PI

C.Sallaud· M.Lorieux· E.Roumen· D.TharreauR.Berruyer· P.Svestasrani· O.GarsmeurA.Ghesquiere· J.-L.Notteghem

Identification offive new blast resistance genes inthehighly blast-resistant rice variety IR64 using aQTL mapping strategy

Received: 10 April 2002 / Accepted: 1 July 2002 / Published online: 29 October 2002©Springer-Verlag 2002

AbstractRice progenies used for the construction ofgenetic maps permit exhaustive identification and char-acterization of resistance genes present in their parentalcultivars. We inoculated a rice progeny derived from thecross IR64 ×Azucena with different Magnaporthe gri-seaisolates that showed differential responses on the pa-rental cultivars. By QTL mapping, nine unlinked lociconferring resistance to each isolate were identified andnamed Pi-24(t) to Pi-32(t). They could correspond tonine specific resistance genes. Five of these resistanceloci (RLs) were mapped at chromosomal locations whereno resistance gene was previously reported, defining newresistance genes. Using degenerate primers of the NBS(nucleotide binding site) motif found in many resistancegenes, two resistance gene analogues (RGAs) IR86 andIR14 were identified and mapped closely to two blastRLs (resistance identified in this study, i.e. Pi-29(t) andPi-30(t) respectively). These two RLs may correspond tothe Pi-11and Pi-ablast resistance genes previously iden-tified. Moreover, the ir86and ir14genes have been iden-tified “in silico” on the indicarice cultivar 93-11, recent-Communicated by H.F. Linskens

C.Sallaud· E.Roumen· D.Tharreau (✉)· R.BerruyerO.Garsmeur· J.-L.Notteghem

CIRAD, TA73/09, 34398MontpellierCedex05, Francee-mail: tharreau@cirad.frFax: +33-4-67-6155-41

M.Lorieux· A.Ghesquiere

IRD, RiceGenomicsResearchUnit, BP5045, 34032MontpellierCedex1, France

P.Svestasrani

KingMongkut’sInstituteofTechnologyLadkrabang,

ChumphonCampus, Mhoo6.Chumco, Pathieu, Chumphon,86160Thailand

Presentaddress:

E.RoumenBayerCropScience, JozefPlateaustraat22,9000Gent, Belgium

Presentaddress:

J.-L.NotteghemENSA.M, 2placeViala, 34060MontpellierCedex1, France

ly sequenced by Chinese researchers. Both genes en-codes NBS-LRR-like proteins that are characteristics ofplant-disease resistance genes.

KeywordsRice Blast · Magnaporthe grisea· Resistancegenes · NBS-LRR · Genetic mapping

Introduction

Rice blast caused by the fungal pathogen, MagnaporthegriseaHebert (Barr), is one of the most devastating ricediseases. In this pathosystem, race-specific resistance isgoverned by the gene-for-gene relationship (Kiyosawa1971; Silué et al. 1992). To-date, more than 40 majorblast resistance genes have been mapped (reviewed inImbe and Matsumoto 1985; Mackill and Bonman 1992;Tohme et al. 1993; Pan et al. 1996; Nagato andYoshimura 1998; Tabien et al. 2000; Fukuoka and Okuno 2001). However, some of these genes could beidentical or allelic, since very few allelism tests wereperformed. Kiyosawa (1984) described differential culti-vars with one or two resistance genes named Pi-, fol-lowed by a different letter for each gene. Some of thesegenes have several alleles, such as loci Pi-zand Pi-tawith two alleles and Pi-kwith five alleles. Some otherresistance genes from tropical cultivars were introducedinto isogenic lines using the susceptible cultivar Co39 asa recurrent parent (Imbe et al. 2000). Allelism tests per-formed on these isogenic lines showed that most of thesegenes were allelic or identical to known Pigenes (Inukaiet al. 1994).

When testing the intensively studied Japanese differ-ential cultivars with exotic isolates having a new aviru-lence gene, Imbe and Matsumoto (1985) discovered anew resistance gene, named Pi-sh. This showed that ad-ditional resistance genes could be detected even in well-characterised differential lines after exposure to a largeset of isolates from diverse geographic or genetic ori-gins. Many comparative inoculations of differential setsand resistant local cultivars were carried out, demon-

strating that almost all rice cultivars possess one, two ormany resistance genes. In most cases, the resistance pat-terns of tested varieties differed from those of standarddifferentials such as the Japanese differentials. Such re-sults indicated the presence of new resistance genes orcombinations of genes in these cultivars. The character-ization of these new genes was difficult to completesince it implies a long task. For each new rice cultivar, itrequires comparison of its resistance pattern with thoseof reference cultivars having known resistance genes to alarge set of differential isolates. Allelism tests must alsobe performed using progeny from crosses with standarddifferential cultivars having similar resistance patterns.This work is more tedious considering the increasingnumber of newly identified resistance genes. Therefore,the use of the molecular identification of resistancegenes is of great interest.

During the last few years, many resistance genes havebeen cloned. A surprising result which has emerged isthat a large number of resistance genes display a similarcharacteristic domain identified as a nucleotide bindingsite (NBS), often associated with a leucine rich repeatmotif (LRR) (Meyers et al. 1999). The NBS motif pos-sesses a highly conserved amino acid domain allowingprimers to be designed for PCR amplification of poten-tial resistance gene homologs. Using this strategy, manyNBS homologs have been cloned in dicotyledous andmonocotyledous plants such as potato (Leister et al.1996), soybean (Kanazin et al. 1996; Yu et al. 1996), Arabidopsis(Aarts et al. 1998), barley (Leister et al.1998) and rice (Mago et al. 1999). This strategy can be apowerful tool to facilitate the isolation of resistancegenes that were previously mapped. Putative resistancegene fragments homologous to the NBS motif [resis-tance gene analogue (RGA) markers] can be rapidlyidentified by PCR and used as molecular markers on ex-isting genetic maps. When co-localisation is observedbetween the RGA and a putative resistance locus, theRGA can be assigned to the resistance gene locus andused to further identify the resistance gene (Aarts et al.1998; Collins et al. 1998; Leister et al. 1999; Shen et al.1998).

