The Fusarium Wilt Resistance Locus Fom-2 of Melon Contains a Single Resistance Gene with Complex Features

Home , R gene

The Plant Journal (2004), 39, 283–297 doi: 10.1111/j.1365-313X.2004.02134.x The fusarium wilt resistance locus Fom-2 of melon contains a single resistance gene with complex features

Tarek Joobeur1, Joseph J. King2, Shelly J. Nolin1, Claude E. Thomas3 and Ralph A. Dean1,* 1Department of Plant Pathology, Fungal Genomics Laboratory, North Carolina State University, Raleigh, NC 27965, USA, 2Seminis Vegetable Seeds, 37437 State Highway 16, Woodland, CA 95695, USA, and 3USDA-ARS, US Vegetable Laboratory, Charleston, SC 29414, USA

Received 19 March 2004; revised 20 April 2004; accepted 10 May 2004. *For correspondence (fax þ1 919 513 0024; e-mail [email protected]).

Summary

The soil-borne fungus Fusarium oxysporum f.sp. melonis causes significant losses in the cultivated melon, a key member of the economically important family, the Cucurbitaceae. Here, we report the map-based cloning and characterization of the resistance gene Fom-2 that confers resistance to race 0 and 1 of this plant pathogen. Two recombination events, 75 kb apart, were found to bracket Fom-2 after screening approximately 1324 gametes with PCR-based markers. Sequence analysis of the Fom-2 interval revealed the presence of two candidate genes. One candidate gene showed significant similarity to previously characterized resistance genes. Sequence analysis of this gene revealed clear polymorphisms between resistant and susceptible materials and was therefore designated as Fom-2. Analysis of susceptible breeding lines (BL) presenting a haplotype very similar to the resistant cultivar MR-1 indicated that a gene conversion had occurred in Fom-2, resulting in a significant rearrangement of this gene. The second candidate gene which shared high similarity to an essential gene in Arabidopsis, presented an almost identical sequence in MR-1 and BL, further supporting Fom-2 identity. The gene conversion in Fom-2 produced a truncated R gene, revealing new insights into R gene evolution. Fom-2 was predicted to encode an NBS-LRR type R protein of the non-TIR subfamily. In contrast to most members of this class a coiled-coil structure was predicted within the LRR region rather than in the N-terminal. The Fom-2 physical region contained retroelement-like sequences and truncated genes, suggest- ing that this locus is complex.

Keywords: Fom-2, melon, complex R locus, ﬁne mapping, BAC end sequence, R gene evolution.

Introduction

The cultivated melon (Cucumis melo L.), an economically viable in the soil as chlamydospores for decades. An important member of the Cucurbitaceae family, includes a effective control for this pathogen is through host resist- diverse group of annual trailing-vine plants such as the ance (Martyn and Gordon, 1996). However, resistant vari- cantaloupe, honeydew, casaba, snake melon, pickling eties often lack the appropriate traits for the commercial melon, mango melon, and snap melon (McCreight et al., market. Traditional artificial inoculation used to evaluate 1993). In the US, over 120 000 acres of melons (cantaloupe resistance to fusarium wilt is a time consuming process and and honeydew) are grown annually, and in 1998 nearly susceptible plants may escape detection (Burger et al., 1 million tons of melons were produced with a total value 2003). Thus, molecular markers tightly linked to fusarium of over 500 million dollars (http://www.usda.gov/nass/). wilt resistance genes are highly valued in melon breeding. However, the full economic value of this crop has not been Currently, four races (0, 1, 2, and 1, 2) of the pathogen are achieved because of diseases caused primarily by fungi, defined by their capacity to incite disease in different vari- such as fusarium wilt, and powdery and downy mildews. eties of melons. Resistance to race 1 and race 2 is conferred Fusarium wilt, caused by Fusarium oxysporum f.sp. mel- by a single dominant gene Fom-2 and Fom-1, respectively. onis, is one of the most difficult diseases to control pri- Both genes also confer resistance to race 0 (Schreuder marily because the pathogen is soil-borne and remains et al., 2000; Zink and Thomas, 1990).

ª 2004 Blackwell Publishing Ltd 283 284 Tarek Joobeur et al.

Typically, race/cultivar-specific resistance is proposed to interval. A single candidate gene with significant similarity involve the recognition of the pathogen avirulence (avr) to previously characterized R genes was found and gene product by the complementary host resistance (R) designated as Fom-2. Evidence for this gene being gene protein. This recognition initiates a signal transduction Fom-2 was provided by the sequence analysis of alleles cascade and the defense response (Hammond-Kosack and from susceptible and resistant cultivars of the two iden- Jones, 1997; Martin et al., 2003). Several R-genes have been tified candidate genes. Sequence analysis of the two BAC isolated by map-based cloning and transposon-tagging clones encompassing Fom-2 showed the presence of strategies (Martin et al., 2003). Interestingly, R genes that several retroelement-like sequences as well as truncated confer resistance to different types of pathogens encode genes, typically found in complex loci. Comparison very similar proteins. A major class of R genes encodes between closely related resistant and susceptible haplo- proteins that contain leucine-rich (LRR) and nucleotide- types strongly indicated that truncated R genes could be binding site (NBS) domains and are likely located in the produced by gene conversion, revealing novel insights cytoplasm. This class can be divided into two subfamilies, into the R gene evolution. the TIR and the non-TIR, depending on the presence of a domain at the N-terminal with similarity to the Toll/interleu- Results kin-1 receptor (TIR) (Meyers et al., 1999). The former subfamily includes the resistance genes M and L6 from flax, N Fom-2 fine mapping from tobacco, and RPP5 from Arabidopsis. The non-TIR subfamily comprises RPS2 and RPM1 from Arabidopsis and In a previous work, we developed two co-dominant PCR- I2 and Prf from tomato. R proteins of the non-TIR class are based markers, FM and AM, that cosegregated with Fom-2 often predicted to have a coiled-coil (CC) structure near their (Wang et al., 2000). In addition, two AFLP markers ACT/CAT1 N-terminus and are referred to as the CC-NBS-LRR class. and AAC/CAT1 were found to flank Fom-2 at 1.7 and 3.3 cM, Other R genes, such as Xa21 in rice, encode a protein that is respectively. We used these markers to map Fom-2 using a likely located in the plasma membrane and presents an population (named Vad375) containing 159 RILs derived extracellular LRR domain and a cytoplasmic kinase domain. from the cultivar ‘Ve´ drantais’ and PI 161375. New primer The Cf proteins (Dixon et al., 1996) in tomato and the pairs were designed for FM and AM, producing smaller HS1Pro-1 (Cai et al., 1997) in sugar beet predominantly fragments that were suitable for analysis with ABI 377s consist of LRR domains. Few R proteins have been described (SSR154 for FM and STS178 for AM). SSR154 and STS178 to lack the LRR motif. One such example is the recently were found to flank Fom-2 at 2 and 4 recombination events, identified RPW8 protein in Arabidopsis that is predicted to respectively (Figure 1). The AFLP markers ACT/CAT1 and contain a CC structure (Xiao et al., 2001). AAC/CAT1 were used to create PCR-based markers for Resistance genes are continuously evolving to produce screening the Vad375 population. The AFLP fragments were new specificities in order to recognize mutated avr genes. isolated, cloned and sequenced. The resulting fragments Understanding the mechanisms governing R gene evolution were utilized to screen a HindIII BAC library of melon (Luo is therefore of major interest. Important insights were et al., 2001). The AAC/CAT fragment was considered repet- obtained from the sequence analysis of R genes loci from itive because several hundred BAC clones were identified. different species and haplotypes (Michelmore and Meyers, However, the ACT/CAT1 marker identified 23 BAC clones, 1998). New specificities are postulated to be the result of indicating this fragment was present as a single copy in the diversifying selection, interallelic recombination and gene genome. Restriction fragment analysis with the FPC pro- conversion. Resistance genes are often organized in tandem gram showed that all the clones belonged to the same repeats of paralogues and this structure is postulated to play contig. Two SSR markers SSR138 and SSR180 were derived a role in their evolution through additional mechanisms from the BAC end sequences. After screening the Vad375 such as unequal crossing-over (Chin et al., 2001; Parniske population, SSR138 was found to cosegregate with Fom-2 et al., 1997). (Figure 1). Interestingly, one recombination event was found Here, we report the map-based cloning and character- between SSR180 and Fom-2 indicating that Fom-2 is located ization of the first R gene in the Cucurbitaceae family, between SSR180 and STS178 (Figure 1). All markers and the Fom-2. Bacterial artificial chromosome (BAC) libraries resistance phenotype followed the expected segregation were screened with probes linked to Fom-2 and a physical ratio (1:1) for a co-dominant marker (P ¼ 0.93). The observed map of the interval was constructed. PCR-based markers frequency of heterozygous RILs was not significantly derived from the BAC end sequences were used to (P ¼ 0.29) different from the expected value, 4.20%. localize Fom-2 into two BAC clones. Fom-2 was further Screening additional populations (SF6 and IndF7; Table 1) delimited to 75 kb size-interval using additional markers with the Fom-2 flanking markers SSR154 and STS178 developed from the sequence of these BAC clones. revealed 15 recombinant plants (Figure 2a). Analysis of the Two candidate genes were found in the Fom-2 physical recombinants with SSR180 and SSR138 confirmed that

