Contrasting Evolutionary Patterns of the Rp1 Resistance GeneFamily in Different Species of Poaceae

Sha Luo,�,1 Junhua Peng,�,2 Kunpeng Li,1 Min Wang,1 and Hanhui Kuang*,1

1Key Laboratory of Horticulture Biology, Ministry of Education, and Department of Vegetable Crops, College of Horticulture andForestry, Huazhong Agricultural University, Wuhan, China2Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture, Wuhan Botanic Garden, Chinese Academy ofScience, Wuhan, Hubei, China

�These authors contributed equally to this work.

*Corresponding author: E-mail: kuangfile@gmail.com.

Associate editor: Neelima Sinha

Abstract

Disease-resistance genes (R-genes) in plants show complex evolutionary patterns. We investigated the evolution of the Rp1R-gene family in Poaceae, and 409 Rp1 fragments were sequenced from 21 species. Our data showed that the commonancestor of Poaceae had two Rp1 loci, but the number of Rp1 locus in extant species varies from one to five. Some wheatand Zea genotypes have dozens of Rp1 homologues in striking contrast to one or two copies in Brachypodium distachyon.The large number of diverse Rp1 homologues in Zea was the result of duplications followed by extensive sequenceexchanges among paralogues, and all genes in maize have evolved in a pattern of Type I R-genes. The high frequency ofsequence exchanges did not cause concerted evolution in Zea species, but concerted evolution was obvious between Rp1homologues from genera Zea and Sorghum. Differentiation of Type I and Type II Rp1 homologues was observed in Oryzaspecies, likely occurred in their common ancestor. One member (Type II R-gene) in the Oryza Rp1 cluster did not changesequences with its paralogues, whereas the other paralogues (Type I R-genes) had frequent sequence exchanges. Thefunctional Pi37 resistance gene in rice was generated through an unequal crossover between two neighboring paraloguesfollowed by four point mutations. The Rp1 homologues in wheat and barley were most divergent, probably due to lack ofsequence exchanges among them. Our results shed more light on R-gene evolution, particularly on the differentiation ofType I and Type II R-genes.

Key words: Rp1, disease-resistance gene, copy number, gene conversion, gene cluster.

IntroductionMore than 100 disease-resistance genes have been clonedfrom different plant species and approximately 80% ofthem encode nucleotide-binding site (NBS) and leucine-rich repeat (LRR) domains (Liu et al. 2007). Many of theNBS–LRR–encoding genes (referred as R-genes in this arti-cle) are clustered in plant genomes (Meyers et al. 2003).The clustering of R-genes might facilitate sequence ex-changes between paralogues and generating new resistancespecificities (Mondragon-Palomino et al. 2002; Kuang et al.2008). Indeed, a group of R-genes, termed Type I R-genes,are extensive chimeras, which were generated through fre-quent gene conversions between paralogues. However, an-other group of R-genes (Type II R-genes) had rare or nosequence exchanges with paralogues though they mightbe located in the same cluster as Type I R-genes (Kuanget al. 2004).

The mechanisms underlying the differentiation of Type Iand Type II R-genes remain unclear. The high conservationof each lineage of Type II R-genes might suggest their in-dispensable function, that is, important resistance specific-ity. However, several pieces of evidence are against the‘‘function’’ hypothesis. First, some Type II R-genes are ap-parent pseudogenes with frameshift indels or premature

stop codon. Paradoxically, some frameshift indels are pres-ent in all genes of a Type II R-gene lineage (Kuang et al.2004). Second, Type II R-genes (as donors) rarely convertedtheir paralogues though such a direction of gene conver-sion would not change the original function of the Type IIR-gene. Third, some Type II R-gene lineages have very lowfrequencies in nature populations (Kuang et al. 2004; Shenet al. 2006). All these facts are against the hypothesis thatthe conservation of Type II R-genes are due to their criticalfunctions.

Physical proximity of homologues should not accountfor the differentiation of Type I and Type II R-genes either.Studies on RPP8 resistance gene homologues in Arabidopsisshowed that frequent sequence exchanges occurred be-tween homologues separated by 2.3 Mb (Kuang et al.2008). In contrast, a Type I and Type II RPP8 homologuerarely had sequence exchanges between them though theyare only 4 kb apart on chromosome V. Furthermore, Type Iand Type II homologues or different lineages of Type I ho-mologues interweave in some R-gene clusters, suggestingthat physical proximity may not be responsible for the dif-ferentiation of Type I and Type II R-genes (Kuang et al.2004, 2005).