As new strategies for resistance genes are being tested(lineage exclusion, Zeigler et al. 1994), characterisationof the resistance gene diversity in rice is important. IR64is one of the most cultivated rice variety in Asia and ishighly resistant to blast disease under irrigated condi-tions (Bonman et al. 1989; Roumen et al. 1992). Howev-er if the blast resistance gene pyramid has been suggest-ed to explain this, the genetics of resistance is not wellknown due, in particular, to its complex genealogic ori-gin (Roumen et al. 1994). In this paper, by exploiting thegenetic diversity of the rice blast fungus, we studied thenumber of major resistance genes present in IR64 andAzucena. A QTL (quantitative trait locus) detection ap-proach was used in order to localize major or minor lociinvolved in the resistance. Several statistical methodswere used to validate the results. Then, using the consen-sus primer of the NBS found in the R gene, we identified

795

several RGAs from IR64 and tested whether these mark-ers could be closely linked to the blast resistance genesdescribed in this study.

Materials andmethods

Rice cultivars andprogeny

A progeny from the reference cross IR64 improved semi-dwarf indica (IR64) obtained by IRRI and an up-×Azucena, between anland japonica from The Philippines (Azucena) was used. A popu-lation of 105 doubled-haploid lines (DH) was obtained by Guiderdoni et al. (1992) and used to map 200 molecular markers(Causse et al. 1994).Rice cultivation

Rice plants (of 40 Oryza sativa1/8 pouzzolane). Ten to 15 seeds of each DH line was sown in×29 ×7cm filled with compost (7/8 Neuhaus compost no.9,L.) were grown in a greenhouse in traysrows, in trays containing 14 lines each. Soil was kept moist withwater and, once a week, with nutritive solution. Nitrogen fertiliza-tion with 8.61 day(s) before inoculation to increase susceptibility to blast.g of nitrogen equivalent was done at 10, 3 and Inoculation method

Inoculations with 3 weeks after sowing either by injection or by spraying with M. griseaHebert (Barr) were performed conidial suspensions. For the spray method, 30conidia.ml–1tray. Then rice plants were stored for 1 night in a controlled cli-suspension with 0.5% gelatin were sprayed on eachml of a 50,000 matic chamber at 24then transferred back to the greenhouse. For the injection method,°C and 95% relative humidity. They wereplants were inoculated by injecting about 0.1conidia.ml–1ml of a 25,000 repetitions of parents and DH lines were grown and inoculated atsuspension with a syringe into the leaf sheaths. Twodifferent times. After 7 days, lesion types on rice leaves were ob-served and scored 1 (resistant) to 6 (susceptible) according to astandard reference scale (Silué et al. 1992). Progenies with scoresbetween 1 and 3 were considered to be resistant and progenieswith scores from 4 to 6 were considered to be susceptible. ForQTL mapping, the exact score of each progeny (1 to 6) was usedas a quantitative variable.Isolates

Twenty nine rice blast isolates from 20 countries (including repre-sentatives from Latin America, Asia and Africa) were initiallychosen from a collection of 1,500 field isolates from 55 countries,to maximize diversity. These isolates were screened on the paren-tal lines (IR64 and Azucena) as well as 13 randomly chosen DHlines. Isolates showing clear differential reactions on parents andDH lines were then inoculated to the whole set of 105 availableDH lines.Molecular markers

Plant DNAs were isolated from lyophilized leaves using theCTAB method (Murray and Thompson 1980). A core map devel-oped at IRRI (Huang et al. 1994) with RFLP, RAPD (random amplified polymorphic DNA) and isozyme markers was used.Marker density for regions of specific interest was increased usingthe following protocols: for RFLP markers, probes from the inter-specific rice map (Causse et al. 1994) were used (kindly providedby Dr. S. McCouch, Cornell University, USA). Southern transfers,

796

hybridization and non-radioactive DNA labeling used for hybrid-ization were done according to IRRI or CIMMYT protocols (Hoisington et al. 1994). For some probes, especially the RGAprobes, radioactive labelling was performed using the MegaprimeDNA labelling system (Amersham Life Science). Six restrictionenzymes were used for the core map: DraI, EcoRI, EcoRV,HindIII, ScaI, and XbaI. Fourteen additional enzymes were used totest for polymorphism with probes showing a monomorphic pat-tern. Several other RFLP markers were placed onto the map bydifferent partners of the EGRAM (European Graminae MappingInitiative) project. These markers mainly corresponded to rice,wheat, oats, barley, sorghum and sugar cane cDNAs with large-spectrum hybridization on the DNA of several grass species lowing the procedure described by Edwards et al. (1991). Degen-erated oligonucleotide primers LM637 and LM638 were synthe-sized according to Kanazin et al. (1996). PCR amplifications wereperformed in a total volume of 25µl with 50ng of genomic DNAand 1.25 Units of TaqGold polymerase (Perking Elmer). DNAtemplate was denatured at 94°C for 10min, followed by 35 cyclesof 1min at 64°C, 30s at 45°C and 30s at 72°C. PCR productswere separated on 1.5% agarose gels, isolated using the GenecleanII kit (BIO101). DNA fragments were cloned into the plasmidpGEMT- (Promega) following the manufacturer’s instructions,and then transformed into Escherichia coliDH5α.