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 Fom-2 is a complex locus 285

Table 1 List of primer pair sequences

TG/CAA2 Expected size Namea Primer pair sequences in MR-1 (bp) SSR178 5 SSR138 GACACGACCTGATCCATGTG 185 CCACATGTTGAATGATGAGGA SSR154 CCCTTCTGTCATTTGGCTTG 299 0.7 CGTCAATTATTAAACATTCTGATGC ACT/CAT1 STS178 TTCGTTCATTACTGCCGTAGG 233 TCTGTGTTCCCTACCCCAAC Fom-2 2 AM SSR180 GGATTTGTTTGCGTCATTTTG 216 SSR138 Fom-2 GGGGCATTTTTGGTATTTTC 0.2 FM SSR180 SSR181 AAATCGAAGCCCAGTGAAAG 275 TCTGGCTGGGAATATGATTG 0.4 4 SSR184 AGCTTATGTCAACGAGGTTGG 298 SSR154 CCTCCAACAAAAGATGACACTG AAC/CA STS259 CATTGATCCGAAACAATCCAG 170 Fom-2 TTCGCCTCCCTTCAAGTATG Fine mapping SSR281 AGTTTGAAGGTGCTGCTTGAC 250 Genetic map for Fom-2 region AGACAAGCCCACAACGAATC Wang et al. (2000) STS296 CCACAAAAAGGAGCTTGACC 380 GCCAATTGCCCAAATCAG Figure 1. Two markers STS178 and SSR154 ﬂank Fom-2. STS303 CAAATTTTGGGGCGTTACAC 350 SSRs markers segregating in the Vad375 population were derived from ACTGGTCATGCTGGTGATTC previously mapped markers (Wang et al., 2000). Genetic distance is expressed in cM, using the Kosombi mapping function. Map distances were based on STS308 TGCAGCTATTCCATGGTCAG 204 SF6 and InF7 population data (see Table 4). Different scales are used in the TAAAATAGGGCCCGAAACTG two maps. STS312 GGAGGATTTGGGAAGTGAG 348 TGTCCATACCTCCTCCAAGC STS411 TTTCTAAAATTTACCATCATTGGAG 346 AATGGCAAATTCAACCTTCAC SSR430 CCATCATGATTTGGAATGAATTAG 341 Fom-2 was located between SSR180 and STS178. No CGTTGCAATTTGATCTTTTTAATG recombination events were found between SSR138 and SSR451 CGTGAGAAAGTACAATGATTGGTG 369 Fom-2 (Figure 2b). When the two populations derived from GCCAAGCTAAGCAATTAGGC MR-1 and AY were analyzed with SSR154 and STS178, no P458 TGAAAACTAAAAAGATGGCATGG 1905 recombination events were detected. TGCAATGGCAATTTCAAGG

aThe marker name includes SSR when a putative SSR was found in the Fom-2 physical interval amplified sequence. P458 primer pair was used to amplify the LRR region of Fom-2. In order to identify the Fom-2 physical interval, chromosome walking toward this gene was initiated from the SSR154, STS178 and SSR138 markers. First, these markers were used Sequence of the two BAC clones encompassing Fom-2 to screen the HindIII BAC library. Three distinct contigs were identified. Second, probes derived from BAC end sequences The BAC clones D09 and ACT11, encompassing the resist- of extending clones of the three contigs were used to design ance gene Fom-2, were subjected to shotgun sequencing. primers and the PCR products (obtained from the corres- Six contigs were assembled from a total of 6569 reads with ponding BAC clone) used to screen the BAC libraries. After sixfold average depth of coverage (ranging from 5.8 to 8.2 two walking steps from each contig, fingerprinting and depending on the contig). The resulting total sequence was analysis with FPC demonstrated that all the clones belonged 217 020 bp and the largest contig being 136 129 bp. The to a single contig. The BAC end sequences of the identified final error rate was less than one base per 10 kb and all the clones were sequenced and used to develop SSR/STS finished sequences had a PHRED value of 25 or more. markers. Two markers STS411 and SSR184 were found to Sequence analysis confirmed that the two BAC clones flank Fom-2 at two and one recombination events, respect- overlapped by approximately 32 kb. All marker sequences ively (Table 2, Figure 2b). Hybridization analysis indicated that were expected to be located on these BACs were iden- that STS411 and SSR184 were located on the BAC clones tified. Additional markers were developed to delimit the D09 and ACT11, respectively. Based on hybridization and position of the closest recombination events to Fom-2; the FPC analysis these two BAC clones were found to overlap recombination event located between Fom-2 and SSR184 and therefore encompass the resistance gene Fom-2 was found within a 1.28 kb interval between markers (Figure 2b). STS296 and SSR451; the two recombination events found

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 286 Tarek Joobeur et al.

Fom-2, the position corresponding to amino acid 46. In addition, no (a) SSR178 SSR138 SSR180 SSR154 XXX XXX start methionine could be found. Taken together these observations strongly indicated that Gene 8 is a pseudo- ACT2 F10 gene. (b) C07 ACT11 X X X D09 XXX X X X X The deduced peptide of Gene 9 presented a high similarity XXX X X X X XX X X SSR180 SSR154 STS178 SSR184 STS312 STS411 SSR181 (E ¼ 0 Identities ¼ 59%, Positives 70%) to AtCPSF73-II of SSR138 Arabidopsis. Close inspection revealed that the exon/intron (c) XX X X X boundaries were highly conserved. Two additional proteins STS411 SSR138 SSR451 SSR184 in Arabidopsis showed significant similarity with Gene 9. SSR430 STS296 SSR281 However alignment scores were much lower (E ¼ 5e ) 75 20 kb and 2e ) 35), indicating that AtCPSF73-II and Gene 9 are Figure 2. Fom-2 map-based cloning. possible orthologs and share the same function. AtCPSF73-II (a) Fom-2 genetic map. X indicates recombination events identified in the was recently found to be an essential gene and is mainly Vad375 population. (b) Fine mapping Fom-2. All markers except STS178 and SSR154 were derived expressed in flowering tissue in Arabidopsis (Xu et al., 2004). from BAC end sequences. SSR312, derived from one end of BAC clone D09, Gene 7 was the only sequence in the two BAC clones with was mapped at 11 recombination events from Fom-2 between STS178 and significant similarity to previously characterized resistance SSR138. It was not possible to develop an STS marker (absence of polymorphism) for the opposite end. However, when used as a probe genes and was thus designated as Fom-2. Further evidence (SSR181), it hybridized to ACT11 that had one end (SSR138) cosegregating supporting the identity of Fom-2 is provided in the following with Fom-2. This end hybridized with D09 confirming the overlap between sections. ACT11 and D09. SSR138 also hybridized to BAC ACT2 that presented one end (SSR180) mapped at three recombination events from Fom-2 between The Fom-2 interval also contained two retroelement-like SSR138 and SSR154. SSR180 hybridized to ACT11 indicating it was contained sequences and three sequences with similarity to other within ACT11. These results indicated that Fom-2 was located between transposable elements (DNA transposons) (see Figure 3). In STS312 and SSR180. These two markers were located on D09 and ACT11, respectively. Thus these two BAC encompassed Fom-2. Additional markers the two BAC clone sequences outside of the Fom-2 interval, (SSR184, STS411) were derived from BAC ends that were expected to be five additional retroelement-like sequences and three DNA between SSR154 and STS178. Fom-2 was located between STS411 and transposon-like sequences were identified. BAC clone D09 SSR184 at 2 and 1 recombination events, respectively. Note: BAC clones are not drawn to scale. also harbored three apparently complete putative genes (c) Analysis of the approximately 75 kb size interval containing Fom-2.All (genes 3–5). Gene 3 showed similarity to the cytochrome sequences corresponding to markers located on these two BAC were p450 protein coding genes. Genes 4 and 5 presented identified, confirming the order of the markers. Only the physical interval STS411-SSR180 containing Fom-2 is represented. Additional markers (STS, similarity with unknown proteins (NM_100901; AY046010) SSR) were developed from the STS411–SSR184 interval, and the Fom-2 in Arabidopsis. interval was reduced to STS296–STS411.

