Marker-assisted pyramiding of two major, broad-spectrum bacterial blight resistance genes, Xa21 and Xa33 into an elite maintainer line of rice, DRR17B

Bacterial blight (BB) disease reduces the yield of rice varieties and hybrids considerably in many tropical rice growing countries like India. The present study highlights the development of durable BB resistance into the background of an elite maintainer of rice, DRR17B, by incorporating two major dominant genes, Xa21 and Xa33 through marker-assisted backcross breeding (MABB). Through two sets of backcrosses, the two BB resistance genes were transferred separately to DRR17B. In this process, at each stage of backcrossing, foreground selection was carried out for the target resistance genes and for non-fertility restorer alleles concerning the major fertility restorer genes Rf3 and Rf4, using gene-specific PCR-based markers, while background selection was done using a set of 61 and 64 parental polymorphic SSR markers respectively. Backcross derived lines possessing either Xa21 or Xa33 along with maximum genome recovery of DRR17B were identified at BC3F1 generation and selfed to develop BC3F2s. Plants harboring Xa21 or Xa33 in homozygous condition were identified among BC3F2s and were intercrossed with each other to combine both the genes. The intercross F1 plants (ICF1) were selfed and the intercross F2(ICF2) plants possessing both Xa21 and Xa33 in homozygous condition were identified with the help of markers. They were then advanced further by selfing until ICF4 generation. Selected ICF4 lines were evaluated for their resistance against BB with eight virulent isolates and for key agro-morphological traits. Six promising two-gene pyramiding lines of DRR17B with high level of BB resistance and agro-morphological attributes similar or superior to DRR17B with complete maintenance ability have been identified. These lines with elevated level of durable resistance may be handy tool for BB resistance breeding.


Introduction
Rice production needed to be increased 42% by 2050to feed the demands of an ever-increasing human population globally [1,2]. Exploitation of heterosis for grain yield through hybrid rice technology is one of the feasible options to enhance rice production and rice hybrids have15-20%yield advantage over inbreeds [3]. Even though rice hybrids were introduced in India in the early 1990s, their adoption has been slow and presently hybrid rice is cultivated in a limited area of 2.5 million ha. One of the principal reasons for slow adoption of rice hybrids in India is their susceptibility to major rice diseases like bacterial blight (BB) and blast [4]. Most of the commercial rice hybrids that have been released and cultivated in India do not possess broad spectrum resistance for BB disease [5].
BB disease is caused by a gram-negative bacterium called Xanthomonas oryzae pv. Oryzae (Xoo). It is one of the most devastating diseases in rice [6]. The bacterium infects rice at maximum tillering stage, leading to water soaking lesions (blighting) on the leaves, which gradually enlarge, wilts and causes yield losses ranging from 74 to 81% [7]. Application of chemicals or antibiotics against is very costly and is not very effective [8,9]. Breeding BB resistant rice varieties and hybrids is the best strategy for managing the BB disease in rice [10]. To date, at least 41 BB resistance genes have been identified,and some of them viz., Xa4, xa5, xa13, Xa21 have been extensively used for development of BB resistant rice varieties [11,12,13,14,15] ( Table 1). However, large scale and long-term cultivation of varieties and hybrids with a single gene results in the breakdown of resistance due to a high degree of pathogenic variation [12,16,17]. Pyramiding of two or three Xa genes can enhance the durability and spectrum of resistance against BB [18,19].
The major BB resistance gene, 'Xa21' was identified from Oryza longistaminata. It is located on chromosome 11 and a tightly linked to gene-specific marker pTA248 [21]. Similarly, 'Xa33' was identified from Oryza nivara. It is located on chromosome 7 and tightly linked to a marker RMWR7.6 [22]. These markers can be used in marker-assisted breeding to introgress Xa21 and Xa33 genes into different rice varieties and hybrid parental lines. These two genes are found to be highly effective against several isolates of Xoo from India and hence, are ideal choices for pyramiding into popular rice varieties or hybrids through marker-assisted breeding.
DRR17B is a fine grain type and medium duration, stable promising maintainer line developed by ICAR-Indian Institute of Rice Research, Hyderabad, India [23]. It is however highly susceptible to BB of rice. In the present study, two major dominant BB resistance genes, Xa21 and Xa33 were introgressed into the genetic background of DRR17B through marker-assisted backcross breeding to develop improved DRR17B lines with broad spectrum resistance against BB.
The popular but BB susceptible maintainer line DRR17B (APMS6B/BPT5204/IR69628B) was used as the recurrent parent. It was developed by ICAR-Indian Institute of Rice Research (IIRR), Hyderabad (17.3200˚N, 78.3939˚E), India.

