Multiple Japonica and Indica transgenic rice plants were generated via Agrobacterium-mediated transformation with four different T-DNAs ( Figure 1 ). The transgenic Japonica lines carrying the Act-1:HvSUT1:NosT and Glb-1:HvSUT1:NosT cassettes were abbreviated to A lines and G lines, respectively. The Indica rice lines containing the Glb-1:HvSUT1:NosT, GluA2:OsNAS2:NosT, and GluA2:OsNAS2:NosT-Glb-1:HvSUT1:NosT were designated as SC1, SC2 and DC1, respectively. All T-DNAs contained a single common sequence of the nopaline synthase (nos) terminator, except for the DC1 T-DNA which contained two nos terminators. Illustrations of the T-DNAs were prepared using SnapGene (GSL Biotech).

The rice genes coding for branching enzyme (OsSBE4) and sucrose phosphate synthase (OsSPS), are appropriate reference genes in the qPCR assay as they are each present as two copies at a single locus in the rice genome (Mizuno et al., 2001; Jiang et al., 2009). All oligonucleotide primers were designed using the Primer3 software (bioinfo.ut.ee/primer3-0.4.0/primer3/). Some parameters were modified from the default setting to achieve similar amplification efficiencies between the reference genes and T-DNAs in the qPCR assay, including product size (80-200 bp) and primer Tm (59-65 °C). The selected primer pairs were checked by BLASTn against representative genomes for Oryza sativa ssp. Japonica and Indica to check the specificity of each primer pair. Folding template of the reference genes and T-DNAs were analysed by OligoArchitect™ Primer (offered by Sigma-Merck, USA) to eliminate primers that matched with predicted secondary structures in the templates. UNAFold software (http://sg.idtdna.com/UNAFold) was used to predict whether amplicons generated any secondary structure at the annealing temperature. Primer pairs that passed all of these requirements were synthesized by Sigma-Merck ( Supplementary Table S1 ).

Rice genomic DNA (gDNA) used for end-point PCR and the qPCR assay was isolated from leaves (100 mg) of transgenic lines (T₀, T₁ or T₂) using an Isolate II plant DNA kit (Bioline, USA). The gDNA was suspended in 50 μl of elution buffer (5 mM Tris-HCl, pH 8.5) and the DNA samples were diluted to achieve an average DNA concentration of around 4.26 ng/ul using sterilized MiliQ water. The quantity of gDNA was estimated using a Nanodrop 2000 instrument (Thermo Fisher Scientific, USA). For end-point PCR ( Supplementary Table S2 ), a 25 µl reaction consisted of 1 × GoTaq Green Reaction Buffer (Promega, USA), 2 mM MgCl₂, 0.4 mM dNTPs, 0.4 µM for each primer and 2 µl of 4.26 ng/µl DNA template. The PCR cycling protocol was as follows; initial denaturation for 10 min at 95 °C, followed by 30 cycles of 95 °C for 30 sec, 64 °C for 30 sec and 72 °C for 30 sec, ending with 10 min at 72 °C. PCR products were visualized on a 1.5% agarose gel.

An efficient, reproducible, and dynamic range qPCR assay was determined by making a standard curve using a 10-fold serial dilution of the target sequence (10⁴, 10³, 10², 10¹ and 10⁰ copies/μl). The absolute quantification of the copy number of the target sequence per haploid genome was calculated based on the mass of genomic DNA and the genomic DNA concentration determined by UV absorption at 260 nm based on the protocol of Applied Biosystems (Applied Biosystems, 2003). The mass of a single copy of the rice genome (haploid) was calculated as follows:

where m is mass (g) of genomic DNA and n is genome size (389 Mb) (Sequencing Project International Rice, 2005).

Then mass of the rice genomic DNA needed to achieve 10⁴ copies of the rice genome was calculated as follows

For this case, 4.26 ng/μl was equivalent to 10⁴ copies/μl of the rice genome. The stock concentration of rice gDNA was determined by UV absorption at 260 nm and diluted to achieve 4.26 ng/μl (10⁴ copies/μl). The 10⁴ copies/μl was serially diluted 10-fold in sterile Milli Q water to reach 10³, 10², 10¹ and 10⁰ copies/μl of the rice gDNA.