Fom-2 Characterization between STS411 and Fom-2 were conﬁned to 5.5 kb be- Fom-2 presented high similarity to previously characterized tween the markers STS411 and SSR430. Thus Fom-2 was R genes. When searched against protein databases using assigned to 75 kb-size interval between the markers STS411 BLASTX, I2 (conferring resistance to F. oxysporum f.sp. and STS296, located on the largest contig (Figure 2c). lycopersici in tomato) was the closest characterized R gene to Fom-2 (E ¼ 2e ) 88; 29% identities and 49% positives). The Fom-2 sequence was predicted to encode an uninter- Candidate genes rupted open reading frame for a polypeptide of 1073 amino Sequence analysis of the two BAC clones ACT11 and D09, acids. A comparison of the Fom-2 predicted protein (FOM-2) revealed the presence of 10 putative genes (genes 1–10, against the InterPro database revealed features characteris- Figure 3). Three putative genes were present in Fom-2 tic of the NBS-LRR class of R proteins: NB-ARC domain interval between STS411 and STS296 (genes 7–9; Figure 3). (PF00931) (Van der Biezen and Jones, 1998) and C terminal However, upon careful inspection only two genes (gene 7 LRRs (PF00560). While four LRRs were found by Pfam and 9) appeared to be complete. analysis, an additional 17 possible LRRs were observed by Gene 8 showed some similarity (E value ¼ 6e ) 10; manual alignment (Figure 4). Nine LRRs conformed with the BLASTX) with a hypothetical transmembrane protein consensus motif LxxLxxLxxLxLxx(N/C/T)x(x)L that is (AL033503.1) in Candida albicans but the region presenting observed in cytoplasmic R gene products (Jones and Jones, similarity was not predicted to be part of an exon. Analysis 1997). No apparent N-terminal signal sequence was found, using FASTX (Pearson, 1990) revealed that the similarity was indicating that the encoded protein is likely located in the limited to the interval between amino acid 1 and 160, cytoplasm. As no TIR domain was found, Fom-2 was however the predicted polypeptide contained seven stop considered to belong to the non-TIR subfamily of R genes. codons in this region. The ﬁrst stop codon was observed at Analysis with the program COILS showed the presence of a

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 Fom-2 is a complex locus 287

Table 2. Recombinants events identiﬁed between STS178 and SSR154

Genotypec

Populationa Recombinantb STS178 STS312 Fom-2 SSR138 SSR184 SSR180 SSR154

Vad375 (159) V375-117 A A A AAAB V375-129 B B A AAAA V375-136 A A B BBBB V375-183 A A B BBBB V375-185 B B A AAAA V375-238 B B B B H H H SF6 (157) SF6-5905 H H B BBBB SF6-5945 H H B BBBB SF6-5931 H H H H H B B SF6-6020 H H H HHHA SF6-5983 H H NS HHHB SF6-6000 H H A AAAA IndF7 (346) 5891-4 H H H HHHB 5916-1 H H B BBBB 5923-1 A A H HHHH 5924-3 B H NS HHHH 5926-3 A A A AAAH 5943-2 B B H HHHH 5977-1 H H B BBBB 5988-1 A A A AAAH 6006-8 H H H H H B B aPopulation used to fine map the resistance gene Fom-2. In brackets is indicated the population size. Vad375 is an RIL population. SF6 is a population derived by self-pollination of individual plants of two V375 RILs heterezygous for Fom-2 and flanking markers SSR154 and STS178. IndF7 population comprises individual plants of families segregating for the flanking markers and derived by self-pollination of SF6 plants. bName of plant/RIL having a recombination event between SSR154 and STS178. cGenotype of the recombinants for Fom-2 and markers used for fine mapping; A ¼ allele from the susceptible parent ‘Vedrantais‘ and B ¼ allele from the resistant parent (PI 161375) and H ¼ both alleles are present. The genotype for Fom-2 was determined by screening 20 plants from each family/line. NS ¼ no seed were obtained.

Figure 3. Organization of putative genes in the Fom-2 locus: black rectangles represent the two BAC clones encompassing Fom-2. Contiguous sequences (contigs) are represented by white rectangles. Gaps between these contigs for which paired reads are found were colored in black. Putative genes (labeled 1–10) are represented by rectangles with different shading pattern depending on the description of the closest match identiﬁed by BLASTX search (E-value > e)10). Asterisk indicates truncated genes. Forward slash indicates continuity of D09 sequence without any physical gap.

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 288 Tarek Joobeur et al.

following plant material; three susceptible cultivars Ve´ drantais, AY (parent of our mapping population) and Durango and the resistant PI 161375 (one parent of the Vad375 population). We also included two breeding lines (BL) randomly selected from a group of 45 that were considered to be relevant for the following reason. All of these lines were susceptible to fusarium wilt race 1 and 0 but analysis with markers within the Fom-2 interval (STS411– STS296) revealed that they presented the same genotype as the resistant cultivar MR-1 (Table 3). Hence the BL and MR-1 should have the same origin for Fom-2 interval and any difference in gene content and characteristics should be highly correlated to susceptibility/resistance to fusarium wilt. We hypothesized that these BL might have been derived from an MR-1-like haplotype in which an alteration had occurred in the resistance gene Fom-2. Interestingly, when the LRR containing fragment was amplified from these BLs the same product size was obtained in all of them, but was approximately 500 bp larger, compared with the product obtained for the two resistant materials MR-1 and PI 161375. The sequences of the LRR fragment from the resistant material (MR-1 and PI 161375) were identical, except at three nucleotides. These differences resulted in the substitution of two residues V and K in MR-1 with M and E in PI 161375, respectively (Figure 5a). The amino acid sequences from the susceptible cultivars (Ve´ drantais, AY and Durango) were identical. However, when compared with the amino acid sequence deduced from resistant genotypes several differences were observed (Figure 5a); 25 amino acids out of 541 were different. When the nucleotide sequences of resistant Figure 4. FOM-2 protein reveals typical NBS-LRR features: underlined and and susceptible alleles were compared, the ratio between boldface sequences correspond to conserved motifs in the NB-ARC domain; non-synonymous (Ka) and synonymous nucleotide substi- in this order, p loop, Kinase 2, Kinase 3a, HD motif and motif 5. The ordinal number of each LRR is indicated on the left. LRR elements identified by tution rates was greater than 1 (Ka/Ks ¼ 4.1) indicating this searching the Pfam database are shaded. Underlined sequences in the LRR region is subject to positive selection (see Discussion). region correspond to residues forming the b-strand/b-turn. Residues that fit Analysis of the b-strand/b-turn region separately showed the cytoplasmic LRR consensus LxxLxxLxxLxLxx(N/C/T)x(x)LxxIP (where x is any residue) are in boldface. Dashes were introduced for better alignment that this region has an even higher ratio (Ka/Ks ¼ 11) than with the consensus and do not represent gaps. Asterisk indicates LRRs that the non-b-strand/b-turn region. However, the latter region clearly conform to the cytoplasmic consensus. Sequence predicted to form a also presented evidence of positive selection, Ka/Ks > 1 CC structure is in italic. (Figure 5b). When the nucleotide sequence of the LRR fragment from the cultivar MR-1 was compared with the susceptible BLs, the two sequences were essentially identical for the first possible CC motif between residues 691 and 718 within the 1077 bp and the last 745 bp. However, the middle part of LRR region (P ¼ 0.87). However, no evidence was found for the BL fragment was clearly different from the MR-1 the CC structure before the P-loop motif as is often observed sequence (Figure 6a). These observations are important in the non-TIR subfamily. for two reasons. First, the identity of sequence stretches between MR-1 and BL provide additional evidence that these plant materials are from very close haplotypes. Fom-2 alleles in resistant and susceptible materials Second, this patchwork structure in the BL (Figure 6a) To provide additional evidence for Fom-2 identity, we indicates that an important rearrangement has occurred analyzed its LRR domain sequences from alleles of resist- within Fom-2 that may have altered its specificity/function, ant and susceptible plant material. The LRR region of which is in agreement with our hypothesis. The deduced Fom-2 was amplified and the PCR product cloned from the amino acid sequence for the BL revealed a clear difference

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 Fom-2 is a complex locus 289

Table 3 Analysis of different plant material with markers in the Fom-2 region

STS178 SSR312 SSR411 SSR138 SSR303 SSR308 SSR259 SSR296 SSR184 SSR180

MR-1 251 367 364 204 368 222 187 397 316 233 PI 161375 274 367 364 204 368 285 189 483 307 196 AY 377 322 366 207 – – – 496 316 233 V 377 322 366 207 – – – 496 316 233 Durango 377 322 366 207 – – – 496 316 233 BLa 274 367 364 204 368 222 187 399 316 233