(Garsmeur et al., in preparation). For RAPD markers, PCR ampli-fications were carried out in 250.02gene), 150U/µl of Taqµl, as follows: 0.4µM of primer,(95PCR products were separated on 1.5% agarose gels and stained°C – 1µmin; 35M of dNTP. Conditions were 95polymerase (Appligene), 1 ×buffer mix (Appli-°C – 1min; 72°C – 2min); 72°C – 5°C – 6min; 45 min.×with ethidium bromide. STSs (sequence tagged sites) published byInoue et al. (1994) were also used. These primers correspond toprobes from the high density map of the Rice Genome ResearchProgram – Japan (Kurata et al. 1994). PCR conditions were thesame as in Inoue et al. (1994). When polymorphism between par-ents was not detected directly, amplification products were digest-ed with restriction enzymes recognizing 4-bp sequences. Separa-tion of PCR/digestion products were carried out on 2–3% agaroseor 8% polyacrylamide gels and stained with ethidium bromide.Map construction

The genetic map of molecular markers and RGAs was computedusing the multipoint functions performed by MapMaker/EXP v.3.0 (Lander et al. 1987). The two-point LOD score threshold wasset to 5, and rthe ‘order’, ‘try’ and ‘ripple’ commands, which calculate likeli-maxto 0.3. Ordering of markers was achieved usinghood ratios for the different possible multipoint orders. Conver-sion of recombination fractions into centimorgans (cM) was ob-tained with Kosambi’s mapping function (Kosambi 1944). The fi-nal map was drawn using MapDisto v. 1.2 (http://www.mpl.ird.fr/mapdisto).

Cosegregation analysis between markers andresistance traitsA QTL detection approach was employed in order to localize lociwith a major or minor effect on resistance. Several statisticalmethods were used. We first performed as the criteria using MapDisto v. 1.3. Quasi-simple interval map-F-testswith the markersping (Q-SIM) was then performed using Qgene v. 3.0.6u (Nelson1997). Although it is based on an properties to the simple interval mapping method based on theF-test, this analysis has similarLOD-scoreposite interval mapping (CIM) was performed in order to resolvetest (Lander and Botstein 1989). When needed, com-unclear QTL positions or to try to identify linked QTLs. CIM wascomputed using QTL Cartographer package v. 1.15c (Basten et al.1994, 2001), with model 6 activated in the Zmapqtl module. De-fault parameters were used for choosing QTLs involved as cofac-tors and window size. QTL or major gene detection was also con-firmed by the distribution-free Kruskall and Wallis test usingMapQTL for Unix v.2.4 (Van Ooijen 1992). As a summary of re-sults, we choose to present three statistical parameters for each de-tected QTL: the quasi-Qgene software, and the LOD-scoreby MapDisto.

R-squared(thereafter (R2) and additivity (LOD) given by thea) givenPCR amplification andcloning ofRGAs

Rice cultivar IR64 was used for PCR-based isolation of resistancegene candidates. Genomic DNA was extracted from leaves fol-RGAs classification based onrestriction analysis andsequencingPositive clones were detected directly from colonies by PCR am-plification using M13 universal and reverse primers. Clones weregrouped by restriction enzymes recognizing 4-bp sequences (Hingested with ten units of restriction enzymes, and separated on 3%fI, RsaI and AluI). Typically, 3.5µl of the PCR product was di-TaqI,agarose gels. After classification, one member of each group was sequenced from both strands using T7 and SP6 primers byGenome express (Grenoble, France).

Results

Resistance evaluation

The inoculation of the 26 M. griseaisolates on parentsand on the subset of 13 DH lines from the cross IR64 ×Azucena revealed an important variability in DH-line re-sistance patterns. None of the isolates were virulent onboth parents. Fourteen isolates were avirulent on bothparents. Among them, seven were avirulent on all 13 DHlines, and seven were virulent on at least some of the DHlines. The 12 remaining isolates were avirulent on IR64and virulent on Azucena. These 12 isolates were virulentto some DH lines and avirulent to others. All 12 isolatesdiffered for their virulence to the subset of DH lines test-ed. These results strongly suggest the existence of acombination of avirulence genes in M. griseaisolatesand of the corresponding resistance genes in the DHlines. Among the 12 isolates tested, six were kept forfurther studies since they showed clear susceptible reac-tions on at least about 25% of the DH lines (see Fig.1).The observed resistant:susceptible ratios observed forthese isolates could correspond to a genetic model in-volving one ore two major resistance genes and one orseveral QTLs with variable effects. These six isolateswere inoculated on all the 105 available DH lines.Resistance loci mapped intheIR64 ×Azucena crossThe distribution frequency of the lesion scores obtainedon the 105 DH lines with the six blast isolates are illus-trated on Fig.1. The score on the standard 1–6 scale ofeach DH line was used as a quantitative variable forQTL mapping on the IR64 ×Azucena map previouslydeveloped. When considering resistance QTLs with aLOD score higher than 2.5, we detected nine resistanceloci (RLs) mapped on 8 of the 12 rice chromosomes

797

Fig.1Distribution frequencyofthelesion scores obtainedontheprogeny of105 DH lines(IR64 ×Azucena cross) afterinoculation withblast isolatesBR26, PH68, CD69, CH66,CH72 andCL6. Lesion typeswere scored 1 (resistant) to6(susceptible) according toastandard reference scale (Siluéet al. 1992). Frequency iscal-culated inpercentage, indicat-ing theratio between thenum-ber ofplants withthesame le-sion score onthetotal numberofplants tested