Strategy for marker-assisted introgression of Xa21 and Xa33 into DRR17B
Marker-assisted backcross breeding strategy was adapted for targeted introgression of Xa21 and Xa33 genes into the genetic background of the elite maintainer line of rice, DRR17B. Each of these genes was separately introgressed into DRR17B through two sets of crosses, i.e., Cross I, viz., DRR17B/ISM and Cross II, viz., DRR17B/FBR1-15 (Fig 1). The F 1 s obtained from the two crosses were analysed by extracting DNA through the method described by [24] and using that DNA by keeping Polymerase Chain reaction with gene-specific markers pTA248 (specific for Xa21; [21]) and RMWR7.6 (specific for Xa33; [22]) to identify 'true' heterozygotes. The 'true' F 1 s were backcrossed with the recurrent parent DRR17B to generate BC 1 F 1 s, which were then screened for the presence of the target resistance genes using the gene-specific markers. The positive plants for Xa21 and Xa33 were selected and further screened for the non-presence of major fertility restorer genes, Rf4 and Rf3 using tightly linked markers, viz., DRCG-RF4-14 and DRRM-RF3-10, respectively [25]. BC 1 F 1 plants possessing BB genes and a non-restoring allele concerning Rf4 and Rf3 in homozygous condition were selected following the procedure described by [23]. These plants were later screened with a set of parental polymorphic SSR Zone I, Zone VII AZ121-AZ127 Tamil Nadu 11 (1,5,6,7,9,11,12,14,17,19 & 21) xa5+xa13+Xa21 & Xa4+xa5+xa13+Xa21 markers (61 markers specific to the cross DRR17B/ISM and 64 markers specific for the cross DRR17B/FBR1-15EM) through background selection to identify a single BC 1 F 1 plant from each cross possessing maximum recovery of the recurrent parent genome. The selected plant was backcrossed once again with DRR17B. The process of marker-assisted backcrossing was repeated until BC 3 generation, and BC 3 F 1 plants of DRR17B possessing either Xa21 or Xa33 and maximum recovery of recurrent parent genome were then selfed to obtain BC 3 F 2 s. Plants homozygous for either Xa21 or Xa33 were identified among the BC 3 F 2 plants and the best plants from the two crosses were intercrossed to obtain intercross F 1 s (i.e., ICF 1 s). 'True' ICF 1 plants were identified by screening with molecular markers specific for Xa21 and Xa33 and were then selfed to generate intercross F 2 s (i.e., ICF 2 s). Plants homozygous for both Xa21 and Xa33 were identified among the ICF 2 plants using the gene-specific markers. The identified plants were advanced through the pedigree method of selection (involving selfing followed by morphological trait-based visual selection) up to ICF 4 generation. Marker-assisted selection procedures were followed as recommended by [21] and [22] for detection of Xa21 and Xa33 genes, while the protocol described by [23] was adopted for background selection and detection of non-restoring alleles of Rf4 and Rf3.

