Marker-Assisted Introgression of Quantitative Resistance Gene pi21 Confers Broad Spectrum Resistance to Rice Blast

Rosalyn B. ANGELES-SHIM , Vincent P. REYES Marilyn M. del VALLE Ruby S. LAPISJunghyun SHIM , Hidehiko SUNOHARA, Kshirod K. JENA Motoyuki ASHIKARI Kazuyuki DOI

(1Plant Breeding Division, International Rice Research Institute, DAPO Box 7777, Metro Manila, the Philippines; 2Bioscience and Biotechnology Center, Nagoya University, Chikusa-ku Furo-cho, Nagoya, Aichi 464-8601, Japan; 3Graduate School of Bioagricultural Sciences, Nagoya University, Chikusa-ku Furo-cho, Nagoya, Aichi 464-8601, Japan; 4Department of Plant and Soil Science, College of Agricultural Sciences and Natural Resources, Texas Tech University, Lubbock, Texas 79409, USA)

Abstract: The quantitative resistance gene pi21 from Sensho was introgressed to an indica breeding line IR63307-4B-13-2, a pyramiding line IRBB4/5/13/21, and a tropical japonica line Kinandang Patong by marker-assisted backcrossing. A total of 192 improved lines at the BC4F3 and BC4F4 generations were developed and confirmed to have the gene introgression via genotyping using a pi21-specific InDel marker. Thirteen randomly selected improved lines, representing all the three genetic backgrounds,demonstrated resistance against leaf blast composites in the field and a broader spectrum resistance against individual isolates compared to the recurrent parents in the glasshouse. Specifically, the tested lines exhibited pi21-acquired resistance against 11 leaf blast isolates that elicited susceptible reactions from the recurrent parents. All the tested lines maintained a comparative heading date, and similar or improved panicle length, number of primary branches per panicle and number of total grains per panicle relative to the recurrent parents. The physical grain characteristics of the recurrent parents were also maintained in the 13 lines tested, although variability in the amylose content and chalkiness degree was observed. The successful marker-assisted introgression of pi21 in diverse genetic backgrounds and the resulting broader spectrum resistance of improved lines against leaf blast indicate the potential of pi21 for deployment in cultivars grown across other rice growing regions in Asia.

Key words: rice blast; pi21 gene; marker-assisted breeding; disease resistance

Rice blast caused by Magnaporthe oryzae (anamorph Pyricularia grisea Sacc.) remains the most economically devastating disease of cultivated rice worldwide and is a major threat to rice production (Dean et al, 2005;Sharma et al, 2012; Yasuda et al, 2015). The incidence of the disease has been reported in 85 countries,particularly in the irrigated and rainfed lowlands of temperate and subtropical Asia, as well as in the uplands of tropical Asia, Latin America and Africa (Sharma et al,2012). Annual economic loss due to blast has been estimated to equal yields that can feed 60 million people (Zeigler et al, 1994; Scheuermann et al, 2012).

Although various measures are available to control rice blast, breeding and cultivation of resistant rice cultivars remain the most efficient, economical and environmentally benign means to manage the disease,especially in resource-poor farmers’ fields (Séré et al,2007; Zeng et al, 2015). To date, more than 100 genes for blast resistance have been identified in rice,although less than 20 have been cloned (Sharma et al,2012; Yasuda et al, 2015). Host plant resistance through the expression of these genes has been generally classified as either qualitative or quantitative. Qualitative or complete resistance is often race-specific and is controlled by a single dominant or recessive R gene(Kou and Wang, 2012). In contrast, quantitative, partial or field resistance is usually non-race-specific and is controlled by quantitative trait loci (QTLs) or multiple genes.

