pCAMBIA1300 (provided by the Centre for the Application of Molecular Biology in International Agriculture, Australia) was digested by BstXI and XhoI to remove the CaMV35s promoter and hygromycin phosphotransferase (hpt) gene, and then the cry2A^∗ gene driven by ubiquitin promoter (Christensen et al., 1992) was cloned into the multiple cloning sites to obtain the transformation vector pMF-2A^∗. The T-DNA region is displayed in Supplementary Figure S1. The pMF-2A^∗ and pCAMBIA1301 were introduced into A. tumefaciens EHA105, respectively. The two strains were used to infect an elite indica restorer line MH86 together at the ratio of 3:1. The callus culture and genetic transformation procedures were based on the method described by Lin and Zhang (Lin and Zhang, 2005).

The cry2A^∗ and hpt genes were detected by PCR and Southern blotting. The primers for the amplification of cry2A^∗ were Cry2A-F (5′-CGTGTCAATGCTGACCTGAT-3′) and Cry2A-R (5′-GATGCCGGACAGGATGTAGT-3′). The primers for the amplification of hpt were hpt-F (5′-AGAATCTCGTGCTTTCAGCTTCGA-3′) and hpt-R (5′-TCAAGACCAATGCGGAGCATATAC-3′). The PCR mixture contained 25 ng of template DNA, 2 μl of 10× buffer, 1.5 μl of 2 mM dNTP, 0.2 μl of each primer (10 μM), 0.2 μl of rtaq. The PCR reaction for the amplification of cry2A^∗ was implemented at 94°C for 5 min; 32 cycles of 94°C for 30 s, 57°C for 30 s, 72°C for 30 s; 72°C for 7 min. The PCR reaction to amplify hpt was implemented at 94°C for 5 min; 32 cycles of 94°C for 30 s, 58°C for 30 s, 72°C for 30 s; 72°C for 7 min.

The total genomic DNA of the transformants was isolated by CTAB method for Southern blotting (Murray and Thompson, 1980). There are HindIII sites in the T-DNAs of pMF2A^∗ (Supplementary Figure S1) and pCAMBIA1301, and there is no HindIII site between the probe (cry2A^∗ or hpt) and LB of the T-DNA. Therefore, the DNA samples (8 μg) were digested by HindIII. The plasmids pMF-2A^∗ and pCAMBIA1301 were also digested by HindIII and 0.5 ng of digested plasmids was loaded as positive control, respectively. The digested DNA samples were separated by 1% TAE agarose gel and transferred to nylon membrane. Each sample was hybridized by cry2A^∗ probe (a PCR-amplified fragment of cry2A^∗ by Cry2A-F and Cry2A-R) and hpt probe (a PCR-amplified fragment of hpt by hpt-F and hpt-R). The hybridization procedures were carried out as described by Southern (Southern, 1975).

The 160 T₁ generation transgenic plants derived from the T₀ transgenic plant with single copy of both cry2A^∗ and hpt were planted. The T₁ transgenic lines containing cry2A^∗ and no hpt were selected by PCR analysis, and were harvested individually. Then, the harvested seeds of these T₁ transgenic lines were sown in seedling bed and cry2A^∗ gene was detected in the seedlings (T₂ generation) through PCR analysis. If the cry2A^∗ gene did not segregate in T₂ generation population derived from the same T₁ plant, the T₁ plant was considered as a cry2A^∗ homozygous line. Three homozygous lines were used for further assays and the presence of cry2A^∗ and absence of hpt were confirmed by southern blotting.

In 2016, the concentrations of Cry2A^∗ protein in leaves (at tillering, heading and filling stage), stems (at tillering, heading and filling stage) and endosperms (at filling stage) of three transgenic homozygous lines and their hybrids were measured using the enzyme-linked immune sorbent assay (ELISA) kit AP005 from EnviroLogix Inc. (Portland, ME, USA). These hybrids were obtained by crossing the homozygous lines with II-32A, an elite indica cytoplasmic male sterility (CMS) line, respectively. Fresh leaf samples of approximately 20 mg were collected and homogenized by grinding in 500 μl of extraction buffer, and fresh stem and endosperm samples of approximately 40 mg were collected and homogenized by grinding in 1 ml of extraction buffer. All of the sample extracts were diluted in appropriate proportions. To quantify the Cry2A protein content, the Cry2Aa Calibrators (1, 5, and 10 ppb) provided by the manufacturer were used to create a standard curve. The Cry2A protein immunoassay procedures followed the protocol of the manufacturer. The optical density values were measured with a microplate reader at 450 nm wavelength. The Cry2A^∗ protein concentration (ppb) of each diluted sample was obtained from the standard curve. The final content of Cry2A^∗ protein was calculated according to the following equation: Cry2A^∗ protein content (μg/g) = [concentration value from the standard curve (ppb) × dilution multiple × extraction buffer volume (ml)]/fresh weight of each sample (mg). To calculate the percentage of Cry2A^∗ protein, the Bradford assay (Bradford, 1976) was used to quantify the total protein concentration. One-way analysis of variance and the LSD test were used to analyze the significant difference in the mean values between different stages of transgenic lines.