The band size (bp) of the different markers in the Fom-2 region resolved on ABI 377 sequencer. The fragment size is 17 bp larger than the actual fragment size (see Experimental procedures for details). The marker order in the table is the same as in the physical map. aThe genotype was the same for all 45 breeding lines. SSR303, SSR308 and SSR259 were dominant markers. Dashes indicate absence of band. Fom-2 is located between SSR411 and SSR296 (see Figure 3). MR-1 and PI 161375 are resistant, AY, Vedrantais, Durango and BL are susceptible to fusarium wilt. from the MR-1 sequence; several of these differences were SSR154 at 1 and 4 recombination events, respectively. Two found in the b-strand/b-turn region, altering the terminal 11 observations strongly indicate that we mapped Fom-2 to a LRRs. similar position as previously published. STS178 and It is noteworthy that we cannot reject the possibility that SSR154 were derived from the DNA sequence of the AM and the fragment amplified in the BL is a paralogue of Fom-2. FM markers, previously identified by Wang et al. (2000). AM However, this appears to be unlikely because no clear and FM were found to cosegregate with Fom-2, using a additional fragments were amplified in the BL. When a different population derived from a cross between the different primer pair was used, one single band was resistant cultivars MR-1 and the susceptible AY. Recently, observed in the BLs. Nevertheless, this unlikely alternative one RFLP marker (NBS3) was mapped at 0.7 cM from Fom-2 does not affect the significance of the result observed in the (Brotman et al., 2002). The predicted amino acid sequence of BLs about Fom-2 identity. this probe (publicly available) was identical to that obtained Analysis of the coding sequence for Gene 9 (the other from the BAC end sequence of clone C07 (Figure 2b) that candidate for Fom-2) in the susceptible BL revealed it was was identified using STS178. Our results also showed that very similar to the corresponding sequence from the resist- this marker was 0.7 cM from Fom-2 (Figure 1). ant cultivar MR-1 (only eight out of 2182 nucleotides were different). The predicted amino acid sequence of the Gene 9 Fom-2 fine map from the BL differed by a single residue from the corresponding sequence in MR-1. Taken together, the results of the In order to identify the resistance gene Fom-2 we followed a sequence analysis strongly indicate that Gene 7 is the only positional cloning approach. Markers tightly linked to and candidate for Fom-2. flanking Fom-2 were used to identify BAC clones covering To confirm that Fom-2 cannot be rejected based on the target gene physical interval. The BAC end sequences of available recombination data, the sequence of the LRR these clones were used to develop PCR-based markers and fragment was also determined from the two plants each of to analyze plants presenting recombination events between which presented a recombination event between SSR430 the flanking markers. The positions of the new markers were and SSR411 (this interval includes Fom-2). Using the single determined to narrow down the target gene physical inter- nucleotide polymorphism (SNP) analysis one recombina- val. The success of this method often depends on the level of tion (5923-1; Table 2) was found on the left of Fom-2 recombination frequency in the target region (Durrett et al., (Figure 2) and the second recombination event (5941; 2002). In the Fom-2 region the recombination frequency was Table 2) was confined to an interval that includes 315 bp sufficient for reducing the target interval size to 75 kb using of the C-terminus of Fom-2. The remaining part of the gene 1324 gametes (equivalent to two times the size of the three was included in Fom-2 interval, indicating that Fom-2 populations used for Fom-2 fine mapping). Based on the cannot be eliminated from the candidacy based on recom- 227 kb cM)1 average physical/genetic distance ratio bination. observed in the Fom-2 region (Table 4), 1540 gametes are necessary to find such a target interval size with a probability of 0.95 (Durrett et al., 2002). This is very close to the total of Discussion gametes used in our experiments. Similar to other resistance gene loci, variation in recombination frequency Fom-2 genetic map (Table 4) was observed in the Fom-2 region (approximately Using 159 RILs derived from a cross between ‘Ve´ drantais’ 150 kp cM)1 upstream; STS178–SSR430 and 320 kb cM)1 and the PI 161375, Fom-2 was mapped between STS178 and downstream; SSR430–SSR154). It has been proposed that

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 290 Tarek Joobeur et al.

the lack of recombination could be related to the high divergence between haplotypes. In the two BAC clones containing Fom-2, a low recombination ratio was observed in two relatively large intervals (Table 4), SSR430–SSR451 (70 kb with no recombination) and SSR281–SSR280 (approximately 63 kb with 630 kb cM)1). Interestingly, both regions contained several retroelement-like sequences; three in the ﬁrst region and two in the second. Retroele- ments are often found to be highly polymorphic between resistance locus haplotypes (Noel et al., 1999). Intervals that appear to have a high recombination ratio (hot spot) were also observed in the Fom-2 region, three recombination events were found in the 11 kb interval SSR451–SSR281 and two were found in the 6 kb interval STS411–SSR430 containing Fom-2. Such hot spots for recombination were postulated to be the cause of sequence reshufﬂing between Xa21 homologues/paralogues (Song et al., 1997). Recom- bination frequency is also found to depend on the haplotype combination as observed in the complex RpD-1 locus (Collins et al., 1999). The observation that 45 BLs having at least two different origins, presented the same genotype between the SSR154 and STS178 markers, strongly indicates that their haplotypes have not recombined. The BLs haplotype is very close to MR-1 and this may also explain the absence of recombination in the two analyzed populations derived from MR-1 and AY. In our analysis we screened 120 plants of backcross population and 55 RIL derived from the cross between MR-1 and AY and no recombination event was found between the SSR154 and STS178 markers. Based on the recombination frequency in the Fom-2 region we would expect approximately three recombinants in the backcross population and one to two recombinants in the RIL population in the same interval.

Predicted CC structure within the Fom-2 LRR region

Using fine mapping, the resistance gene Fom-2 was confined to approximately 75 kb interval, containing two candidate genes. Only one candidate gene presented significant similarity to previously characterized R genes. This gene was found to be clearly polymorphic between resistant and susceptible material and was therefore designated as Fom-2. Figure 5. LRR region of Fom-2 in resistant and susceptible materials. The LRR domain from different plant materials was amplified, cloned and Further evidence supporting Fom-2 identity was provided by sequenced. The resulting sequences were compared. the analysis of susceptible BLs that presented a very close (a) Distribution of sequence variation: all three susceptible cultivars present haplotype to the resistant cultivar MR-1; a gene conversion the same protein sequence. The residues that are different in susceptible material (AY, Ve´ drantais and Durango) from MR-1 sequence are indicated altering a large part of the LRR region of Fom-2 was found in under its corresponding residue in MR-1. The two residues that are different in the BLs. In contrast, the sequences of the only other candi- the PI from MR-1 are shaded. Underlined sequences in the LRR region date gene (Gene 9) from the BLs and MR-1 were very similar. correspond to residues forming the b-strand/b-turn. (b) Diversifying selection. Synonymous and non-synonymous substitution The Fom-2 predicted product (FOM-2) contained the rates for the nucleotide codon forming the b-strand/b-turn region (xx(L)x(L)xx) different features of the non-TIR class of NBS-LRR and the remaining LRR region (non-framed) were calculated between MR1 R-proteins. Unlike with most members of this class, no and AY nucleotide sequences. As proposed by Parniske et al. (1997) the codons for the L residue were omitted from the b-strand/b-turn analysis as evidence of CC structure was found in the N-terminal of the they are subject to sequence conservation. Fom-2 protein. However, a CC structure was detected within the LRR repeats (Figure 4). This result is similar to that

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 Fom-2 is a complex locus 291

Figure 6. Gene conversion in susceptible breeding lines: (a) Rearrangement between BL and MR-1 in the Fom-2 LRR region: stretches of nearly identical sequences are indicated with the same shading pattern (95% identities for vertical lines and 96% identities for angled lines). The gray shaded areas show high similarity (85% identities) between the two haplotypes. The borders of these stretches are numbered according to the nucleotide position in the ampliﬁed LRR fragment. The sequence has the same orientation as the Fom-2 ORF. The arrowheads indicate possible gene conversion breakpoints. Asterisk indicates the position of the stop codons. Broken lines indicate the duplicated part of the LRR region in the BL. (b) Model for gene conversion in Fom-2: HDH is a hypothetical donor haplotype. A double-strand break (DSB) within the gray shaded area in MR-1 could have initiated the gene conversion. This gray area is similar to the one in HDH. DSB repair occurs following a model proposed by Richard- son and Jasin (2000) via non-homologous end joining and results in partial tandem gene duplication. (1) Invasion of one end into the similar region on the homologue chromatid. (2) Repair synthesis. (3–4) Repair event completion. The shading patterns used to depict the MR-1 haplotype are conserved only for comparative purposes with panel a. (c) Sequence variation between the predicted amino acid sequences in MR-1 and BL for the LRR region. Only amino acids that differ from MR-1 are indicated for BL. Dashes indicate indels. The number on the left indicates the ordinal number of each LRR in MR-1.

observed in the R protein HERO (Ernst et al., 2002). From the form an N terminal CC structure. Similar to Fom-2, this gene analysis of 31 full-length NBS-LRR R gene homologues Pan contains a predicted CC structure within the LRR region. et al. (2000) identified a single gene (AF017752; a resistance Recently, five NBS-LRR homologues (non-TIR) lacking an gene homologue in Lactuca sativa) that is not predicted to N-terminal CC structure have been identified in Arabidopsis

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 292 Tarek Joobeur et al.