The recessive pi21 allele is a quantitative blast resistance gene that was isolated from the japonica rice cultivar Owarihatamochi on chromosome 4 (Fukuoka et al, 2009). Deletions in the proline-rich motifs of pi21 accelerate the defense response of the plant under blast attack, although the response is neither as strong nor as fast as that of a qualitative resistance gene. This type of non-specific, weaker response presumably contributes to the reduction of selective pressure for pathogens to overcome host resistance, rendering pi21 more durable and broad-spectrum (Fukuoka and Okuno, 2001; Fukuoka et al, 2009, 2012). Previous studies on the introgression of pi21, alone or in combination with other quantitative resistance genes in select Japanese rice cultivars confirmed the effectivity of pi21 in conferring durable resistance against blast isolates in Japan (Fukuoka et al, 2009, 2015; Yasuda et al, 2015; Horo et al, 2016). Combining pi21 with another quantitative resistance gene, Pi35, in the japonica rice cultivar Koshihikari, results in the resistant reaction of the cultivar against the Japanese blast isolate Ao-92-06-2. Even without Pi35, pi21 is able to significantly reduce the diseased leaf area caused by the blast isolate (Yasuda et al, 2015).Similarly, near-isogenic lines of pi21 in the japonica rice cultivar Aichiasahi exhibit significant reduction in leaf lesion size when challenged with nine Japanese blast isolates (Fukuoka et al, 2015). Although the results of these studies strongly indicate the efficiency of pi21 in conferring blast resistance in rice, the effects have only been assessed in the genetic background of temperate japonica cultivars against the blast isolates from Japan.

The genetic diversity and evolution of fungal pathogens are spatially dependent on their native hosts.Because of this dependence, blast resistance genes have been reported to exhibit differential effectivity against fungal pathotypes from various geographical origins (Kiyosawa and Ling, 2001; Séré et al, 2007). It is likely, therefore, that both the spectrum and degree of resistance of pi21 against blast isolates from Japan will differ from those in different geographical sources. To establish the spatial suitability of pi21 for targeted deployment, its expression in different genetic backgrounds in response to diverse pathogen populations under multiple environments needs to be evaluated. The aim of this study was to introgress pi21 by marker-assisted backcrossing into indica and tropical japonica backgrounds, and assess the spectrum of resistance of the gene in these genetic backgrounds against blast isolates maintained at the International Rice Research Institute (IRRI) in the Philippines.Preliminary evaluations of the effects of pi21 introgression on the agronomic performance and grain quality of the improved rice lines were also carried out.

MATERIALS AND METHODS

Rice materials

The Japanese upland rice cultivar Sensho shares the recessive pi21 allele of Owarihatamochi (Kawasaki-Tanaka and Fukuta, 2014; Yasuda et al, 2015) and has been shown to exhibit field resistance to blast isolates in Japan (Goto, 1970; Kato et al, 2002). In this study,Sensho was used as the donor of pi21, whereas an advanced breeding line IR63307-4B-13-2 (IR63307),a pyramiding line IRBB4/5/13/21 (IRBB) and a landrace Kinandang Patong (KP) were selected as breeding targets for leaf blast resistance improvement(Fig. 1-A). IR63307 is a medium-maturing indica line bred for salt tolerance at IRRI. IRBB is an indica pyramiding line bred to express Xa4, xa5, xa13 and Xa21 genes for bacterial blight resistance (Huang et al,1997). KP is a tropical japonica landrace (Uga et al,2013) from the Philippines with reported tolerance to drought (Yue et al, 2006; Uga et al, 2011, 2013) and is mostly grown in the uplands.

Introgression of pi21 allele from Sensho to different genetic backgrounds by marker-assisted backcrossing (MAB)

A graphical representation of the marker-assisted backcrossing scheme used to transfer pi21 to various genetic backgrounds is presented in Fig. 1-B. Initial crosses (F1) between Sensho and the recurrent parents,as well as early backcross populations in the backgrounds of IRBB and KP (BC1F1), and IR63307 (BC2F1), were generated at the Togo field of Nagoya University,Aichi, Japan, from 2011 to 2012. In January 2013,BC1F1and BC2F1seeds were exported to IRRI for further backcrossing and generation advance. Due to the limited number of seeds, only 12 and 24 BC1F1plants in the IRBB and KP backgrounds, respectively,and 48 BC2F1plants in the IR63307 background were grown during the initial planting of the materials in IRRI in the wet season of 2013. Ten plants that approximate the gross morphology of the recurrent parents were then selected from each population for blind backcrossing. During the following dry season(2014), a total of 240 BC2F1and 144 BC3F1plants were established in the field for marker-assisted selection (MAS) using an InDel marker (forward primer:GATCCTCATCGTCGACGTCTGGC and reverse primer: AGGGTACGGCACCAGCTTG) targeting the pi21 locus in Sensho. Seven plants from each cross combination that are heterozygous for the pi21 allele were selected for backcrossing to the recurrent parents.The rest of the plants that are segregating for the pi21 allele were tagged and allowed to self-pollinate. Each of 120 BC3F1plants in the backgrounds of IRBB and KP were genotyped for the pi21 introgression and 7 plants that are heterozygous for the target allele were again backcrossed to their respective recurrent parents.Sixty BC4F1plants each in the genetic backgrounds of IRBB, KP and IR63307 were advanced up to BC4F3generation by self-pollination following MAS. Because the breeding lines in the IR63307 background were already advance by one backcross generation upon exportation to IRRI, we were able to self them by one more generation (BC4F4). DNAs of all the materials used in the MAB were extracted following the method of Miura et al (2009), using a buffer composed of Tris-HCl, potassium chloride and EDTA salt. PCR analysis was carried out using a standard PCR profile for simple sequence repeats (Shim et al, 2015).

