T₅ seeds (15 in number) for each of two transgenic chickpea lines, namely, BS 100B and BS 100E expressing cry1Ac gene under the control of the Arabidopsis Rubisco small subunit gene promoter and tobacco SSU terminator (Supplementary Figure S1), were procured from the Department of Biotechnology-Assam Agricultural University Centre, Assam Agricultural University, Jorhat, Assam, India, during 2013. The transgenic lines carrying cry1Ac at a single locus were used as donor (male) parents in chickpea backcrossing program; the lines are reported to accumulate a high level of Cry1Ac protein (˃ 50 μg g⁻¹ leaf tissue) that causes 80–100% neonatal H. armigera larval mortality (Hazarika et al., 2019). The high-yielding commercial cultivars PBG7 (desi) and L552 (kabuli) were used as recipient (female) parents. F₁ plants of PBG7 × BS 100B (designated as Cross A), L552 × BS 100E (Cross B), and PBG7 × BS 100E (Cross C) were backcrossed with their respective recipient parents to obtain BC₁F₁ seeds that were sown to generate BC₁F₁ populations. F₁ plants, BC₁F₁, BC₁F₂, BC₁F₃, BC₂F₁, BC₂F_2, and BC₂F₃ populations were raised under contained conditions (Supplementary Figure S2; Supplementary Table S1) in a net house (30-mesh screen) at Experimental Farms, Department of Plant Breeding & Genetics, Punjab Agricultural University (PAU), Ludhiana. The populations were grown in plots comprising 25 rows with 2 m length and row-to-row distance of 40 cm following normal agronomic practices during the month of October. The schematic overview of marker assisted-backcross breeding of commercial chickpea cultivars with transgenic lines is shown in Figure 1.

Genomic DNA was extracted from tender twigs of 20-day-old BC₁F₁, BC₁F₂, BC₂F₂, and BC₂F₃ populations, transgenic donor parents BS 100B and BS 100E, and non-transgenic recipient parents PBG7 and L552 according to the miniprep method. For quantification, the extracted DNA was electrophoresed on 0.8% (w/v) agarose gel using PowerPacHC (Bio-Rad, United States) at 50 V for 2 h; ethidium bromide-stained gel was visualized under UV light and photographed on a 110 V AlphaImager HP imaging system (ProteinSimple, United Kingdom).

The foreground selection of backcross populations was carried out through PCR using cry1Ac-specific (Accession Number M11068, Hazarika et al., 2019), internal forward 5-TAT​CTT​TGG​TCC​ATC​TCA​ATG​GG-3 and reverse 5-GTG​TCC​AGA​CCA​GTA​ATA​CTC-3 primers to amplify 757 bp transgene. PCR mixture (20 µl) contained 50 ng genomic DNA (2 µl), 10 µM of each primer (0.6 µl), 1 mM dNTPs (4 µl), 25 mM MgCl₂ (1.5 µl), 5 × Green GoTaq Flexi buffer (4 µl), 5 units GoTaq DNA polymerase (1 µl) [Promega, United States] and nuclease-free water (6.3 µl). The reaction mixtures were placed in a GeneAmp PCR System 9700 (Thermo Fisher Scientific, United States) programmed for an initial denaturation at 94 C for 4 min, followed by 35 cycles of denaturation at 94 C for 50 s, annealing at 58 C for 1 min, extension at 72 C for 1 min, and concluded by a final extension at 72 C for 7 min and held at 4 C prior to storage. The amplicons were resolved on 1.5% (w/v) agarose gel, visualized, and photographed. The statistical significance for cry1Ac segregation data was determined by Chi-square analysis using the formula: χ² = (O-E)²/E, where O is the observed value and E is the expected value.

Cry1Ac expression in BC₁F₁ plants and transgenic donor and non-transgenic recipient parents was quantified through ELISA using Cry1Ac QuantiPlate kit (EnviroLogix, United States) according to the manufacturer’s instructions. A total of two leaflets (10 mg) of each plant were homogenized in an Eppendorf grinding tube for 20–30 s by adding 500 µl of 1 × extraction buffer. Each leaf tissue sample (50 µl) was diluted in a 1:11 ratio by adding 550 µl of 1 × extraction buffer; thereafter, 100 µl each of diluted sample, negative control, and positive calibrator was dispensed in the ELISA plate, followed by parafilm masking and incubation at an ambient temperature for 15 min. The assay was performed in triplicate. Cry1Ac-enzyme conjugate (100 µl) was added to each well, and the plate was again covered with parafilm and incubated for 1 h. After incubation, the parafilm mask was removed and well contents were agitated vigorously to decant the wells. The vacant wells were flooded with washing buffer and agitated to decant; the washing step was performed thrice. Then substrate (100 µl) was added to each well and mixed thoroughly, followed by plate covering with parafilm and incubation for 20 min. The reaction was terminated by adding 100 µl of stop solution to each well. The ELISA plate was read in a 96-well ELISA plate reader Infinite 200 Pro (Tecan, Switzerland) at 450 and 600 nm. The optical density (OD) values of samples and positive calibrators were analyzed using a Microsoft Excel sheet to generate a linear scale graph of the mean OD of each calibrator against its Cry1Ac concentration (Supplementary Table S2). The amount of Cry1Ac protein in each leaf tissue sample (µg g⁻¹) was determined using the formula {(OD of sample - mean OD of negative control) - 0.425/0.127} × dilution factor 1 (38.46) × dilution factor 2 (11)/1,000 (Supplementary Table S2). The data were analyzed for mean ± standard deviation using Microsoft Excel 2007 software at default settings.