Table 4 Distribution of the recombination rate across the Fom-2 of the LRR is subjected to selection for amino acid diversity region that is demonstrated to be associated with regions involved Physical Genetic Number in ligand binding (Hughes and Nei, 1988; Parniske et al., Intervala distance (kbp) distanceb of recombinants 1997). Analysis of the LRR region in Fom-2 revealed strong evidence for diversifying selection acting in this region. As 312–411 100 0.5 9 observed in the RPP13 resistance gene, diversifying selec- 411–430 6.5 0.2 2 tion was not limited to the solvent exposed residues of the b- 430–249 19 0 0 249–451 51 0 0 strand/b-turn (Figure 5b) (Bittner-Eddy et al., 2000). These 451–296 1 0 1 observations suggest Fom-2 is an active gene and the LRR 296–184 2.6 0 0 region is involved in ligand binding. 184–281 7.6 0.2 2 It has been proposed that intragenic recombination is an 281–180 12.2 0 0 important player in resistance gene evolution by combining 180–154 96 0.4 6 parts of different alleles to create new specificities (Van der Total 295.3 1.3 20 Hoorn et al., 2001). Unequal crossing over between different aNumbers refer to marker name. LRR units was suggested to produce duplications and indels bThe genetic distance was determined based on 503 F2 population by and thus may alter or create a new specificity (Anderson joining the population SF6 (157 plants) and IndF7 (346 plants). Fom-2 et al., 1997; Noel et al., 1999). It was thus proposed that LRR was located between SSR411 and SSR430. The approximate physical repeats are more labile compared with other regions distance is indicated when sequence gaps are present. These gaps, (Mondragon-Palomino et al., 2002). In our analysis, two except in the 312–411 interval, were estimated to be less than 2 kbp because of paired reads identified between the adjacent contigs. A recombination events did occur in the 3¢ end of Fom-2 312–411 interval size was estimated based on the overall BAC clone indicating that this region may be a recombination hot spot. size (140 kb). Marker 154 was present in a clone overlapping with Furthermore, we identified a group of BLs that appear to ACT11 and the size of the interval 180–154 was estimated from its have undergone a rearrangement in the LRR region partial sequencing (result not shown). (Figure 6b). The identification of this rearrangement and the relatively high frequency of non-synonymous variation (Meyers et al., 2003). This class of resistance genes homo- in the b sheet of Fom-2 LRRs compared with susceptible logues appears to be more abundant in rice. An analysis of alleles, indicates this gene is subject to mechanisms often 100 random NBS-LRR homologues revealed that only 47% linked to resistant gene evolution. are predicted to contain a CC structure (Bai et al., 2002). The difference in structure observed in Fom-2 and Hero from the Fom-2 is a single resistance gene homologue in a complex majority of other members of the non-TIR class may be locus related to different disease response mechanisms. For instance, RPM1 and RPS2 appear to have a particular form Resistance genes are usually found in a cluster of tandem of CC structure, a leucine zipper. These genes are NDR1 array of paralogues (Hulbert et al., 2001). It is thought that dependent in contrast to most NBS-LRR Arabidopsis R this structure has a major implication in the evolution of R genes that are EDS1 dependent (Aarts et al., 1998). RPM1 genes; as unequal crossing over may cause a rapid and RPS2 present an additional common feature; they both expansion/contraction of the cluster and recombination guard a key virulence target RIN4 (Mackey et al., 2002; between different members could give rise to a new Marathe and Dinesh-Kumar, 2003). specificity (Michelmore and Meyers, 1998; Parniske et al., 1997). These clusters mainly contain resistance gene homologues and retroelements (Michelmore, 2000). The N The Fom-2 LRR region is subject to positive selection locus in flax comprises three tandem copies within 30 kb R proteins are postulated to interact specifically with separated by protein kinases. The RPP5 cluster in Arabid- pathogen elicitors and trigger the hypersensitive response opsis comprises eight to nine homologues and two to (HR) (Hammond-Kosack and Jones, 1997). Several obser- three (depending on the haplotype) retroelements in a vations strongly suggest that the LRR domain of the NBS- 90-kb interval (Noel et al., 1999). The Cf-4/9 cluster contains LRR class of resistance proteins mediates direct or indirect five R-gene homologues with a single retrotransposon interaction with the pathogen molecules. The LRR structure (Parniske and Jones, 1999). By contrast, the Fom-2 locus has been shown to participate in protein–protein interaction contains a single copy of a resistance gene homologue (Kobe and Deisenhofer, 1995). Natural or induced variations and six retroelement-like sequences. However, our results in the LRR region alter gene function or produce new spe- are similar to those obtained in maize for the complex cificity (Dodds et al., 2001; Warren et al., 1998). Comparison locus rp1 (a susceptible haplotype for the Rp1-D locus). of non-synonymous and synonymous nucleotide substitu- Two rp1 homologues were separated by four retroele- tion rates in R genes revealed that the b-strand/b-turn motif ment-like sequences (two contiguous and two separated

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 Fom-2 is a complex locus 293 by truncated genes) (Ramakrishna et al., 2002). The lack of over gene conversion was also identified in the terminal LRR Fom-2 paralogues in the MR-1 haplotype could be the region of the Dm3 R gene (Chin et al., 2001). The dissociation result of the absence of any duplication that had included between crossing-over and gene conversion after DSB was Fom-2 or that the duplicate had ‘died’. The latter was suggested as a crucial mechanism to reduce unequal proposed as an additional mechanism that could shape the crossing-over events and their deleterious effects (Inbar structure of resistance genes (Meyers et al., 1998). In the et al., 2000; Johnson and Jasin, 2001). rust-resistance locus RPS5, a deletion was proposed for A model similar to the birth-and-death process described the lack of a second copy (RFL1; 1 kb apart from RPS5)in for the histocompatibility complex in mammalian cell some haplotypes. This deletion was followed with an (Gu and Nei, 1999) was proposed for R genes in plants insertion of an Ac-like transposable element (Henk et al., (Michelmore and Meyers, 1998). This model postulates that 1999). Strikingly, in the MR-1 haplotype a transposable interallelic recombination, gene conversion and diversifying element-like sequence is adjacent to Fom-2 (Figure 3) and selection are the main forces in R gene evolution. Based on could be associated with a deletion of a Fom-2 paralogue, the observation that the R gene alleles are more similar than as was suggested for the RPS5 locus. paralogues, the unequal crossing-over events that tend to Truncated genes are often observed in resistance gene homogenize paralogues (concerted evolution) would be complexes. In the RPP5 cluster, eight sequences with limited. The Fom-2 locus may constitute an example where similarity to serine/threonine protein kinase were found unequal crossing over occurs rarely. dispersed between the resistance gene homologues (Noel et al., 1999). The Cf-4/9 cluster contains dispersed fragments of Lox genes (Parniske and Jones, 1999). In the rp1 locus, Experimental procedures and contrasting with these two loci, clusters of different truncated genes were found. Similar results were observed Plant material in the Fom-2 region; three truncated genes (gene 6, 8 and 10) Three populations, segregating for the resistance to F. oxysporum were observed in the Fom-2 region. Two distinct forces may f.sp. melonis race 1, were used to map Fom-2. Two were derived produce these different loci structures observed in Fom-2 from the cross between MR-1 and Ananas Yokneum (AY) and and rp1 versus RPP5 and Cf-4/9. In the latter, it appears that were represented by a population of 45 RILs (named MR1-RIL) and a backcross population of 125 plants (named MR1-BC). MR-1 is unequal crossing over within the Lox/protein kinase genes resistant to downy and powdery mildew, F. oxysporum f.sp. produced several truncated forms of these genes and melonis race 0, 1 and 2 (Thomas et al., 1992). AY is susceptible to simultaneously the duplication of the R genes (Parniske downy and powdery mildew and all races of fusarium wilt. The and Jones, 1999). In the former group, the truncated genes third population (named Vad375) contains 159 F6/F7 RILs and was appear to have been acquired as a filler sequence associated derived from the cultivar ‘Ve´drantais’ and the PI 161375 by single- seed-descent from individual plants of the F2 progeny after six/ with gene conversions (Ramakrishna et al., 2002). Gene seven selfings with no conscious selection in any generation. conversion and crossing over result after the repairing of the ‘Ve´drantais’ is susceptible to F. oxysporum f.sp. melonis race 0 Double-Strand Breaks (DSB) in the DNA (Petronczki et al., and 1. PI 161375 is a Korean line resistant to F. oxysporum f.sp. 2003). In recent studies the occurrence and the size of the melonis race 0 and 1. filler sequence acquired after DSB was found to be genome To fine map Fom-2, two additional populations (SF6 and Indf7) segregating for Fom-2 were screened with Fom-2 flanking mark- dependent (comparing Arabidopsis and tobacco; Kirik et al., ers. SF6 population contained 157 plants from two F7 RILs (from 2000) and could have also been locus dependent. This Vad375 population) segregating for Fom-2. To determine the observation may explain in part the difference between the genotype of this population, plants were self-pollinated and bulks two groups of loci (Fom-2, rp1 versus RPP5 and Cf-4/9). In of leaves from 12 plants of each family were used for DNA yeast, the association between crossing over and gene extraction and marker development. A total of 346 individual plants randomly selected from 73 families derived from the SF6 conversion is locus-specific (reviewed in Allers and Lichten, population and found to be still segregating for the flanking 2001). We thus propose that in the Fom-2 locus these gene markers composed the second population (IndF7). For this conversions were not associated with unequal crossing over population the genotype was determined using DNA extracted and therefore did not produce duplication of Fom-2. Fur- from each individual plant and the identified recombinants were thermore, in our analysis, an alteration of the Fom-2 allele in self-pollinated and their progeny were used to determine their resistance to fusarium wilt. the susceptible BL can be explained by gene conversion after DSB repair via non-homologous end-joining (NHEJ; Figure 6b) (Gorbunova and Levy, 1999; Richardson and Fusarium wilt resistance test Jasin, 2000). This gene conversion also gave rise to a truncated gene (Figure 6b). Strikingly, the gene conversion Resistance to F. oxysporum f.sp. melonis race 1 was identified as described by Wang et al. (2002). Twenty individuals derived by self- in Fom-2 appears to be uncoupled from a crossing-over pollination of each member of the SF6 and IndF7 population were event, as all analyzed markers in this region appear to be inoculated to determine the genotype of these populations. The identical to the MR-1 haplotype. Recently, a non-crossing same protocol was used for Vad375 RILs.

