Effect of the Matrix and Target on the Accurate Quantification of Genomic and Plasmid DNA by Digital Polymerase Chain Reaction

Abstract

In polymerase chain reaction (PCR)-based nucleic acid quantification, the DNA template type, primer/probe sequence, and instrument platform such as real-time quantitative PCR (qPCR) and digital PCR (dPCR) affect the accuracy and reliability of quantitative results. In this study, a plasmid DNA (pDNA) pBI121-screening, genetically modified (GM) rice SDrice genomic DNA (gDNA), and GM rapeseed SDrape gDNA, all carrying the same 11 screening elements, were used to prepare samples of different levels of gDNA and pDNA in a non-GM gDNA background. The comparison of the dPCR assays targeting the 11 screening elements revealed that the primer/probe set is a key factor that affects the accuracy of dPCR quantification. The optimal PCR method for the 11 screening elements was screened out from among the validated qPCR methods. The accuracy of the qPCR quantification of the low-level pDNA and gDNA test samples was low when pDNA was used as a calibrator, whereas that of the dPCR quantification was high and not affected by variations in template type and detection target. The validated dPCR assays targeting one or two elements can be randomly selected to characterize multiple-target pDNA reference materials (RMs). Low-level pDNA RMs with certified values can be used as quality controls for dPCR assays to avoid significant bias in gDNA quantification.

1. Introduction

Accurate quantification of the copy number concentration and ratio of target nucleic acid is important in the fields of disease diagnosis, genetically modified organism (GMO) testing, and oncology. Both real-time quantitative polymerase chain reaction (qPCR) and digital PCR (dPCR) are currently widely used quantitative methods. qPCR is a relative method that is dependent on calibration curve and reference materials (RMs). dPCR can achieve absolute quantification of DNA molecule copy number concentration by partitioning the PCR mix across thousands of small individual partitions without calibration [1]. It shows improved precision and reduced interference of PCR inhibitors and was demonstrated to have the potential to replace qPCR for the quantification of target DNA molecules [2,3,4]. Numerous dPCR methods have been reported and applied for diagnostic or other routine testing purposes [2,3,4].

Practical measurements by dPCR are affected by many factors such as the sequences of primers and probes, instruments, operators, reagents, and environmental conditions. qPCR requires RMs to calibrate quantitative results, and dPCR also requires at least one RM with a defined property value as quality control to monitor and control the operation of dPCR in addition to negative controls [5,6,7]. In this situation, DNA RMs are essential to guarantee the reliability and traceability of the quantitative results for both qPCR and dPCR analysis. In GMO quantification, a RM with a certified value close to the legal thresholds for GMO labeling should be used as a quality control [8]. If significant bias is observed from the quality control, we can conclude that the measurement results from real-life samples are not sufficiently reliable. Thus, the potential causes that result in the significant bias of the quality control should be identified.

Owing to the difficulty of obtaining real-life GMO materials for RM preparation, many plasmid RMs such as those for BCR-ABL1 mRNA quantification [9], Shiga toxin-producing Escherichia coli [10] and GM canola detection [11,12] were developed as quality controls of qualitative PCR or for calibration of qPCR results. By contrast, plasmid DNA (pDNA) and gDNA often show different behaviors known as matrix-matching errors on the qPCR platform, which result in significant deviations in measured results from true values if the plasmid RM is used to construct standard curves of qPCR [13]. A plasmid RM for GM Kefeng6 rice was quantified using dPCR, and no matrix effect associated with plasmid was found on the Fluidigm dPCR platform [14]. By contrast, a dPCR analysis of gDNA and pDNA revealed a greater dPCR technical variation in gDNA than in linearized plasmid DNA containing the same targets [15]. This means that the feasibility of using pDNA as a quality control to assess or control the dPCR performance of gDNA should be further evaluated.