Screening for agro-morphological traits
Improved lines (ILs) of DRR17B (ICF 4 ) were field evaluated in randomized complete block design in Kharif 2014 (i.e. July-October/Wet season 2014) for the following agro-morphological traits involving days to 50% flowering (days), plant height (cm), number of productive tillers (No.), panicle length (cm), grains per panicle (No.) and spikelet fertility. Each entry was planted in 20 rows of 4m length with a spacing of 15 x 20 cm between rows and within rows. Days to 50 percent flowering was recorded based on number of days from sowing to 50% population flowering on a whole plot basis. Plant height (cm), number of productive tillers (No.) and panicle length (cm) were recorded from 5 competitive plants from each plot chosen at random and the mean values computed for different lines. Five individual panicles harvested separately from five plants were collected to compute for the average grain number per panicle (No.). The ILs were crossed with IR58025A line and evaluated for spikelet fertility based on seed setting of each cross. The percentage was calculated based on seed setting per panicle as described in [23].

Statistical analysis
Agro-morphological and phenotypic BB screening data were analysed using standard procedures by calculating Mean, significant standard error of Mean (S.E.M ±), Analysis of variance (ANOVA) and Least Significance Difference (LSD) [29]. Analysis of variance (ANOVA) and Duncan's multiple range test (DMRT) and Least Significance Difference (LSD) at 5% level of significance, significant standard error of Mean (S.E.M ±) were calculated by using MS Excel and Statistical computer software Statistix8.1 [30] software to analyze the variation between ILs and parents.

Marker-assisted transfer of Xa21 and Xa33 into DRR17B
The true F 1 s derived by crossing DRR17B with 'ISM' (i.e., Cross I) and FBR1-15 (i.e., Cross II) were backcrossed with DRR17B to obtain BC 1 F 1 s, which were then screened with the genespecific markers. A total of 61 and 65 BC 1 F 1 plants were observed to be positive for the target genes in Cross I and Cross II, respectively. The positive plants were screened with markers specific for Rf3 and Rf4,and a total of 15 and 11 plants were identified to be devoid of both the fertility restorer genes concerning Cross I and Cross II, respectively. These plants were then subjected to background selection using a set of polymorphic SSR markers (61 markers for Cross I and 64 for Cross II). Plant # IIRRGP3 from Cross I, with a recurrent parent genome (RPG) recovery of 73.7% and Plant # IIRRGP22 from Cross II, with a RPG recovery of 75% were identified to be the best ones (i.e. having a maximum recovery of DRR17B genome) and were used for further backcrossing. The process of marker-assisted backcrossing was carried out until BC 3 F 1 generation (details given in Table 2). At BC 3 F 1 , plant # IIRRGP3-87-64 from Cross I with RPG recovery of 93.4% and plant # IIRGP22-73-10 with RPG recovery of 93.7% were identified to be superior and were selfed to generate BC 3 F 2 s. With regards to the BC 3 F 2 s produced from Cross I and Cross II, 39 and 52 plants were identified to be homozygous for Xa21 and Xa33, respectively. Among these, a solitary plant, which was morphologically similar to DRR17B, was identified from Cross I (i.e., plant # IIRRGP 3-87-64-22 and Cross II (i.e., plant # IIRRGP 22-73-10-15) and intercrossed with each other to generate intercross F 1 s (i.e., ICF 1 s). Out of 68 ICF 1 s, 63 were identified to be heterozygous for both Xa21 and Xa33 (i.e. true intercross F 1 s), and they were selfed to obtain ICF 2 generation. At ICF 2 , a total of 309 plants were screened with markers specific for Xa21 and Xa33 and 18 were identified to be double homozygotes (Table 2; Fig 2). A total of nine plants out of the 18, which were identified to be phenotypically similar to DRR17B, were further advanced until ICF 4 generation through phenotype-based pedigree selection. At ICF 4 generation, six promising lines which were similar to the recurrent parent were identified (line #IIRRIC 10-8-94, IIRRIC 10-19-138, IIRRIC 102-26-7, IIRRIC 123-34-84, IIRRIC 123-58-3 and IIRRIC 172-77-12) and analysed for their resistance to BB, sterility maintenance ability and also characterized for important agro-morphological traits. Among the six ILs, line # IIRRIC102-26-7 exhibited the highest recurrent parent genome recovery with more than 95% along with minimal linkage drag on carrier chromosomes (Fig 3).

