Kerala, a state in India, has a distinctive food culture, particularly its unique preference for rice. The widespread use of parboiled, red rice grains (matta rice) makes it a healthier dietary choice for the people in the region. Uma (Mo. 16) is one among the popular high-yielding red-kernelled rice varieties grown in the state and is estimated to occupy more than 60% of the paddy cultivation area in the state. It is a semi-dwarf, non-lodging, medium tillering; exhibits resistance against brown plant hoppers and gall midge biotype-5 and produces a yield of about 6 to 6.5 tonnes/ha (Devika et al., 2004; Estelitta et al., 2016). Of late, severe yield reductions are commonly observed in this elite rice cultivar owing to the occurrence of bacterial blight (BB) disease caused by Xanthomonas oryzae p.v. oryzae (Xoo) (Jerish et al., 2022; Laha et al., 2023). Among the various disease management tactics, pyramiding resistance genes through the marker-assisted backcross breeding (MABB) approach is found to be an effective method to ensure durable resistance. According to Priyadarisini and Gnanamanickam (1999), the rice line NH56 carrying four R genes (Xa4 + xa5 + xa13 + Xa21) was found to be resistant to the Kerala isolate of the Xoo pathogen. However, the breakdown of resistance of cultivars with Xa4 has been reported earlier in the Philippines and India (Pandey et al., 1986; Mew, 1992). Therefore, efforts are currently being undertaken to pyramid the resistance genes xa 5, xa13 and Xa21 into the genetic background of Uma using the MABB approach, with Improved samba mahsuri (ISM) serving as the donor parent. Advanced backcross progenies were screened using marker-assisted selection (MAS) to identify plants carrying all three resistance genes while minimising linkage drag from the donor genome.

The study was conducted in the Department of Plant Breeding and Genetics, College of Agriculture, Kerala Agricultural University (KAU), Vellanikkara, Thrissur, Kerala. The rice varieties Uma (Mo16) and ISM were used as the parents in the present backcross breeding programme. The rice variety ISM, an Essentially Derived Variety (EDV), served as the donor for bacterial blight resistance genes xa 5, xa13 and Xa21, while Uma, was used as the recipient parent. The backcross lines (BC₁F₁s) generated from the cross were evaluated through MAS and the three BC₁F₁ lines (Plant no. 8.3.2, 8.3.3 and 8.3.9) pyramided with xa5, xa13 and Xa21 genes were identified (Joseph 2016). However, these lines exhibited lower recovery of the recurrent parent genome (21.80 to 23.90%); hence, they were further advanced to develop BC₂F₁ and BC₁F₂ populations, forming the base material for the study (Fig 1).


 
Genotyping of the BC₂F₁ population
 
Of the BC₂F₁ seeds sown, only 21 successfully germinated, attained maturity and were subsequently utilised for screening, whereas the parental lines were raised in replicated plots to facilitate comparison. For cellular DNA isolation, fresh leaf samples were collected from BC₂F₁s and parents. The modified CTAB method was used for the extraction of DNA from the collected samples (Dellaporta et al., 1983). The quality and quantity (µg/ml) of the DNA isolated were analysed using Nanometre (JH BIO Innovations, India). The absorbance ratio (A260/A280) for the BC₂F₁ plants and the parents ranged from 1.89 to 1.90. Foreground selection was carried out using three STS markers, RG556, RG136 and pTA248, closely linked to the BB resistance genes xa5, xa13 and Xa21, respectively (Huang et al., 1997; Sundaram et al., 2008). Further, the presence of resistance genes xa5 and xa13 in the BC₂F₁ progenies was also confirmed using functional markers xa5S/R and xa13 promoter, respectively (Chu et al., 2006; Sundaram et al., 2008).
       