The standard curve was generated from Ct values plotted against the logarithm of the template concentration at each dilution with three replicates. Amplification efficiency was calculated from the slope of the standard curve (Bio-Rad Laboratories, 2006). The amplification efficiencies of the primer pairs for the reference genes and the T-DNAs were then compared with each other and chosen when within 5% of each other, and as close to 100% as possible. Based on the standard deviation of five dilutions in the standard curve, a concentration of DNA template was chosen with the lowest standard deviation and then used in the qPCR assay for zygosity analysis.

Reactions were done in a CFX96™ Real-time PCR detection instrument (BioRad, USA). A 10 μl reaction consisted of 5 μl (1 ×) KiCqStart SYBR Green qPCR ReadyMix (Sigma-Merck, USA), 300 nM for each primer and 2.5 μl of gDNA template (0.426 ng/μl ~10³ copies/μl). A standard two-step protocol was used as follows: 1 cycle for DNA denaturation at 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 sec and 60 °C for 30 sec and a melting curve was generated in 0.5 °C increments starting at 60 °C. Reactions for both reference gene (OsSBE4) and the T-DNAs were run in triplicate.

As the amplification efficiency of the reference gene and T-DNAs were similar and nearly 100% and within 5% of each other, the zygosity of transgenic plants was determined by the Livak method, 2^-ΔΔCt (Livak and Schmittgen, 2001). The difference between the Ct values of the T-DNAs and reference gene for the test plant was calculated (ΔCt), then normalized to the ΔCt for a calibrator plant to obtain copy number calculation (ΔΔCt). The resulting ΔΔCt value was incorporated to determine zygosity of the test plant. The 2^-ΔΔCt method was calculated using the following steps.

First, the Ct values of the T-DNAs (for example HvSUT1 and Nos terminator) were normalized to that of the reference gene (OsSBE4) for both test plants and calibrator:

Second, the ΔCt of test plant was normalized to that of the calibrator to obtain ΔΔCt:

Finally, the 2-fold difference was then found by 2 to the power of -ΔΔCt:

If the calibrator was hemizygous (that is a single-insert T₀ plant), the 2^-ΔΔCt of a homozygous single-insert plant will be twice that of the calibrator.

Southern blot analysis was conducted to provide a confirmation of T-DNA insert number independent to the qPCR assay. Rice gDNA of T₀ transgenic Japonica lines was digested with HindIII or EcoRI (New England BioLabs, USA) in 30 μl reactions which consisted of 3 μl buffer (10 ×), restriction enzyme (30 units) and 3 μg gDNA. The reaction was incubated at 37 °C for at least 4 hrs. Digested fragments were separated on a 0.8% agarose gel by electrophoresis at 30-40 V overnight. After the separation of the DNA fragments was completed, the DNA was depurinated by incubating the gel in 0.25 M HCl for 10 min. The gel was rinsed in MilliQ water three times before the DNA was denatured in the high salt denaturation solution for 30 min. Afterward the DNA was transferred directly to Biodyne B nylon membrane (Pall Life Sciences, USA) by capillary transfer overnight in 20 × SSC buffer (pH 7) containing sodium chloride (3 M) and sodium citrate buffer (0.3 M). The membrane was then dried at 80 °C for 15 min, and then fixed with UV cross-linker (GS Gene linker UV chamber, BioRad, USA) at 150 mJoule for 10 sec. After fixation, the membrane was processed with hybridization and detection.

A 498-bp fragment amplified from the HvSUT1 plasmid was used as a probe template for radioactive ³²P-dCTP labelling using random primers (Geneworks, Australia), which consisted of 3 μl of random primer (40 μM), 5 μl of 20 ng/μl probe template, 12.5 μl of 2 × oligo labelling buffer, 1.5 μl of DNA polymerase (Klenow fragment) (2 units/μl) and 5 μl of ³²P-dCTP in a 27 μl reaction. The labelled probe was incubated at 37 °C for 1-2 hrs, and then isolated by ethanol precipitation. Before hybridization, the membrane was mixed well and incubated with 5 × SSC overnight in the hybridization oven at 65 °C. The hybridization and detection processes were carried out according to Weiss (1992).

