Development of flash-flood tolerant and durable bacterial blight resistant versions of mega rice variety ‘Swarna’ through marker-assisted backcross breeding

Bacterial blight (BB) disease and submergence due to flash flood are the two major constraints for achieving higher yield from rainfed lowland rice. Marker-assisted backcross breeding was followed to develop submergence tolerant and durable BB resistant variety in the background of popular cultivar ‘Swarna’. Four BB resistance genes viz., Xa4, xa5, xa13, Xa21 and Sub1 QTL for submergence tolerance were incorporated into the mega variety. Foreground selection for the five target genes was performed using closely linked markers and tracked in each backcross generations. Background selection in plants carrying the target genes was performed by using 100 simple sequence repeat markers. Amongst backcross derivatives, the plant carrying five target genes and maximum recurrent parent genome content was selected in each generation and hybridized with recipient parent. Eighteen BC3F2 plants were obtained by selfing the selected BC3F1 line. Amongst the pyramided lines, 3 lines were homozygous for all the target genes. Bioassay of the 18 pyramided lines containing BB resistance genes was conducted against different Xoo strains conferred very high levels of resistance to the predominant isolates. The pyramided lines also exhibited submergence tolerance for 14 days. The pyramided lines were similar to the recurrent parent in 14 morpho-quality traits.

Rice, the queen of cereals is life for millions of global population. Rice grain is a source of carbohydrate, protein, specific oils, dietary fibre, vitamins, many minerals and other disease-fighting phyto-compounds for which, also known as golden cereal. The crop is cultivated in diverse agro-ecology starting from high elevation to below sea level. Globally, rice is cultivated in 163.2 million hectares of which approximately 45% area is under rainfed ecology with low productivity due to various abiotic and biotic stresses 1 . In India, rainfed lowland rice occupies around 16 million hectares area of which 92% is located in the eastern region of the country. Rice cultivation in rainfed lowland ecosystem is associated with major biotic and abiotic stresses that reduce the productivity. Improved varieties are essentially needed that combine high grain yield with submergence tolerance along with in-built resistance to major diseases and insect pests.
Among the biotic stresses, bacterial leaf blight causes considerable yield loss in this ecology. This disease caused by Xanthomonas oryzae pv. oryzae (Xoo) is also a destructive disease of rice in many rice growing countries of the world. The estimated yield loss due to attack of this disease is to the tune of 20-80% depending on location, season and variety [2][3][4][5][6][7] . Globally, 42 BB resistance genes have been reported from diverse sources 8 . Most of these resistance genes are tagged by closely linked molecular markers 3,[9][10][11][12][13][14] . Future demand for staple food grain requirement is increasing in India. The country needs an incremental rice production of more than 2 million tons per year to meet the projected demand of 135-140 million tons by 2030. This production increase should be obtained under constraints like less land, less water, less labor and fewer chemicals, constant battle against new

Results
Foreground and background selections in backcross progenies. Closely linked molecular markers for Sub1 and BB resistance genes Xa4, xa5, xa13 and Xa21 were deployed for screening in each of backcross segregating generations for selecting the plants carrying five target genes (Fig. 1). True hybridity was checked in the F 1 generation plants using the Xa21 marker and all the plants raised from crossed seeds were found to be pure F 1 plants. One of the pure F 1 plants was hybridized with the third donor parent 'Swarna-Sub1' to combine submergence and four BB resistance genes. True multiple F 1 plant was hybridized with the recurrent parent to produce BC 1 F 1 seeds. The BC 1 F 1 seeds were raised and foreground selection was performed in 525 BC 1 F 1 plants to select plants carrying Sub1, Xa4, xa5, xa13 and Xa21 genes in the lines (Fig. 2). The molecular markers used for selecting plants carrying the target genes in the derived progenies were validated first in the parental lines (Table 1). In Figure 2. PCR amplification of markers linked to BB resistance genes Xa21, xa13, xa5, Xa4 using primers pTA248, Xa-13 prom, RM122, MP-Nbp-131 and submergence tolerance QTL, Sub1 using primers Sub1-A203, Sub1-BC2 Sub1-C173 and RM8300 in BC 1 F 1 derivatives. Lanes on the top of the gel indicate the BC 1 F 1 plants designation; L-Molecular weight marker (50 bp plus ladder). addition, the parental polymorphism survey was performed using 1058 rice microsatellite markers covering all chromosomes, of which 100 were polymorphic that were used for background selection ( Table 2).