The background selection of cry1Ac-positive BC₂F₂ plants was carried out using Simple Sequence Repeat (SSR) markers. As a preliminary step, polymorphism analysis was undertaken on parents PBG7 and BS 100E using 210 markers belonging to the following series: CGMM, CaM, GA, GAA, TA, TAA, TS, TR, NCPGR, H, and CaSTMS 11 (Supplementary Table S3). The amplified products were resolved on 6% (w/v) PAGE, and marker data were scored based on differential separation of amplicon(s). BC₂F₂ plants possessing maximum recurrent parent genome were identified with reproducible polymorphic SSR markers (Supplementary Table S3). The percent recurrent parent genome recovery in a BC₂F₂ plant was calculated as the sum of the number of alleles corresponding to recurrent parent detected by polymorphic markers divided by the total number of alleles detected by polymorphic and cry1Ac-specific markers.

The agronomic performance of BC₂F₂ population was assessed for plants analyzed for recurrent parent genome recovery, and of BC₂F₃ population was based on three progeny plants (from each BC₂F₂ plant) having phenotype similar to the recurrent parent. The data were recorded on days to 50% flowering, number of branches per plant, days to maturity, plant height, number of pods per plant, number of seeds per plant, 100-seed weight, biological yield, seed yield per plant, and harvest index, compared with the recurrent parent and analyzed for percent phenome recovery in BC₂F₂ plants and for mean ± standard deviation in BC₂F₃ plants.

Four-month-old morphologically healthy plants were analyzed for toxicity to H. armigera using two approaches, i.e., detached leaf bioassay and whole plant bioassay given by Sharma et al. (2005a), Sharma et al. (2005b) with modifications. Detached leaf bioassay: The terminal twigs having fully expanded leaflets were plucked from F₁, backcross population (BC₁F₁, BC₁F₂ and BC₂F₂), transgenic donor parent and non-transgenic recipient parent plants, and placed on 3% (w/v) agar (HiMedia, India) medium slants in sterile 500 ml plastic cups and used for bioassay. H. armigera larvae (3rd to 4th instar) collected in February from chickpea fields were reared individually in bioassay cups and maintained initially on non-transformed tender chickpea twigs, followed by growth on a semi-synthetic diet (Armes et al., 1992) until pupation. The pupae were kept on moist sponges covered with filter paper (Whatman, United States) in plastic containers till the emergence of adults that were paired in oviposition chambers i.e., cell pots wrapped in black paper on all sides and covered with muslin cloth on top. The adults were fed on 5% (v/v) honey solution by hanging honey-soaked cotton swab inside each oviposition chamber. Subsequently, egg laying occurred on the muslin cloth that was shifted to bioassay cup containing semi-synthetic diet for egg hatching, thereafter, neonates were used for bioassay of plant twigs. Ten neonate larvae were released in each bioassay cup and incubated in a growth chamber (Saveer Biotech Limited, India) maintained at 25 ± 2 C, 14 h light: 10 h dark period and >65 ± 5 percent relative humidity. The bioassay was replicated thrice and performed in the Pulses Entomology Laboratory, Department of Plant Breeding & Genetics, PAU, Ludhiana. The assayed plants were visually scored for the damage caused by neonate larvae after 96 h of release on a scale of 1–9 (1 = < 10% leaf area damaged, and 9 = > 80% leaf area damaged) given by Sharma et al. (2005a) for detached leaf assay. The larval mortality rate was compared among plants of backcross populations and donor and recipient parents to monitor the relationship between percent larval mortality and Cry1Ac protein concentration. The data were analyzed for mean ± standard deviation.