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 294 Tarek Joobeur et al.

DNA extraction from melon Tris–HCl (pH 8.4), 50 mM KCl], 1.5 mM of MgCl2, 0.1 mM of each dNTP, 5 lM of each primer, 50 ng of genomic DNA and 0.625 DNA was isolated following the CTAB method as described by units of Taq DNA polymerase (Invitrogen, Carlsba, CA, USA). The Wang et al. (1997). Leaves from at least 12 plants from each RIL or following program was used for both the SSR and STS markers; families (SF6 or IndF7) were used for DNA extraction. one cycle for 2 min at 94; 30 cycles for 15 s at 94, 15 s at 52, 2 min at 72, with a final extension of 30 min at 72. All PCR amplifications were performed in a GeneAmp PCR System 9700 Cloning AFLP fragments thermocycler (PE Biosystems). Gel images were analyzed using TM AFLP analysis was elaborated as indicated by Wang et al. (1997). GenScan (PE Biosystems) analysis software version 2.0.2. Hybridization signals on autoradiograms were matched with the corresponding areas in sequencing gels to identify targeted-AFLP fragments. The fragments were then cut out from the dried gel, Genetic mapping soaked overnight at 4Cin40llofH2O and amplified using the The MAPMAKER (Lander et al., 1987) program was used for linkage Mse1 and EcoRI primers. The resulting PCR products were isolated analysis and identification of markers flanking the resistance gene. from 2% agarose gels and cloned into the pGEM-T Easy vector The Kosambi function was used to convert recombination (Promega, Madison, WI, USA). Co-migration of the cloned frag- frequency into genetic distance (Kosambi, 1944). Plants with cros- ments and the original AFLP fragment on sequencing gels was used sing over between flanking markers were used to develop a high- to confirm the identity of the cloned fragment. Two confirmed col- resolution map for Fom-2. onies from each cloning experiment were used for sequencing the targeted AFLP fragments. BAC library screening Fingerprinting analysis and BAC end sequencing We previously constructed two BAC libraries using the multiple disease resistance line MR-1 (Luo et al., 2001). The HindIII library To isolate BAC DNA, 5 ml of LB broth (Sigma, St Louis, MO, USA) contains 67 968 clones with an average insert size of 118 kb, while )1 containing 12.5 mg ml of chloroamphenicol (Sigma) was inocu- the EcoRI library contains 85 248 clones with an average size of lated with a single BAC clone followed by growth at 37C with agi- 114 kb. These libraries provide a coverage of 15 and 18 genome tation at 250 rpm for 20 h. BAC DNA was extracted according to an equivalents, respectively. Probes used to screen the BAC libraries alkaline lysis method described by Woo et al. (1994). Agarose gel were PCR fragments derived from MR-1 genomic DNA or BAC electrophoresis and restriction fragment data analysis with the clones. In both cases, the PCR product was run on a 1% agarose program FPC 3.2 (Soderlund et al., 1997) for contig construction gel, the band cut out and cleaned using the QIA quick gel was performed as described by Zhu et al. (1999). BAC end sequence extraction kit (Qiagen, Valencia, CA, USA). Screening the BAC was determined using SP6 and T7 vector primers using Big dye library was realized by hybridization as described by Budiman terminator chemistry v2.0 (PE Biosystems, Foster City, CA, USA) et al. (2000). and analyzed with an ABI 3700 automatic DNA sequencer (PE Bio- systems). The resulting sequences were utilized to design PCR primers, obtain probes for BAC library screening and develop sim- Shotgun sequencing of BAC clones ple sequence repeat (SSR) or sequence tagged sites (STS) markers for fine mapping Fom-2. BAC DNA, isolated with the Qiagen large-construct kit (Qiagen), was used for constructing shotgun libraries for sequencing. Two hundred microliters of BAC DNA was sheared with GeneMachine SSR and STS analysis Hydroshear using a speed code of 15 for 20 cycles. DNA fragments of 1.5–3.0 kb were agarose-gel purified. Three micrograms of the Putative SSRs were identified using the program Sputnik (http:// obtained DNA was end-repaired using End-It kit (Epicenter, Madi- espressosoftware.com/pages/sputnik.jsp). The minimum number son, WI, USA) and ligated to dephosphorylated EcoRV-pBluescript II of the core motifs was set to six for dinucleotides and four for KS overnight at 16C. One microliter of the ligation solution was trinucleotides. To select primers that flank the SSRs we analyzed transformed into 25 llofEscherichia coli electroMax DH10B TM the target sequence with the program Primer 3 (Rozen and Ska- competent cells (Invitrogen) by electroporation. A Genetix Q-bot letsky, 2000) using the default settings. Primers expected to pro- was used to transfer recombinants colonies to 384-well microtiter duce fragments of sizes ranging between 100 and 400 bp were plates for storage at )80C. Both ends of resulting clones were selected and custom synthesized by IDT (Coralville, IA, USA). sequenced using standard techniques and the sequences assem- When a putative SSR was not identified in a BAC end sequence a bled into contigs using the Phred and Phrap programs (Ewing et al., random sequence was chosen to develop an STS marker. The list 1998; Gordon et al., 1998). of primer sequences and the expected size for the PCR products from MR-1 is indicated in Table 1. PCR products were labeled with TM a fluorescent dye and size separated using ABI Prism 377 DNA Allele sequencing of the candidate genes sequencers as described below. For the PCR-product labeling the method described by Oetting et al. (1995) was used. The se- To obtain sequence from the alleles of the two candidate genes quence of the extension added to the 5¢ end of the forward primer (gene 7 and 9), we used the following method. Specific primers was AACAGCTATGACCATGA. In order to detect the PCR product, were designed to amplify the overlapping 400–600 bp fragments a labeled primer having the same sequence as the extension was from different alleles. The PCR products were then cloned as pre- added in the PCR reaction. This primer was labeled with NED, viously indicated and sequenced from both ends using standard HEX or 6-FAM, (PE Biosystems) allowing the multiplexing of techniques. Sequencing primers were designed to complete the several PCR products in the same gel lane. The PCR amplification DNA sequence as needed. At least two clones for each PCR frag- was performed in a 12 ll reaction with 1· reaction buffer [20 mM ment were used for sequencing.