The grass family (Poaceae) is ideal for comparative ge-nomics. Numerous genetic and genomic data are available

© The Author 2010. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, pleasee-mail: journals.permissions@oup.com

Mol. Biol. Evol. 28(1):313–325. 2011 doi:10.1093/molbev/msq216 Advance Access publication August 16, 2010 313

Research

article

for different species of Poaceae. Noticeably, the rice (Oryzasativa) genome has been completely sequenced (Goff et al.2002); the draft genome sequences of maize (Zea may ssp.may) (Schnable et al. 2009), sorghum (Sorghum bicolor)(Paterson et al. 2009), and Brachypodium distachyon (Vogelet al. 2010) have been released; and genomic bacterial ar-tificial chromosome (BAC) libraries and physical maps ofseveral genotypes of cultivated and wild rice are available(Ammiraju et al. 2006). All these data are valuable for stud-ies on comparative and evolutionary genomics of R-genes.

We investigated the evolutionary patterns of the Rp1gene family in Poaceae. The Rp1-D gene, a member ofthe Rp1 family clustered on the short arm of chromosome10 of maize, encodes resistance against rust caused byPuccinia sorghi (Collins et al. 1999). The Rp1 locus in maizewas unstable due to frequent unequal crossovers betweenparalogues (Hulbert 1997; Sun et al. 2001). Unequal cross-over between paralogues was shown to be the main geneticevent for the spontaneous loss of the Rp1-D and generatingnovel resistance specificity (Sun et al. 2001; Smith andHulbert 2005). Unequal crossover was also likely responsi-ble for the large variation of the copy number of Rp1 genefamily in maize, which varied from one to more than 50 indifferent maize haplotypes (Smith et al. 2004). The Rp1homologues were present in many species in the grass fam-ily (Ayliffe et al. 2000). The Rp1 locus in sorghum, calledrph1 locus, was sequenced and five Rp1 homologueswere discovered (Ramakrishna et al. 2002). The Rp1 homo-logues were duplicated to several chromosomes in barley(Hordeum vulgare L.) (Ayliffe et al. 2000; Rostoks et al.2002). In rice cultivar Nipponbare, genes most similar toRp1-D are four tandem repeated homologues located onthe long arm of chromosome 1. The Rp1 locus in rice cul-tivar St. No. 1 has a similar structure to that in Nipponbare,and one (the Pi37 gene) of the four homologues in St. No. 1encodes resistance against rice basic local alignment searchtool (Blast) (Lin et al. 2007).

We comprehensively studied the structures of the Rp1clusters in Oryza, Triticum, Zea, B. distachyon, Sorghum, etc.A total of 164 full-length or nearly full-length and 245partial-length Rp1 homologues were sequenced from21 species of the grass family. The evolutionary patternsof Rp1 homologues in different species and subfamiliesof Poaceae were compared. The differentiation of Type Iand Type II Rp1 homologues in Poaceae was addressed.

Materials and Methods

Materials Used to Clone Rp1 HomologuesRp1 homologues were obtained from 72 genotypes of 21species of the grass family (Poaceae), representing its threemajor subfamilies, Bambusoideae, Panicoideae, and Pooi-deae (supplementary table S1, Supplementary Material on-line). Forty genotypes of the Bambusoideae subfamily wereincluded in this study: 14 cultivars of O. sativa ssp. indica, 10cultivars of O. sativa ssp. japonica, 2 genotypes of O. alta, 4genotypes from O. nivara, 4 genotypes from O. rufipogon, 1genotype each from O. glaberrima, O. punctata, O. minuta,

O. australiensis, O. coarctata, and O. brachyantha. Eightgenotypes from the Panicoideae subfamily were used in thisstudy: three genotypes of Z. may ssp. parviglumis, one ge-notype each of Z. luxurians and Z. diploperennis, and threegenotypes from sorghum (S. bicolor). Twenty-four geno-types from the Pooideae subfamily were included in thisstudy: 3 cultivars of wheat (Triticum aestivum), 17 geno-types of B. distachyon, and 1 genotype each of T. monocum,T. urartu, T. durum, and Aegilops tauschii. Another eightgenotypes of B. distachyon were also included in this study,but no Rp1 homologues were sequenced from thesegenotypes.