The resistance of the transgenic lines against SSB and YSB was tested in the laboratory. The SSB were fed in our lab, and the eggs of yellow rice borers were collected in the field and hatched in laboratory. The resistance of the three transgenic homozygous lines (MH86 as the susceptible control) and their hybrids (II Youming86 produced by crossing MH86 with II-32A as the susceptible control) was tested. The tested rice stems were collected from field at elongation stage and dissected into 8 cm in length. One glass tube (12 cm in length and 2 cm in diameter), which contained three dissected stems, was set as a biological repeat. Fifteen first-instar larvae of SSB of YSB were placed into the tube and sealed with cotton and black gauze. Three biological repeats were performed for each tested rice line. All of the glass tubes were incubated at 28°C and 80% relative humidity. After 6 days the larvae mortality was investigated. One-way analysis of variance and the LSD test were used to compare the significant difference in the mean values between transgenic lines and corresponding controls.

Insect resistance of the three transgenic homozygous lines in the field was evaluated at Huazhong Agricultural University experimental field in 2011, Wuhan, China. The transformation receptor MH86 was planted as susceptible control. The seeds of all tested materials were sown on rice seedling bed in June 6, 2011 and the seedlings were transplanted to paddy field in July 2. The seedlings of the four rice materials were distributed according to a randomized block design with three replications. Each replication consisted of 24 plants of each material in two rows, with a planting distance of 20 cm within a row and 26.7 cm between rows. The experimental plots were bordered by three rows of non-transgenic indica rice plants as protection. The experimental field was not sprayed with chemical pesticides during the entire growth period.

To evaluate the resistance of transgenic lines against stem borers in the field, each rice plant was infested with about 15 first-instar larvae of YSB at the booting stage. Rice dead hearts caused by stem borers was investigated after two weeks of artificial infestation. Meanwhile, leaf folds caused by natural infestation of rice leaf folder were counted. The insect resistance of transgenic hybrids and II Youming86 was evaluated in the field in 2013 using the similar method in 2011 except for that the planting distance was 17 cm within a row and 26.7 cm between rows. Because the eggs of YSB were not collected in this year, the insecticidal activity of the hybrids was evaluated under natural infestation in the experimental paddy. Non-repeated two-factor analysis of variance and LSD test were used to compare the significant difference in the mean values between transgenic lines and corresponding controls.

The design of experiments for evaluating the agronomic performance of the transgenic homozygous lines and their hybrids were conducted as described above in the insect assay in the field in 2011 and 2013, respectively. Chemical insecticides were sprayed during the growth period to protect the crop. The plant height of 20 plants in the middle of each plot was measured in the field at maturity, and these plants were harvested to record the number of panicles per plant, panicle length, number of grains per panicle, 1000-grain weight, seed setting rate and yield per plant. Non-repeated two-factor analysis of variance and LSD test were also used to compare the significant difference in the mean values between transgenic lines and corresponding controls.

Thermal asymmetric interlaced polymerase chain reaction (TAIL PCR) was adopted for isolating the flanking sequences of the T-DNA integration site in the transgenic plant. Three nested specific primers (SP1, SP2, and SP3) near the T-DNA right border (RB) and degenerate primer (AD8) were designed (Supplementary Table S1). The PCR mixture and reaction programs were based on the method described by Liu and Chen (2007). The tertiary amplification product was cloned to pEasy-T3 vector and sequenced. The flanking sequence of T-DNA RB flanking region was obtained by analysis of the sequencing results, and then the sequence was used for blastn in the bioinformatic website¹ in order to determine the T-DNA integration site in rice genome. Meanwhile, the putative flanking sequence of T-DNA left border (LB) flanking region was obtained through querying in bioinformatic website ² according to the blastn results. The genome primer 2AH2-tF based on the putative flanking sequence of LB and primer Tail-L1 in vector pMF-2A^∗ were designed (Supplementary Table S1), and the PCR product of the two primers was sequenced to confirm the flanking sequence of LB.

Only two positive co-transformation plants, which were named as 2AH1 and 2AH2, respectively, were obtained in T₀ generation by co-transformation of MH86 callus. The copy number of transgenic plants was confirmed by southern blotting. The results showed that both cry2A^∗ and hpt genes were single-copy in the 2AH2 family (Supplementary Figure S2). Three marker-free homozygous lines containing cry2A^∗ (named as 7-61, 8-30, and 8-62) derived from 2AH2 were selected by PCR and southern blot analysis (Figure 1).

In 2016, the Cry2A protein contents and the percentage of Cry2A protein in soluble protein of transgenic homozygous lines were measured at tillering, heading and filling stage (Figures 2A-C; Supplementary Table S2). In leaves, the concentrations of Cry2A protein ranged from 5.01 μg g^-1 to 26.79 μg g^-1 fresh weight, and the percentage of Cry2A protein ranged from 0.016 to 0.099% in soluble protein. In stems, the Cry2A protein concentrations ranged from 0.70 μg g^-1 to 2.35 μg g^-1 fresh weight, and the percentage of Cry2A protein ranged from 0.018 to 0.039% in soluble protein. In endosperms, the contents of Cry2A protein ranged from 3.34 μg g^-1 to 4.09 μg g^-1 fresh weight, and the percentage of Cry2A protein ranged from 0.026 to 0.031% in soluble protein. These results indicated that the Cry2A protein was stably expressed in the three transgenic homozygous lines at each growth stage.