(Table1, Fig.2). One isolate (CD69) revealed one RL,three isolates (BR26, CH72 and CL6) revealed two RLsand the last two (CH66 and PH68) revealed three RLs.Each RL was detected by one (six RLs), two (two RLs)or three isolates (one RL). None of these RLs were effi-cient against all the isolates, showing that all were iso-late-specific and could be considered as putative specif-ic resistance genes for blast. Six RLs were inheritedfrom the indica cultivar IR64 and three RLs from thejaponica cultivar Azucena. For CD69 and PH68, the re-sults seem to fit well with the expected genetic model.The observed resistant:susceptible ratios were close to1:1 and thus clearly showed the predominant effect ofone major gene, which was evidenced by a very highLOD scoreand Rsqstatistics (LOD= 19.43 and 17.64,Rsq= 0.54 and 0.47, respectively). For PH68, two addi-tional QTLs with smaller effects were also found. ForCH66 and CH72, the resistant:susceptible segregationis close to 3:1, indicating a possible action of two majorgenes. The results fit well to this model for CH72, astwo loci were detected. For CH66, the accordance isless clear since three loci were evident. This showsthat, at least for CH66, we cannot consider that the re-sistance segregation pattern is resulting in independantmajor genes. It could be considered as the result of thesegregation of several QTLs, i.e. genes with less impor-tant individual effects on the trait. The observed LODvalues, which reside between 2.33 and 4.28, and theRsqvalues, which reside between 0.1 and 0.16, seem tosupport this model. For CL6 and BR26, the segregationis intermediate between a 1:1 and a 3:1 segregation.This seems to be in good accordance with the observedLODsand Rsqs, which indicate that these two traits are

controlled by two genes, with one of a predominant ef-fect.

Five RLs were mapped in chromosomal locationswhere no specific resistance gene to blast was previouslydescribed, leading to the identification of five new blastresistance genes. Following the current blast resistancegene nomenclature, they were named Pi-24(t) to Pi-28(t)(Table1, Fig.2).

Four RLs were mapped in chromosomal locationswhere other specific resistance gene(s) to blast were lo-cated (Fig.2). These may correspond to the specific re-sistance genes already described to these map positionsor to different alleles. Pi-29(t), identified on chromo-some 8, mapped closely to the Pi-11(t) (previouslynamed Pi-Zh, Zhu et al. 1993) blast resistance gene. Pi-30(t), located on chromosome 11, mapped closely tothe Pi-ablast resistance gene (Kiyosawa 1972; Goto etal. 1981). Pi-31(t), located on chromosome 12, mappedclosely to the Pi-6(t), Pi-157and Pi-ta, (=Pi-4; allelic or closely linked to Pi-ta2) blast resistance genes (Kiyosawa 1972; Mackill and Bonman 1992; McCouchet al. 1994; Naqvi and Chattoo 1996). Pi-32(t), locatedon chromosome 12, mapped closely to the Pi-12(t), Pi-tq6and Pi-21(t) blast resistance genes (Inukai et al.1996; Tabien et al. 2000; Fukuoka et al. 2001).RGAs isolation andmapping

Using degenerated primers LM637 and LM638, corre-sponding to the P-loop (GGVGK/NTT) and HD (GLPLT)motif (Kanazin et al. 1996) of the conserved NB-ARCdomain, we amplified an expected 500-bp band from

798

Table1Summary oftheloci identified forresistance tosix blastisolates byQTL analysis inthepresent study. Foreach resistancelocus, theclosest marker onthegenetic map isindicated. Threestatistics are shown: LOD= maximum LOD score forthepresenceofalocus controlling theanalysed trait estimated bytheSimpleChromo-Closestsomemarker12568

K5RG520RG313Est-2RZ617

RGA-IR86RZ500OpZ11-fRGA-IR14O10-800

LOD = 17.64LOD = 19.43Rsq = 0.47Rsq = 0.54LOD = 2.99Rsq = 0.10a = –0.73

LOD = 4.28Rsq = 0.12a = 0.53

LOD = 3.24Rsq = 0.10a = 0.42

LODa= 4.65Rsqb= 0.21ac= 0.68

LOD = 2.70Rsq = 0.10a = 0.47

LOD = 2.33Rsq = 0.10a = 0.43

BR26

PH68

CD69

Interval Mapping method; Rsq= percentage ofvariance explainedbythelocus, a= additivity forthelocus. a> 0 indicates afavor-able allele fromIR64 parent, whereas a< 0 indicates afavorableallele fromtheAzucena parentCH66

CH72

CL6LOD = 4.2Rsq = 0.16a = –0.69

ResistancePossible locus nameidentityPi-24(t)Pi-25(t)Pi-26(t)

LOD = 3.55Rsq = 0.16a = 0.56

Pi-27(t)

LOD = 6.30

Rsq = 0.22Pi-29(t)a = 1.06

Pi-28(t)Pi-30(t)Pi-31(t)

Pi-aPi-6(t),Pi-157, Pi-ta, Pi-ta2Pi-12(t),Pi-tq6, Pi-21(t)Pi-11(t)

LOD = 2.83Rsq = 0.11a = –0.80

1011

12

12AF6

LOD = 8.65Rsq = 0.29a = 0.90

a = 1.54a = 1.61

Pi-32(t)

aLOD= SIM LOD (Qgene)bRsq= R-squared (MapDisto)

ca

= additivity (MapDisto); a> 0, favourable allele from IR64; a< 0, favourable allele from Azucena

IR64 genomic DNA. One hundred clones were analyzedand grouped into 19 classes using 4-bp restriction en-zymes. One member of each class was sequenced. Se-quence comparison leads to the identification of onlyseven different RGA families having a partial putativeORF (open reading frame) with a significant homologyto the NB-ARC domain of the resistance genes. This re-sult indicated that seven putative RGAs had been identi-fied (C. Sallaud, unpublished data). These RGAs weremapped to assess their genetic linkage to the RLs identi-fied in this study. Using 105 DH lines of the IR64 ×Azucena population, one member of each RGA familywas successfully mapped. They were located on chromo-somes 3, 7, 8 and 11. Two RGAs, named IR86 and IR14,revealed a co-localization with two RLs identified in thisstudy.