In this study, a plasmid harboring 11 screening elements (pBI-screening) was developed to prepare a common pDNA RM for the quality control of GMO screening or quantification [16]. In addition, a construct harboring the same 11 screening elements was transformed into the recipients of rice and rapeseed to obtain a single-copy GM rice (SDrice) and a single-copy GM rapeseed (SDrape) [17,18]. The SDrice and SDrape containing the same 11 screening elements were also used to assess the impact of template type on the technical variability of dPCR. As dPCR does not rely on RMs, it has become a reference method for the characterization of RMs [19,20,21]. It has not been determined whether the selection of a detection target affects the characterization result of multiple-target RMs by dPCR. In this study, one plasmid pBI-screening and two gDNA of SDrice and SDrape were used as materials to assess the effect of the matrix and target on the accurate quantification of gDNA and pDNA by dPCR. The variability of the dPCR results of 11 linked elements between different type templates was estimated to investigate the behavior differences of pDNA and gDNA on dPCR platforms and to evaluate the feasibility of using pDNA as a quality control in the dPCR analysis of gDNA.

2. Materials and Methods

2.1. Materials

A plasmid pBI121-screening, GM rice (SDrice), and GM rapeseed (SDrape) were developed by our laboratory. The plasmid pBI121-screening harbors 11 commonly used screening elements (three promoters of P-CaMV35S, P-FMV35S, and P-NOS; four marker genes of Bar, NPTII, HPT, and Pmi; and four terminators of T-NOS, T-CaMV35S, T-g7, and T-e9) and the reference genes of maize, soybean, rapeseed, rice, cotton, and wheat [16]. Both SDrice and SDrape are homozygous single-copy events developed by transforming a constructed binary vector pBI121-ELEMENTS (Genbank No. HM047294) containing the 11 commonly used genetic elements in the T-DNA region [17,18].

2.2. DNA Extraction

A TIANprep Mini Plasmid Kit (Tiangen, Beijing, China) was used to extract pDNA. The extracted plasmid was first linearized by digestion with the endonuclease of Not I and then purified using CyCle-PURE Kit (OMEGA, Norcross, GA, USA). The gDNA from leaves were extracted and purified with the CTAB (Hexadecyl trimethyl ammonium Bromide) method. The purity of both gDNA and pDNA was evaluated by examining the OD260/OD280 and OD260/230 ratios using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). The rice and rapeseed gDNA concentrations were adjusted to approximately 50 ng/μL using 0.1 × TE buffer.

2.3. Sample Preparation

The copy number concentrations of SDrice and non-GM rice DNA were measured using a dPCR assay targeting the rice reference gene of PLD; those of SDrape and non-GM rapeseed DNA, using a dPCR assay targeting the rapeseed reference gene of CruA; and those of plasmid pBI121-screening, using the dPCR assays of PLD and CruA. The mixture preparation and reaction program of dPCR are detailed in the dPCR assay section (4.6). An average of 1.59 target copies per partition (λ) was demonstrated to lead to the most precise estimates [7]. The default droplet volume of 0.85 nL defined by Quantasoft 1.7.4.0917 of QX200 (Bio-Rad, Pleasanton, CA, USA) was taken into account, and the optimal λ value of 1.59 corresponds to 37,412 copies of target DNA per 20-μL reaction. The copy number concentrations of plasmid and gDNA were adjusted to approximately 3.7 × 10⁴ copies/μL. Four series of gradient samples were prepared to correspond to GMO contents of 5%, 1%, 0.5%, and 0.1% in copy number ratios. SDrice DNA was mixed with non-GM rice DNA to prepare SDrice gradient samples (S1-G-rice~S4-G-rice) as follows: first, 50 μL of SDrice DNA was thoroughly mixed with 950 μL of non-GM rice DNA to prepare the 5% solution (S1-G-rice) in copy number ratio. After that, 200 μL of the 5% solution was thoroughly mixed with 800 μL of non-GM rice DNA to prepare the 1% solution (S2-G-rice), and then 500 μL of the 1% solution was diluted twofold with 500 μL of non-GM rice DNA to prepare the 0.5% solution (S3-G-rice). Finally, 200 μL of the 0.5% solution was further diluted fivefold with 800 μL of non-GM rice DNA to prepare the 0.1% solution (S4-G-rice). The same dilution strategy was used to prepare pBI121-screening gradient samples by mixing the plasmid solution with non-GM rice DNA (S1-P-rice~S4-P-rice) and non-GM rapeseed DNA (S1-P-rap~S4-P-rape), respectively. Rapeseed is an allotetraploid species, with the theoretical copy number ratio of transgenes to the reference gene of CruA being 1/2 for single-copy homozygous GM rapeseed. The preparation of SDrape gradient samples (S1-G-rape~S4-G-rape) requires two times the volume of SDrape DNA solution compared with the preparation of SDrice gradient dilutions.