The PCR amplification was performed using the thermal cycler eppendorf master cycler (Eppendorf, Germany, Model: Hamburg 22331) and the reaction mixture of 15 µl was prepared using 2 µl of 10x Taq buffer, 1 µl of dNTP mix, 1.5 µl of MgCl₂ (25 mM), 0.3 µl of Taq DNA polymerase (1 U), 1 µl each of forward and reverse primers, 3 µl of DNA sample and 5.2 µl of distilled water. The PCR reaction profile was as follows: initial denaturation (hot start) at 94°C for 5 min, 35 cycles of denaturation (94°C for 30 sec), annealing (55°C for 30 sec), extension (72°C for 1 min) and final extension at 72°C for 7 min. Restriction digestion was performed for the PCR products of STS markers RG556 and RG136 using enzymes Dra1 and Hind1, respectively, for detecting polymorphism. The amplified PCR products were separated by running them in a 1.5% agarose gel stained with ethidium bromide and the bands obtained were visualised using gel documentation software, UVITEC Fire Reader software (Merck, UK). The location of the amplicon position and its molecular weight were assessed in comparison to the known molecular weight marker. The BC₂F₁ lines having amplicon sizes corresponding to xa5, xa13 and Xa21 genes were selected for further evaluation. For finding the recurrent parent genome recovery, the selected lines of BC₂F₁s were further subjected to background selection. A total of 364 rice microsatellite (RM) markers were screened between Uma and ISM, of which only 22 markers were polymorphic. These markers, covering 12 rice linkage groups, were selected for background selection. PCR amplification of samples with background markers, agarose gel separation and documentation were performed as per the protocol described earlier.
 
Morphological characterisation of introgressed BC₂F₁s and parents
 
As observations were based on single plants, morphological characterisation of individual BC₂F₁ plants was conducted in a non-replicated experimental block, whereas the parental lines were evaluated in ten replications. All observations were recorded following the standard descriptors for rice developed by the International Rice Research Institute (2002). The traits observed include plant height (cm), days to flowering, leaf blade width (cm), leaf blade length (cm), productive tillers, panicle length (cm), spikelets/panicle, grain length (mm), grain width (mm), decorticated grain length (mm) and decorticated grain width (mm). Mean and standard deviation (SD) values were computed for BC₂F₁s and parental lines.
 
Phenotypic screening of the BC₁F₂ population for BB resistance
 
The BC₁F₂ lines (106 nos) produced by selfing of R gene pyramided BC₁F₁ lines (8.3.2, 8.3.3 and 8.3.9) were phenotypically screened for bacterial blight resistance following the leaf clip method (Kauffman et al., 1973; Jalil et al., 2023; Kanipriya et al., 2024) for identification of resistant and susceptible genotypes. The isolate of Xanthomonas oryzae pv. oryzae (Xoo) collected from the Regional Agricultural Research Station, Pattambi, Kerala, was used to screen the BC₁F₂ population and parents to evaluate their reactions to pathogen infection. The BB susceptible recurrent parent, Uma, was used as the susceptible check in this study. The local strains of the bacterial blight pathogen were isolated from infected rice leaves. After 4-5 days of incubation, typical round, yellowish colonies of Xoo were selected, subcultured to obtain pure cultures and multiplied on sucrose agar medium. The cultures were then preserved as glycerol stocks at 4°C for further use. The identity of the pathogen was also confirmed using species specific primers. For inoculation, bacterial suspension was prepared in sterile distilled water to get a final concentration of 10⁸ CFU/ml and was immediately used for inoculation following the leaf clipping technique at the crop’s maximum tillering stage (Kauffman et al., 1973). In each plant, a minimum of three leaves were inoculated early in the morning by cutting away 1-2 cm of the leaf tip with scissors that were dipped in bacterial suspension on the cutting edge. Individual plants of the BC₁F₂ population and parents were evaluated for field infection of bacterial blight and the disease incidence was scored 15 days after inoculation. Lesion length was measured and the disease severity was scored 1-9 as per the IRRI standard evaluation system, SES (IRRI, 2002).
 
Production of advanced generations of BB-resistant lines
 
For the production of BC₂F₂s and BC₃F₁s, half the total number of panicles in the R-gene-pyramided BC₂F₁ line was self-pollinated and the other half was backcrossed to the recurrent parent, Uma, respectively. Concurrently, all the panicles of the BC₁F₂s exhibiting resistance or moderate resistance to BB pathogen infection during phenotypic screening were self-pollinated to obtain BC₁F₃s. At maturity, the seeds were harvested and dried to 13 per cent moisture to aid prolonged storage.
 