RGA-IR86 was mapped on chromosome 8 betweenmarkers G104 and C225b. A RL [Pi-29(t)] identified inthis study, and the blast resistance gene Pi-11(t) (Zhu etal. 1993), are also located at the same position. The IR86nucleotide sequence of 501 bp shows 86% and 85% iden-tity with two others RGAs, p8558-3 (NID: Y09812, Xueet al. 1998) and NBA3 (NID: AF159886, Zhou et al., un-published) respectively. Using the blastN search on thenew draft sequence of the rice (O. sativassp.indica) ge-nome recently published by Yu et al. (2001), we identi-fied a unique sequence with more than 98% nucleic acididentity with RGA-IR86. The sequence is located withincontig10362. This contig is 8,554bp in length, and con-tains an intron-less ORF of 3,111bp starting at position425 and ending at position 3,538 (Fig.3a). In view of thehigh homology (98% nucleic acid identity) betweenRGA-IR86 and the ORF present within contig 10362, itis tempting to assume that this ORF represents an ir86gene; ir86encodes a putative protein of 1,037 amino ac-ids having a nucleotide-binding site and a leucing-rich re-peat domain that are characteristic of the NBS-LRR typeof plant disease-resistance proteins (data not show). The

799

IR86 protein show 56% and 55% identity with two puta-tive rice NBS-LRR proteins YR9 (NID: AAK93796) andNLP1 (NID: AAK58606) respectively. These two pro-Fig.2Genetic localisation ofseveral loci controling resistanceteins are deduced amino-acid sequences of cDNAs fromtosix blast strains inrice. Thegenetic map was obtained usingseedling leaf (Yang et al., unpublished data) and root li-adoubled-haploid population derived fromanther culture ofthebraries (Zhou et al., unpublished data).F1 (IR64 ×Azucena). Thedifferent types ofmarkers were placed

RGA-IR14 was mapped on chromosome 11, betweenonthemap byseveral laboratories (see text fordetails) andare in-dicated ontherightofthelinkage groups. Theblast resistance locimarkers OpZ11-f and CSU50. OpZ11-f and IR14 are the

which were identified either intheliterature orinthepresent studymarkers closest to the RL Pi-30(t) identified in this(inbold underlined) are indicated ontheleftofthelinkage groups.study. The Pi-ablast resistance gene was previouslyAtentative name was given totheresistance loci identified

inthepresent study. Probes corresponding totheRGAs ofthemapped in this position (Kiyosawa 1972; Goto et al.present study are underlinedandare named as RGA-IRxx. 1981). The IR14 nucleotide sequence exhibits more thanMapping function: Kosambi99% homology with two putative RGAs identified as

800

Fig.for3Schematic representation showing putative ORFs encodingsequences fromNBS-LRR-like proteins. Contig sequences are nucleic acid

AtheO. sativassp. indicaforContig 10362 withaputative ORF of1,037 amino acids codingcultivar 93-11 genome. ORFs coding forNBS-LRR-like proteins. BContig 1837 withthree putativeorduced manually todeletions atpositions 2,826, 6,801 andNBS-LRR-like proteins. Three base insertions7,651 have been intro-like proteins ofof912 andobtain ORF1 and907 amino acids, respectively. PositionORF2 encoding NBS-LRR-withthecate theablack vertical arrowRGA sequence homologous to. Horizontal black shaded boxesthecontig isindicatedindi-barshaded arrowsindicates nucleotide binding-site domains (NBS), leucine-rich repeat protein motif (LRR), thevertical blackindicate tourist MITE-type transposable elements

horizontal

RGA8 (NID: AF074889, Mago et al. 1999) and h359-1

(NID: Y09807, Xue et al. 1998). RGA 8 was isolatedfrom the indicarice variety Phalguna and was mapped atthe same position on chromosome 11. Using the blastNsearch on the draft sequence of the Chinese indicaricegenome (cultivar 93-11), we found only one sequencewith high homology to IR14 (97% nucleic acid identity).This homology obtained with a different cultivar sug-gests that it corresponds to the ir14gene. This sequencebelongs to contig1837 which is 16,882bp in length. Thiscontig contains two putative ORFs (ORF1 and ORF2)encoding NBS-LRR-like proteins of 912 and 907 aminoacids respectively (Fig.3b). Both proteins share a high

degree of homology (70% identity), and a significant ho-mology with RPR1(64% identity), a NBS-LRR-like pro-tein as well as to other NBS-LRR proteins (data notshown). Rpr1is located within another contig of the in-dicarice genome. One and two frameshift mutationscaused by nucleotide insertions or deletions in ORF1 andORF2 respectively have to be manually corrected in thesequence at positions 2,826, and 6,801 and 7,651, to ob-tain the two putative ORFs. We do not know if these in-sertions or deletions correspond to sequencing errors, orif they leads to inactive alleles, or pseudogenes. A third,although partial ORF (ORF3) encoding a putative pro-tein having a homology to LRR proteins is located at theend of the contig in an opposite orientation.

Discussion andconclusion

This study showed that genetically fixed progeny likedoubled-haploid lines, constitute a very useful tool forthe characterization of resistance alleles originating fromtwo parental lines. Such permanent populations permitthe exploitation of the genetic diversity of the pathogento detect specific resistance genes. When more than oneresistance allele is combined in the segregating popula-tion, the QTL mapping method can be more powerfulthan Mendelian analysis to map these genes, even if theresistance observed in the population is not quantitative(Fig.2).

We inoculated a limited set of 13 DH lines from thecross IR64 ×Azucena with 26 M. griseaisolates.Among them, each of the 12 isolates giving a polymor-phic reaction between the two cross parents gave a dif-ferent resistance pattern on the set of DH lines. Thiscomplex pattern reflects the different combinations of atleast nine segregating resistance loci (RLs).Five out of the nine RLs identified in this studymapped in chromosomal regions where no specific resis-tance gene to blast has been previously described. Thus,these five loci can be considered to be new resistance

genes to the blast fungus. This is a significant numberwhen compared with the 40 rice resistance genes to blastalready described. This is all the more surprising sinceonly six blast isolates and two rice cultivars were used inthis study. These data suggest that the number of resis-tance genes yet to be discovered remains large.

