Quantification of genetically modified soybeans using a combination of a capillary-type real-time PCR system and a plasmid reference standard.

Abstract

Because the labeling of grains and feed- and foodstuffs is mandatory if the genetically modified organism (GMO) content exceeds a certain level of approved genetically modified varieties in many countries, there is a need for a rapid and useful method of GMO quantification in food samples. In this study, a rapid detection system was developed for Roundup Ready Soybean (RRS) quantification using a combination of a capillary-type real-time PCR system, a LightCycler real-time PCR system, and plasmid DNA as the reference standard. In addition, we showed for the first time that the plasmid and genomic DNA should be similar in the established detection system because the PCR efficiencies of using plasmid DNA and using genomic DNA were not significantly different. The conversion factor (Cf) to calculate RRS content (%) was further determined from the average value analyzed in three laboratories. The accuracy and reproducibility of this system for RRS quantification at a level of 5.0% were within a range from 4.46 to 5.07% for RRS content and within a range from 2.0% to 7.0% for the relative standard deviation (RSD) value, respectively. This system rapidly monitored the labeling system and had allowable levels of accuracy and precision.

Full text

Page 1

Quantification of Genetically Modified Soybeans
Using a Combination of a Capillary-Type Real-Time PCR System
and a Plasmid Reference Standard
Akie TOYOTA,1;4Hiroshi AKIYAMA,2;yMitsunori SUGIMURA,1Takahiro WATANABE,2
Hiroyuki KIKUCHI,2Hisayuki KANAMORI,1Akihiro HINO,3
Muneharu ESAKA,4and Tamio MAITANI2
1Hiroshima Prefectural Institute of Public Health and Environment,
1-6-29 Minami-machi, Minami-ku, Hiroshima 734-0007, Japan
2National Institute of Health Sciences, 1-18-1 Kamiyoga, Setagaya-ku, Tokyo 158-8501, Japan
3National Food Research Institute, 2-1-12 Kannondai, Tsukuba, Ibaraki 305-8642, Japan
4Graduate School of Biosphere Sciences, Hiroshima University,
1-4-4 Kagamiyama, Higashi-Hiroshima, Hiroshima 739-8528, Japan
Received August 23, 2005; Accepted December 15, 2005
Because the labeling of grains and feed- and food-
stuffs is mandatory if the genetically modified organism
(GMO) content exceeds a certain level of approved
genetically modified varieties in many countries, there is
a need for a rapid and useful method of GMO
quantification in food samples. In this study, a rapid
detection system was developed for Roundup Ready?
Soybean (RRS) quantification using a combination of a
capillary-type real-time PCR system, a LightCycler?
real-time PCR system, and plasmid DNA as the
reference standard. In addition, we showed for the first
time that the plasmid and genomic DNA should be
similar in the established detection system because the
PCR efficiencies of using plasmid DNA and using
genomic DNA were not significantly different. The
conversion factor (Cf) to calculate RRS content (%)
was further determined from the average value ana-
lyzed in three laboratories. The accuracy and reprodu-
cibility of this system for RRS quantification at a level of
5.0% were within a range from 4.46 to 5.07% for RRS
content and within a range from 2.0% to 7.0% for the
relative standard deviation (RSD) value, respectively.
This system rapidly monitored the labeling system and
had allowable levels of accuracy and precision.
Key words:genetically modified soybean; Roundup
Ready?Soybean; capillary-type real-time
PCR system
The production of genetically modified (GM) crops,
especially Roundup Ready?Soybean (RRS), has in-
creased in the United States over the past several years.1)
In some countries, controversial issues still exist
regarding the acceptance of GM crops, and concerns
about their safety persist in public opinion. In many
countries, the labeling of grains and feed- and foodstuffs
is mandatory if the genetically modified organism
(GMO) content exceeds a certain level of approved
GM varieties. For instance, the European Union (EU),
Japan, Korea, and Taiwan have set threshold values at
0.9% (EU Regulation No. 1830/2003), 5%, 3%, and 5%
respectively of GMO material in a non-GM background
as the basis for labeling.2–4)The enforcement of these
threshold values has created a demand for the develop-
ment of reliable GMO analysis methods of a rapid and
inexpensive character.
Most of the established analytical methods for
detecting the GMO identification and quantification in
foods are based on the polymerase chain reaction (PCR),
due to its sensitivity, specificity, and applicability to the
analysis of complex food matrices.5–18)Furthermore,
many real-time PCR systems based on fluorescent
detection, such as TaqMan chemistry, have been
developed to identify and quantify GM soybeans, GM
maize, and GM varieties of other agricultural commod-
ities.19–28)Real-time PCR systems using TaqMan chem-
istry are based on the use of a fluorescent TaqMan probe
that monitors the formation of the PCR product during
each cycle of the reaction. In addition, most commonly,
GMO quantification by quantitative real-time PCR
methods is calculated from the ratio of the target
transgenic specific DNA sequence copy number vs. the
DNA sequence copy number of the respective target
plant species (taxon gene sequence). Determination of
the copy number by real-time PCR methods involves the
establishment of calibration curves based on analysis of
yTo whom correspondence should be addressed. Tel: +81-3-3700-9397; Fax: +81-3-3707-6950; E-mail: akiyama@nihs.go.jp
Biosci. Biotechnol. Biochem., 70 (4), 821–827, 2006