The purpose of background selection was to know the recovery of recurrent parent's genome content in the backcross derived lines. Background selection was performed in BC 1 F 1 to BC 3 F 2 generations in the foreground positive plants for the five target genes. In each generation, the genotype possessing maximum genome recovery of the recurrent parent was selected for hybridization in next backcross. Background genotyping was performed using 100 SSR markers in BC 1 F 1 , BC 2 F 1 , BC 3  Four hundred eighty seven BC 2 F 1 plants were subjected to foreground selection by using Sub1, Xa4, xa5, xa13 and Xa21 specific markers. Fifty five, sixty one, forty eight, sixty two and eighty three progenies possessed Xa21, xa13, xa5, Xa4 and Sub1 genes, respectively. Only fifteen plants were positive for all four BB resistance genes and Sub1 QTL (Fig. 3). Background recovery among the 15 plants with all five target genes ranged from 82-93% showing average of 88.4%. The plant (SSB 121-28) containing 93% of Swarna genome was considered for next back crossing.
In BC 3 F 1 generation, a total of 492 progenies were generated from hybridization between the BC 2 F 1 plant (SSB 121-28) and recurrent parent 'Swarna' . Foreground screening for five target genes was performed using the linked markers in BC 3 F 1 progenies. A total of 63 plants positive for Xa21, 59 for xa5, 54 for xa13 and 71 for Xa4 were detected. When counted for BB resistance gene combinations in these progenies, 43 plants showed presence of Xa21 and xa13; 38 plants with Xa21 and xa5 and 45 plants with Xa21 and Xa4. However, 11 plants carried the target genes, Sub1, Xa21, xa13, xa5 and Xa4 (Fig. 4). The background analysis in these plants using 100 polymorphic SSR markers detected 93 to 97% recurrent genome recovery showing an average recovery of 94.27% (Table 3). During BC 3 F 2 foreground analysis, eighteen pyramided lines were found to carry homozygous alleles of 4 target genes viz., Sub1, Xa21, xa13 and Xa4 based on gene specific markers. However, 3 plants were detected to carry all the target genes in homozygous condition by genotyping 620 BC 3 F 2 progenies from a total of 1924 by deploying the gene specific markers (Fig. 5). Eighteen pyramided lines were evaluated for yield and other agro-morphologic traits in the BC 3 F 4 generation. The dendrogram obtained by using the SSR data could classify the pyramided and parental lines into two major clusters (Fig. 6A). Cluster I accommodated 20 genotypes including the recipient parent 'Swarna' , submergence tolerance donor parent 'Swarna-Sub1' and 18 pyramided lines while the donor parent for BB was located in cluster II. Almost all the pyramided lines in cluster I were similar in Jaccard's coefficient value as that of the recurrent parent 'Swarna' . Amongst the 18 BC 3 F 2 pyramided lines, 100% similarity was noticed in SSB-121-28-13-2, SSB-121-28-13-5, SSB-121-28-13-6, SSB-121-28-13-7, SSB-121-28-13-10, SSB-121-28-13-12, SSB-121-28-13-14, SSB-121-28-13-15 and SSB-121-28-13-17 based on 100 background primers used (Fig. 6B). Grain yield and morpo-quality traits of the pyramided lines carrying Sub1 and BB resistance genes. Eighteen pyramided lines including 3 homozygous carrying five target genes in the background of Swarna in BC 3 F 4 generation along with the donor and recipient parents were evaluated during wet season, 2017 at NRRI, Cuttack. The recurrent parent, Swarna produced mean grain yield of 5.35 t/ha. Many test entries with five target genes showed grain yield higher than the recurrent parent, Swarna ( Table 5). Most of the pyramided lines were similar in various morpho-quality traits like the recipient parent, Swarna ( Table 5). The genotype-by-trait biplot diagram generated using 14 morphologic and quality traits of the 18 pyramided lines along with parents also indicated clearly the morphologic and quality traits similarity among pyramided lines from the placement pattern in the quadrants (Fig. 8). All the pyramided lines present in the first quadrant were superior in grain quality, grain yield and other parameters similar to the popular variety, Swarna. These lines in 1 st quadrant and the lines present in the 2 nd quadrant may be considered for further evaluation and release as cultivars in the country. The first principal component explained 60.2% of variation while second component showed 15.6% of the total variation. Amongst the 14 studied morpho-quality traits, spikelet fertility and plant height contributed maximum towards diversity (Fig. 8).