Phenotypic evaluation of ILs for BB resistance
The recurrent parent, DRR17B (11) (with lesion lengths ranging from 18.8 to 33.1 cm) and susceptible check TN1 (12) (with lesion lengths ranging from 20.9 to 33.8 cm) showed a disease score of 9 against all the eight isolates of the Xoo ( Table 3 (Table 3; Fig 4).

Characterization of ILs for maintenance ability and agro-morphological traits
The current study screened the six ILs for their maintenance ability. Out of six, three lines showed partial spikelet fertility, while the remaining three lines (viz., line # IIRRIC102-26-7, IIRRIC123-34-84, and IIRRIC172-77-12) showed complete spikelet sterility when crossed with the WA-CMS line, IR58025A ( Table 4). Comparison of five agro-morphological parameters (days to 50% flowering, plant height, number of productive tillers, panicle length and number of grains per panicle) revealed thatall the six ILs are isophenic in their panicle length and number of productive tillers to DRR17B, while significant differences were observed with respect to the number of grains per panicle. The ILs viz., IIRRIC10-8-94, IIRRIC102-26-7, IIR-RIC123-58-3 and IIRRIC172-77-12 possessed more number of grains per panicle with respective to DRR17B viz., 301.6, 360.4, 308 and 317 respectively (Fig 5A and 5B). However, all selected six lines showed comparatively shorter plant height than recurrent parent. While panicle length of, line # IIRRIC102-26-7 was observed longest among all six panicle (24.16 cm), the remaining five ILs exhibited equal or less than the recurrent parent DRR17B (average length of 23.24 cm: Table 4). Line # IIRRIC102-26-7 exhibit highest numbers of productive tillers per plant (average of 12), all remaining five ILs were similar to thir recurrent parent (10-11.2). The to 50% flowering, of all the six ILs flowered earlier (92-102 days), as compared to DRR17B (105 days).

Discussion
Several studies indicate that global rice production needs to be doubled by 2050 to meet the demands of ever growing population [2]. However, rice grain yield is badly affected by biotic and abiotic stresses [31]. The present study was taken up to improve, an elite maintainer of rice, DRR17B, for its resistance against BB resistance. DRR17B is a fine grain type and medium     duration maintainer line of rice, possessing stable maintenance ability was developed by ICAR-Indian Institute of Rice Research, Hyderabad, India [23]. As DRR17A and its maintainer parent-DRR17B are highly susceptible to BB disease, considering this deficiency in the elite maintainer line, the current study was carried out with an objective to introgress two major dominant BB resistant genes, viz., Xa21 and Xa33 through MABB in order the make the maintainer line durably resistant to BB. These two selected genes are known to confer resistance against multiple isolates of the BB pathogens for large rice cultivated area; hence, the hybrids developed from ILs of DRR17A will also be sustainable resistant against this disease. Introgression of BB resistance genes through conventional breeding involving patho-phenotypic selection which is very laborious, time and resource consuming process and its success  Marker-assisted pyramiding of broad-spectrum bacterial blight resistance genes into an elite maintainer line significantly depends on accurate disease scoring, the existence of environmental conditions which favour disease development and the availability of appropriate virulent strains of the pathogen causing the disease [11]. As compared to conventional breeding, marker-assisted selection (MAS) breeding strategy is more useful for targeted introgression of resistance genesas it does not depend on the availability of virulent strains or existence of ideal environmental conditions, since the selections are indirect, and are based on the presence or absence of specific alleles of molecular markers linked to the resistance genes. Previous studies, [23,32,33] reported on successfully developed bacterial blight resistant versions of hybrid rice parental lines PRR78 and IR58025B, through marker-assisted selection for target traits in the initial stages and phenotype-based selection at later stages and hence at the same methodology was adopted in the current study. So far, at least 41 genes conferring resistance against BB have been identified in rice [11,12,13]. Among them, the wild rice derived gene, Xa21 encoding a receptor kinase-like protein has been successfully deployed by many research groups across the world, as it has been documented to confer broad-spectrum resistance against the BB disease [17,18,23,32,34,35,36,37,38]. The commonly used BB resistance gene Xa21 has been tagged and mapped on chromosome 11 with a tightly-linked PCR-based marker pTA248 [21].Xa33, the wild rice derived BB resistance gene has been reported to confer broad spectrum resistance [22] and the gene has been deployed by the research group at Tamil Nadu Agricultural University, Coimbatore, India and the breeding lines possessing Xa33 were observed to be very effective in terms of their BB resistance [39,40]. Hence, these two broad spectrum resistance genes were selected for introgression into the DRR17B.