All the improved lines generated in this study were part of the breeding project ‘Wonder Rice Initiative for Food Security and Health’ (WISH) and hence, are referred to hereinafter as ‘WISH’ lines.

Screening of WISH lines for leaf blast resistance under nursery and glasshouse conditions

Sensho, IR63307, IRBB and KP, along with 15 randomly selected WISH lines representing all the three recurrent backgrounds were screened for their resistance to natural blast composites in the uniform blast nursery (UBN) of IRRI, as well as to 20 virulent isolates of the pathogen in the glasshouse. The indica rice varieties Lijiangxintuanheigu (LTH) and CO39 were used as susceptible controls (Teleblanco-Yanoria et al, 2011; Vasudevan et al, 2014), whereas the cultivar IR72 and/or breeding line IR65482-4-136-2-2 were used as resistant controls during the course of evaluation. IR72 and IR65482-4-136-2-2 carry the blast resistance genes, Pib and Pi40, respectively(Fujita et al, 2009; Prahalada et al, 2017). Leaf blast screening of the parental and the WISH lines in UBN was carried out following the methods of Suh et al(2009). Seedlings of each line (40-50 plants) were planted in two replications in nursery beds during the dry season of 2016. Spreader rows composed of a mixture of cultivars (IR72, IR36, CO39, IR50 and IR42) with varying levels of resistance to leaf blast were planted around each replicate to maintain the diverse pathogen population. Alongside this, susceptible controls were planted in parallel fields. Leaf blastinfected seedlings of the susceptible controls were used as natural sources of inoculum. Ten days after sowing,controls exhibiting leaf lesions due to blast were uprooted from parallel fields and spread in between the experimental materials to allow natural infection by the pathogen. The nursery beds were watered 3-4 times a day and covered with a plastic sheet every afternoon(5:00 pm) to maintain the high humidity requirement for pathogen sporulation. Scoring of the test materials for leaf lesions due to blast was carried out at 14 (initial or spot reading) and 24 d after inoculation (final scoring). Lesion type was scored using the 0-9 scale of the standard evaluation system (SES) of IRRI where scores of 0-3 indicate resistance, 4-6 indicate moderate resistance (quantitative resistance), and 7-9 indicate susceptibility (IRRI, 2014).

Isolate-specific screening for resistance to 20 known virulent blast isolates from the Philippines (Teleblanco-Yanoria et al, 2008; Selisana et al, 2017) was also carried out in two replicates in the glasshouse facility of IRRI during the dry season of 2016 following Jeung et al (2007) (Supplemental Table 1). The WISH lines(24 plants/line) were grown in plastic trays (10 rows ×2 columns) in 20 batches for inoculation with 20 blast isolates. Seedlings of each line (21-day-old) were spray-inoculated with 20 mL spore suspension (1.5 ×105spores/mL) of individual isolates. The inoculated plants were maintained in the glasshouse at 12 h day /12 h night photoperiod and 90% relative humidity for 7 d. One week before the assessment of the reaction, the plants were transferred to the incubation chamber set at 25 °C ± 2 °C. Scoring was carried out following the standards for leaf blast scoring of the Japan International Research Center for Agricultural Sciences (Hayashi et al,2009). Leaf lesion scores ranging from 0 to 2.9 were considered resistant, whereas scores of 3-5 were considered susceptible.