Page 2

a set of calibrators such as genomic DNA (gDNA) or
plasmid DNA (pDNA). Especially, plasmid DNA
markers containing a cloned transgenic sequence have
been used and are increasingly promoted as the stand-
ards of choice for GMO analysis, because pDNA is
semi-infinitely available at the same quality.
We have already developed quantification methods
based on the TaqMan Chemistry of real-time PCR for
five lines of GM maize and a GM soybean, Roundup
Ready?, using pDNA as the reference molecules, and
our methods have been validated by interlaboratory
testing.23,24)These methods were performed using an
ABI PRISM 7700?real-time PCR system, titer-plate
type equipment. However, if these methods were to be
applied to other real-time PCR equipment, beforehand
we had to set up the PCR conditions and determine the
conversion factor (Cf), and the coefficient value to
calculate the GMO amount (%). Because titer-plate type
PCR equipment is time-consuming in PCR and com-
paratively expensive, the application of rapid and
inexpensive equipment is desirable. The LightCycler?
real-time PCR system, capillary-type real-time PCR
equipment, is one of the most widely used types of
equipment in biomaterial tests and is more rapid due to
the temperature change using air media, and is also
comparatively inexpensive.29,30)
In the present study, we developed a detection system
for RRS quantification using a combination of reference
plasmid DNA, a LightCycler?real-time PCR system,
and capillary-type real-time PCR equipment, and eval-
uated the method developed, including determination of
the Cf, to calculate the RRS content (%) system. In
addition, we showed the equivalence of pDNA and
gDNA as reference standards in the method developed
using the LightCycler?real-time PCR system.
Materials and Methods
Materials. Non-GM soybean grains were obtained
from the Ministry of Health, Labour, and Welfare
(MHLW) of Japan. RRS seeds were kindly provided by
Monsanto (St. Louis, MO). Soybean powder certified
reference material (CRM) IRMM-5, 5% RRS, was
purchased commercially (Fluka, Buchs SG, Switzer-
land). Three soybean grain samples labeled as non-GM
were purchased commercially in Tokyo. As a standard
material for the calibration curve, RRS Detection
Plasmid Set-ColE1/TE (Nippon Gene, Toyama, Japan)
was used.
Preparation of test samples. To prepare GM mixed
test samples, soybeans (GM seeds and non-GM seeds)
were separately milled to a fine powder using grinders
(Retsch, Haan, Germany), and then mixed with 1% and
5% GM soybeans weight per weight, respectively,
according to a previously reported procedure.23)
Extraction and purification of gDNA. DNA extraction
and purification was carried out using an anion exchange
resin-type kit (Genomic-tip 20/G; Qiagen, Hilden,
Germany) according to the manufacturer’s manual, with
the following modification: Powdered samples (500mg)
were vortexed in 15ml of Digestion Buffer G2 (800mM
guanidine HCl; 30mM Tris–Cl, pH 8.0; 30mM EDTA,
pH 8.0; 5% Tween-20; 0.5% Triton X-100) and 200ml
of (1mg/ml) ?-amylase, and then incubated at 37?C for
1h. One hundred ml of (20mg/ml) Proteinase K and
20ml of (100mg/ml) RNase A were added to the
samples and then incubated at 50?C for 2h. The
supernatant generated after centrifugation at 8;000 ? g
for 25min at 4?C was filtered through Millex-HV
(Millipore, Billerica, MA). Two ml of the supernatant
was applied three times to a Genomic-tip equilibrated
with 1ml of Buffer QBT (750mM NaCl; 50mM MOPS,
pH 7.0; 15% isopropanol; 0.15% Triton X-100). The tip
was washed three times with 2ml of Buffer QC (1.0 M
NaCl; 50mM MOPS, pH 7.0; 15% isopropanol), and
then eluted twice with 1ml of Buffer QF (1.25 M NaCl;
50mM Tris–Cl, pH 8.5; 15% isopropanol; at 50?C). The
eluate was mixed with 0.1 volume of 3 M sodium acetate
at pH 4.8 and 2.5 volumes of ethanol. The mixture was
centrifuged at 8;000 ? g for 20min at 4?C. The super-
natant was discarded and the pellets washed with 70%
ethanol. Finally, the DNA was dissolved in 200ml of TE
buffer (10mM Tris–Cl, pH 8.0; 1mM EDTA). The DNA
concentration in the solutions was determined by
measuring UV absorption at 260nm using a GeneQuant
pro spectrophotometer (Amersham Biosciences, Piscat-
away, NJ). The purity of the extracted DNA was
evaluated using a ratio of 260/280nm, and the ratio was
between 1.7 and 2.0 for most of the test samples. The
extracted DNA was diluted with an appropriate volume
of DW to a final concentration of 10ng/ml and stored at
?20?C until used. These DNA samples were used for
the subsequent PCR analysis.
Real-time quantitative PCR of the endogenous lectin
gene (Le1) and the RRS specific gene (RRS). Fluores-
cence resonance energy transfer (FRET) hybridization
probes were used for quantification in the LightCycler?
real-time PCR system (Roche Diagnostics, Mannheim,
Germany). The lectin gene (Le1) was chosen as the
reference DNA gene for quantitative analysis, because
Le1 has been reported as a single copy gene in soybean
and is widely used in the previously reported methods.
The GM Soybean (RRS) Detection Le1 Oligonucleotide
Set and GM Soybean (RRS) Detection RRS Oligonu-
cleotide Set (Nippon Gene) were used for Le1 and RRS
specific gene (RRS) assays as the primers and probes.
PCR was performed in glass capillary tubes (Roche).
Unless otherwise specified, the total reaction volume of
20ml contained 50ng of the DNA template, 0.25mM of
each primer, 0.2mM of the probe, 4.0mM of MgCl2, and
a LightCycler-FastStart DNA Master Hybridization
Probes Kit (Roche), which contained FastStart Taq
DNA polymerase, hybridization probes for detection by
822A. TOYOTA et al.