IX-133
Phenotypic screening for BB resistance was carried out in this study among selected single gene containing BC 3 F 6 lines possessing either Xa21 or Xa33 and two-gene containing intercross derived lines at ICF 4 generation possessing Xa21+Xa33 along with the donor and recurrent parents ('ISM', FBR1-15, and DRR17B, respectively) using eight virulent isolates of Xoo. All the ILs possessing Xa21+Xa33 were observed to show significantly higher level of resistance against BB when compared to the donor parents, 'ISM' and FBR1-15. Single gene containing lines of DRR17B (i.e. possessing either Xa21 or Xa33), the recurrent parent DRR17B and the BB susceptible check TN1 (Table 3; Fig 4).It is earlier known that Xa21 confers broad Marker-assisted pyramiding of broad-spectrum bacterial blight resistance genes into an elite maintainer line spectrum resistance against many of the virulent pathotypes of Xoo in India [17,18] and several studies have indicated the suitability of Xa21 in BB resistance gene pyramiding programmes [10,18,34,41,42]. However, in this study, a few isolates of the pathogen were observed to be compatible with Xa21 containing lines of DRR17B indicating that Xoo isolates, which are capable of overcoming Xa21 conferred resistance are fast-developing [17,43,44]. Interestingly, the ILs of DRR17B possessing Xa33were observed to show a better level of resistance as compared to the lines having Xa21. Furthermore, DRR17B lines possessing both Xa21 and Xa33 were observed to be highly resistant against all the eight virulent isolates of Xoo, thus, indicating the suitability of deployment of Xa33 either singly or in combination with Xa21. Earlier, two elite restorer lines, KMR3R, and RPHR1005 were improved for BB resistance by introducing Xa21 [23,33,36,38]. Similarly, Xa33 has been deployed in different genetic backgrounds by different research groups [19,22,39,40]. However, this is the first report wherein Xa21 has been combined with Xa33 in the genetic background of an elite maintainer line, i.e., DRR17B and the gene-pyramid lines demonstrated a higher level of resistance as compared to lines possessing a single resistance gene (Table 3; Fig 4).
It is a known fact that long term cultivation of rice varieties possessing single resistance gene can result in the breakdown of resistance by faster development of virulent pathogens [43,44,45].Hence, pyramiding of multiple resistance genes has been advocated to be one of the best strategies to achieve durable dual-resistance [18,46,47]. In our present study, the genotype 'ISM' (with Xa21 + xa13 + xa5) has displayed satisfactory level of resistance with a score of 3 against all eight isolates. Interestingly, when Xa21 gene was combined with another major dominant gene-Xa33, such breeding lines exhibited the highest level of resistance with a score of 1. This indicates that the gene combination Xa21 + Xa33 displayed a broad spectrum of resistance and hence can be recommended for deployment in hybrid rice improvement programs as both Xa21 and Xa33 are both dominant and will express in the F 1 hybrid.
Similar to the approach adopted in the current study, several earlier studies also resorted to phenotype-based selection for key agro-morphological traits along with marker-assisted selection while improving elite varieties and parental lines for one or more target traits through MABB [18,23,33,35,36,37,38,48].The approach of deployment of MABB strategy for the target resistance genes along with negative selection for major fertility restorer genes, Rf3 and Rf4, coupled with phenotype-based selection for certain key agronomic characters helped in near-complete recovery of good features of DRR17B along with identification of few ILs with complete maintenance ability ( Table 4). In addition to improving BB resistance of DRR17B, The current study also focused on