Statistical analysis
 
In the above study, the recurrent parent genome contribution (G) in the R gene pyramided lines of BC₂F₁ based on SSR data of background selection was estimated by using the formula:
 

G = [(X + 1/2Y) * 100]/N

 
Where,
N= The total number of parental polymorphic markers screened.
X= The number of markers showing homozygosity for the recurrent parent allele.
Y= The number of markers showing heterozygosity for parental alleles (Sundaram et al., 2008).
       
The graphical representation of the parental genome contribution using molecular marker data was carried out using Graphical Geno Types (GGT) software version 2.0 (van Berloo 1999). The software generates a similarity matrix following the method of Sneath and Sokal (1973) and performs clustering based on the default similarity coefficient. A dendrogram is then generated to visualise the relationships among the samples. The variability in morphological characters of the backcross progenies (BC₂F₁s) was assessed through statistical measures, viz., mean and SD. Clustering of BC₂F₁s and parents was also carried out based on morphological traits.

Presently, our study evaluated BC₂F₁ and BC₁F₂, the advanced segregating generations of backcross Uma × ISM, for BB resistance through molecular marker-based genotypic and phenotypic screening approaches, respectively.
 
Foreground selection in BC₂F₁s
 
The presence of the R-gene xa5 in BC₂F₁s was confirmed through PCR amplifications of STS marker RG556 and functional marker xa5SR. The analysis revealed a monomorphic banding pattern in all BC₂F₁ plants as well as the parents for both the markers assayed (Fig 2a and 2b). These results indicated the endogenous presence of the recessive resistant allele of the xa5 gene in homozygous condition (xa5/xa5) in both parents and all BC₂F₁s screened. Such an endogenous presence of the xa5 gene in several genotypes has been reported by earlier workers; i.e., Singh et al., (2015) reported the presence of the xa5 gene in 15 cultivars out of 35 wild rice accessions evaluated for the presence of BB-resistant genes. Another widely cultivated high-yielding variety in Kerala, Jyothi, has been reported to naturally possess the xa5 gene, conferring BB resistance (Gorakhanath et al., 2017). It is important to note that, despite carrying the resistant allele of the R-gene xa5, Uma remains susceptible to the BB pathogen. This suggests that xa5, when deployed singly, does not provide resistance against the BB pathogen isolates prevalent in Kerala. It was reported that xa5 provides only moderate resistance to the BB pathogen and possesses only a relatively small but consistent residual effect (Li et al., 2001; Sundaram et al., 2008) and also tends to break down over time (Khan et al., 2014; Carpenter et al., 2020). Hence, pyramiding multiple R-genes was recommended to achieve durable resistance against the BB pathogen (Bharathkumar et al., 2008; Yugander et al., 2017).


       
The introgression of the recessive xa13 R-gene in BC₂F₁ plants was confirmed using the STS marker RG136 and the functional marker xa13 promoter. Similar to xa5, xa13 is also a recessive R-gene, but in contrast, xa13 had no residual effects and showed prominent race specificity (Li et al., 2001). MAS is particularly useful for the selection of recessively inherited R-genes like xa5 and xa13 (Sundaram et al., 2008; Singh et al., 2015). The STS marker pTA248 was used to detect the introgression of the dominant Xa21 R-gene. Amplification results for both R-genes (xa13 and Xa21) showed that all BC₂F₁ plants, except plant no. 8.3.9.10, exhibited a monomorphic banding pattern identical to the recurrent parent, Uma, for their respective markers screened. However, plant No. 8.3.9.10 displayed bands corresponding to both Uma and ISM, indicating its heterozygous allelic status for both xa13 (Fig  2c and 2d) and Xa21 (Fig 2e). Hence, the results of the foreground selection concluded that among the 21 BC₂F₁ plants screened, only plant no. 8.3.9.10 was introgressed with BB resistance genes xa13 and Xa21 from the donor parent, ISM. The xa5 gene was found endogenously present in both parents and was inherited by all BC₂F₁s. The resistant alleles of the xa13 and Xa21 genes were detected only in plant no. 8.3.9.10, where they were introgressed in a heterozygous state (xa13/Xa13 + xa21/Xa21). The effectiveness of the Xa21 R-gene in providing broad-spectrum resistance to BB in rice has been well established through numerous studies (Sundaram et al., 2008; Shanti et al., 2010). Lines carrying Xa21 in combination with xa5, xa13, or both have confirmed strong resistance, highlighting the effectiveness of Xa21 in enhancing BB resistance levels in rice (Huang et al., 1997; Sanchez et al., 2000; Singh et al., 2001; Pradhan et al., 2016; Sagar et al., 2018). The broad-spectrum resistance conferred by Xa21 and xa13 is attributed to their synergistic effect in combating the BB pathogen. However, in the present study, the alleles of xa13 and Xa21 being in a heterozygous condition will hinder the complete resistance expression of the R-gene-pyramided plant. Hence, further advancing the generation (BC₂F₂) was recommended for attaining homozygosity for the alleles concerned.
 