Discussion
Deployment of molecular markers for selection of target gene carrying segregating genotypes in breeding programs enhances the precision of target gene transfer to the recipient variety, as well as reduces the duration of selection cycle compared to conventional breeding program. In this breeding program, we could achieve transfer of four BB resistance genes and Sub1 QTL into mega variety, 'Swarna' by integrating molecular markers with the phenotypic selections. The pyramided lines were developed in three backcrosses followed by two selfing generations to achieve the desired product. Control of bacterial blight disease using resistance genes is economical www.nature.com/scientificreports www.nature.com/scientificreports/ and environment friendly. Again, Sub1 QTL was introduced into the background so as to confer tolerance to submergence during flash flood stress. Therefore, it is a significant achievement in lowland rice crop improvement program by gene stacking of two important traits through molecular breeding. Development of varieties through marker-assisted selection has been reported earlier with less duration, more precision and more environment friendly approaches 4-7,27,28,34-38 . However, by adopting this precision breeding, Sub1 QTL along with 4 BB www.nature.com/scientificreports www.nature.com/scientificreports/ resistance genes were stacked and few Swarna type pyramided lines suitable for the flash flood and BB endemic areas of the country could be developed.
Many successful examples of research results on gene pyramiding have been reported [4][5][6][7]27,28,[34][35][36][37][38] . However, development of cultivar with gene stacked for broad spectrum resistance against BB pathogen and submergence tolerance in the mega variety, Swarna is clearly different from other's results. Previous research publications on gene stacking of submergence tolerance and BB resistance in rice varieties such as improved Tapaswini and improved Lalat have been reported 39,40 . These two varieties are early to mid early type and are not suitable for lowland ecology. But, our gene stacking results are on a popular variety that is a late maturing type and hence suitable  Table 3. Number of heterozygotes identified for five target genes and estimation of recurrent parent genome contribution. a As per Mendelian ratios for independent gene action. The recovery of higher percentage of recipient parent genome with less drag from donor parents IRBB60 and Swarna-Sub1 using 100 polymorphic primers is also an important achievement. Similar rice breeding work for transfer of various traits have been reported earlier [5][6][7]14,[34][35][36][37]41 . Using this molecular breeding approach, it was possible to develop plants carrying five target genes with recurrent parent genome of more than 95% in the pyramided lines. The results of this breeding program resulted in development of 18 pyramided high yielding lines homozygous for Sub1, Xa21, xa13 and Xa4 exhibiting maximum recurrent parent genome recovery. However, 3 pyramided lines were found to carry all five target genes. There may be a "pull" for introgression of the Xa21, xa13 and xa5 genes during selection that helps in dragging of additional unlinked loci from the donor genome in backcross generations 35 . However, we could detect such pull effects only in few pyramided lines during the transfer of Sub1, Xa21, xa13, xa5 and Xa4 genes into Swarna background in different backcross generations. The graphical genotyping data of the pyramided lines also indicated the linkage drag in the carrier chromosomes carrying the five target genes (Fig. 9). No donor segment linkage drag was seen in xa5 carrier chromosome 5. Similar types of results were observed in earlier publications with foreground and background selections with less linkage drag by involving more numbers of background markers 5,6,35,36,42 . In our selected plants, very less genetic linkage drag  www.nature.com/scientificreports www.nature.com/scientificreports/ was observed for transfer of Sub1, Xa21, xa13, xa5 and Xa4 genes (Fig. 9) as the donor parent was an improved genotype already. These improved donor lines IRBB60 and Swarna-Sub1 are likely to contribute less undesirable effects than using a wild or landrace type of donor source for BB durable resistance breeding program. Earlier publications suggest that improved variety as a source of donor was expected to give less undesirable linkage drag compared to the wild and landraces as donors 6,39,40,42 .
Genotype-trait-biplot diagram places the donor parent quite away from the origin and in a separate quadrant from the derived lines. This shows that much less or no drag from donor parent except target genes recovered in the pyramided lines (Fig. 4). However, the recipient parent was placed close to the pyramided lines and in the same quadrant. Thus, the pyramided lines were expected to be similar in phenotype. Many pyramided lines were located near the origin and hence these lines indicated better stability as compared to the distant ones. Evaluation results of 18 pyramided lines for various morpho-quality traits revealed similarity of the lines with the recipient parent while a few were better than the recipient parent for yield, quality and morphologic traits (Figs 7 and 8; Table 5). The genotyping result and its further analysis indicated that the recovery of recipient genome was higher in few backcross derivatives than the theoretical expected value of 75% in backcross generations. In BC 3 F 2 , the recovery of Swarna genome in all homozygous target genes containing derived lines was more than 95 percent (Fig. 6B). The background selection with more number of molecular markers helped in recovery of the recipient genome content in the marker-assisted backcross breeding program. Besides, it revealed that presence of Sub1 gene with other four BB resistance genes (Xa21, xa13, xa5 and Xa4) in the same background may not show antagonistic effects for yield and other traits 5,6,36,42,43 .