Page 3

the LightCycler?real-time PCR system, reaction buffer,
and dNTPs. The cycling conditions were as follows: pre-
incubation at 95?C for 10min, denaturation at 95?C for
15s (20?C/s), annealing at 59?C for 30s (1?C/s), and
pre-incubation at 95?C for 10min, denaturation at 95?C
for 15s (1?C/s), and annealing at 59?C for 30s (20
?C/s). Both cycles were repeated 50 times. Standard
curves were calibrated using five concentrations of
control plasmids 40, 250, 3,000, 40,000, and 500,000
copies per reaction. A no-template control was also
prepared as the negative control. Each control sample
was run in one capillary for each target, and unless
otherwise specified, each soybean sample was run in
duplicate for each target. The data were analyzed using
LightCycler?real-time PCR system Data Analysis
software version 3.5.5 and the ‘‘Fit Points’’ algorithm.
The cycle at which the amplification curve crosses the
threshold was defined as Ct (cycle of threshold), and the
standard curve was constructed from the mean Ct values
of the triplicate determination.
Measurement of PCR efficiency. The PCR efficiency
was calculated using the slope of the standard curve
according to Formula1 as follows.28)
Formula1:
PCR efficiency ¼ 10ð?1=slopeÞ
Measurement of conversion factor and calculation of
GMO amount. The copy number of each sample was
obtained as the mean value of triplicates compared to the
optimal standard curve.23,24)The ratio of the copy
number of RRS and Le1 in each genuine seed was
calculated using Formula2 below and was defined as the
conversion factor (Cf). The GMO amounts (%) were
calculated using Formula3 below and the defined Cf.
Formula2:
Cf¼ ðcopies of RRS in the DNA extracted from GM
seedsÞ=ðcopies of Le1 in the DNA extracted
from GM seedsÞ
Formula3:
GMO amount (%)
¼ ðcopies of RRS in the DNA extracted from an
unknown sample ? 100Þ=ðcopies of Le1 in the
DNA extracted from an unknown sample ? CfÞ
Statistical analysis. All values are expressed as
means ? standard deviation of the mean. Statistical
comparisons were performed by Student’s t-test. In all
cases, probability (P) values below 0.05 were considered
significant.
Results and Discussion
Optimization of PCR conditions
To optimize real-time PCR conditions using the
LightCycler?real-time PCR system, specific primers,
probes, and pDNA, we examined several factors such as
MgCl2concentration and the speed of the temperature
change between the annealing temperature and the
denaturing temperature under the PCR conditions. Le1
and RRS of the gDNA extracted from soybean grain
flour were determined using the control pDNA as the
reference standard. The optimal MgCl2 concentrations
in the PCR reaction were investigated at a concentration
of 2.0mM to 4.0mM. With increasing MgCl2 concen-
tration in the PCR reaction, the Ct of both Le1 and RRS
gradually decreased and the detection range appeared to
be expanded. Hence, we concluded that the optimal
MgCl2concentration is 4.0mM in each reaction buffer.
To optimize the amplification curves, we examined
the speed of the temperature change between the
denaturing temperature and the annealing temperature
under the PCR conditions using the LightCycler?real-
time PCR system. As shown in Fig. 1, among the
several conditions examined, satisfactory amplification
curves were observed when the speed of the rising step
from the annealing temperature to the denaturing
temperature was slightly rapid (1?C/s) and the speed
of the cooling step from the denaturing temperature to
the annealing temperature was even faster (20?C/s).
Analysis of PCR efficiencies for gDNA and pDNA
using the established real-time quantitative system
We have reported the real-time PCR detection method
for construct-specific quantification of GM soybeans and
maize using the reference pDNA as a standard molecule
and an ABI PRISM 7700 system as the real-time PCR
equipment,23)but comparable studies between pDNA
and gDNA have not yet been performed. We concluded
that these studies are important, because comparable
behavior between pDNA and gDNA in the PCR is
reflected in the characteristics of the standard curves
obtained.