Preliminary field evaluation of WISH lines for yield components

Preliminary evaluation of the agronomic performance of the WISH lines with the pi21 introgression was carried out in a field experiment conducted in IRRI during the wet season of 2016. Seeds used for the agronomic testing came from the same lot used for leaf blast resistance screening. Twenty-four plants per line were grown to maturity in 2-row plots spaced at 20 cm × 20 cm following the standard agronomic practices in IRRI. Data on yield components including heading date, plant height, tiller number per plant,panicle length, number of primary branches per panicle,total number of grains per panicle, seed-setting rate,and 100-grain weight were recorded from five plants each line, as well as from the recurrent parents used to generate the introgression lines. Heading date was determined as the number of days from seeding until 50% of the plants per line were flowering. Seeds were air-dried in the glasshouse for 7 d or until seed moisture content reached 13%-14% and then weighed.

Comparative grain quality testing of WISH lines having pi21

Each WISH line (24 plants) and the parents were planted in the field during the dry and wet seasons of 2017. The seeds came from the same lot were used for leaf blast resistance screening and agronomic testing.At harvest, seeds for each line were bulk-harvested,and 200 g seeds per line from each season were submitted to the Grain Quality and Nutrition Center of IRRI for comparative grain quality testing. All lines were screened for physicochemical properties including grain dimension and overall grain chalkiness degree using the CervitecTM1625 Grain Inspector (Foss,Denmark). Measurement of amylose content was conducted using the routine method of ISO6647(International Standardization Organization, 2007).

Statistical analysis

All agronomic data were analyzed using the Statistical Tool for Agricultural Research software v2.01 (STAR v2.01) (IRRI, 2013). Significant differences in the values of the parameters measured between the advanced generation lines relative to those of the recurrent parents were determined by one-way analysis of variance and post hoc comparison of means using the Tukey’s test at 95% confidence level (P ＜ 0.05).

RESULTS

Development of WISH lines with pi21 introgression by MAB

A total of 72 WISH lines in the background of IR63307 and 60 WISH lines each in the backgrounds of IRBB and KP having the pi21 allele for blast resistance from Sensho were generated as part of the WISH Breeding Project. Repeated backcrossing of the early breeding lines to their respective recurrent parents was carried out to recover as much as 95%(96.875% theoretically in BC4Fn) of the genome of the recurrent parent, as well as to increase the chances of recombination within the pi21 locus that will allow the dissociation of the latter from genes that negatively affect grain quality of rice. Backcross generations segregated into 1:1 ratio for the pi21 locus.

Reaction of WISH lines to blast isolates

Seedlings of two WISH lines in the KP background were damaged by pests during the blast screening in the field and hence were removed from the initial set of materials that were selected for leaf blast resistance screening and agronomic evaluation (Table 1).

Sensho recorded resistance to the leaf blast composite in the field, with average resistant scores of 1.1 and 1.5 during the initial and final readings,respectively. The resistance scores of Sensho were comparable to those obtained for the resistant control,IR65482-4-136-2-2. Similarly, the recurrent parents IR63307 and KP exhibited resistant reactions to the naturally occurring leaf blast composite in the UBN,with final reading scores ranging from 0 to 0.5. IRBB showed moderate resistance, with average scores of 4.0 and 5.0 during the initial and final readings,respectively. All the WISH lines scored resistant to the leaf blast composite in the nursery, with average scores ranging from 0 to 3.5. The susceptible control,CO39 scored higher than 7 during both the initial and final readings, whereas LTH plants scored moderately resistant during the initial reading but died before the final reading.

In the glasshouse, the donor line, the recurrent parents, as well as all the WISH lines exhibited differential reaction to the 20 leaf blast isolates tested(Supplemental Table 2). Sensho was resistant to 16 highly virulent leaf blast isolates and susceptible to 4 isolates namely V850256, BN209, BN111 and P06-6.The resistant scores of Sensho against the 16 virulent strains ranged from 0 to 2, whereas the susceptible scores against the remaining 4 strains ranged from 3 to 4. Although used as susceptible controls, CO39 and LTH exhibited resistance to isolate 43, with CO39 also recording resistance to isolate BZ64-1. However, the resistant controls IR65482-4-136-2-2 and IR72 showed susceptibility to isolates JMB8401 and M39-1-3-8-1,respectively. IR63307 exhibited susceptibility to only three isolates (51671, P06-6 and 9475-1-3), whereas IRBB exhibited susceptibility to nine isolates(JMB840610, V850256, M39-1-2-21-2, Pi9-G7-2K-1,JMB8401, IK813, P06-6, M64-1-3-9-1 and IK81-25),and KP also to nine isolates (V850256, V806010,BN209, BN111, Pi9-G7-2K-1, M101-1-2-9-1, 51671,M64-1-3-9-1 and 9475-1-3). Lesion scores caused by leaf blast isolates that elicited susceptible reactions from IR63307, IRBB and KP ranged from 3 to 5.