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 Fom-2 is a complex locus 295

Sequence analysis and database search Brotman, Y., Silberstein, L., Kovalski, I., Perin, C., Dogimont, C., Pitrat, M., Klingler, J., Thompson, G.A. and Perl-Treves, R. (2002) To identify putative genes in the Fom-2 interval, the DNA sequences Resistance gene homologues in melon are linked to genetic loci were analyzed with the programs GENSCAN (Burge and Karlin, 1997) conferring disease and pest resistance. Theor. Appl. Genet. 104, and GeneMark.hmm (Lukashin and Borodovsky, 1998) using Ara- 1055–1063. bidopsis settings followed by manual editing. The sequences were Budiman, M.A., Mao, L., Wood, T.C. and Wing, R.A. (2000) A deep- also searched against publicly available non-redundant protein and coverage tomato BAC library and prospects toward development EST databases using the BLASTX and TBLASTX (Altschul et al., 1990) of an STC framework for genome sequencing. Genome Res. 10, programs, respectively. The criteria used to define a gene were as 129–136. described by Ramakrishna et al. (2002). Putative genes were Burge, C. and Karlin, S. (1997) Prediction of complete gene struc- considered pseudogenes when BLASTX analysis detected similarity tures in human genomic DNA. J. Mol. Biol. 268, 78–94. to only limited regions of the corresponding protein match Burger, Y., Katzir, N., Tzuri, G., Portnoy, V., Saar, U., Shriber, S., (Ramakrishna et al., 2002) and manual inspection revealed the lack Perl-Treves, R. and Cohen, R. (2003) Variation in the response of of suitable ATG as well as the presence of indels/base substitutions melon genotypes to Fusarium oxysporum f.sp. melonis race 1 resulting in stop codons. determined by inoculation tests and molecular markers. Plant The CC structure prediction was performed using the COILS Pathol. 52, 204–211. program with variable window size (21–28) and the MTIDK matrix Cai, D., Kleine, M., Kifle, S. et al. (1997) Positional cloning of a gene for (Pan et al., 2000). nematode resistance in sugar beet. Science, 275, 832–834. Upon request, all novel material described in this article will be Chin, D.B., Arroyo-Garcia, R., Ochoa, O.E., Kesseli, R.V., Lavelle, made available in a timely manner for non-commercial research D.O. and Michelmore, R.W. (2001) Recombination and sponta- purposes. neous mutation at the major cluster of resistance genes in lettuce (Lactuca sativa). Genetics, 157, 831–849. Accession numbers Collins, N., Drake, J., Ayliffe, M., Sun, Q., Ellis, J., Hulbert, S. and Pryor, T. (1999) Molecular characterization of the maize Rp1-D GenBank accession number for the sequences reported in this art- rust resistance haplotype and its mutants. Plant Cell, 11, 1365– icle is AY583855. 1376. Dixon, M.S., Jones, D.A., Keddie, J.S., Thomas, C.M., Harrison, K. and Jones, J.D. (1996) The tomato Cf-2 disease resistance locus Acknowledgements comprises two functional genes encoding leucine-rich repeat proteins. Cell, 84, 451–459. We wish to express our appreciation to Rita Mia for technical sup- Dodds, P.N., Lawrence, G.J. and Ellis, J.G. (2001) Six amino acid port, Jeff Mills from Seminis Vegetable Seeds for DNA extraction, changes confined to the Leucine-Rich Repeat b-strand/b-turn Thomas Houfek for sequence data management. We also gratefully motif determine the difference between the P and P2 rust resist- acknowledge Dr Michel Pitrat (INRA, Station de Ge´ ne´tique et ance specificities in flax. Plant Cell, 13, 163–178. d’Ame´lioration des fruits et le´ gumes, Montfavet France) for provi- Durrett, R.T., Chen, K. and Tanksley, S.D. (2002) A simple formula ding us with seed from the RIL population (Ve´ drantais · PI 161375) useful for positional cloning. Genetics, 160, 353–355. and Dr Pere Arus (IRTA centre de Cabrils, Spain) for probes linked to Ernst, K., Kumar, A., Kriseleit, D., Kloos, D.U., Phillips, M.S. and Fom-2. This work was supported by funds from the USDA-ARS Ganal, M.W. (2002) The broad-spectrum potato cyst nematode Charleston SC and Seminis Vegetable Seeds. resistance gene (Hero) from tomato is the only member of a large gene family of NBS-LRR genes with an unusual amino acid repeat References in the LRR region. Plant J. 31, 127–136. Ewing, B., Hillier, L., Wendl, M.C. and Green, P. (1998) Base-calling Aarts, N., Metz, M., Holub, E., Staskawicz, B.J., Daniels, M.J. and of automated sequencer traces using Phred. I. Accuracy assess- Parker, J.E. (1998) Different requirements for EDS1 and NDR1 by ment. Genome Res. 8, 175–185. disease resistance genes define at least two R gene-mediated Gorbunova, V.V. and Levy, A.A. (1999) How plants make ends signaling pathways in Arabidopsis. Proc. Natl Acad. Sci. USA, 95, meet: DNA double-strand break repair. Trends Plant Sci. 4, 263– 10306–10311. 269. Allers, T. and Lichten, M. (2001) Differential timing and control of Gordon, D., Abajian, C. and Green, P. (1998) Consed: a graphical tool noncrossover and crossover recombination during meiosis. Cell, for sequence finishing. Genome Res. 8, 195–202. 106, 47–57. Gu, X. and Nei, M. (1999) Locus specificity of polymorphic alleles Altschul, S.F., Gish, W., Miller, W., Meyers, E.W. and Lipman, D.J. and evolution by a birth-and-death process in mammalian MHC (1990) Basic local alignment search tool. J. Mol. Biol. 215, 403– genes. Mol. Biol. Evol. 16, 147–156. 410. Hammond-Kosack, K.E. and Jones, J.D.G. (1997) Plant disease Anderson, P.A., Lawrence, G.J., Morrish, B.C., Ayliffe, M.A., Finne- resistance genes. Annu. Rev. Plant Physiol. Plant Mol. Biol. 48, gan, E.J. and Ellis, J.G. (1997) Inactivation of the flax rust resist- 575–607. ance gene M associated with loss of a repeated unit within the Henk, A.D., Warren, R.F. and Innes, R.W. (1999) A new Ac-like leucine-rich repeat coding region. Plant Cell, 9, 641–651. transposon of Arabidopsis is associated with a deletion of the Bai, J.F., Pennill, L.A., Ning, J.C. et al. (2002) Diversity in nucleotide RPS5 disease resistance gene. Genetics, 151, 1581–1589. binding site-leucine-rich repeat genes in cereals. Genome Res. 12, Hughes, A.L. and Nei, M. (1988) Pattern of nucleotide substitution at 1871–1884. major histocompatibility complex class I loci reveals overdomi- Bittner-Eddy, P.D., Crute, I.R., Holub, E.B. and Beynon, J.L. (2000) nant selection. Nature, 335, 167–170. RPP13 is a simple locus in Arabidopsis thaliana for alleles that Hulbert, S.H., Webb, C.A., Smith, S.M. and Sun, Q. (2001) Resist- specify downy mildew resistance to different avirulence deter- ance gene complexes: Evolution and utilization. Annu. Rev. minants in Peronospora parasitica. Plant J. 21, 177–188. Phytopathol. 39, 285–312.

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 296 Tarek Joobeur et al.

