

464.05

54.37

143.36

4.02

120.74

67.88

1.47

79.03

54.57

92.83

177.71

245.02

4.54

30.10

98.83

15.16

23.17

711

109

777

42

265

627

14

1247

399

274

403

990

43

328

172

111

174

206.35

29.66

126.98

4.64

123.00

66.83

1.53

95.37

40.01

85.76

279.77

301.17

4.17

31.10

89.27

15.26

17.37

1691.52

312.14

554.02

17.57

492.20

266.57

6.00

411.02

209.90

319.35

946.69

1041.74

20.14

132.78

428.73

56.29

84.42

Average
standard
standard
standard
standard of RPKM
average
average
average
average
deviation
deviation
deviation
deviation
of RPKM
of RPKM
of RPKM
of RPKM
of RPKM
of RPKM
of RPKM
of RPKM
258

Gene
name

Unigene 17158_All ACT7

Unigene ID

July

Table 3. Expression and coefficient variation of each gene from July to October based on RNASeq database

286.39

49.57

12.72

0.51

3.15

0.95

0.03

23.89

10.17

15.93

43.70

27.31

1.15

6.48

15.97

1.37

2.63

Standard
deviation
of RPKM

Ranks
of CV

0.169 17

0.159 16

0.023

0.029

0.006

0.004

0.005

0.058 15

0.048 11

0.050 13

0.046 10

0.026

0.057 14

0.049 12

0.037

0.024

0.031

CV

840
ZHOU et al.

2013

SELECTION OF REFERENCE GENES FOR QUANTITATIVE REALTIME PCR

levels were then calculated from Ct values using the

formula: 2gCt, in which gCt is the lowest Ct value sub
tracted from the corresponding Ct value of each sam
ple for every gene. According to the analysis by
geNorm and NormFinder, ten most suitable genes
were chosen to be further analyzed by BestKeeper
based on untransformed t values and amplification
efficiencies.
Reference gene validation. Squalene synthase
(SQS); [29, 30] and squalene epoxidase (SQE); [31]
are enzymes that catalyze the synthesis of squalene,
sterols and terpenoids (which are extremely important
ingredients in oiltea camellia seeds and play a crucial
role in biological growth and metabolism). It was
found that the expression of SQS and SQE changed
considerably during different seed developmental
stages according to the RNASeq database (Table. 3).
Therefore, the expression values of SQS and SQE were
compared as target genes in different species and tis
sues of interest to test the validity of the reference
genes analyzed by the three software tools in the cur
rent study.
RESULTS
Characters of Candidate Reference Genes
Fifteen unigenes were chosen to be candidate ref
erence genes because of their low fluctuations of their
RPKM values during different seed development. The
raw reads and RPKM values of each month, RPKM
values average, standard deviation and coefficient
variation are listed in Table 3. Nine of them were tra
ditional reference genes; others were not commonly
used. It was shown that different candidate reference
genes showed different RPKM values. The sequences
of the unigenes were then aligned using BLASTn in
NCBI, this alignment revealed that the sequences
were highly similar to each particular gene, and the
conserved regions were used for primer design. The
gene names, descriptions, primer sequences, ampli
con lengths, amplification efficiencies, Tm values and
correlation coefficients are listed in Table 4. The Tm
values varied from 79.9C (GARY) to 85.9C (CYP),
and the amplicon lengths were about 200 bp, The
amplification efficiencies were between 89.4%
(COP1) and 101.8% (PGK), and the correlation coef
ficients were all larger than 0.99. Agarose gel electro
phoresis (2%) showed unique amplicons of expected
length without primer dimers. The melting curve anal
ysis showed a single peak for each gene, and no non
specific products were detected (Fig. 2).
Comparison of Candidate Reference Gene
Expression Stability
Realtime qRTPCR assays of the fifteen candi
date reference genes were performed using the newly
designed primer pairs, and cycle threshold (Ct) data
was obtained for all samples. The Ct values revealed
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

841

differences in transcript levels between the various

candidate genes (Fig. 3). The Ct values of the fifteen
genes ranged from 19 to 23 cycles. Two genes (PGK
and TEF1) showed the lowest Ct values (mean Ct =
18.3 and 17.8), indicating that they had the most
abundant transcript levels. The gene GARY had the
lowest transcript abundance (mean Ct = 22.9). Three
software tools, geNorm, NormFinder and Best
Keeper, were used to calculate the expression stability
of the candidate reference genes.
GeNorm Analysis
GeNorm is a Visual Basic application tool for
Microsoft Excel that operates on the assumption that
the expression ratio of two ideal reference genes is
constant throughout the different groups of templates.
Gene expression stability values () for all genes are
calculated and genes with an value below the
threshold of 1.5 are considered stable [32]. In this
analysis, most genes in different species and tissues
had values less than 1.5, which meant that all genes
were stable according to geNorm. When the results of
all 30 samples were combined, CESA and TUB3 had
the highest expression stability (the lowest value),
GAPDH was the least stable, and the other twelve genes
varied between the two extremes (Fig. 4). The analysis
results changed when the samples were classified into
different species and tissues (Fig. 4); CAC and PGK had
the highest expression stability in Zhejiang, TUB3 and
RNAPII were the most stable in Guangning, CESA and
TUB3 were the most stable in both Xiaoguo and
Putong, CESA and ACT7 had the highest expression
stability in Wantian, and CAC and COP1 were the
most stable in Youxian. GAPDH was the least stable
gene in Zhejiang, Guangning and Wantian, while the
least stable genes in Xiaoguo, Youxian and Putong
were GARY, PGK, and COP1, respectively. Further
more, the reference gene expression stability also var
ied in the five different tissue samples. PGK and
GAPDH had the lowest value (the highest expression
stability) in both roots and stems, CAC and COP1 had
the highest stability in leaves, CESA and TUB3 were
the most stable in flowers, and PGK and TBP were the
most stable in seeds. GAPDH was the least stable gene
in both leaves and flowers, CYP, TEF1 and GARY
were the least stable genes in roots, stems and seeds,
respectively.
Evaluation of the optimal number of reference
genes required accurate normalization. The pairwise
variation (Vn /Vn + 1) between consecutively ranked
normalization factors was calculated using geNorm.
NFn and NFn + 1 were used to determine the number of
genes required for reliable normalization. It has been
suggested that if the pairwise variation is below the
threshold value of 0.15, then there is no need for an
additional internal control gene [32]. In the current
research, six genes were needed when all 30 samples
were analyzed together. However, different species