Evaluation of pyramided lines revealed that lines SSB-121-28-13-1, SSB-121-28-13-2, SSB-121-28-13-3, SSB-121-28-13-4, SSB-121-28-13-5, SSB-121-28-13-6, SSB-121-28-13-7, SSB-121-28-13-8 and SSB-121-28-13-9 exhibited higher yield than the recipient parent Swarna and both the donor parents IRBB60 and Swarna-Sub1. The higher yield obtained might be due to higher level of resistance to BB disease, more tolerance to submergence than the sensitive parent and absence of any yield penalty due to pyramiding of BB resistance genes with Sub1 QTL in the derived lines. Similar results were also observed in earlier publications 5,6,36,42 . Thus, it created confidence in integrating molecular markers for selection of the desired trait(s) and recovery of the recipient parent's genome to the derived line in conventional breeding program. Deployment of five genes in a popular variety like  www.nature.com/scientificreports www.nature.com/scientificreports/ Swarna could achieve higher level of tolerance in many BB endemic in lowland rice growing areas of the country. The study clearly establishes the use of marker-assisted selection for conferring resistance/tolerance to biotic and abiotic stresses which is very much important under climatic change situations.

Materials and Methods
Plant materials. The donor parent IRBB60, containing BB resistance genes Xa4, xa5, xa13 and Xa21 was used as male parent in the hybridization program with recipient variety 'Swarna' . The recurrent parent, Swarna is a popular variety of eastern India but highly susceptible to bacterial blight disease and flash flood. Swarna was hybridized with IRBB60 during wet season, 2013 and the true F 1 plants were backcrossed with third parent Swarna-Sub1 during dry season, 2014 (Fig. 1). During wet season 2014, the true F 1 plant was again hybridized with recurrent parent 'Swarna' . Variety Swarna was used as recipient parent than Swarna-Sub1 for better preference by farmers due to its grain coloration, grain quality and ideal maturity duration for shallow lowlands. All 525 BC 1 F 1 seeds were grown and foreground positive plants for submergence tolerance and four bacterial blight resistance genes were selected by using molecular markers (Table 1). All the foreground positive plants in BC 1 F 1 generation were subjected to background selection. BC 1 F 1 line with maximum recurrent genome content was hybridized with Swarna to generate BC 2 F 1 . A Total of 487 BC 2 F 1 plants were generated during dry season, 2015. The foreground positive plants in BC 2 F 1 containing maximum recurrent genome content was again hybridized to produce BC 3 F 1 seeds during wet season, 2015. Background selection was continued in BC 3 F 1 generation during dry season, 2016. Genotyping was performed to confirm homozygous lines for target gene combinations in BC 3 F 2 generation. For bioassay and evaluation trial, seeds of the plant carrying homozygous target genes were increased during dry season, 2017 as per the breeding scheme (Fig. 1). Evaluation and bioassay trials were conducted during wet season, 2017 and 2018.

DNA isolation and PCR amplification.
Isolation of mini scale DNA preparation was performed using standard protocol 44 . The PCR reaction mixture contained 30 ng templates DNA, 5 pico mole of each of the primers, 200 μM dNTPs, 1 X PCR buffer (10 mM Tris-HCl, pH 8.3, 50 mM KCl, 1.5 mM MgCl 2 , and 0.01 mg/ml gelatin) and 1 unit of Taq DNA polymerase in a volume of 20 μl and amplification of target sequences were done as per earlier reports ( Table 1). The PCR products were separated by gel electrophoresis and imaged on gel documentation system (SynGene, Germany).

Marker analysis.
The publicly available nine tightly linked and gene based markers of the five target genes were used for foreground selection (Table 1). Amongst the 1058 SSRs markers used for parental polymorphism survey, 100 were found to be polymorphic between the parents and used for background selection. Data were  www.nature.com/scientificreports www.nature.com/scientificreports/ analyzed and similarity matrix was constructed from binary data with Jaccard's coefficients and dendrogram was generated by unweighted pair group method arithmatic average (UPGMA) algorithm, using software [45][46][47] . Graphical Geno Types (GGT) Version 2.0 software program was used for the assessment of the genomic contribution of the parent in the selected recombinants based on SSR marker data 48 .