WISH lines in the IR63307 background were resistant to all the 20 isolates despite the susceptibility of both donor and recurrent parent lines to isolate P06-6. All the IRBB-derived lines were resistant to 15 isolates and completely susceptible to 2 isolates(P06-6 and M64-1-3-9-1). WISH48:1-3-2-1 in the genetic background of IRBB was resistant to leaf blast isolate V850256, which elicited a susceptible reaction from both IRBB and Sensho. All WISH lines in the background of KP were resistant to 19 leaf blast isolates, 8 of which elicited a susceptible reaction from KP. Resistances to V850256, BN209 and BN111 of WISH lines derived from KP were observed despite the susceptibility of both Sensho and KP to theseisolates. All the 18 WISH lines were resistant to 8 blast isolates namely JMB840610, M39-1-2-21-2,BN209, BN111, M101-1-2-9-1, JMB8401, IK81-3 and 5167-1, despite the susceptibility of either parents to some of the isolates (Supplemental Table 2).

Table 1. Reaction of WISH lines to natural leaf blast infection in the field based on SES scores (IRRI, 2014).

Agronomic performance of improved lines with pi21 for blast resistance

The WISH lines resembled their respective recurrent parent for most of the agronomic traits examined,although a few lines recorded significantly different values for specific agronomic traits. For example, WISH lines in the genetic background of IR63307 exhibited significantly higher plants (WISH110:1-1-11-12-1,WISH110:1-1-11-5-4 and WISH110:2-5-2-1-1), longer panicles (WISH110:2-5-2-1-1), more number of primary branches per panicle and total grain number per panicle(WISH110:1-1-11-5-4 and WISH110:2-5-2-1-1), higher seed-setting rate (WISH110:1-1-11-12-1 and WISH110:2-5-2-1-1), and significantly lower 100-grain weight(WISH110:2-5-2-12-1) compared to the recurrent parent (Fig. 2). All WISH lines in the background of IRBB recorded a significantly lower values for at least one agronomic trait compared to the recurrent parent.Only WISH48:1-3-20-1 exhibited a significantly higher value for total grain number per panicle compared to IRBB (Fig. 2). All WISH lines in the KP background approximated the gross morphology of the recurrent parent except for one line (WISH40:1-3-7-7)that exhibited significantly lower plant height (Fig. 2).Heading dates in all WISH lines were similar to those of the respective recurrent parents.

Grain quality of improved lines having pi21 allele

Analysis of variance showed no significant differences in the brown and cooked grain length and width, and grain shape of the WISH lines compared to their respective recurrent parents (Supplemental Table 3).However, variation in the chalkiness degree in the grains was observed across the WISH lines tested regardless of the genetic background and season (Fig.3). A lower grain chalkiness degree was recorded for all the WISH lines in the genetic background of IR63307 compared to the recurrent parent (Fig. 3-B).

IRBB recorded 0.3%-0.4% chalkiness degree, whereas the WISH lines derived from IRBB recorded chalkiness degree ranging from 0.3%-7.9% during both seasons(Fig. 3-C and Supplemental Table 3). Similarly, KPderived WISH lines recorded variable chalkiness degrees that ranged from 6.6% to 17.0% compared to the 7.5%-8.5% chalkiness degree of the recurrent parent during both seasons (Fig. 3-D). WISH lines derived from IR63307 and KP had comparable amylose content with those of the recurrent parents. Grains of WISH lines in the IRBB background recorded a generally higher range of amylose content (15.3% to 21.0%)compared to that of the recurrent parent (12.9%-13.0%) during the dry and wet seasons (Supplemental Table 3).

DISCUSSION

The differential response of the same blast resistance genes to fungal pathotypes from various geographical origins has been attributed to the spatial dependence of the pathogen to its native hosts (Kiyosawa and Ling,2001; Séré et al, 2007). To establish the spatial suitability of a resistance gene such as pi21 for targeted deployment, it needs to be evaluated for disease response under multiple environments, preferably in the target location, using diverse races of the pathogen.