842

ZHOU et al.

ACT7 CYPb TUB3 TBP CYCD42TEF1 PGK GAPDHRNAP II CESA COP1 NPL4 VSR3 GARY CAC

SQS

SQE

500 bp
200 bp

Derivative

TBP
0.32
0.6
CYPB
TUB3
1.4
ACT7
1.4
0.28
0.5
1.2
1.2
0.24
1.0
0.4
1.0
0.20
0.8
0.8
0.3
0.16
0.6
0.6
0.12
0.2
0.4
0.4
0.08
0.1
0.2
0.2
0.04
0
0
0
0
0.2
0.04
0.1
0.2
60 65 70 75 80 85 90
60 65 70 75 80 85 90
60 65 70 75 80 85 90
60 65 70 75 80 85 90 95

Derivative

1.0
1.0
0.7
1.2
CYCD42
TEF1
PGK
GAPDH
0.6
1.0
0.8
0.8
0.5
0.8
0.6
0.6
0.4
0.6
0.3
0.4
0.4
0.4
0.2
0.2
0.2
0.2
0.1
0
0
0
0
0.2
0.2
0.1
0.2
60 65 70 75 80 85 90
60 65 70 75 80 85 90
60 65 70 75 80 85 90
60 65 70 75 80 85 90 95

Derivative

1.2
0.7
0.8
0.5
RNAP II
CESA
COP1
NPL4
0.7
0.6
1.0
0.4
0.6
0.5
0.8
0.5
0.3
0.4
0.6
0.4
0.3
0.2
0.3
0.4
0.2
0.2
0.1
0.2
0.1
0.1
0
0
0
0
0.2
0.1
0.1
0.1
60 65 70 75 80 85 90
60 65 70 75 80 85 90
60 65 70 75 80 85 90
60 65 70 75 80 85 90 95

Derivative

Derivative

1.0
1.4
1.8
1.2
VSR3
GARY
CAC
SQS
1.6
1.2
1.0
0.8
1.4
1.0
0.8
1.2
0.6
0.8
1.0
0.6
0.4
0.6
0.8
0.4
0.6
0.4
0.2
0.4
0.2
0.2
0.2
0
0
0
0
0.2
0.2
0.2
0.2
60 65 70 75 80 85 90
60 65 70 75 80 85 90
60 65 70 75 80 85 90
60 65 70 75 80 85 90 95
T, C
T, C
T, C
1.0
SQE
0.8
0.6
0.4
0.2
0
0.2
60 65 70 75 80 85 90 95
T, C
Fig. 2. Specificity and melting curves of candidate genes. 2% agarose gel electrophoresis indicated unique amplicons of expected
length and no primer dimmers. The melting curve analysis showed a single peak in each gene, and no nonspecific products were
detected.
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

SELECTION OF REFERENCE GENES FOR QUANTITATIVE REALTIME PCR

843

30
28
26

Ct

24
22
20
18
16
14
12
1

6
7
8
9 10 11
Candidate reference genes

12

13

14

15

Fig. 3. Ct values of each reference genes. Expression data displayed as Ct values for each reference gene in all samples from 1 to
15 respectively. 1, ACT7; 2, CYPB; 3, TUB3; 4, TBP; 5, CESA; 6, CYCD42; 7, TEF1; 8, PGK; 9, GAPDH; 10, RNAPII;
11, GARY; 12, NPL4; 13, VSR3; 14, CAC; and 15, COP1. A line across the box is depicted as the median. The box indicates the
25th and 75th percentiles, whisker caps represent the maximum and minimum values, dots represent outliers.