Hence, to assess the validity of the reference pDNA
for quantitative analysis of GM soybeans, we compared
the PCR efficiencies of the pDNA and gDNA (For-
mula1). The standard curves of pDNA were calibrated
by the pDNA series (five concentrations of the control
plasmids 40, 250, 3,000, 40,000, and 500,000 copies per
reaction). The standard curves of gDNA were calibrated
by the gDNA series (five concentrations of diluted DNA
extracted from 100% RRS soy 0.005, 0.05, 0.5, 5, and
50ng per reaction). As shown in Table 1 and Fig. 2, the
average calculated R2values of the standard curves
using pDNA for Le1 and RRS were 0.9993 and 0.9999
respectively. The average calculated R2values of the
standard curves using gDNA for Le1 and RRS were
0.9980 and 0.9999 respectively. Since approximately
50,000 copies of Le1 are contained in 50ng of soybean
DNA, 40 copies of RRS in 50ng of soybean DNA
correspond to a GMO content of approximately 0.1%.
These results suggest that the detection limit is 0.1%. As
shown in Fig. 2, the pDNA and gDNA standard curves
show nearly equal slopes for Le1 and RRS. Table 2
Quantification of Genetically Modified Soybeans823

Page 4

shows a statistical comparative analysis of the calibra-
tion curves set up with pDNA and gDNA using the
established detection system. As shown in this table, it is
clear that the slopes of gDNA and pDNA are not
significantly different for either Le1 or RRS. The average
PCR efficiencies for pDNA and gDNA were 1.964 and
1.953 in Le1 and 1.980 and 1.996 in RRS respectively.
These results suggest that pDNA and gDNA behave in a
similar way in the established detection system.
Determination of Cf
To calculate the RRS content (%) according to
Formula2, described in ‘‘Materials and Methods,’’ we
had to determine the conversion factor (Cf) as described
in Formula2.23)The measurements of Cf in DNA
extracted from 100% RRS were performed by different
researchers from three laboratories using the Light-
Cycler?real-time PCR system equipment set in each
laboratory, and were run in triplicate for Le1 and RRS.
The Cfvalues determined in the three laboratories were
0.78, 0.88, and 0.84. The average value obtained from
the three laboratories was defined as the Cf for
calculation of RRS content (%) in the soybean samples.
The averaged Cf for the RRS quantification was 0.83.
The relative standard deviation (RSD) value of the Cf
was 5.5%.
Accuracy, repeatability, and reproducibility for the
quantification of RRS in soybean samples
The accuracy, repeatability, and reproducibility of
RRS quantification using the established detection
method were assessed. The DNA extracted from a 5%
RRS mixed sample was repeatedly amplified five times
a day. Each sample was run in duplicate for Le1 and
RRS. The RRS content (%) of each sample was
calculated using Formula3, described in ‘‘Materials
and Methods.’’23)Table 3 shows the results of this
reproducibility study of the calculated RRS content (%)
for the same day and three different days using the
established method. As can be seen, the calculated RRS
Cycle number
Cycle number
B
C
D
0.1
0.06
10
10
50
10
10
50
50
A
0.1
0.0610
10
50
a
b
c d e
a bc d e
f g
hi
j
f g hi
j
Fluorescence F1/F2
Fluorescence F1/F2
Fig. 1.Amplification Plots of PCR Product from pDNA and gDNA Using the LightCycler Real-Time System.
Le1 amplification plots from pDNA (A) and gDNA (B); RRS amplification plots from pDNA (C) and gDNA (D). a, 500k copies; b, 40k
copies; c, 3k copies; d, 250 copies; e, 40 copies; f, 50ng; g, 5ng; h, 0.5ng; i, 0.05ng and j, 0.005ng.
Table 1. Quantitative Results for the Control Plasmid and the RRS Genome
Sample TargetStandard (copies) Cta
SDb
SampleTargetStandard (ng) Cta
SDb
pDNA Le14035.02
32.66
28.79
24.83
21.34
0.63
0.82
0.65
0.77
0.85
gDNA Le10.005
0.05
0.5
5
50
37.58
34.83
30.99
27.59
24.20
1.16
1.00
0.38
0.38
0.35
250
3,000
40,000
500,000
pDNA RRS40 32.93
30.39
26.58
22.82
19.17
0.83
0.74
0.58
0.61
0.69
gDNA RRS 0.005
0.05
0.5
5
50
35.94
32.64
29.31
25.90
22.43
1.10
0.65
0.38
0.33
0.21
250
3,000
40,000
500,000
aCt, cycle of threshold (mean of six replicates for standard curves).bSD, standard deviation.
824A. TOYOTA et al.