2.4. Primers and Probes

The reported primer/TaqMan probe sets targeting the 11 commonly used exogenous elements and two reference genes of PLD and CruA were collected. Some elements, such as P-FMV35S and P-CaMV35S, have several primer/probe sets (Table S1). The collected qPCR methods for the exogenous elements and reference genes have been validated and reported in published papers or standards. All primers and fluorescent probes were synthesized by Sangon Biotech Co., Ltd. (Shanghai, China). The 5′ and 3′ ends of the TaqMan probes were labeled with a reporter dye 6-carboxy-fluorescein and a quencher dye Black Hole Quencher 1 (BHQ1), respectively. After initial assessment, the primers and probes used in this study were further selected and outlined in

Table 1. Selected primer/probe sets for digital polymerase chain reaction (dPCR) assays.

Gene/Element	Primer/Probe	Sequence	Amplicon Size, bp	References
NPTII	qNPTⅡF63	CTATGACTGGGCACAACAGACA	101	Lu et al., 2012 [22]
	qNPTⅡR163	CGGACAGGTCGGTCTTGACA
	qNPTⅡFP90	CTGCTCTGATGCCGCCGTGTTCCG
pFMV35S	pFMV35S-F	CAAAATAACGTGGAAAAGAGCT	78	ISO/TS 21569-5:2016 [23]
	pFMV35S-R	TCTTTTGTGGTCGTCACTGC
	pFMV35S-P	CTGACAGCCCACTCACTAATGC
Bar	RapB-F1	ACAAGCACGGTCAACTTCC	60	Grohmann et al., 2009 [24]
	RapB-R1	GAGGTCGTCCGTCCACTC
	RapB-S1	TACCGAGCCGCAGGAACC
T-NOS	180-F	CATGTAATGCATGACGTTATTTATG	84	Reiting et al., 2007 [25]
	180-R	TTGTTTTCTATCGCGTATTAAATGT
	Tm-180	ATGGGTTTTTATGATTAGAGTCCCGCAA
HPT	qHPTF286	CAGGGTGTCACGTTGCAAGA	110	Lu et al., 2012 [26]
	qHPTR395	CCGCTCGTCTGGCTAAGATC
	qHPTFP308	TGCCTGAAACCGAACTGCCCGCTG
pCaMV35S	p35s-F	ATTGATGTGATATCTCCACTGACGT	101	Fu et al., 2017 [27]
	p35s-R	CCTCTCCAAATGAAATGAACTTCCT
	p35s-P	CCCACTATCCTTCGCAAGACCCTTCCT
PMI	PMIF240	ACTGCCTTTCCTGTTCAAAGTATTAT	96	Lu et al., 2012 [22]
	PMIR335	TCTTTGGCAAAACCGATTTCAGAA
	PMIP267	CGCAGCACAGCCACTCTCCATTCAGG
TE9	TE9-F	TGAGAATGAACAAAAGGACCATATCA	87	Debode et al., 2013 [27]
	TE9-R	TTTTTATTCGGTTTTCGCTATCG
	TE9-P	TCATTAACTCTTCTCCATCCATTTCCATTTCACAGT
Tg7	Tg7-F	ATGCAAGTTTAAATTCAGAAATATTTCAA	97	Debode et al., 2013 [27]
	Tg7-R	ATGTATTACACATAATATCGCACTCAGTCT
	Tg7-P	ACTGATTATATCAGCTGGTACATTGCCGTAGATGA
T35S	T35SM-F	CCCTTAGTATGTATTTGTATTTGTAAAATACTTC	83	Pansiot et al., 2011 [28]
	T35SM-R	GGATTTTAGTACTGGATTTTGGTTTTAG
	T35S-P	TATCAATAAAATTTCTAATTC
P-NOS	P-NOS-F	GTGACCTTAGGCGACTTTTGAAC	79	Debode et al., 2013 [27]
	P-NOS-R	CGCGGGTTTCTGGAGTTTAA
	P-NOS-P	CGCAATAATGGTTTCTGACGTATGTGCTTAGC
CruA	qCruAF	GGCCAGGGCTTCCGTGAT	101	Jacchia et al., 2009 [29]
	qCruAR	CCGTCGTTGTAGAACCATTGG
	qCruFP	AGTCCTTATGTGCTCCACTTTCTGGTGCA
PLD	KVM-159	TGGTGAGCGTTTTGCAGTCT	68	Mazzara et al., 2006 [30]
	KVM-160	CTGATCCACTAGCAGGAGGTCC
	TM-013	TGTTGTGCTGCCAATGTGGCCTG

[22,23,24,25,26,27,28,29,30].