In this study, pi21 was introgressed by MAB into three different rice backgrounds that are preferentially grown by farmers for favorable traits inherent in each line. Disease screening established the resistance of the pi21 donor cultivar, Sensho, to leaf blast composites in field nurseries, as well as to 16 out of 20 leaf blast isolates in the glasshouse. Similarly, IRBB, KP and IR63307 recorded varying levels of resistance to blast composites in the field and against individual isolates.

Genes for leaf blast resistance have not been directly identified from IR63307 and IRBB although the pedigrees of each cultivar hint on the origin of their resistance to the blast isolates used in the study.IR63307, for example, is a product of a cross between a somaclonal variant of the salt-tolerant landrace Pokkali and the breeding line IR51511-B-B-34-B.IR51511-B-B-34-B was selected from crosses between the breeding line IR8909 and the cultivar IR34. The resistance genes Pib, Pik-s, Piz-t and Pia that have been identified in IR34 (Imbe et al, 1998) may have provided the inherent resistance of IR63307 to 17 leaf blast isolates tested in the study. However, because Pib, Pik-s, Piz-t and Pia have not been reported to exhibit resistance to all isolates used in this study, the resistant reaction of IR63307 to 17 leaf blast isolates may also be attributed to the presence of other unidentified genes (Fukuta et al, 2004; Teleblanco-Yanoria et al, 2011). Likewise, IRBB benefits from having the genetic background of the rice cultivar IR24, which is a product of multi-crossing of at least 28 landraces (Li and Yuan, 1985). Previous studies showed that IR24 has resistance to the Philippine blast isolates BN209, V850196, BN111 and B9002 as conferred by the Pib, Pik-s, Pi20t and Pia genes,respectively (Imbe et al, 1997, 1998; Sallaud et al,2003). In the present study, IRBB also exhibited resistance to BN209 and BN111, indicating the probable presence of Pib and Pi20t in the cultivar. The presence of other undetermined genes from IR24,alone or in combination with Pib and Pi20t, and possibly Pia and Pik-s, may account for the resistance of IRBB to 11 Philippine blast isolates. KP is a landrace that is highly adapted to the Philippines uplands. It has been reported to exhibit weak resistance to both upland and lowland blast isolates(Bonman et al, 1986). Isolate-specific screening of KP in the glasshouse demonstrated the resistance of the landrace to 11 out of 20 blast isolates, confirming previous reports of blast resistance in KP.

Given the inherent resistance of the recurrent parents used in the study, the resistant scores of the WISH lines against the field leaf blast composites may be attributed to either the action of pi21 from Sensho, alone or in combination with the unidentified or putative resistance genes present in the background of the recurrent parents. Under glasshouse conditions,pi21 introgression provided resistance to a wider range of leaf blast isolates in the WISH lines compared to the recurrent parents. Differences in the response of the WISH lines to the pathogen under field and glasshouse conditions may be attributed to the absence of isolates that elicited susceptible reactions from the WISH lines under glasshouse conditions in the composites of the pathogen that are present in the field. Identification of the individual isolates that comprise the blast composite in the field would be necessary to confirm such assumption.

The acquired resistance of the WISH lines to specific isolates showed that pi21 can confer a broader spectrum of resistance against blast isolates coming from geographically different origins. Conversely, the observed resistance of WISH lines to isolates that elicited susceptible reactions from Sensho substantiates the presence of inherited blast resistance genes in IR63307 and IRBB, and the presence of unidentified resistance gene in KP. The isolate-specific resistance of IR63307, IRBB and KP indicated their potential as sources of new genes/QTLs conferring resistance to blast disease.

Incidentally, the WISH lines recorded resistance to specific blast isolates that elicited susceptible reactions from both the donor and the recurrent parents.Transgressive segregation for biotic stress resistance has been commonly reported in plants. In fact, susceptible parental lines producing highly resistant progenies have been reported in various crops including tomato,cotton, wheat and barley (Vega and Frey, 1980;Wallwork and Johnson, 1984; Cherif and Harrabi,1993; de Vicente and Tanksley, 1993; Wang et al, 2012).Classical genetic studies have provided evidences for the role of rare recessive alleles, non-additivity of allelic effects within (over-dominance) and between(epistasis) loci and complementary action of additive alleles to explain transgressive segregation in crops(Rick and Smith, 1953; Vega and Frey, 1980).Marker-based QTL analysis in tomato has shown that different species or parental lines are often fixed for sets of alleles with opposing effects (de Vicente and Tanksley, 1993). The recombination of alleles with opposing effects in the crosses in this study might have brought together complementary alleles in the WISH lines that conferred resistance to specific blast isolates(Rieseberg et al, 1999).