A transgenic Japonica rice line overexpressing the Act-1:HvSUT1:NosT (designated A5.1) was confirmed to carry a single insert in its genome by Southern blot, following digestion of genomic DNA with HindIII, which cuts once in the T-DNA construct ( Figures 1A, B ). The genomic DNA of this single copy line was used to optimize conditions of the qPCR protocol. For the expression construct Glb-1:HvSUT1:NosT (transformed plants designated G lines), there was a single EcoRI recognition site in hygromycin phosphotransferase (hpt) gene near on the left T-DNA border and two HindIII recognition sites close to the HvSUT1 gene near the right border of the T-DNA construct. Two Southern blot analyses of the G lines were carried out with HindIII and EcoRI digested gDNA. The G1.4 line showed two bands in EcoRI digested DNA and a single band in HindIII digested DNA, while the G3.1 line had two hybridising fragments in EcoRI digested DNA and one fragment in the HindIII digestion ( Figure 1A ). The two G3.1 and G1.4 transgenic lines were predicted to contain two inserts of the T-DNA in a tandem repeat, but inverted orientation ( Figures 1F, G ).

To reliably determine the zygosity of transgenic lines using the qPCR protocol, the comparative 2^-ΔΔCt method was used for copy number estimation (Bubner and Baldwin, 2004). This method requires similar PCR efficiencies of the T-DNA amplicons and an endogenous reference gene present as two copies at a single locus in the diploid rice genome. Thus, specific and well-matched sets of primer pairs for the T-DNAs and reference genes were designed and it was verified that their amplification efficiencies were close to 100% ± 5%. Three primer pairs for two endogenous reference genes, namely the starch branching enzyme 4 (OsSBE4) and the sucrose phosphate synthase (OsSPS), five primer pairs for the HvSUT1 gene (HvSUT1m, HvSUT1j, HvSUT1a, HvSUT1b, and HvSUT1c), and three primer pairs for the nos terminator (NosTY, NosTYA and NosTh), all present in the T-DNAs, were designed and screened in this study. All selected primer pairs were used in the qPCR assay with 10-fold serial dilutions of the genomic DNA of the A5.1 line (10⁰, 10¹, 10², 10³ and 10⁴ copies/μl) to determine their relative efficiencies ( Table 1 ). Three primer pairs (SBE4, NostYA and HvSUTj) showed high specificity, generating single PCR products (as seen on the 1.5% agarose gel; Figure S1 ) and a single peak in the melt curve ( Figure 2 ). Their reproducible amplifications had similar efficiencies close to 100%. The R² values for SBE4, HvSUTj and NosTYA primers were 0.997, 0.989 and 0.989, respectively. It was interesting that the three regression lines derived from the qPCR assays with SBE4, HvSUTj and NosTYA primers were parallel. This indicates that the reaction efficiencies of the two primer pairs are very well-matched and comparable within the gDNA template concentration range of 10⁰ – 10⁴ copies/μl. Two well-matched sets of primer pairs (SBE4/HvSUTj and SBE4/NosTYA) were selected for further analyses.

Based on a logarithmic standard curve generated from the serial dilution of the genomic DNA from the A5.1 transgenic line, an optimized qPCR assay was established to apply in the determination of homozygous and hemizygous plants, as described above. A dilution of 10³ or 10² copies/μl was within the dynamic range of the reaction and considered as the optimal range of starting concentrations of genomic DNA template in the qPCR assay. It was noted that the dilutions with DNA concentration lower than 10² copies/μl or higher than 10³ copies/μl had increased variation between technical replicates. Therefore, a concentration of 10³ copies/μl of genomic DNA was used as the working concentration in all subsequent experiments.