2.5. Real-Time Quantitative PCR

The qPCR assays were performed on a CFX384 Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA) in a final volume of 10 μL that contained 1 μL gDNA, 1 × TaqMan Universal PCR Master Mix (Applied Biosystems, Foster City, CA, USA), 0.4 μL of each primer (10 μM), and 0.2 μL of probe (10 μM). The final concentrations of the primers and probes were identical to each other for both the exogenous elements and reference genes. The qPCR assays were performed as follows: 95 °C initial denaturation for 10 min; 50 cycles of 95 °C denaturation for 15 s and 60 °C annealing and extension for 1 min; and fluorescence measurement after the annealing and extension. The fluorescence signals were monitored and analyzed using the CFX Manager software 3.1 (3.1.1517.0823, Bio-Rad, Hercules, CA, USA).

2.6. Droplet Digital PCR

The dPCR assays were performed on the QX200 platform (Bio-Rad, Pleasanton, CA, USA) using dPCR Supermix for Probes kit (Bio-Rad, Pleasanton, CA), with the same final concentration of both the primer and probe as that of qPCR in 10 μL of the mixture. The prepared dPCR mixture with 70 μL of droplet generation oil was transferred into a Bio-Rad DG8 cartridge, which was placed in a QX200 Droplet Generator to generate water-in-oil droplets. Forty microliters of droplets was transferred into a 96-well plate for PCR amplification on a C1000 Touch thermal cycler (Bio-Rad, Pleasanton, CA, USA), in accordance with the following program: 95 °C for 10 min; 50 cycles of 94 °C for 30 s and 60 °C for 1 min; 98 °C for 10 min; and final cooling to 4 °C. After amplification, the 96-well plate was loaded onto a QX200 droplet reader (Bio-Rad) to read droplets using the Bio-Rad QuantaSoft software (v1.7.4.0917, Bio-Rad, Pleasanton, CA, USA). The default setting for thresholds was initially adopted to discriminate between positive and negative droplets. If the default setting failed to work, the threshold was manually set up to analyze the droplets.

3. Results

3.1. Effect of Method Sensitivity on dPCR Assays

The GMO content is also expressed as the copy number ratio of transgenes to taxon-specific reference genes. The quantification of GMO content requires accurate measurements of the copy numbers of transgenes and reference genes by dPCR. Therefore, dPCR assays should be well designed and sufficiently optimized, with primer/probe set screening, to guarantee that the transgenes have an amplification ability identical to the reference genes on a dPCR platform. When screening optimal primer/probe sets for dPCR assays using homozygous DNA as the quality control, some primer/probe sets were found to yield quantitative results that deviated significantly from the theoretical value (1.0). The insufficient sensitivity of the primer/probe set was speculated to be the reason for the deviation of the quantitative results. To assess the effect of method sensitivity on dPCR assays, the detection sensitivity of the 23 primer/probe sets collected was tested using serial diluted plasmid pBI121-screening solutions of 20, 10, 5, and 1 copy/μL as templates. The amplification results are shown in Table S2. Except for the primer/probe set NOS-QF/R/P of T-NOS and PMI-F/R/P of PMI, the lowest copy number that the other primer/probe sets can detect reached 5–20 copies per reaction (Table S2). The sensitivity test demonstrated that the validated methods had sufficient sensitivity to perform dPCR assays.