Whole plant bioassay: The assay was carried out under net house conditions on plants grown in plots with row to row distance of 40 cm and plant to plant spacing of 10 cm, according to the method given by Sharma et al. (2005b) for screening chickpea against H. armigera under greenhouse conditions with modifications. The healthy plants at the flowering stage from BC₁F₃ and BC₂F₃ populations and donor and recipient parents were covered with cages sized 25 × 25 × 75 cm³. The cages made of galvanized iron wire (2 mm in diameter) were supported by four vertical bars and covered with a muslin cloth bag. The experiment was performed in triplicate by caging three plants of each population, parent individually and releasing 10 H. armigera neonatal larvae on each plant, and terminated after 120 h when significant leaf area was damaged in recipient parents. The plants were scored for leaf-feeding visually on a 1-9 scale (where 1 = ˂ 10%, 2 = 11–20%, 3 = 21–30%, 4 = 31–40%, 5 = 41–50%, 6 = 51–60%, 7 = 61–70%, 8 = 71–80% and 9 = > 80% leaf area and/or pods damaged). The number of surviving larvae was recorded and individually placed in 25 ml plastic cups to express the data as percent larval mortality that were analyzed using Microsoft Excel 2007 software. No insecticide was applied in the experiment.

Data on Cry1Ac protein concentration, leaf-feeding, and larval mortality in backcross populations are presented as mean ± SD of three replicates. Statistical significance for the segregation data was determined using Chi-square analysis; calculated Chi-square value > table value was considered statistically significant at 5 percent level of significance.

F₁ plants developed from Cross A, Cross B, Cross C, and transgenic donor parent and non-transgenic recipient parents were analyzed for toxicity to H. armigera. F₁ plants obtained from Cross A (seven in number), Cross B (seven), Cross C (three), and donor parent displayed 100% H. armigera neonatal larval mortality and negligible (<10–20%) leaf-feeding damage, whereas recipient parents exhibited 23.33–30% larval mortality on an average with significant (51 to 70%) leaf-feeding damage (Table 1). F₁ plants toxic to H. armigera were backcrossed to generate BC₁F₁ populations.

The foreground selection of two BC₁F₁ populations derived from Cross A and Cross B, comprising 130 and 50 plants, respectively, was carried out through PCR using cry1Ac-specific primers. An amplicon corresponding to cry1Ac was detected in 46 (35.38%) BC₁F₁ plants obtained from Cross A (Supplementary Figure S3; Supplementary Table S1). The transgene segregation in a ratio of 1:1.8 deviated significantly from the 1:1 ratio expected if transgene was inserted at a single locus (Table 2). BC₁F₁ raised from Cross B segregated for the transgene in an expected Mendelian ratio of 1:1, as 25 (50.0%) plants were found to be cry1Ac positive (Supplementary Figure S4).

The recombinant protein concentration was estimated in 13 healthy BC₁F₁ plants derived from Cross A showing amplification of cry1Ac, nine Cross B plants along with transgenic donor parents BS 100B and BS 100E, and non-transgenic recipient parents PBG7 and L552 through ELISA (Supplementary Table S2). The average Cry1Ac protein concentration in both populations (11.03 to 11.71 μg g⁻¹ leaf tissue) was at par with donor parents (11.35 to 11.64 μg g⁻¹), whereas recipient parents did not exhibit any Cry1Ac concentration (Table 1). The BC₁F₁ plants (13 obtained from Cross A, nine from Cross B, and five from Cross C) had a phenotype similar to the recurrent parent, their bioassay for toxicity to H. armigera revealed that 13, 7, 4 plants from respective crosses, donor parents showed 100% H. armigera mortality and minor (<10–20%) leaf-feeding damage; in contrast, recipient parents exhibited 23.33–30% larval mortality with significant (51 to 70%) leaf-feeding damage (Table 1; Supplementary Figure S5). BC₁F₁ plants displaying toxicity to H. armigera were advanced for raising BC₁F₂ populations.

The foreground selection was carried out on two BC₁F₂ populations: the first comprising of 190 plants derived from Cross A, and the second consisting of 17 plants obtained from Cross B; cry1Ac amplification was detected in 16 (8.4%) and 13 (76.5%) plants (Supplementary Figures S6, S7; Supplementary Table S1), exhibiting non-Mendelian (1:10.9) and Mendelian (3.3:1) segregation ratios in the two populations, respectively (Table 2).

The insect bioassay was performed on nine BC₁F₂ plants raised from Cross B displaying phenotypic growth similar to the recurrent parent, along with transgenic donor and non-transgenic recipient parents. In the BC₁F₂ plants, donor parent displayed 100% H. armigera mortality and negligible (<10%) leaf-feeding damage; however, recipient parent showed 23.33% larval mortality with significant (61 to 70%) leaf-feeding damage (Table 1). BC₁F₂ plants showing toxicity to H. armigera were used to raise BC₁F₃ population.