Inbar, O., Liefshitz, B., Bitan, G. and Kupiec, M. (2000) The rela- NBS-LRR gene family of Arabidopsis thaliana. Genome Res. 12, tionship between homology length and crossing over during the 1305–1315. repair of a broken chromosome. J. Biol. Chem. 275, 30833– Noel, L., Moores, T.L., van der Biezen, E.A., Parniske, M., Daniels, 30838. M.J., Parker, J.E. and Jones, J.D.G. (1999) Pronounced intraspe- Johnson, R.D. and Jasin, M. (2001) Double-strand-break-induced cific haplotype divergence at the RPP5 complex disease resist- homologous recombination in mammalian cells. Biochem. Soc. ance locus of Arabidopsis. Plant Cell, 11, 2099–2111. Trans. 29, 196–201. Oetting, W.S., Lee, H.K., Flanders, D.J., Wiesner, G.L., Sellers, T.A. Jones, D.A. and Jones, J.D.G. (1997) The role of leucine-rich repeat and King, R.A. (1995) Linkage analysis with multiplexed short proteins in plant defences. Adv. Bot. Res. Adv. Plant Pathol. 24, tandem repeat polymorphisms using infrared fluorescence and 89–167. M13 tailed primers. Genomic, 30, 450–458. Kirik, A., Salomon, S. and Puchta, H. (2000) Species-specific double- Pan, Q.L., Wendel, J. and Fluhr, R. (2000) Divergent evolution of strand break repair and genome evolution in plants. EMBO J. 19, plant NBS-LRR resistance gene homologues in dicot and cereal 5562–5566. genomes. J. Mol. Evol. 50, 203–213. Kobe, B. and Deisenhofer, J. (1995) A structural basis of the inter- Parniske, M. and Jones, J.D.G. (1999) Recombination between di- actions between leucine-rich repeats and protein ligands. Nature, verged clusters of the tomato Cf-9 plant disease resistance gene 374, 183–186. family. Proc. Natl Acad. Sci. USA, 96, 5850–5855. Kosambi, D.D. (1944) The estimation of map distances from Parniske, M., HammondKosack, K.E., Golstein, C., Thomas, C.M., recombination values. Ann. Eugen. 12, 172–175. Jones, D.A., Harrison, K., B.B.H., W. and J.D.G., J. (1997) Novel Lander, E.S., Green, P., Abrahamson, J., Barlow, A., Daly, M.J., disease resistance specificities result from sequence exchange Lincoln, S.E. and Newburg, L. (1987) MAPMAKER an interactive between tandemly repeated genes at the Cf-4/9 locus of tomato. computer package for constructing primary linkage genetic Cell, 91, 821–832. maps of experimental and natural populations. Genomics, 1, Pearson, W.R. (1990) Rapid and sensitive sequence comparison 174–181. with FASTP and FASTA. Methods Enzymol. 183, 63–98. Lukashin, A.V. and Borodovsky, M. (1998) GeneMark.hmm: new Petronczki, M., Siomos, M.F. and Nasmyth, K. (2003) Un Meńage a` solutions for gene finding. Nucl. Acids Res. 26, 1107–1115. Quatre: the molecular biology of chromosome segregation in Luo, M., Wang, Y.H., Frisch, D., Joobeur, T., Wing, R. and Dean, R.A. meiosis. Cell, 112, 423–440. (2001) Melon BAC library construction using improved methods Ramakrishna, W., Emberton, J., Ogden, M., SanMiguel, P. and and identification of clones linked to the locus conferring resist- Bennetzen, J.L. (2002) Structural analysis of the maize Rp1 com- ance to melon Fusarium wilt (Fom-2). Genome, 44, 154–162. plex reveals numerous sites and unexpected mechanisms of local Mackey, D., Holt, B.F., Wiig, A. and Dangl, J.L. (2002) RIN4 interacts rearrangement. Plant Cell, 14, 3213–3223. with Pseudomonas syringae type III effector molecules and is Richardson, C. and Jasin, M. (2000) Coupled homologous and required for RPM1-mediated resistance in Arabidopsis. Cell, 108, nonhomologous repair of a double-strand break preserves ge- 743–754. nomic integrity in mammalian cells. Mol. Cell. Biol. 20, 9068– Marathe, R. and Dinesh-Kumar, S.P. (2003) Plant defense: one post, 9075. multiple guards?! Mol. Cell, 11, 284–286. Rozen, S. and Skaletsky, H.J. (2000) Primer3 on the WWW for Martin, G.B., Bogdanove, A.J. and Sessa, G. (2003) Understanding general users and for biologist programmers. In Bioinformatics the functions of plant disease resistance proteins. Annu. Rev. Methods and Protocols: Methods in Molecular Biology (Krawetz, Plant Biol. 54, 23–61. S. and Misener, S., eds). Totowa, NJ: Humana Press, pp. 365–386. Martyn, R.D. and Gordon, T.R. (1996) Fusarium wilt of melon. In Schreuder, W., Lamprecht, S.C. and Holz, G. (2000) Race deter- Compendium of Cucurbit Diseases (Zitter, T.A., Hokins, D.L. mination and vegetative compatibility grouping of Fusarium and Thomas, C.E., eds). Minneapolis, MA: APS Press, pp. 14– oxysporum f.sp. melonis from South Africa. Plant Dis. 84, 231– 15. 234. McCreight, J.D., Nerson, H. and Grumet, R. (1993) Melon Cucumis Soderlund, C., Longden, I. and Mott, R. (1997) FPC: a system for melo L. In Genetic Improvement of Vegetable Crops (Kalloo, G. building contigs from restriction fingerprinted clones. Comput. and Berch, B.O., eds). Oxford: Pergamon Press, pp. 267–294. Appl. Biosci. 13, 523–535. Meyers, B.C., Chin, D.B., Shen, K.A., Sivaramakrishnan, S., Lavelle, Song, W.Y., Pi, L.Y., Wang, G.L., Gardner, J., Holsten, T. and Ronald, D.O., Zhang, Z. and Michelmore, R.W. (1998) The major resistance P.C. (1997) Evolution of the rice Xa21 disease resistance gene gene cluster in lettuce is highly duplicated and spans several family. Plant Cell, 9, 1279–1287. megabases. Plant Cell, 10, 1817–1832. Thomas, C.E., McCreight, J.D., Zink, F.W. and Jourdain, E.L. (1992) Meyers, B.C., Dickerman, A.W., Michelmore, R.W., Sivaramakrish- Inheritance of disease resistance in Muskmelon line MR-I. In Fifth nan, S., Sobral, B.W. and Young, N.D. (1999) Plant disease EUCARPIA Cucurbitaceae Symposium, Warsaw, Poland, pp. 143– resistance genes encode members of an ancient and diverse 147. protein family within the nucleotide-binding superfamily. Plant J. Van der Biezen, E.A. and Jones, J.D.G. (1998) The NB-ARC domain: 20, 317–332. a novel signalling motif shared by plant resistance gene products Meyers, B.C., Kozik, A., Griego, A., Kuang, H. and Michelmore, R.W. and regulators of cell death in animals. Curr. Biol. 8, 226–227. (2003) Genome-wide analysis of NBS-LRR-encoding genes in Van der Hoorn, R.A.L., Kruijt, M., Roth, R., Brandwagt, B.F., Arabidopsis. Plant Cell, 15, 809–834. Joosten, M.H.A.J. and De Wit, P.J.G.M. (2001) Intragenic recom- Michelmore, R.W. (2000) Genomic approaches to plant disease bination generated two distinct Cf genes that mediate AVR9 resistance. Curr. Opin. Plant Biol. 3, 125–131. recognition in the natural population of Lycopersicon pimpinel- Michelmore, R.W. and Meyers, B.C. (1998) Clusters of resistance lifolium. Proc. Natl Acad. Sci. USA, 98, 10493–10498. genes in plants evolve by divergent selection and a birth-and- Wang, Y.H., Thomas, C.E. and Dean, R.A. (1997) A genetic map death process. Genome Res. 8, 1113–1130. of melon (Cucumis melo L.) based on amplified fragment Mondragon-Palomino, M., Meyers, B.C., Michelmore, R.W. and length polymorphism (AFLP) markers. Theor. Appl. Genet. 95, Gaut, B.S. (2002) Patterns of positive selection in the complete 791–798.

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297 Fom-2 is a complex locus 297

Wang, Y.H., Thomas, C.E. and Ralph, A.D. (2000) Genetic mapping Xiao, S.Y., Ellwood, S., Calis, O., Patrick, E., Li, T.X., Coleman, M. of a fusarium wilt resistance gene (Fom-2) in melon (Cucumis and Turner, J.G. (2001) Broad-spectrum mildew resistance in melo L.). Mol. Breeding, 6, 379–389. Arabidopsis thaliana mediated by RPW8. Science, 291, 118– Wang, Y.H., Choi, W.B., Thomas, C.E. and Dean, R.A. (2002) Cloning 120. of disease-resistance homologues in end sequences of BAC Xu, R., Ye, X. and Li, Q.Q. (2004) AtCPSF73-II gene encoding an clones linked to Fom-2, a gene conferring resistance to fusarium Arabidopsis homolog of CPSF 73 kDa subunit is critical for early wilt in melon (Cucumis melo L.). Genome, 45, 473–480. embryo development. Gene, 324, 35–45. Warren, R.F., Henk, A., Mowery, P., Holub, E. and Innes, R.W. (1998) Zhu, H., Blackmon, B.P., Sasinowski, M. and Dean, R.A. (1999) A Mutation within the leucine-rich repeat domain of the Arabid- Physical map and organization of chromosome 7 in the rice blast opsis disease resistance gene RPS5 partially suppresses multiple fungus, Magnaporthe grisea. Genome Res., 9, 739–750. bacterial and downy mildew resistance genes. Plant Cell, 10, Zink, F.W. and Thomas, C.E. (1990) Genetics of resistance in 1439–1452. muskmelon to Fusarium oxysporum f.sp. melonis race 0, 1, and 2 Woo, S., Jiang, J., Gill, B., Paterson, A. and Wing, R. (1994) Con- in muskmelon line MR-I. Phytopathology, 80, 1230–1232. struction and characterization of a bacterial artiﬁcial chromosome library of Sorghum bicolor. Nucl. Acids Res. 22, 4922–4931.

ª Blackwell Publishing Ltd, The Plant Journal, (2004), 39, 283–297