needed different numbers of genes; Zhejiang, Youxian,

Xiaoguo and Putong needed two genes, Wantian
needed three genes, and Guangning needed four
genes. The optimal number of reference genes also
varied in the different tissues. Two genes were needed
in leaves and flowers, three in stems and seeds, and five
in roots (Fig. 5).
NormFinder Analysis
NormFinder is another Excel application that uses
a modelbased approach to identify the most stable
reference genes by combining samples into groups.
More stably expressed genes should show lower aver
age expression stability values ( values). The stability
value of each gene was calculated by NormFinder
(Table 5), and the results indicated that TUB3 and
CESA were the most appropriate for use as reference
genes over the 30 samples. The least stable genes were
GAPDH and GARY. The most stable and least stable
reference genes were different between each species
and tissues. PGK was most stable in Zhejiang, GARY in
Guangning, TEF1 in Wantian, RNAPII in Youxian,
TUB3 in Putong, and CESA in Xiaoguo. The least
stable gene in Guangning, Zhejiang and Wantian was
GAPDH, while in Youxian, Putong and Xiaoguo were
PGK, TBP and GARY respectively. For the tissues,
TUB3 was the most stable in roots, PGK in stems,
COP1 in leaves, CAC in flowers, and RNAPII in seeds.
The least stable gene in leaves and flowers was GAPDH,
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

while in roots, stems and seeds it was COP1, TBP and

GARY, respectively.
BestKeeper Analysis
BestKeeper, also an Excelbased tool, estimates the
intergene relationships between possible reference
gene pairs by performing numerous pairwise correla
tion analyses using the raw t values of each gene.
Most importantly, all genes may be included in the
calculations of the BestKeeper index, which is the
geometric mean of the Ct values of all candidate ref
erence genes and can be used to rank the best refer
ence genes because the stable reference genes show a
strong correlation with the BestKeeper index. Best
Keeper also calculates the coefficient of variance
(CV) and the standard deviation (SD) of the Ct values
using the whole data set, which includes all Ct values.
Reference genes are identified as the most stable
genes, i.e., those that exhibit the lowest coefficient of
variance and standard deviation (CV SD). Genes
with SD values greater than 1 are considered to be
unacceptable [33, 34].
According to the analysis of geNorm and Norm
Finder, the five least stable genes including TEF1,
VSR3, CYCD42, GARY and GAPDH were abandoned.
The left ten candidate genes were ranked according to
the CV SD of each species and tissue (Table 6). The
analysis revealed that all ten genes were acceptable in the
Xiaoguo and Putong species and the seeds tissues. Several
genes (NPL4, PGK, COP1, CAC, RNAPII) were not

844

ZHOU et al.
1.6

(a)

1.1
0.6
0.1
GAPDH GARY VSR3 RNAP II NPL4 CYPB ACT7 TEF1 TBP TUB3 CYCD COP1 CESA
42

1.8

PGK
CAC

(b)

1.3
0.8
0.3

GAPDH CAC

VSR3 COP1 CESA CYCD TBP

42

CYPB TEF1 NPL4 PGK

GARY ACT7 TUB3

RNAP II

(c)

1.6
Average expression stability,

1.1
0.6
0.1

GAPDH GARY COP1 CAC

VSR3

PGK

NPL4 CYCD TBP CYPB RNAP II TEF1TUB3 ACT7

42
CESA

(d)

1.6
1.2
0.8
0.4
0

PGK TEF1 GARY CYCD CESA ACT7 GAPDH TUB3 CYPB RNAP II TBP
42

VSR3 NPL4

CAC
COP1

(e)

0.9
0.7
0.5
0.3
0.1

GARY RNAP II TBP TEF1 NPL4 ACT7 CYCD PGK

42

CYPB VSR3 GAPDH CAC COP1 TUB3

CESA

(f)

1.2
1.0
0.8
0.6
0.4
0.2
COP1RNAP II TBP

PGK

CYPB ACT7 GARY NPL4 TEF1 VSR3

Least stable genes

CAC CYCD GAPDH TUB3

42
CESA

Most stable genes

Fig. 4. Average expression stability values (M) calculated by geNorm of each species (Lower average expression stability (M value)
indicates more stable expression). (a) Zhejiang oiltea camellia; (b) Guangning oiltea camellia; (c) Wantian oiltea camellia;
(d) Youxian oiltea camellia; (e) Xiaoguo oiltea camellia; (f) Putong oiltea camellia; (g) roots; (h) stems; (i) leaves; (j) flowers;
(k) seeds; (l) all samples.

acceptable in some species and tissues as their SD val

ues were greater than 1. ACT7, TUB3 and CESA
were among the three most stable genes when all sam
ples were combined. These three genes were also stable
among different species and tissues, and only varied in

their rank positions. NPL4, PGK, COP1 and CAC

exhibited the high SD in most species and tissues, indi
cating that these were the least stable reference genes,
when all tissues and species under analysis were com
bined, these four genes were not acceptable to be ref
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

SELECTION OF REFERENCE GENES FOR QUANTITATIVE REALTIME PCR

845

(g)

1.6
1.2
0.8
0.4
COP1 CYCD GARY TEF1 RNAP II CAC
42

VSR3

TBP

NPL4 ACT7 CESA TUB3 CYPB PGK

GAPDH

(h)

1.2
1.0
0.8
0.6
0.4

Average expression stability,

TBP ACT7 RNAP II CYCD CAC

42

VSR3 TEF1 NPL4 COP1 CYPB CESA TUB3 GARY PGK

GAPDH

(i)

1.0
0.8
0.6
0.4
0.2
GAPDH GARY CYCD CYPB CESA
42

TBP TEF1 VSR3 ACT7TUB3 PGK RNAP II NPL4

CAC
COP1

(j)

1.6
1.1
0.6
0.1
1.7

GAPDH GARY COP1 ACT7 CYCD TEF1 CYPB VSR3 RNAP II NPL4
42

TBP

(k)

PGK

CAC TUB3
CESA

1.2
0.7
0.2

GARY VSR3 NPL4

CAC COP1 GAPDH ACT7 CYCD CYPB TEF1 RNAP II CESA TUB3 PGK
42
TBP

(l)