For the four RLs that mapped in chromosomal re-gions where specific resistance genes were already de-scribed, allelism tests between reference cultivars withknown resistance genes and some DH lines having onlyone RL will be necessary to demonstrate whether theseRLs are new resistance gene alleles. As more DH linesare produced and more markers are mapped, a more ac-curate localisation of the existing resistance genes willbe possible. Moreover, as more avirulence genes will beidentified, the use of isolates with only one avirulencegene will be another powerful tool to simplify the identi-fication of RLs.

The presence of six RLs in IR64 is unexpected, sincemost rice cultivars seem to have one or two known resis-tance genes. Such an unplanned pyramiding of resistancegenes has only been found in the rice cultivar Hama Asa-hi (Kiyosawa et al. 1991). The origin of the six RLs fromIR64 is not known since the genealogy of this cultivar isvery complex and involves different resistant parentalcultivars such as Tetep, Tadukan, Peta, Taichung Native1, CP-SLO, Chow Sung, BPI 76, NM S4, PTB 18,TKM6, GP 15, MUDGO, 17-1LT, W1263, Dee-geo-woo-gen and Oryza nivara. Progeny of such crosseswere tested in IRBN nurseries (with susceptible spread-

ers and a natural local population of M. grisea). This re-sistance screening is likely to have led to the selection ofprogeny that had accumulated different resistance genes.IR64 is resistant to M. griseain many countries under ir-rigated conditions. In Asia, its resistance in the field isconsidered to be good, even in areas where virulent iso-lates on IR64 are present.

The NBS motif belongs to a larger domain family, theNB-ARC domain, which is shared by proteins involvedin the regulation of cell death in animals and resistancegenes in plants is often associated with the LRR domain(known as NBS-LRR-like proteins). More than 70% ofthe approximately 40 resistance genes that have beencloned up to now, possess a NB-ARC domain. More-over, recent genome sequence data reveal that the NBS-LRR class represents as much as 1% of the Arabidopsis genome. In this paper, we have shown thatwith a limited set of degenerate primers homologous tothe NB-ARC domain, two out of seven distinct RGAswere mapped very close to the blast resistance genes.Moreover, another RGA identified in this study, IR19, islocated at the end of chromosome 11 where multipleblast resistance genes have been identified (Pi-1, Pi-k,Pi-18, Pi-sh, Pi-f). These results confirmed other studiesin various plants such as Arabidopsis, Rice, Maize, Lettuce and Potato (Leister et al. 1996, 1999; Botella etal. 1997; Aarts et al. 1998; Collins et al. 1998; Shen etal. 1998), showing that RGAs are localized where knownresistance genes had been previously mapped. These ex-amples show the utility of the candidate gene approachto facilitate the cloning of resistance genes. The genomeof the japonicarice cultivar Nipponbareis being se-quenced by an international rice consortium and 50% ofthe sequence is already available. More recently, a draftsequence of the indicacultivar 93-11, which covers 97%of the genome, has been published by Chinese research-ers (Yu et al. 2001). By data mining on rice genome se-quences, the number of NBS-LRR genes have been esti-mated to be more than 500. It is feasible to design spe-cific primers that will allow the identification of RGAsalleles in different rice varieties. With the rice physicalmap being available for the Nipponbarecultivar, it willbe easier to co-localize the putative resistance gene,with the RL, found by QTL analysis. In this paper, wedemonstrate that this strategy could be promising. Bytaking advantage of the available rice genome data,(http://btn.genomics.org.cn/rice), we have identified “insilico” two genes corresponding to two RGA markers(IR14 and IR86) which co-localized with blast diseaseRLs. Both genes encode a putative NBS-LRR-like pro-tein. One of them (ir14) is highly homologous with rpr1which has been identified by differential display in ascreen to find genes induced by probenazole, a chemicalinducer of acquired resistance to M. grisea(Sakamoto etal. 1999). On the same contig, we also identified twoputative NBS-LRR-like proteins, one of which is highlyhomologous to ir14, suggesting gene duplication. Inter-estingly, rpr1, which shows significant homology toir14, mapped on chromosome 11 close to RGA-IR14.

801

This seems to indicate that a resistance gene cluster ispresent in this region. This has not been previouslyfound. By screening an IR64 BAC library with RGA-IR14 and RGA-IR86 probes we have identified severalBAC clones (data not shown) that will be used to confirm whether rpr1belong to this cluster, and the precise physical organization of this resistance genecluster.

We have presented here a consensus map for mostblast resistance genes identified to-date (Fig.2). Whileco-localization with a gene candidate is effective, thework is far from complete. The next step is to use themas molecular markers in resistance gene segregationanalysis. The same strategy is being used to identifyQTLs using rice ESTs with defense-response gene-likesequences (Wang et al. 2001). Finally, mutagenesisand/or complementation analysis will be necessary forthe final proof of the resistance gene function. This workis in progress for the putative resistance genes, ir14andir86, found in this study.

Acknowledgementsproviding us with RFLP map data. Henri Adreit and Joëlle We thank Dr. N. Huang (IRRI) for kindlyMilazzo are acknowledged for their excellent technical support.We also thank Claudia Kaye, Nathalie Chantret and Marc Gibandfor a critical review of the manuscript.

References

Aarts MGM, Te Lintel HekkertStiekemaB, HolubEB, BeynonJL,

mologous DNA fragments genetically linked to disease resis-WJ, PereiraA (1998) Identification of R-gene ho-tance loci in Arabidopsis thaliana. Mol Plant Microbe InteractBasten11:251–258

SmithC, WeirGibsonC, GavoraB, ZengJS, BenkelZ (1994) Zmap-QTL cartographer. In:

B, ChesnaisJ, FairfullCongress on Genetics Applied to Livestock Production: Com-JP, KennedyBW, BurnsideEB (eds) Proc 5th WorldW, puting Strategies and Software. Published by the organizingCommittee, 5th World Congress on Genetics Applied to Live-Bastenstock Production, Guelph, Ontario, Canada, pp65–66

Department of Statistics, North Carolina State University, C, WeirB, ZengZ (2001) QTL cartographer, version 1.15.