The dPCR assays of the 25 primer/probe sets collected were performed using three template types of linearized pDNA, SDrice gDNA, and SDrape gDNA to quantify the screening element/PLD (or CruA) copy number ratio. The measurement of pDNA provided two sets of data corresponding to the screening element/PLD copy number and screening element/CruA copy number ratios (Table S3). The theoretical copy number ratio between the screening element and the reference gene is 1.0 for pDNA and SDrice gDNA, and 0.5 for SDrape gDNA, as rapeseed is an allotetraploid species. The most primer/probe sets yielded accurate measurement results close to the theoretical value for both the pDNA and gDNA templates (Table S3). For the pDNA template, the measured copy number ratio between each element and PLD was equal to that between the corresponding element and CruA. This demonstrated that the reference gene PLD and CruA had the same amplification ability in plasmid dPCR assays. By contrast, some primer/probe sets with similar sensitivity generated quantitative results different in accuracy. For example, FMV35-QF/R/P of P-FMV35, P-nos-F1/R/Tm of P-NOS, NOS-QF/R/P of T-NOS, and NPTII-F1/R1/P1 of NPTII showed accurate pDNA quantitative results but significantly smaller gDNA quantitative results. The primer/probe set sF/sR/35S of P-35S generated significantly smaller quantitative results for both pDNA and gDNA. The quantitative results demonstrated that the method sensitivity did not exert an obvious effect on the accuracy of dPCR quantification. The validated real-time quantitative PCR method cannot be used directly for dPCR assay until it is further validated on the dPCR platform.

At least one primer/probe set for each screening element can accurately quantify both pDNA and gDNA (Table S3), and the annealing elongation temperature of 60 °C was tested to be compatible with the selected 11 primer/probe sets in dPCR assays (Figure S2). Therefore, one optimal primer/probe set of each element that yielded a measured average closest to the theoretical value (1.0 or 0.5) with the highest precision was selected for the subsequent analysis (Table 1). The selected primer/probe sets showed amplification abilities similar to those of the reference gene PLD and CruA and can be used to quantify the copy number ratio between the screening elements and reference genes. The increased structural complexity of gDNA might have affected the matching efficiency between the template and primer/probe, and the longer amplicons and uneven GC-content in primer sequence might also have produced biased copy number ratios on gDNA, all these factors led to the decreased amplification ability of some primer/probe sets. The primer/probe set is a key factor that affects the quantitative accuracy of dPCR. The optimal PCR method should be screened out through comparison if multiple validated methods are available for the same target.

3.2. Quantification of Low-Level Samples by qPCR

The homozygous gDNA (SDrice and SDrape) and pBI121-screening plasmid were used as calibrators to quantify the GMO content of four series of test samples corresponding to SDrice gDNA in non-GM rice gDNA (S1-G-rice–S4-G-rice), SDrape gDNA in non-GM rapeseed gDNA (S1-G-rape–S4-G-rape), pBI121-Screening in non-GM rice gDNA (S1-P-rice–S4-P-rice), and pBI121-Screening in non-GM rapeseed gDNA (S1-P-rape~S4-P-rape) on the qPCR platform by T-NOS and P-CaMV35s qPCR. The amplification plots and constructed standard curves of T-NOS, P-CaMV35S, reference gene PLD for rice, and CruA for rapeseed using different calibrator types are shown in Figure S1, the obtained regression equations and amplification efficiencies are listed in Table S4. The measured GMO contents of all samples are shown in Table S5. The gDNA samples (S1-G-rice–S4-G-rice, S1-G-rape–S4-G-rape) were accurately quantified using the gDNA calibrators but were significantly overestimated using the plasmid calibrators (Figure 1). By using both the gDNA and pDNA calibrators, the pDNA content in gDNA for samples S1-P-rice–S4-P-rice and S1-P-rape–S4-P-rape was significantly overestimated from the expected values, and the relative bias of the measured values exceeded the acceptable range of ±25% (Figure 1). The accuracy of qPCR was severely affected by the inherent technical variability expressed as the Ct readout. The pDNA template outputted a relatively smaller Ct value than the equivalent amount of gDNA template, resulting in the overestimation of the pDNA content in the gDNA background when using either gDNA or pDNA as calibrator. We speculate that the simpler the template structure, the more efficient it is to match the primer/probe in the annealing stage, which thus generates relatively small Ct values. The difference in structural complexity is speculated to be the cause of the matrix matching error, which led to a lack of commutability of the gDNA calibrator to the pDNA calibrator on the qPCR platform.