1.5
1.3
1.1
0.9
0.7

GAPDH GARY CYCD VSR3 TEF1 COP1 CAC

42

PGK

Least stable genes

NPL4 CYPB RNAP II TBP ACT7 TUB3

CESA

Most stable genes

Fig. 4. Contd.

erence genes. The results of BestKeeper analysis

showed only a small difference from the results
obtained by geNorm and Normfinder.
Reference Gene Validation
The expression values of SQS and SQE were com
pared among the 5 different tissues of Putong camellia
and among seeds of the 6 different oil camellia species
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

using the most stable and least stable candidate refer

ence genes as internal controls. CESA, TUB3 and
PGK were the most stable reference genes, and GARY
was the least stable one for seeds of different oiltea
camellia species, TUB3 and CES were the most sta
ble, and COP1 was the least stable for different tissues
of Putong oiltea camellia. The results showed that
relative expression trends were mainly the same
between the stable reference genes, but there were

846

ZHOU et al.
V2/3 V3/4

V4/5 V5/6 V6/7 V7/8 V8/9 V9/10

V10/11 V11/12 V12/13

V13/14 V14/15

0.25

Pairwise variation

0.20
0.15
0.10
0.05
0
Zhejiang CuangningWantian Youxian Xiaoguo Putong Roots

Stems Leaves Flowers Seeds

All

Fig. 5. Optimal number of reference genes in each species and tissues. Pairwise variation (Vn /Vn+1) calculated by geNorm, and
the NFn and NFn+1 were used to determine the number of genes required for reliable normalization. It has been suggested that if
the pairwise variation is below the threshold value of 0.15, then there is no need for an additional internal control gene. The results
show that different numbers of reference genes are needed in different species and tissues.

considerable differences between the results obtained

with stable reference genes versus unstable ones. For
example, the expression levels of SQE in different tis
sues of Putong camellia were almost the same with
TUB3 and CES, and both of the relative expression
levels were highest in seeds, followed by flowers, roots,
stems and leaves. However, the results changed when
COP1 was used as the internal reference, the relative
expression in flowers was lower than in roots. Notable
discrepancy of the results of SQS was also shown
between stable references genes and unstable ones
(Fig. 6). Relative expression of SQS and SQE in seeds
of six species showed greater difference between stable
reference genes and unstable one. For example, the
relative expression of SQS showed little difference
among the six species by stable reference, but showed
great difference by the least stable reference gene:
about 150 in Xiaoguo oiltea camellia, but greater than
7000 in Putong oiltea camellia. Thus, the application
of different reference genes resulted in different rela
tive expression levels. Unsuitable references could
cause great deviation from the actual target gene
expression levels. Thus, it is important to validate the
reference genes before experimental application.
DISCUSSION
Recently, the quantification of RNA transcripts has
become increasingly rapid and precise because of
advances in gene quantification strategies. To remove
sampling differences and identify real genespecific
variation, especially when studying samples with subtle
gene expression differences, normalization becomes
necessary for accurate gene expression quantification
by qRTPCR. Normalizing to a reference gene is a
simple and popular method for this. However, the

expression of many traditional housekeeping genes

changes considerably under different conditions or in
different tissues, which biases the analysis of target
genes [35]. For example, 18SrRNA, ACTB and
RNAPII were shown to be the most stable genes among
six leaf samples of different citrus genotypes, but when
further analyzed in five other tissues the results indi
cated that they were not completely stable [36]. GAPDH
and ACT7 were shown to have unacceptable variability
in peach [37], and two of the most commonly used ref
erence genes, TUB3 and ACT7, are unsuitable for
different tissues of Chinese cabbage [11].
Oiltea camellias are a very important woody oil
crop of the genus Camellia. So far, studies on gene
expression in Camellia have been carried out using
single reference genes. For example, the expression of
the B function CjDEF1 gene in Camellia japonica
Hongshibaxueshi [38] and the chalcone isomerase
gene in Camellia nitidissima [39] were analyzed across
different parts of the flowers. All used 18S rRNA as the
reference gene, but none validated their results with
any preliminary expression stability analysis. Further
more, the expression analysis of two calmodulin genes
of Camellia oleifera did not use any reference genes
[21]. To our knowledge, this is the first time suitable
reference genes have been assessed for qRTPCR in
six oiltea camellia species and their different tissues.
Therefore, gene expression normalization by qRT
PCR in camellia can now be put into practice based on
this selection of reference genes. Pairwise analysis
using geNorm reveals that the optimal number of ref
erence genes varies from two to six depending on the
particular sample set. Our results suggest that there is
no particular gene that is expressed across all species
and tissues in a consistently stable way, and also show
that the more complex the samples are, the more ref
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

Relative expression

SELECTION OF REFERENCE GENES FOR QUANTITATIVE REALTIME PCR

9000
8500
8000
7500
7000
550
500
450
400
350
300
250
200
150
100
50
0

(a)
CESA
PGK
TUB3
CESA + PGK + TUB3
GRAY

1600
1400
1200
1000
400

847

(b)
CESA
PGK
TUB3
CESA + PGK + TUB3
GRAY

350
300
250
200
150
100
50
0

Zhejiang
Wantian
Putong
Cuangning
Youxian
Xiaoguo
(c)
200
CESA
TUB3
CESA + TUB3
COP1
150