Page 5

content ranges from 4.35 to 4.95% for three different
days, the RSD values range from 2.01 to 4.51% for one
day, and the RSD value is 7.00% for three different
days. These results indicate that we detected the RRS
content of samples with a good accuracy and precision
by the detection system developed in this study.
Quantitative results of RRS content in the soybean
samples
To attempt to apply this method further to other level
samples, GMO content (%) was measured using the
established detection method and calculated for 0, 1.0,
and 5.0% of RRS mixed samples and an IRMM-5
(5.0%) sample of a certified reference sample. As shown
in Table 4, the calculated GMO contents of the 0, 1.0,
and 5.0% RRS mixed samples and the IRMM-5 sample
were 0, 0.86, 4.74, and 4.62% respectively. The RSD
values of the 1.0% mixed sample, the 5.0% mixed
sample, and the IRMM-5 sample were 22.28, 7.00, and
12.42% respectively. These results indicate that all the
calculated RRS contents for these samples were very
close to the expected theoretical values and are
reasonable.
Le1 (pDNA)
20
25
30
35
40
0246
Log copies
Ct
y = −3.3876x + 40.575
R2 = 0.9993
E=1.96
A
Le1 (gDNA)
20
25
30
35
40
−3
−2
−1
012
Log ng
Ct
y = −3.3987x + 30.014
R2 = 0.998
E=1.95
B
RRS (pDNA)
15
20
25
30
35
0426
Log copies
Ct
y = −3.3767x + 38.384
R2 = 0.9999
E=1.98
C
RRS (gDNA)
15
20
25
30
35
40
−3
−2
−1
012
Log ng
Ct
y = −3.3763x + 28.229
R2 = 0.9999
E=2.00
D
Fig. 2.Standard Curves of PCR Products Amplified from pDNA Series and RRS gDNA Series.
Le1 standard curves obtained from pDNA (A) and gDNA (B), RRS standard curves obtained from pDNA (C) and gDNA (D). Six standard
curves are indicated in each column. The standard curves for A and C were obtained from 40 to 500,000 copies of pDNA, and those for B and D
were obtained from 0.005 to 50ng of gDNA.
Table 2.Statistical Analysis of the Results of Calibration Curves Set up with pDNA and gDNA Using the Established Detection System
Target Criterium
Type of
calibrator
Mean of
data set
Variance of
data set
n
T statistical
data
P (T < t)
two-sided
Le1slopepDNA
gDNA
pDNA
gDNA
?3:416
?3:463
1.964
1.953
0.013
0.073
0.002
0.011
6
6
6
6
?0:379
0.720
E0.2270.829
RRS slopepDNA
gDNA
pDNA
gDNA
?3:377
?3:344
1.980
1.996
0.019
0.035
0.003
0.005
6
6
6
6
0.5020.637
E
?0:606
0.571
Table 3.
tive Results of 5% RRS Mixing Sample
Accuracy, Repeatability, and Reproducibility of Quantita-
Same day
(n ¼ 5)
2
Three days
(n ¼ 15)
13
GMO Amount (%)
RSD (%)
4.95
2.00
4.35
2.01
4.91
4.51
4.74
7.00
Quantification of Genetically Modified Soybeans825