Background selection for recurrent parent genome recovery
 
The efficiency of MABB can be greatly enhanced through background selection, facilitating the earlier development and release of BB-resistant cultivars (Joseph et al., 2004). The R-gene introgressed BC₂F₁ plant (8.3.9.10) showed similarity to the recurrent parent Uma at 13 out of 22 microsatellite markers, while it matched the donor parent ISM at four loci and remained heterozygous at five loci. The recurrent parent genome recovery, computed following Sundaram et al., (2008), was 70.45%, as against the expected recovery of 87.5%. This lower magnitude of recurrent parent genome recovery may be attributed to the relatively low recovery (21.80%) observed in the corresponding R-gene introgressed BC₁F₁ (8.3.9) parent, as previously reported by Joseph (2016). Furthermore, given the size of the rice genome (approximately 400-450 Mbp), the use of a relatively small number of markers (only 22) for background selection may have influenced the accuracy of the estimated recurrent parent genome recovery in the R-gene introgressed plant. The genome recovery in the R-gene introgressed BC₂F₁ plant was also represented through Graphical Genotyping Software (GGT) version 2.0 (Fig 3a). Additionally, the dendrogram generated grouped the parents and Plant No. 8.3.9.10 into two clusters, where cluster 1 contained only the donor parent ISM (monogenic), while cluster 2 included the recurrent parent Uma and plant no. 8.3.9.10, reaffirming their similarity (Fig 3b). Therefore, advancing the backcross generation is expected to enhance the recurrent parent genome contribution.


 
Morphological characterisation of R-gene-pyramided BC2F1 and parents
 
To assess the phenotypic resemblance of the R-gene-pyramided BC₂F₁ to its parents, morphological characterisation was conducted (Table 1). Based on morphological traits, clustering of the BC₂F₁ plants and parents was performed. At an 80% similarity coefficient, they were grouped into 10 major clusters, with the donor parent ISM forming a separate monogenic cluster (Fig 4). The R-gene introgressed BC₂F₁ plant (8.3.9.10) exhibited 45.39% similarity to the recurrent parent Uma and 26.86% similarity to the donor parent ISM. Consistent with the molecular clustering pattern, morphological clustering also placed the R-gene introgressed plant closer to the recurrent parent Uma than to the donor parent ISM. The R-gene-introgressed BC₂F₁  plant (8.3.9.10) was early flowering (115 days), showing greater similarity to Uma (120 days) than to ISM (148 days). However, the BC₂F₁ plant (8.3.9.10) exhibited a significant reduction in the number of tillers per plant and spikelets per panicle compared to rice variety Uma, possibly owing to poor adaptation of the backcross progenies arising from linkage drag. Notwithstanding these reductions, the grain length, decorticated grain length and kernel colour of the BC₂F₁ plant (8.3.9.10) closely resembled those of Uma, indicating successful retention of the Matta (Red kernelled rice) rice characteristics preferred by the consumers of Kerala (Fig 5a and 5b). In contrast the grain width and decorticated grain width in the R-gene introgressed line were lower than those of rice variety Uma resulting in comparatively slender grains resembling those of the donor parent ISM. These observations suggest incomplete recovery of the recurrent parent phenotype, thereby necessitating further backcrossing to enhance recurrent parent traits recovery in the R-gene introgressed line. Integrating agro-morphological evaluation with molecular genotyping is more effective for the rapid recovery of desirable genotypes possessing a higher proportion of the recurrent parent genome (Olalekan et al., 2019; Chukwu et al., 2020).