250

200

Zhejiang
Wantian
Putong
Cuangning
Youxian
Xiaoguo
(d)
CESA
TUB3
CESA + TUB3
COP1

150
100
100
50
50

0
Roots

Leaves
Stems

Seeds
Flowers

Species

Roots

Leaves
Stems

Seeds
Flowers

Species

Fig. 6. Relative expression levels of SQE and SQS in different tissues/organs of Putong oiltea Camellia and seeds of different spe
cies. Expression levels of both SQS (a and c) and SQE (b and d) were normalized to stable and unstable reference genes respec
tively. Error bars show the mean standard error calculated from two biological replicates.

erence genes are needed. Thus, it is necessary to select

reference genes according to the samples being
researched.

downregulated when 18S rRNA was used as a refer

ence gene [41]. Based on these results, 18S rRNA was
not chosen for the current study.

18S rRNA has been used as a reference gene in

many gene expression studies of the genus Camellia,
but it has been reported that 18S rRNA is an unsuit
able reference gene since its synthesis is executed by
RNA polymerase I, whereas mRNA transcription is
carried out by RNA polymerase II. Moreover, the
reverse transcription reaction is an oligo(dT) reaction,
but rRNA contains no poly(A) tail, so oligo(dT)
primed cDNA cannot be synthesized for rRNA [40].
It was also reported that target gene expression was

The expression results of fifteen candidate genes

were analyzed for each sample with three different
software tools: geNorm, NormFinder and Bestkeeper.
These three software tools use different algorithms and
can give different results. However, the results showed
only a small difference among them, and all of the
results indicated that TUB3 and CESA were the
most stable reference genes, though they were not the
most stable among all of the fifteen candidate genes
based on comparison of the RPKM values in the

MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

MOLECULAR BIOLOGY

Vol. 47

No. 6

SQE

SQS

COP1

CAC

VSR3

NPL4

GARY

CESA

RNAP II

GAPDH

PGK

TEF1

CYCD42

TBP

TUB3

CYPB

ACT7

Gene
Symbol

Primer sequence (5' 3')

(forword/reverse)

CAACTTCGCTGGTGTCTTCA
ACCCTCTACGCAGAAGCAAA
Cyclophilin B
ACAGGGAGCTCACCACATTC
TCTTAGCATGGCAAATGCAG
Tublin alpha3
CCATGCCTTGGATCACATTT
TGGGGCCATTAATGTAGACG
TATA box binding protein component
GAAAAGGCACCATGGGAATA
of TFIID and TFIIIB
GGAAGATGGTTTGCACTGGT
CyclinD42
GGATTGAGGAATGGGGATTT
ATAAACAGGCCACAGCCAAC
Translation elongation factor 1alpha
TCCAGGAGCATCAATGACAG
ACCACCACTGGTCACCTCAT
Phosphoglycerate kinase
CCCAAGGGTACTCAGTTGGA
CCATCCAACCATCAGGGATA
Glyceraldehyde3phosphate dehydrogenase TCAATCACCCGATTGCTGTA
CTGCTATCAAGGAGGCTTCG
RNA polymerase II assessory factor
AATGCTCGCTCTCACAACCT
CGAAATCGTTGTCGTCATTG
Cellulose synthetase A
AAGGACCGCTGATACTCGAA
ACACCATGGCCTGGAAATAA
Glyoxylate reductase
TGCGGTTCTTGTGGATGATA
GCACTCATGCTTTCCTGACA
Nuclear protein localization protein 4
GGCCATGGACTCAATTAGGAA
TCATCTGGACCGAACAAGG
Vacuolarsorting receptor 3
GCACAAATGGCCTTCAAAAC
GGTGACCCAAATGCTGATTC
Clathrin coat associated protein AP2 comlex GGCATTCCAGAAAGAAAGCA
subunit
AGGAAGGAGTACGCTCACCA
Coatomer protein subunit epsilon1
GCCTTTCCATTCAGGATCAA
ATGCGGAAAAACAGTTGAGG
Squalene synthetase
TTTCGCCCTCGTAATTCAAC
CATGAAAAATGCCAGTCACG
Squalene epoxidase
AAAGAGCAGACCACCACCAC
TCGGGCTCTGTCAAATCTCT