BonmanRaleigh, North Carolina, USA

resistance in tropical lowland rice cultivars. Plant DisJM, EstradaBA, BandongJM (1989) Leaf and neck blast

Botella73:388–390

JDG, SomervilleMA, ColemanMJ, DouglasEH, NishimuraMT, Jones

sequences with sequence similarity to disease resistanceSC (1997) Map positions of 47 ArabidopsisCaussegenes. Plant J 12:1197–1211

XiaoMA, FultonTM, ChoYG, AhnSN, ChunwongseJ, WuK,

McCouchJ, Yuof the rice genome based on an interspecific backcross popula-SR, TanksleyZ, RonaldSD (1994) Saturated molecular mapPC, HarringtonSE, SecondG,Collinstion. Genetics 138:1251–1274

(1998) The isolation and mapping of disease resistance geneNC, WebbCA, SeahS, EllisJG, HulbertSH, PryorA

Edwardsanalogs in maize. Mol Plant Microbe Interact 11:968–78

method for the preparation of plant genomic DNA for PCRK, JohnstoneC, ThompsonC (1991) A simple and rapid

Fukuokaanalysis. Nucleic Acids Res 19:1349–52

a recessive gene for field resistance to rice blast in JapaneseS, OkunoK (2001) QTL analysis and mapping of pi21,

upland rice. Theor Appl Genet 103:185–190

802

GotoI, JawYL, BaluchAA (1981) Genetic studies on the resis-tance of rice plants to the blast fungus.IV. Linkage analysis offour genes, Pi-a, Pi-k, Pi-z, and Pi-i. Ann Phytopathol Soc Japan 47:252–254

GuiderdoniE, GalinatoE, LuistroJ, VergaraG (1992) Anther cul-ture of tropical japonica x indica hybrids of rice (Oryza LeisterD, KurthJ, LaurieDA, YanoM, SasakiT, GranerA,

Schulze-LefertP (1999) RFLP- and physical-mapping of re-sistance gene homologues in rice (O. sativa) and barley (H. vulgare). Theor Appl Genet 98:509–520

MackillDJ, BonmanJM (1992) Inheritance of blast resistance in

near-isogenic lines of rice. Phytopathology 82:746–749

HoisingtonsativaL.). Euphytica 62:219–224

tory protocols: CIMMYT applied molecular genetics laborato-D, KhairallahM, González-de-LeónD (1994) Labora-Huangry 2nd edn. CIMMYT, Mexico City, Mexico

Development of an RFLP map from a doubled-haploid popu-N, McCouchSR, MewT, ParcoA, GuiderdoniE (1994)

Imbelation in rice. Rice Genet Newslett 11:134–137

eties to the blast fungus strains virulent to the variety “Reiho”.T, MatsumotoS (1985) Inheritance of resistance of rice vari-ImbeJpn JBreed 35:332–339

of blast resistance in IR varieties and resistance breeding strat-T, TsunematsuH, KatoH, KushG (2000) Genetic analysis

egy. In: Tharreau(eds) Advances in Rice Blast Research. Kluwer AcademicD, LebrunMH, TalbotNJ, NotteghemJLInouePress, Dordrecht, The Netherlands, ppMiyaderaT, Zhong(1994) Sequence-tagged sites (STSs) as standard landmarkersN, NagamuraHS, MiyaoA, Ashikawa1–8

Y, KurataN, SasakiI, MonnaL, FukuokaT, MinobeS,

YInukaiin the rice genome. Theor Appl Genet 89:728–734

BonmanT, NelsonRJ, ZeiglerRS, SarkarungS, Mackillresistance genes in near- isogenic lines of rice. PhytopathologyJM, TakamureI, KinoshitaT (1994) Allelism of blastDJ,

Inukai84:1278–1283

T, NelsonT, ZeiglerRS, SarkarungS, BronsonM, DungLV, Kinoshita

rice blast-resistance by marker-aided selection from a recom-RJ (1996) Development of pre-isogenic lines forKanazinbinant inbred population. Theor Appl Genet 93:560–567

logues are conserved and clustered in soybean. Proc Natl AcadV, MarekLF, ShoemakerRC (1996) Resistance gene ana-KiyosawaSci USA 93:11,746–11,750

S (1971) Genetical approach to the biochemical nature

Kiyosawaof plant disease resistance. JARQ 6:72–80

Kiyosawabreeding. IRRI, Los Baños, The Philippines, ppS (1972) Genetics of blast resistance. In: IRRI (ed) Rice

203–225

pathogenicity tests of the rice blast fungus. Rice Genet News-S (1984) Establishment of differential rice varieties for

Kiyosawalett 1:53–67

Three examples of yearly changes of unnecessary virulenceS, AndoI, IshiguraK, NemotoF, HashimotoA (1991)

genes under field conditions where resistance genes for riceKosambiblast are not present. Bull Fukushima Prefect Exp Stn 30:23–36DD (1944) The estimation of map distance from recom-Kuratabination values. Ann Eugen 12:172–175

AntonioN, NagamuraY, YamamotoK, HarushimaY, SueN, WuJ,

FukudaKiriharaA, ShimanoBA, ShomuraT, KubokiA, ShimizuY, ToyamaT, LinSY, InoueT, TamuraSasakiYT, Hayasaka, WangZX, MommaK, MiyaoA, MonnaT, MiyamotoT, UmeharaL, ZhongY, YanoHS, Y, M,map of rice including 883 expressed sequences. Nature GenetT, MinobeY (1994) A 300 kilobase interval geneticLander8:365–372

lying quantitative traits using RFLP linkage maps. GeneticsES, BotsteinD (1989) Mapping Mendelian factors under-Lander121:185–199