Six out of 26 BC₁F₃ plants having comparable phenotype to the recurrent parent developed from Cross B, transgenic donor parent, and non-transgenic recipient parent were analyzed for toxicity to H. armigera. The plants revealed 16.67–93.33% larval mortality and variable (˂ 10–60%) leaf and pod feeding damage; donor parent exhibited 100% H. armigera mortality with negligible (<10%) leaf and pod feeding damage, whereas recipient parent showed 23.33% larval mortality and significant (61 to 70%) damage to leaves and pods (Table 1). Two BC₁F₃ plants were observed to display 93.33% insect mortality.

The foreground selection of BC₂F₂ population derived from Cross C and comprising of 83 plants led to the identification of 10 (12.05%) plants showing amplification of cry1Ac (Supplementary Table S1). The population deviated significantly for transgene segregation (1:7.3) from the Mendelian ratio (3:1) for a single insertion site (Table 2).

The donor and recipient parents were assessed for polymorphism using 210 SSR markers leading to the identification of 25 (11.9%) polymorphic markers (Supplementary Figure S8; Supplementary Table S3). The background selection using reproducible polymorphic markers on cry1Ac-positive BC₂F₂ plants demonstrated amplification pattern in ten BC₂F₂ plants (designated as 1, 2, 8, 9, 12, 20, 26, 33, 39, and 44) to be similar to recurrent parent “PBG7” profile (Figure 2), and the average recurrent parent genome recovery in these plants after two backcrosses was calculated to be 91.3% (Supplementary Table S4). The comparison of agronomic traits in BC₂F₂ plants with PBG7 revealed an average recurrent parent phenome recovery of 90.94% in BC₂F₂ plants (Table 3; Supplementary Table S5).

The randomly selected BC₂F₂ plants (designated as 2, 8, 20, 33, and 39) were bioassayed for toxicity to H. armigera. The results revealed that the selected plants and transgenic donor parent exhibited 100% larval mortality and negligible (<10%) leaf-feeding damage, whereas non-transgenic recipient parent was vulnerable to H. armigera with no larval mortality and significant (51 to 60%) leaf-feeding damage (Table 1). Subsequently, seeds of all ten BC₂F₂ plants were sown to obtain BC₂F₃ population.

BC₂F₃ population obtained from Cross C, consisting of 128 plants was subjected to foreground selection for identifying BC₂F₂ plants homozygous for cry1Ac through recognition of BC₂F₃ plants carrying cry1Ac gene. The results revealed that on an average, 82.81% BC₂F₃ plants carried cry1Ac, and three (30%) BC₂F₂ plants designated as 26, 39, and 44 were homozygous for the transgene as all progeny plants (16 of plant no. 26, 10 of plant no. 39, and 20 of plant no. 44) contained the transgene (Table 2: Figure 3). On the contrary, the remaining seven BC₂F₂ plants designated as 1, 2, 8, 9, 12, 20, and 33 were hemizygous for cry1Ac with their BC₂F₃ progeny plants segregating in a ratio of 6:1, 2.3:1, 1.3:1, 3:1, 1.6:1, 4:1, and 3:1, respectively for transgenes that were found to fit in Mendelian 3:1 ratio expected for a selfed population (Table 2).

BC₂F₃ progeny plants belonging to seven BC₂F₂ plants, namely 2, 8, 20, 26, 33, 39, and 44, were assessed for agronomic performance. The results showed that the mean number of pods and seed yield of BC₂F₃ progeny plants derived from BC₂F₂ plant no. 20 and homozygous BC₂F₂ plants, namely 26, 39, and 44 were 53.67 ± 1.53, 54.00 ± 3.00, 55.00 ± 4.00, 53.67 ± 3.78, and 15.64 ± 0.51 g, 15.33 ± 0.98 g, 15.74 ± 0.62 g, 14.84 ± 0.53 g, respectively were statistically similar to mean number of pods (54.67 ± 3.05) and seed yield (15.90 ± 0.56 g) of recurrent parent (PBG 7) (Table 3; Supplementary Table S5).

The bioassay of BC₂F₃ progeny plants revealed 53.33–100% H. armigera larval mortality and variable (<10–40%) leaf and pod feeding damage; amongst these, the progeny of homozygous BC₂F₂ plants displayed 100% mortality with negligible (<10%) leaf and pod feeding damage (Table 1; Figure 4A,B). The larval mortality in transgenic donor parent and BC₂F₃ progeny plants was similar, whereas the non-transgenic recipient parent displayed no larval mortality and significant (51 to 60%) leaf, pod feeding damage (Figure 4C).