Preliminary field evaluation showed that the WISH lines approximated the agronomic performance of the recurrent parents. These results validated previous findings that pi21 has no negative effects on the yield and yield-related traits of rice (Fukuoka et al, 2009).The observed variations in the agronomic performance of the WISH lines may be due to specific recombinations that can be eliminated by selection for target agronomic traits at earlier generations.

Linkage drags that reduce the overall fitness of an individual have been commonly associated with breeding for disease resistance not only in rice but also in other cereals such as wheat and barley (Piffanelli et al,2002; Chen et al, 2016). Initial studies on the introgression of pi21 from Sensho to Koshihikari confirmed the strong association between blast resistance and poor eating quality in rice. However,grain quality evaluation of different recombinants showed that lines having the chromosome sequence of Koshihikari from a point less than 2.4 kb fragment downstream of pi21 has similar eating quality as that of the recurrent parent, whereas those having the chromosome fragment of Sensho upto 37 kb downstream of pi21 have poor eating quality. The gene Os04g0401400 located within this 37 kb region is associated with poor eating and grain quality(Fukuoka et al, 2009).

In the present study, backcrossing of early breeding lines to their respective recurrent parents at least four times and selfing up to three times not only served to recover the overall gross morphology of the recurrent parents but also increase the chances of recombinations that can possibly dissociate pi21 with downstream genes causing poor grain quality.

Rice grain chalkiness has low heritability and is highly affected by environmental factors such as high temperature during grain ripening and high moisture during harvesting (Ram and Mishra, 2010). The environmental differences during the wet and dry seasons most probably contributed to the seasonal variation in the chalkiness degree of the WISH lines.

Selection for leaf blast resistance improvement was carried out using an InDel marker that tightly cosegregates with pi21. While marker selection ensured the identification of plants that are heterozygous for the pi21 locus from a minimum number (24 plants) of breeding materials for backcrossing, it also limited the number of sibs to select from for good grain quality. To ensure the recovery of the grain quality of the recurrent parents, grain quality testing in early backcross progenies,generation of a larger population of breeding lines or marker-assisted recombinant selection to dissociate pi21 from genes responsible for poor grain quality can be carried out.

WISH lines carrying pi21 showed resistance to a wider range of blast isolates from outside of Japan,indicating the suitability of pi21 for targeted deployment outside of the country. To further establish the spectrum of resistance of pi21, screening for the resistance of the gene against more isolates would be necessary. The WISH lines also need to be tested in multiple environments for a prolonged duration to establish the durability of pi21 against the blast isolates. IR63307 and IRBB are adapted to the tropical rice ecosystems of Southeast Asia while KP is adapted to upland and rainfed ecosystems. Adoption of WISH lines with acquired resistance to the most number of isolates tested has the potential to reduce losses caused by blast disease, while benifitting from the favorable traits inherent to each background.

ACKNOWLEDGEMENTS

The research was supported by the Japan International Cooperation Agency as part of the Wonder Rice Initiative for Food Security and Health (WISH) Project,and in part by the Canon Foundation and the SATREPS project entitled ‘Improvement in Productivity and Yield Stability of Rice under Kenya’s Biotic and Abiotic Stress Conditions Through Tailor-made Breeding and Development of Cultivation Methods’.

SUPPLEMENTAL DATA

The following materials are available in the online version of this article at http://www.sciencedirect.com/science/journal/16726308; http://www.ricescience.org.

Supplemental Table 1. Twenty blast isolates from the Philippines used in the leaf blast isolate-specific screening of WISH lines carrying the pi21 allele.

Supplemental Table 2. Reaction of WISH lines to twenty leaf blast isolates in the glasshouse.

Supplemental Table 3. Grain quality of WISH lines having the pi21 allele for leaf blast resistance from Sensho measured during the dry (DS) and wet season (WS) of 2017.