3.3. Quantification of Low-Level Samples by dPCR

Four series of test samples were quantified using the primer/probe sets listed in Table 1 that target 11 screening elements and 2 reference genes on the dPCR platform: the copy number of the transgenes in each sample was repeatedly measured using dPCR assays targeting the 11 screening elements, and the copy number of total DNA in the test samples was measured using a dPCR assay targeting the rice PLD gene or rapeseed CruA gene, with each dPCR assay for each sample set with three parallels. Each sample was given 11 sets of quantitative data of GMO content (Table S6), and the precision and trueness of each set of quantitative data were assessed for all samples. For the gDNA and pDNA samples at different levels, the dPCR assay of each element resulted in a quantitative value close to the expected value, with satisfactory precision (coefficient of variation (CV) < 25%) and trueness in the range of ±25% bias. The measured mean value of each ddPCR assay together with standard deviation (SD) was shown in Figure 2 for each test sample. Therefore, all dPCR assays fulfilled the requirements for trueness and precision according to the acceptance criteria for GMO analytical methods [31]. For the samples at the lower levels, some dPCR assays generated fluctuating measurement results expressed as relatively large CV and bias, but within the acceptable ranges of the GMO performance parameters. The decreased precision afforded by dPCR at low GMO content might have resulted from the cumulative errors incurred by pipetting and the low-concentration template heterogeneity. The 11 primer/probe sets selected obtained accurate quantitative results from four series of test samples, regardless of the gDNA or pDNA samples can be accurately quantified by dPCR assays. The accuracy of the dPCR quantitative analysis was not affected by the template types.

Each sample obtained 11 sets of measurement data targeting 11 elements. One-way analysis of variance was performed to evaluate the difference in significance between the data sets obtained using different dPCR assays targeting different elements for the same sample. F_test values were calculated by estimating the mean squares within a dPCR assay (MS_within) and the mean squares between dPCR assays (MS_between). The calculated F_test and p values at 95% confidence level for all test samples are shown in

Table 2. Statistical analysis of one-way analysis of variance of the quantitative data of the test samples by dPCR assays.

Sample		Variation Source	Sum of Squares (SS)	Degrees of Freedom (df)	Mean of Squares (MS)	p Value	F value	F0.05(10,22)
Type	No.
SDrape gDNA in non-GM rapeseed gDNA	S1-G-rape	Within-dPCR	0.482	10	0.048	0.097	1.919	2.297
		Between-dPCR	0.553	22	0.025
	S2-G-rape	Within-dPCR	0.052	10	0.005	0.529	0.925	2.297
		Between-dPCR	0.123	22	0.006
	S3-G-rape	Within-dPCR	0.011	10	0.0011	0.517	0.942	2.297
		Between-dPCR	0.026	22	0.0012
	S4-G-rape	Within-dPCR	0.003	10	0.0003	0.289	1.302	2.297
		Between-dPCR	0.006	22	0.0003
plasmid in non-GM rapeseed gDNA	S1-P-rape	Within-dPCR	0.898	10	0.090	0.067	2.133	2.297
		Between-dPCR	0.926	22	0.042
	S2-P-rape	Within-dPCR	0.059	10	0.006	0.082	2.017	2.297
		Between-dPCR	0.064	22	0.003
	S3-P-rape	Within-dPCR	0.016	10	0.002	0.922	0.419	2.297
		Between-dPCR	0.083	22	0.004
	S4-P-rape	Within-dPCR	0.004	10	0.0004	0.435	1.054	2.297
		Between-dPCR	0.008	22	0.0004
SDrice gDNA in non-GM rice gDNA	S1-G-rice	Within-dPCR	1.467	10	0.147	0.063	2.161	2.297
		Between-dPCR	1.493	22	0.068
	S2-G-rice	Within-dPCR	0.110	10	0.011	0.059	2.201	2.297
		Between-dPCR	0.110	22	0.005
	S3-G-rice	Within-dPCR	0.039	10	0.004	0.112	1.839	2.297
		Between-dPCR	0.047	22	0.002
	S4-G-rice	Within-dPCR	0.005	10	0.0005	0.282	1.317	2.297
		Between-dPCR	0.008	22	0.0004
plasmid in non-GM rice gDNA	S1-P-rice	Within-dPCR	0.589	10	0.059	0.069	2.110	2.297
		Between-dPCR	0.614	22	0.028
	S2-P-rice	Within-dPCR	0.055	10	0.006	0.502	0.960	2.297
		Between-dPCR	0.126	22	0.006
	S3-P-rice	Within-dPCR	0.013	10	0.001	0.465	1.012	2.297
		Between-dPCR	0.029	22	0.001
	S4-P-rice	Within-dPCR	0.001	10	0.0001	0.514	0.945	2.297
		Between-dPCR	0.003	22	0.0002