SE, NewburgES, GreenP, Abrahamsonpackage for constructing primary genetic linkage maps of ex-L (1987) Mapmaker: an interactive computerJ, BarlowA, DalyMJ, Lincoln

Leisterperimental and natural populations. Genomics 1:174–181

approach for isolating pathogen resistance genes from potatoD, BallvoraA, SalaminiF, GebhardtC (1996) PCR-based

with potential for wide application in plants. Nature GenetLeister14:421–429

GranerD, KurthJ, LaurieDA, MasahiroY, TakujiS, Devossistance gene homologues in cereal genomes. Proc Natl AcadA, Schulze-LefertP (1998) Rapid reorganization of re-K,

Sci USA 95:370–375

Magoform rice: cloning, sequencing and mapping. Theor Appl GenetR, NairS, MohanM (1999) Resistance gene analogues

McCouch99:50–57

blast resistance genes in rice. In: ZeiglerSR, NelsonRJ, TohmeJ, ZeiglerRS (1994) Mapping of

TengRS, LeongSA,MeyersWallingford,PS (eds) Rice blast disease. CAB International/IRRI,S, SobralBC, DickermanUK, pp167–186

genes encode members of an ancient and diverse protein fami-BW, YoungAW, MichelmoreND (1999) Plant disease resistanceRW, Sivaramakrishnan

ly within the nucleotide-binding superfamily. Plant J 20:317–Murray332

molecular-weight plant DNA. Nucleic Acids Res 8:4321–MG, ThompsonWF (1980) Rapid isolation of high-Nagato4325

symbolization, nomenclature and linkage groups. Rice GenetY, YoshimuraA (1998) Report of the committee on gene

NaqviNewslett 15:13–74

quence-characterized amplified region-assisted selection ofNI, ChattooBB (1996) Molecular genetic analysis and se-blast resistance in rice. In: IRRI (ed) Rice genetics III. IRRI,NelsonManila, The Philippines, ppJC (1997) QGene: Software for marker-based genomic

570–576

Pananalysis and breeding. Mol Breed 3:239–245

new blast resistance gene in the indica rice cultivar KasalathQ, WangL, IkehashiH, TanisakaT (1996) Identification of a

using Japanese differential cultivars and isozyme markers.RoumenPhytopathology 86:1071–1075

tance in rice cultivars to virulent isolates of the blast pathogen.EC (1992) Small differential interactions for partial resis-RoumenEuphytica 64:143–148

resistance to blast disease in rice within a conventional breed-EC (1994) A strategy for accumulating genes for partial

ing program. In: Zeiglerdisease. CAB International/IRRI, Madison, USA, ppRS, LeongS, TengP (eds) Rice blastSakamoto265

245–T, IchikaswaK, TadaY, YokozekiY, Akagitance in rice is correlated with the expression of a gene encod-N (1999) Chemical induction of disease resis-H, HayashiN, Fujimura

ing a nucleotide binding site and leucine-rich repeats. PlantShenMol Biol 40:847–855

MichelmoreKA, MeyersBC, Islam-FaridiMN, ChinDB, StellyDM,

by PCR with degenerate oligonucleotide primers map to clus-RW (1998) Resistance gene candidates identifiedters of resistance genes in lettuce. Mol Plant Microbe InteractSilué11:815–823

for-gene relationships in the D, NotteghemJL, TharreauOryza sativa-Magnaporthe griseaD (1992) Evidence of a gene-Tabienpathosystem. Phytopathology 82:577–580

SRM (2000) Mapping of four rice blast resistance genes RE, LiZ, PatersonAH, MarchettiMA, StanselJW, Pinson

from ‘Lemont’ and ‘Teqing’ and evaluation of their combina-torial effect for field resistance. Theor Appl Genet 101:1215–Tohme1225

RocaJ, MontenegroMV, CorreaF, MartinezC, ZeiglerRS,

for Colombian lineages of rice blast. In: Proc 6th AnnualW (1993) Implications of molecular markers in breedingMeeting International Program on Rice Biotechnol, ChiangVan OoijenMai, Thailand

JW (1992) Accuracy of mapping quantitative trait loci

Wangin autogamous species. Theor Appl Genet 84:803–811

WangZ, TaraminoG, YangD, LiuG, TingeySV, MiaoGH,

defense-response gene-like sequences mapped to regions con-GL (2001) Rice ESTs with disease-resistance gene- ortaining major resistance genes or QTLs. Mol Genet Genomics265:302–310

803

XueY, TangD, ZhangY, LiW (1998) Isolation of candidate R

disease resistance genes from rice. Chinese Sci Bull 43:497–500

YuYG, BussGR, MaroofMA (1996) Isolation of a superfamily

of candidate disease-resistance genes in soybean based on aconserved nucleotide-binding site. Proc Natl Acad Sci USA21:751–759

YuJ, HuS, WangJ, LiS, WangKG, LiuB, DengY, DaiL, et al

(2001) A draft sequence of the rice (Oryza sativassp. indica)genome. Chinese Sci Bull 46:1937–1942

ZeiglerRS, TohmeJ, NelsonR, LevyM, Correa-VictoriaFJ

(1994) Lineage exclusion: a proposal for linking blast popula-tion analysis to resistance breeding. In: ZeiglerRS, LeongSA,TengPS (eds) Rice blast disease. CAB International/IRRI,Wallingford,UK, pp267–292

ZhuLH, ChenY, XuYB, XuJC, CaiHW, LingZZ (1993) Con-struction of a molecular map of rice, and gene mapping usinga double-haploid population of a cross between Indica andJaponica varieties. Rice Genet Newslett 10:132–135