Actin7 alpha

Gene name

Table 4. Description of candidate genes from Camellia for qRTPCR

80.15

81.43

81.7

82

82.3

83.8

79.9

83.9

85.55

82.25

83.82

83.75

82.25

81.8

82.63

85.9

82.4

Tm, C

208

180

178

235

167

176

220

195

196

199

192

219

196

193

195

210

212

Amplicon
length, bp

100.8

105.3

89.4

92.3

98.3

98.9

100.9

99.1

93

96.9

101.8

99.4

90.7

95.6

94

93.9

101.1

Amplification
efficiency, %

0.9941

0.9966

0.9900

0.9920

0.9904

0.9935

0.9927

0.9939

0.9958

0.9993

0.9995

0.9987

0.9993

0.9955

0.9949

0.9989

0.9969

R2

848
ZHOU et al.

2013

MOLECULAR BIOLOGY

NPL4
0.422

TUB3
0.564

COP1
0.620

CESA
0.639

TEF1
0.685

CYPB
0.686

CYCD42
0.809

TBP
0.860

VSR3
0.861

CAC
0.996

GAPDH
1.174

CESA
0.366

TUB3
0.447

CYCD42
0.452

NPL4
0.477

ACT7
0.580

TBP
0.609

CYPB
0.676

RNAPII
0.728

VSR3
0.773

GARY
1.005

GAPDH
1.606

10

11

12

13

14

15

ACT7
0.385

CAC
0.213

PGK
0.416

RNAPII
0.371

COP1
0.188

TEF1
0.346

GARY
0.302

PGK
0.061

Guangning

Zhejiang

Rank

Vol. 47

No. 6

2013

GAPDH
1.383

GARY
1.174

COP1
1.166

CYCD42
0.975

CAC
0.771

NPL4
0.733

TBP
0.727

VSR3
0.674

PGK
0.498

CYPB
0.465

ACT7
0.438

RNAPII
0.430

CESA
0.411

TUB3
0.390

TEF1
0.331

Wantian

PGK
1.069

TEF1
0.905

GARY
0.849

CYCD42
0.834

CESA
0.726

ACT7
0.669

VSR3
0.586

GAPDH
0.512

TBP
0.483

TUB3
0.469

NPL4
0.430

COP1
0.422

CAC
0.393

CYPB
0.356

RNAPII
0.109

Youxian

TBP
0.643

CYPB
0.383

PGK
0.340

GARY
0.339

RNAPII
0.292

COP1
0.290

ACT7
0.246

TEF1
0.245

NPL4
0.240

CESA
0.219

CYCD42
0.216

GAPDH
0.205

VSR3
0.201

CAC
0.194

TUB3
0.144

Putong

GARY
0.756

TEF1
0.625

RNAPII
0.622

CYCD42
0.517

TBP
0.510

PGK
0.442

NPL4
0.432

ACT7
0.386

CYPB
0.374

VSR3
0.340

GAPDH
0.332

CAC
0.159

COP1
0.159

TUB3
0.080

CESA
0.052

Xiaoguo

COP1
1.244

CYCD42
1.129

GARY
1.029

TEF1
1.016

RNAPII
0.804

CAC
0.775

VSR3
0.767

GAPDH
0.711

NPL4
0.606

TBP
0.530

ACT7
0.524

PGK
0.524

CESA
0.523

CYPB
0.366

TUB3
0.263

Roots

TBP
0.852

RNAPII
0.778

ACT7
0.752

CAC
0.737

VSR3
0.726

CYCD42
0.711

TEF1
0.628

COP1
0.589

CYPB
0.529

NPL4
0.527

GARY
0.443

GAPDH
0.354

TUB3
0.345

CESA
0.225

PGK
0.163

Stems

GAPDH
1.092

GARY
0.676

CYCD42
0.588

CYPB
0.557

CESA
0.519

VSR3
0.517

TBP
0.416

TEF1
0.402

ACT7
0.303

CAC
0.274

NPL4
0.253

TUB3
0.242

RNAPII
0.227

PGK
0.199

COP1
0.130

Leaves

Table 5. Ranking of candidate reference genes in order of their expression stability as calculated by NormFinder

GAPDH
2.142

GARY
0.760

CYCD42
0.724

TEF1
0.670

CYPB
0.632

VSR3
0.607

COP1
0.411

RNAPII
0.405

ACT7
0.393

NPL4
0.264

TBP
0.175

PGK
0.163

CESA
0.062

TUB3
0.062

CAC
0.026

Flowers

GARY
1.293

VSR3
1.249

NPL4
0.933

ACT7
0.890

CYCD42
0.804

GAPDH
0.767

CAC
0.727

COP1
0.718

CYPB
0.717

TEF1
0.645

TBP
0.562

PGK
0.483

CESA
0.459

TUB3
0.270

RNAPII
0.261

Seeds

GAPDH
1.231

GARY
0.917

CYCD42
0.850

VSR3
0.749

TEF1
0.708

COP1
0.641

CYPB
0.631

TBP
0.627

RNAPII
0.587

PGK
0.557

ACT7
0.552

CAC
0.547

NPL4
0.524

CESA
0.492

TUB3
0.358

All

SELECTION OF REFERENCE GENES FOR QUANTITATIVE REALTIME PCR

849

850

ZHOU et al.

RNAseq database of Camellia oleifera seeds from July

to October. CV values of TUB3 and CESA were
ranked as the ninth and the fifteenth respectively
(Table 3). So it seems that the most stable reference
genes are not the genes with the lowest CV values of
RPKM. This may be because RNAseq is focused on
gene expression during the development of the seed,
but not on different species and tissues, furthermore,
all of the candidate genes were relatively stable uni
genes, and there was only a little CV difference
between them. So RNAseq is still valid and it may be
supposed that more suitable reference genes will be
found using RNAseq data in future research.
Finally, reference gene validation was performed.
The results showed many relative expression discrep
ancies between the stable and unstable reference
genes, indicating that the inappropriate use of refer
ence genes without validation may produce bias from
the real expression.
Fifteen genes were selected as candidate reference
genes from the RNAseq database, all of which were
stable among the different seed developmental stages.
Ten genes were expressed at moderate levels, three
genes expressed at low levels (GARY, CAC and CYCD42),
and the other two genes (TEF1 and PGK) were
expressed at a much higher level. Thus, their Ct values
were lower than those of the others. Based on the ref
erence gene validation analysis, we speculated that if
the reference genes were highly expressed, it would
lower the relative expression and lead to transcript
abundance discrepancies of the target gene. The two
highestabundance genes, TEF1 and PGK, were the
least stable genes in most species and tissues according
to the three different software tools. Although PGK was
found to be a stable gene in seeds, the relative expres
sion of SQE in the seeds of Guangning, Wantian and
Youxian was almost zero when PGK was used as the
reference gene, so differences in target gene expres
sion were hard to distinguish. This result verified our
assumption that a highly expressed reference gene
could lead to discrepancies in target gene expression.
CONCLUSION
In summary, fifteen reference genes were analyzed
in different tissues of six oiltea camellia species with
geNorm, NormFinder and BestKeeper. The results
indicated that TUB3 and CESA were the most stably
expressed when all samples were analyzed together.
Suitable reference genes were also evaluated in each
tissue and species specifically. Although no gene was
stable across all of the different species and tissues,
TUB3, CESA and ACT7 were comparatively stable
in many species and tissues. The optimal number of
reference genes required for accurate normalization
varied from 26.