In 25 offspring of the A5.1 line analysed by this qPCR assay, the prediction was that there were four homozygous plants with the mean 2^-ΔΔCt value of 2.00 ± 0.107, 14 hemizygous plants (the mean 2^-ΔΔCt value of 1.01 ± 0.041) and seven null plants (the mean 2^-ΔΔCt value of 0.00 ± 0.008). These data, when analysed by Chi-square (χ² _0.05) test for goodness of fit, showed no significant difference between the expected segregation ratio (1:2:1) and the observed ratio with a P value of 0.583 ( Table 2 ). The results in Table 2 showed that Ct value differences (ΔCt values) for homozygous plants were from 0.008 to 0.17 cycles between Ct values of NosTYA and SBE4, while the ΔCt values of hemizygous plants was from 0.92 to 1.20. For the set of SBE4 and HvSU1j primer pairs, the ΔCt values were more variable with around 0.41 for homozygotes and from 1.20 to 1.46 for hemizygotes ( Supplementary Table S4 ). Moreover, on analysis the standard curves for the two sets of primer pairs, the reactions with NosTYA and SBE4 primer pairs had Ct values for the same dilutions differing by around 1 cycle, while reactions with the HvSUT1j and SBE4 primers had Ct values differing by approximately 0.6 cycles with the same dilutions ( Figure 2G ). This indicates that the primer set of SBE4 and NosTYA reflected the copy number of the OsSBE4 reference gene (two copies) doubling the copy number of the T-DNA (one copy) in the diploid genome. In the case of this A5.1 line with a single T-DNA insertion, the quantitative PCR assay reliably distinguishes the zygosity of the transgene utilising the set of primer pairs (NosTYA and SBE4).

To verify the accuracy of zygosity identification by the qPCR assay, T₂ plants of the A5.1 line with a single T-DNA insert were chosen for further segregation analysis. Around 100 progeny from one homozygous plant (No. 6), three hemizygous plants (No. 1, 2 and 5), and one null plant (No. 4), as predicted by the qPCR assay, were analysed for their segregation using hygromycin selection ( Supplementary Table S5 ). Progeny from the plants predicted by the qPCR to be homozygous were all resistant to hygromycin, whereas progeny from the null plant were all sensitive. Segregation of the T-DNA in the progeny of hemizygous plants showed no significant difference from the expected segregation ratio (3:1 for resistant: sensitive) ( Supplementary Table S5 ). The result of the hygromycin sensitivity assay thus verified the 100% accuracy of the zygosity determined for the A5.1 line as predicted by the qPCR assay.

This optimized qPCR assay using the hemizygous plant from the A5.1 line as a calibrator, was then adapted to determine the zygosity of other transgenic lines. Two transformed lines which contained two T-DNA inserts at the same locus in the rice genome (G1.4 and G3.1) were analysed to determine their zygosity by using the optimized qPCR assay. The qPCR assay identified 2^-ΔΔCt values of homozygous plants with four copies as 3.9 to 4.18, and 1.93-1.98 for plants containing two copies ( Table 3 and Supplementary Data S2 ). Only plants with four, two or zero copies of the T-DNA were detected in a 1:2:1 ratio, and the segregation ratios of T-DNA inserts in the G1.4 and G3.1 lines determined by the qPCR assay suggested a single insert locus. The results of the qPCR assay were therefore consistent with the Southern blot analyses. The use of the qPCR assay to confidently detect homozygous plants where two T-DNAs had integrated at the one locus is a powerful and useful application of this tool.

Having optimized the qPCR assay, this method was successfully applied to fast-track the identification of T-DNA copy number and zygosity of transgenic Indica rice plants in the T₁ generation. In this experiment, three populations of transgenic Indica rice lines carrying T-DNAs with a single transgene of the HvSUT1, the rice nicotinanamine synthase (OsNAS2), and both the transgenes (HvSUT1 and OsNAS2) were developed and designated as SC1, SC2 and DC1 lines, respectively ( Figures 1C–E ). All the transgenic lines contained the same nos sequence element as the A5.1 line ( Figure 1B ), so the qPCR assay with the same primer pairs for T-DNAs and reference gene could be used without any adaptation. The hemizygous plant from the A5.1 line was used as a calibrator for all qPCR runs. To achieve the fastest route for developing a homozygous line, we could detect single-insert lines in the T₀ generation.