the identification of ILs of DRR17B possessing plant height which is significantly shorter than DRR17B, as shorter plant is preferred for deployment as good maintainers [23]. Significant differences in plant height were observed in many improved DRR17B lines viz., RMSIC 10-8-94, RMSIC 10-19-138, RMSIC 102-26-7, RMSIC 123-34-84, RMSIC 123-58-3 and RMSIC 172-77-12 and these lines could serve as better maintainers as compared to DRR17B. Interestingly, when compared to DRR17B, some of the ILs exhibited advantage concerning grain number per panicle. These lines include RMSIC 10-8-94, RMSIC 102-26-7, RMSIC 123-58-3 and RMSIC 172-77-12 (Fig 5A and 5B).For the panicle length, all the ILs showed values equivalent to DRR17B, except one line viz., RMSIC 102-26-7, a wherein slight improvement over the recurrent parent was noticed. Significant differences (i.e., shorter duration) were observed concerning number of days to 50% flowering in some of the backcross derived plants ( Table 4). No significant differences were observed between improved versions of DRR17B and recurrent parent DRR17B concerning other agro-morphological characters or grain type and the lines mostly resembled the original recurrent parent. The approach of coupling of MABB with phenotypic selection adopted in this study helped to regain most of the key agro-morphological traits of DRR17B, while simultaneously helping in the selection of some superior backcross derived segregants of DRR17B possessing BB resistance.
The ILs of DRR17B were test crossed with IR58025A (WA-CMS line) to check their maintainer ability. Three lines (viz., IIRRIC102-26-7, IIRRIC123-34-84, and IIRRIC172-77-12) displayed complete maintainer ability highlighting the necessity of phenotypic confirmation for maintenance ability, despite a rigorous marker-assisted selection for rf3 and rf4 alleles in this study. This could be attributed to the existence of minor fertility restorer genes/QTLs as reported earlier [49].The three ILs of DRR17B, possessing Xa21 + Xa33 are being converted as CMS lines by crossing with DRR17A through MABB.
The six ILs of DRR17B exhibited high level of BB resistance against the BB isolates, when compared with the recurrent parent DRR17B. Whereas in agro-morphological characters like plant height, day to 50% flowering and number of grains per panicle etc, variations were observed. All the improved lines were shorter than the recurrent parent. With regards to Days to 50% flowering all the improved lines were little early (92-103 days) than DRR17B (105 days). Some the improved lines Viz., IIRRIC10-8-94, IIRRIC102-26-7 IIRRIC123-58-3 and IIRRIC172-77-12 were exhibited significantly more number of grains per panicle then DRR17B (280 per panicle). The ILs Viz., IIRRIC102-26-7, IIRRIC123-34-84, and IIRRIC172-77-12 were exhibited complete maintainer ability as like DRR17B and remaining lines were partial maintainers.

Conclusion
The present study has resulted in development of improved versions of an elite maintainer of rice, DRR17B possessing durable resistance against BB through incorporation of two major dominant genes conferring broad-spectrum resistance, Xa21 and Xa33 by marker-assisted backcross breeding (MABB) strategy. The double gene pyramided lines of DRR17B expressed high level of resistance against eight different virulent isolates of Xoo and their resistance levels was comparable with triple resistance gene pyramided rice variety, 'ISM' (possessing Xa21 + xa13 + xa5) and were also significantly better than the single gene containing lines (possessing Xa21 or Xa33). Three promising double-gene pyramided lines of DRR17B with high level of BB resistance, agro-morphological attributes similar to or superior to the DRR17B with complete maintainer ability would be helpful in development of superior rice hybrids with durable, broad-spectrum resistance.