 
Bioassay of BC₁F₂s
 
The BC₁F₂ plants developed in parallel (Joseph 2016) were evaluated for resistance to the BB pathogen using the leaf-clipping method described by Kauffman et al., (1973). The BC₂F₁ plants were not subjected to phenotypic screening, as they may be heterozygous at the bacterial blight resistance loci; in particular, since xa13 confers resistance only in the homozygous recessive state. Phenotyping in this generation may thereby result in loss of plants bearing the desirable allele in the heterozygous state. The recurrent parent, Uma, was used as the susceptible check and the disease severity was scored 1-9 according to the IRRI standard evaluation system (Fig 6a, b and c). The donor parent revealed a score of 1, indicating resistance, while Uma had a score of 9, indicating high susceptibility. The BC₁F₂ plants displayed a higher percentage of resistant (54.72%) and moderately resistant (20.75%) genotypes than of susceptible (3.77%) and highly susceptible (10.38%) genotypes (Table 2). These results indirectly suggest the presence of the appropriate R-gene combinations (xa5 + xa13 + Xa21) in the BC₁F₂ individuals that exhibited resistant and moderately resistant reactions to the BB pathogen, in contrast to the recurrent parent Uma. However, further confirmation through molecular assays and phenotypic evaluation in the BC₁F₃ generation is necessary to conclusively determine the specific order and nature of gene combinations responsible for the observed resistance to BB infection.

The present study successfully evaluated the backcross generations derived from the cross between rice varieties Uma and ISM through integrated molecular marker-based genotypic screening and phenotypic assessment approaches. Genotypic screening of BC₂F₁ plants for bacterial blight resistance identified only a single plant (8.3.9.10) carrying the resistance genes xa13 and Xa21. This limited recovery may be attributed to the reduced population size arising from poor germination of the F₁s, thereby underscoring the need for further evaluation of cross-compatibility between the parental lines. Phenotypic screening for bacterial blight resistance could not be performed in the gene-introgressed BC₂F₁ plant (8.3.9.10) due to the heterozygous status of the resistance loci (xa13/Xa13 and xa21/Xa21), particularly considering the recessive nature of xa13, which may impede phenotypic expression of resistance. Furthermore, the identified plant (BC₂F₁ 8.3.9.10) exhibited a recurrent parent genome recovery of 70.45%, which was substantially lower than the theoretically expected 87.5% at the BC₂ stage. Therefore, in view of the heterozygous status of the resistance genes and the comparatively low recovery of the recurrent parent genome, the breeding programme was further advanced to develop BC₂F₂ and BC₃F₁ generations. In addition, the lower marker density employed in the present study may also have contributed to the reduced estimation of recurrent parent genome recovery, highlighting the necessity for screening with a larger set of polymorphic markers for more precise background selection.
       
In parallel, the BC₁F₂ lines were phenotypically screened for BB resistance using the leaf-clipping method, which revealed a higher proportion of resistant to moderately resistant plants against the BB pathogen. These lines were subsequently advanced to the BC₁F₃ generation. The advanced backcross populations will undergo further evaluation through marker-assisted selection (MAS) and phenotypic screening to identify superior genotypes pyramided with BB resistance genes and exhibiting near-complete recovery of the recurrent parent genome, ‘Uma’.

We are grateful for the financial assistance and contributions of the DBT (Department of Biotechnology) project ‘Rice-Gene Pyramiding to develop cultivars with durable resistance to Bacterial Leaf Blight through Marker Assisted-Selection’ for the commencement of this programme and Tintumol Joseph for providing the experimental material for the conduct of the present study.
 
Author contributions
 
Experimental conception and design were done by Rose Mary Francies. Study material preparation, execution, data collection, analysis and result interpretation were performed by L.M. Megha. The draft manuscript was prepared by L.M. Megha and statistical analysis was done by both Rose Mary Francies and L.M. Megha. The manuscript was reviewed and improved by Rose Mary Francies, Jiji Joseph, P.S. Abida and P. Raji. All authors read and approved the final manuscript. Rose Mary Francies, Jiji Joseph, P.S. Abida and P. Raji provided the necessary facilities for the study.

The authors declare that they have no relevant financial or non-financial interests to disclose.