The T-DNAs in all the T₀ putative transgenic lines exist in a hemizygous state. For single-transgene T-DNA lines (SC1 and SC2), Ct values (~ 25) of the NosTYA primer pair ( Figures 3D, E ) were higher by approximately 1, compared to the Ct values (~ 24) of the reference SBE4 primer pair ( Figures 3A, B ). The results were comparable to the hemizygous calibrator. Based on the 2^-ΔΔCt value of each putative transgenic line in the T₀ generation, the copy number of T-DNA inserts was predicted, with one indicating a hemizygous single insert, two for two hemizygous inserts, etc. The results are summarized in Figures 3G, H , with the putative transgenic events showing a single insertion of the T-DNAs as the most frequent, with seven of the 21 SC1 lines (33.33%) and ten of the 21 SC2 lines (47.62%). The transgenic lines likely to contain a single insertion per genome in the SC1 and SC2 lines had a 2^-ΔΔCt value of 0.8 - 1.0, and 0.7 - 1.1, respectively. For T₀ transgenic lines carrying the DC1 T-DNA, Ct values of the NosTYA primer pair were similar to that of the reference SBE4 primer pair, around 24, because in this case there were two nos terminator regions in the T-DNA ( Figures 3C, F ). The 2^-ΔΔCt value was ~ 2 for single-insert lines, ~ 4 for two-insert lines, etc. The result, shown in Figure 3I , indicates eleven of the 30 DC1 transgenic lines (36.67%) were determined to carry a single insertion of the DC1 T-DNA with 2^-ΔΔCt values of 1.8 – 2.2.

The T₀ lines carrying a single insert for the SC1, SC2 and DC1 T-DNAs were grown and used to screen for homozygous plants in the T₁ generation. The theoretical segregation of the T-DNA in single-insert lines would be consistent with a Mendelian ratio of 1:2:1 for homozygous, hemizygous and null plants. Seeds from 13 independent transgenic events in the T₀ generation, predicted to have a single insertion by qPCR, were grown to determine zygosity in the T₁ generation using qPCR ( Table 3 ). A total of 37 homozygous T₁ plants were predicted from the qPCR assay; 19 homozygous plants from five independent transformation events carrying the DC1 T-DNA with the 2^-ΔΔCt values of 3.78 - 4.41, 11 homozygous plants from three independent SC1 transgenic lines with the 2^-ΔΔCt values of 1.78 - 2.17 and seven homozygous plants from four independent SC2 transgenic lines with the 2^-ΔΔCt values of 1.75 – 2.02. A Chi-squared (χ² _0.05) analysis was used to analyse the segregation ratios of the T₁ progenies based on the qPCR results. In all cases, there was no significant difference between the observed ratios and that expected for Mendelian segregation of a single gene. The frequencies of homozygous plants carrying the SC1 T-DNA (23.9%) were much closer to the expected frequency of 25%. There were lower frequencies in the SC2 lines (only 10.3%) and in the DC1 lines (16.2%). There was no T₁ progeny plant from the single-insert lines with more than two copies identified by the qPCR assay. For the SC1 two-insert lines, homozygous plants were identified with 2^-ΔΔCt values from 4.07 - 4.14, compared to 1 for the hemizygous calibrator.

To verify the reliability and accuracy of the qPCR assay to determine zygosity and number of T-DNA insertions, 105 T₂ progeny from eleven homozygous single-insert plants and one homozygous double-insert line (shown in Table 3 ) were analyzed for the presence of the T-DNA by end-point PCR using the specific primer pairs for the transgenes ( Supplementary Table S2 ). The results ( Table 4 ) indicate that all progeny from the plants predicted to be homozygous for a single T-DNA insert were positive for the presence of the T-DNA by end-point PCR, demonstrating 100% accuracy for identifying homozygous plants attained from this qPCR assay in the T₁ generation. A similar result was also achieved for the homozygous, two-insert transgenic plants determined by the qPCR. Therefore, all materials from those homozygous plants and their progenies could be confidently used for further phenotypic, molecular, and physiological analysis.