Screening for submergence tolerance. Three weeks' old seedlings of the BC 3 F 4 pyramided lines carrying Sub1 and BB resistance genes along with the three parents were transplanted in the screening tank of ICAR-National Rice Research Institute (NRRI), Cuttack, during wet season, 2017. A population of eighteen plants per row with three rows per entry at 15 × 20 cm spacing was planted in a randomized complete block design with two replications. Complete submergence stress was given for 14 days and water depth was maintained upto 1.5 m each day. After 14 days, the tank was de-submerged and regeneration was noted after a week. Recording of observations and scoring of genotypes were done as described in earlier publication 18 . Bioassay against BB resistance. Forty five-day-old pyramided lines carrying BB resistance and Sub1 genes, along with the control, were inoculated with eight isolates of Xoo. Eight highly virulent isolates of BB pathogen identified on their reaction against near isogenic line differentials carrying resistant genes Xa3, Xa4, xa5, Xa7, Xa10, xa13 and Xa21 maintained at ICAR-National Rice Research Institute, Cuttack, Odisha, India were used for inoculation. The reaction of pyramided and parental lines against different Xoo inoculated strains were screened during wet season, 2017 and 2018. The Xoo isolates were prepared by suspending the bacterial mass in sterile water at a concentration of approximately 10 9 cells/ml 49 . Five leaves from five different plants of each entry and replication were clip inoculated at the maximum tillering stage and lesion lengths (LL) were recorded after 15 days. The disease symptoms were scored as resistant (R, LL ≤ 3.0 cm), moderately resistant (MR, 3.0 cm < LL ≤ 6.0 cm), moderately susceptible (MS, 6.0 cm < LL ≤ 9.0 cm) or susceptible (S, LL > 9.0 cm) as in the previous publications 6,42,50 . Characterization of pyramided lines for morphologic, quality and yield traits. Thirty days old pyramided lines carrying Sub1 and BB resistance genes and the parents were transplanted in the lowland field. Plot size of 9.6 m 2 was provided for each entry with forty plants per row and eight rows per entry at 15 × 20 cm spacing in a randomized complete block design with three replications transplanted at NRRI, Cuttack, during wet season, 2017. Data were recorded from ten plants of each entry and replication for agronomic traits viz., plant height, panicles/plant, number of filled grains/panicle, 1000-grain weight, grain length, grain breadth, milling (%), head rice recovery (%), kernel length after cooking (mm), alkali spreading value, gel consistency, amylose content (%) while, days to 50% flowering and plot yield were recorded on whole plot basis. Head rice recovery was calculated as per the earlier described method 51 . Estimation of gel consistency (GC) was done as per the standard procedure 52 . Alkali spreading value was computed according to the procedure of 53 . For analysis of cooking qualities, 25 grains were taken in a test tube. Grains were soaked in 20 ml distilled water for 20 minutes after which test tubes were placed in boiling water for 10 min and then cooled. The average length and breadth of 10 cooked kernels were measured. The amylose content of the pyramided lines was estimated as per the standard protocol 54 . Analysis of variance for various agro-morphologic and quality traits and principal component analysis were performed using SAS statistical software 55 .

Conclusion
Deployment of single resistance gene is risky as there are chances of resistance knock-down by the pathogens. The development of pyramided lines showing higher level of resistance to BB containing four resistance genes namely Xa21, xa13, xa5 and Xa4 along with Sub1 QTL in the mega variety background is an important achievement for the lowland ecology. The products from the gene stacking workin the 'Swarna' background through molecular breeding will provide a solution in the flash-flood and bacterial blight endemic areas of eastern Indian lowland rice ecosystems. The grain and its cooking quality characters viz., milling %, head rice recovery %, kernel length (mm), kernel breadth, kernel length after cooking, alkali spreading value, gel consistency and amylose content (%) remained almost same in the pyramided lines as in the recipient parent. The quality features of the mega variety Swarna remained same along with high grain yield and durable bacterial blight resistance in the selected pyramided lines. The diverse agro-climatic zones of our country may offer scope for creation of many virulent Xoo strains for which these BB pyramided lines with Xa21, xa13, xa5 and Xa4 along with Sub1 QTL are expected to provide good substitute to the existing susceptible array of varieties used by the farmers in the targeted region of the country where majority of the areas are rainfed lowlands.