ACKNOWLEDGMENTS
This work was supported by the Creation of High
Yield and Quality New Oiltea camellia Germplasm
(no. 2009BADB1B01) and the Research and Devel
opment of Crucial Technology for Oiltea camellia
Industrial Upgrading (no. 2009BADB1B00).
REFERENCES
1. Chang Z., Ling C., Yamashita M., et al. 2010. Microar
raydriven validation of reference genes for quantitative
realtime polymerase chain reaction in a rat vocal fold
model of mucosal injury. Anal. Biochem. 406 (2), 214
221.
2. Umenishi F., Verkman A.S., Gropper M.A. 1996.
Quantitative analysis of aquaporin mRNA expression
in rat tissues by RNase protection assay. DNA Cell
Biol.15, 475480 .
3. Zhang L., Zhou W., Velculescu V.E., et al. 1997. Gene
expression profiles in normal and cancer cells. Science.
276 (5316), 12681272.
4. Demidenko N.V., Logacheva M.D., Penin A.A. 2011.
Selection and validation of reference genes for quanti
tative realtime PCR in Buckwheat (Fagopyrum escu
lentum) based on transcriptome sequence data. PloS
ONE. 6 (5), e19434.
5. Huggett J., Dheda K., Bustin S., et al. 2005. Realtime
RTPCR normalization: Strategies and considerations.
Genes Immun. 6 (4), 279284.
6. Paolacci A.R., Tanzarella O.A., Porceddu E., Ciaffi M.
2009. Identification and validation of reference genes
for quantitative RTPCR normalization in wheat.
BMC Mol. Biol. 10 (1), 11.
7. Maccoux L.J., Clements D.N., Salway F., Day P.J.
2007. Identification of new reference genes for the nor
malization of canine osteoarthritic joint tissue tran
scripts from microarray data. BMC Mol. Biol. 8 (1), 62.
8. Migocka M., Papierniak A. 2011. Identification of suit
able reference genes for studying gene expression in
cucumber plants subjected to abiotic stress and growth
regulators. Mol. Breeding. 28 (3), 343357.
9. Dheda K., Huggett J.F., Chang J.S., et al. 2005. The
implications of using an inappropriate reference gene
for realtime reverse transcription PCR data normal
ization. Anal. Biochem. 344 (1), 141143.
10. Andersen C.L., Jensen J.L., rntoft T.F. 2004. Nor
malization of realtime quantitative reverse transcrip
tionPCR data, a modelbased variance estimation
approach to identify genes suited for normalization,
applied to bladder and colon cancer data sets. Cancer
Res. 64 (15), 5245.
11. Qi J., Yu S., Zhang F., Shen X., Zhao X., Yu Y., Zhang D.
2010. Reference gene selection for realtime quantita
tive polymerase chain reaction of mRNA transcript levels
in Chinese cabbage (Brassica rapa L. ss. pekinensis).
Plant Mol. Biol. Rep. 28 (4), 597604.
12. Hruz T., Wyss M., Docquier M., et al. 2011. RefGenes,
identification of reliable and condition specific refer
ence genes for RTqPCR data normalization. BMC
Genomics. 12 (1), 156.
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