To evaluate the commercial potential of the three transgenic homozygous lines, three hybrids, namely H_7-61, H_8-30, and H_8-62, were produced from the crossing of II-32A with the transgenic homozygous lines, respectively. The Cry2A protein content and its proportion in soluble protein in the Bt hybrids were also measured in 2016. In the hybrids, the concentrations of Cry2A protein in leaves ranged from 1.79 μg g^-1 to 14.56 μg g^-1 fresh weight, and the proportion of Cry2A protein ranged from 0.007 to 0.049% in soluble protein. In stems, the Cry2A protein concentrations ranged from 0.27 μg g^-1 to 0.95 μg g^-1 fresh weight, and the proportion of Cry2A protein ranged from 0.008 to 0.017% in soluble protein. In endosperms, the contents of Cry2A protein ranged from 1.56 μg g^-1 to 1.92 μg g^-1 fresh weight, and the proportion of Cry2A protein ranged from 0.012 to 0.013% in soluble protein (Figures 2D-F and Supplementary Table S3). These results suggested that the Cry2A contents in the hybrids were lower than in the homozygous lines.

In the laboratory assay of the insecticidal activity of the transgenic homozygous lines, the larvae mortality of SSB or YSB fed on non-transgenic control MH86 stems was approximately 22% after 6 days, and the surviving larvae showed normal development. In addition, the MH86 stems were severely damaged and plenty of frass was observed. The larvae mortality of SSB fed on transgenic rice stems was about 50%, and the surviving larvae showed developmental retardation: their body length was significantly shorter than that of survival from non-transgenic control MH86. Besides, there was a smaller amount of frass (Figures 3A,C). All YSB larvae fed on transgenic rice stem were killed after 6 days, and the stems suffered from slight damage and little frass (Figures 3A,D). These results revealed the high resistance of transgenic homozygous lines against YSB.

In the insect-resistance assay of the hybrids, the stem cuttings of II Youming86 showed similar level of damage as the stems of MH86, and the mortality of SSB larvae was about 25%. Besides, the surviving larvae developed normally. However, the three transgenic hybrids could kill about 50% of the SSB larvae, and the development of the surviving larvae was retarded (Figures 3B,E). The results indicated that the Cry2A protein content of the transgenic hybrids was lower than that of transgenic homozygous lines, but the hybrid stems showed equivalent insecticidal activity with their parents.

In the insect-resistance assay of the transgenic homozygous lines in the field in 2011, the leaves of non-transgenic control MH86 showed serious folds caused by rice leaf folders. Averagely two leaves were recorded as being folded per tiller and the percentage of tillers suffering from damage per plant was up to 99%. However, the percentage of tillers damaged in the three transgenic homozygous lines ranged from 4.78 to 8.25% and the leaves were only slightly damaged (0.06-0.11 leaves per tiller). Very few white heads were observed in the transgenic homozygous lines in contrast to the white heads in non-transgenic control MH86 per plant (11.84%) (Table 1; Figure 4A).

The resistance of the transgenic hybrids against target insects in the field was evaluated under natural infestation in 2013. The damage caused by rice leaf folders was not severe in this year. 15.46% of plant tillers suffered from damage in susceptible control II Youming86, and the transgenic hybrids showed only slight damage (Table 2). Some dead hearts caused by rice stem borers were observed in II Youming86 (Figure 4B), while almost no damage was observed in the transgenic hybrids. These results revealed that the transgenic rice was highly resistant to rice leaf folder and rice stem borer in the field.

Some agronomic traits of the three transgenic homozygous lines showed different degrees of variation. The panicle length and 1000-grain weight were significantly lower than those of non-transgenic control MH86 (P < 0.01), but the numbers of grains per panicle of line 7-61 and line 8-30 were significantly larger than that of MH86 (P < 0.05). All the transgenic homozygous lines showed no change in yield compared with wide-type MH86 (Table 3). The three transgenic hybrids showed little difference with non-transgenic control II Youming86 in agronomic performance, especially the hybrid of line 8-30 (Table 4).

The sequence analysis results of TAIL-PCR and subsequent routine PCR revealed that one complete T-DNA copy containing cry2A* gene was inserted into rice genome chromosome 2 in line 8-30. The information of the genes nearby T-DNA integration site was obtained by querying the flanking sequence in the bioinformatics website³. The results indicated that the T-DNA LB and transcription initiation site of gene Os02g43680 had a distance of about 1.5 kb, and the T-DNA RB was 2.99 kb away from the transcription initiation site of gene Os02g43690 (Figure 5A); besides, the T-DNA insertion led to a 7-bp deletion of rice genome (ctcgcgc) (Figure 5B).