. The calculated F_test values were lower than the critical F_0.05(11,22) (2.297) value, with p values of >0.05. The statistical analysis indicated that the differences between the 11 dPCR assays were not significant compared with the differences within the same dPCR assay. The dPCR assays using different screening elements as targets all obtained accurate quantitative results for both the pDNA and gDNA samples. The selection of any one element accurately quantified the GMO content of the gDNA and pDNA samples in the background of non-GM gDNA. In the practical quantification of GMO content or characterization of RM harboring multiple linked targets, one or two targets can be randomly selected for DNA quantification using dPCR under the condition that the corresponding dPCR method is sufficiently optimized or validated.

4. Discussion

qPCR or dPCR methods should use RMs as quality controls to evaluate the overall quality and reliability of the data produced [6,7]. Replicate measurements of a certified RM are required to perform bias control and estimate the corresponding uncertainty contribution to be taken into account [8]. Owing to the difficulty in obtaining real-life GMO materials for RM preparation, the development of plasmid RMs provides an efficient strategy to solve the lack of GMO RMs in quantification. On the qPCR platform, gDNA samples can be accurately quantified by qPCR using homozygous gDNA as calibrants but cannot be accurately quantified using pDNA as calibrants; and pDNA samples cannot be accurately quantified using either homozygous gDNA or pDNA as calibrants. The quantitative results of qPCR further demonstrated the presence of matrix-matching errors between gDNA and pDNA [13]. The differences in conformation states between gDNA and pDNA were supposed to result in different PCR amplification efficiencies, especially during the early stages of PCR [32,33]. pDNA generated relatively small Ct values compared with the equivalent amount of gDNA, causing the overestimation of the quantification of pDNA samples by qPCR, regardless of whether homozygous gDNA or pDNA was used as calibrators. dPCR is an end-point method unaffected by the typical variation in Ct value [34,35]. Both gDNA and pDNA samples were accurately quantified using dPCR assays, with satisfactory precision and trueness, and the quantitative results did not show significant differences between the different dPCR assays. The matrix differences between gDNA and pDNA did not influence the accurate measurements of the gDNA and pDNA samples by dPCR. Therefore, no matrix-matching error was observed on the dPCR platform between gDNA and pDNA. The matrix effect between gDNA and pDNA prevents the use of pDNA RM to replace gDNA RM in calibrating or controlling the gDNA quantification on the qPCR platform. However, it allows the pDNA RM to be used as a quality control to monitor or control the dPCR performance of gDNA quantification.

The available powder or gDNA RMs are usually developed focusing on special GM events and can only be used for the detection of corresponding GM events. Plasmid molecules can be synthesized without relying on raw materials. Multiple detection targets can be co-integrated into the same vector to construct a common pDNA RM; thus, many plasmid RMs have been successfully developed [9,10,11,12]. A RM at the level as close as possible to the threshold stipulated in the legislation is preferred for bias control in DNA quantitative analyses [8]. Therefore, RMs at low GMO content levels must be developed and certified. Using the same method used in this study for preparing plasmid samples, we can produce low-level pDNA RMs suitable for different GM crops by diluting the same plasmid into different non-GM gDNA backgrounds. This study revealed that the 11 dPCR assays targeting 11 different elements obtained consistent quantitative results for both gDNA and pDNA samples. Thus, one or two elements can be selected as detection targets to characterize multiple-target pDNA RMs.

dPCR technology would take the place of qPCR for DNA quantification in the future owing to its advantages over qPCR, including its higher precision and independence from calibrants and standard curves. The validated primer/probe sets in qPCR methods are not necessarily suitable for dPCR analysis. Further validation on the dPCR platform is recommended by analyzing samples with defined contents or certified values. The key factor affecting the accuracy of dPCR assays was demonstrated to be the primer/probe set. In this study, 11 optimal primer/probe sets of 11 different elements were screened out. Any one primer/probe set can be directly used for GMO quantification and characterization of pDNA or gDNA RMs on the dPCR platform. The preparation of low-level pDNA RMs would provide alternative quality control for performing bias control in gDNA quantification by dPCR.