SELECTION OF REFERENCE GENES FOR QUANTITATIVE REALTIME PCR

13. Zhuang R.L., Yao X.H. 2008. OilTea Camellia of
China, 2nd ed. Beijing: Chinese Forestry Publ., pp. 110
(in Chinese).
14. Vijayan K., Zhang W.J., Tsou C.H. 2009. Molecular
taxonomy of Camellia (Theaceae) inferred from nrITS
sequences. Am. J. Bot. 96 (7), 13481360.
15. Tan X.F., Hu F.M., Xie L.S., et al. 2006. Construction
of EST library and analysis of main expressed genes of
Camellia oleifera seeds. Sci. Silvae Sinicae (China). 42 (1),
4348.
16. Lin P., Cao Y.Q., Yao X.H., et al. 2011.Transcriptome
analysis of Camellia oleifera Abel seed in four develop
ment stages. Mol. Plant Breeding (China). 19 (4), 498
505.
17. Hu X., Tan X., Tian X., et al. 2008. cDNA cloning,
sequence analysis and physiological role speculation of
a dehydrinlike protein from Camellia oleifera. Acta Bot.
BorealiOccidentalia Sinica (China). 28 (8), 15411548.
18. Jiang Y., Tan X.F., Zhang D.Q., et al. 2009. Cloning
and sequence analysis of a metallothionein gene from
Camellia oleifera. Acta Agric. Univ. Jiangxiensis (China).
31 (4), 699705.
19. Luo Q., Xie L.S., Tan X.F., et al. 2008. Cloning of full
length cDNA of FAD2 gene from Camellia oleifera. Sci.
Silvae Sinicae (China). 44 (3), 7075.
20. Zhang D., Tan X., Chen H., et al. 2008. Fulllength
cDNA cloning and bioinformatic analysis of Camellia
oleifera SAD. Sci. Silvae Sinicae. 44 (2), 155159.
21. Wang B., Tan X.F., Chen Y., et al. 2012. Molecular
cloning and expression analysis of two calmodulin
genes encoding an identical protein from Camellia ole
rfera. Pak. J. Bot. 44 (3), 961968.
22. Shao G., Tan X.F., Chen H., et al. 2012. Isolation and
characterization of an aldoketo reductase cDNA from
Camellia oleifera seed. Adv. Sci. Lett. (China). 10 (1),
153157.
23. Han X.J., Lu M., Chen Y., et al. 2012. Selection of reli
able reference genes for gene expression studies using
realtime PCR in tung tree during seed development.
PloS ONE. 7 (8), e43084.
24. Li R., Zhu H., Ruan J., et al. 2010. De novo assembly
of human genomes with massively parallel short read
sequencing. Genome Res. 20, 265272.
25. Mortazavi A., Williams B.A., McCue K., Schaeffer L.,
Wold B. 2008. Mapping and quantifying mammalian
transcriptomes by RNASeq. Nature Methods. 5 (7),
621628.
26. Czechowski T., Stitt M., Altmann T., et al. 2005.
Genomewide identification and testing of superior
referencegenes for transcript normalization in Arabi
dopsis. Plant Physiol. 139, 517.
27. ExpsitoRodrguez M., Borges A.A., BorgesPrez A.,
et al. 2008. Selection of internal control genes for quan
MOLECULAR BIOLOGY

Vol. 47

No. 6

2013

851

titative realtime RTPCR studies during tomato devel

opment process. BMC Plant Biol. 8, 131.
28. Chang E., Shi S., Liu J., et al. 2012. Selection of refer
ence genes for quantitative gene expression studies in
Platycladus orientalis (Cupressaceae) using realtime
PCR. PloS ONE. 7 (3), e33278.
29. Do R., Kiss R.S., Gaudet D., et al. 2009. Squalene syn
thase, a critical enzyme in the cholesterol biosynthesis
pathway. Clin. Genet.75 (1), 1929.
30. Beytia E., Qureshi A.A., Porter J.W. 1973. Squalene
synthetase III. Mechanism of the reacion. J. Biol.
Chem. 248 (5), 18561867.
31. MBaya B., Fegueur M., Servouse M., et al. 1989.
Regulation of squalene synthetase and squalene epoxi
dase activities in Saccharomyces cerevisiae. Lipids.
24 (12), 10201023.
32. Vandesompele J., De Preter K., Pattyn F., et al. 2002.
Accurate normalization of realtime quantitative RT
PCR data by geometric averaging of multiple internal
control genes. Genome Biol. 3 (7), res. 0034.
33. Pfaffl M.W., Tichopad A., Prgomet C., et al. 2004.
Determination of stable housekeeping genes,differen
tially regulated target genes and sample integrity: Best
KeeperExcelbased tool using pairwise correlations.
Biotechnol. Lett. 26 (6), 509515.
34. Livak K.J., Schmittgen T.D. 2001. Analysis of relative
gene expression data using realtime quantitative PCR
and the 2Ct method. Methods. 25 (4), 402408.
35. Benn C.L., Fox H., Bates G. 2008. Optimisation of
regionspecific reference gene selection and relative
gene expression analysis methods for preclinical trials
of Huntingtons disease. Mol. Neurodegener. 3 (1), 17.
36. Yan J., Yuan F., Long G., et al. 2011. Selection of refer
ence genes for quantitative realtime RTPCR analysis
in citrus. Mol. Biol. Rep. 39 (2), 18311838.
37. Tong Z., Gao Z., Wang F., et al. 2009. Selection of reli
able reference genes for gene expression studies in
peach using realtime PCR. BMC Mol. Biol. 10 (1), 71.
38. Zhu G.P., Li J.Y., Fan Z.Q., et al. 2011. Isolation of B
function CjDEF1 gene involved in floral development
in Camellia japonica Hongshibaxueshi and its expres
sion analysis. J. Agric. Biotechnol. 19 (3), 442448.
39. Zhou X.W., Li J.Y., Fan Z.Q. 2012. Cloning and expres
sion analysis of chalcone isomerase gene cDNA from
Camellia nitidissima. Forest Res. (China). 25 (1), 9399.
40. Radonic A., Thulke S., Mackay I.M., Landt O., Siegert W.,
Nitsche A. 2004. Guideline to reference gene selection
for quantitative realtime PCR. Biochem. Biophys. Res.
Commun. 313 (4), 856862.
41. Raaijmakers M.H., van Emst L., de Witte T., Mensink E.,
Raymakers R.A. 2002. Quantitative assessment of gene
expression in highly purified hematopoietic cells using
realtime reverse transcriptase polymerase chain reac
tion. Exp. Hematol. 30 (5), 481487.