Cloning Rp1 Homologues from Oryza speciesThe Rp1 locus in Nipponbare has four homologues(Os01g57270, Os01g57280, Os01g57310, and Os01g57340),and specific primers were designed to amplify threeof them. Eight polymerase chain reaction (PCR) primercombinations specific to Os01g57270 were used to investi-gate the presence/absence of this gene in Oryza species(supplementary table S2, Supplementary Material online).If present, the PCR products of Os01g57270-like genes weresequenced directly. The ‘‘alleles’’ of Os01g57280 were PCRamplified using primer combination 57280-up-K and Rp-R2-K, which are located 711 bp upstream and 3,842 bpdownstream of the start codon of geneOs01g57280, respec-tively. The ‘‘alleles’’ ofOs01g57340were PCR amplified usingprimer combination Rp1-F1K and 57340-dn-R3, which arelocated 146 and 4,218 bp downstream of the start codon ofgene Os01g57340, respectively. The PCR products weretreated with Exo-SAP (Fermentas, Ontario, Canada) andthen sequenced directly.

When available, BAC clones were used as templates toPCR amplify Rp1 homologues. In the sequenced genomesof rice cultivars Nipponbare and 9311 and B. distachyon,the Rp1 locus is flanked by a gene encoding vacuolar pro-tein sorting 28 and a gene encoding sphingosine kinase.Therefore, we delimited the Rp1 locus in Oryza by thesetwo genes. First, BAC clones spanning the Rp1 locus wereidentified using their BAC end sequences. Rp1 homologuesfrom these BAC clones were amplified using primers con-served in all available Rp1 homologues (supplementarytable S2, Supplementary Material online). The PCR prod-ucts were sequenced directly using one primer afterExo-SAP treatment. If the sequencing showed a clean chro-matogram, the whole fragment was sequenced throughprimer walking. If multiple peaks were present in the chro-matogram, the PCR products were cloned and individualclones were sequenced.

Cloning and Sequencing Rp1 Homologues from ZeaSpecies, Triticum Species, B. distachyon, A. tauschii,and S. bicolorRp1 fragments of �3.8 kb were amplified from three Zeaspecies using primer combination Maize-F2 and Maize-R(supplementary table S2, Supplementary Material online),which are located at þ6 and þ3798 of the Rp1-D gene,

Luo et al. · doi:10.1093/molbev/msq216 MBE

314

respectively. PCR was performed using TaKaRa LA Taq(TaKaRa, Dalian, China), with a reaction of 25 ll, treated5 min at 94 �C, followed by 32 cycles of 94 �C for 30 s,55 �C for 45 s, and 72 �C for 4 min. The PCR productswere gel purified using the Gel Purification Kit (Generay,Shanghai, China) and cloned into pCR2.1 vector usingthe TOPO TA cloning kit (Invitrogen, Carlsbad, CA). Thismethod was first validated by comparing sequences of frag-ments amplified from two independent PCRs for Z. mayssp. parviglumis genotype PI384062. After validation, thismethod was applied to other Zea genotypes.

A sequencing method was used to estimate the Rp1copy number in Zea species, which was shown to be themost reliable approach to estimate the copy number ofa large gene family in a genome (Kuang et al. 2004). A con-served primer combination For2/Low4 was used to PCRamplify Rp1 fragments (�2 kb), the PCR products werecloned into TOPO TA cloning kit, individual clones weresequenced, and distinct fragments were counted to esti-mate the copy number of Rp1 homologues in a genotype.PCR products were amplified using Taq polymerase, witha reaction of 25 ll, 5 min at 94 �C, followed by 32 cycles of94 �C for 30 s, 55 �C for 30 s, and 72 �C for 2 min. Theprocedure was first validated by comparing sequences offragments obtained from four independent PCRs for Z.may ssp. parviglumis genotype Ames21889. Two cloneswere considered to be derived from the same gene if theyexhibited more than 99.4% nucleotide identity. If an Rp1gene is present in two or more PCRs, the gene is consideredas ‘‘verified.’’ For the Rp1 genes present in one PCR only,primers specific to some of them were designed and usedto amplify and sequence PCR products to confirm theirpresence in the genome. The method is considered validif above experiments show no PCR artifacts. The validatedmethod was used to estimate the copy number of Rp1 ho-mologues in other Zea genotypes.

Several primer combinations were designed to amplifynearly full-length Rp1 homologues from subfamilyPooideae, but only primer combination CS-F3/Brach-R2was partially successful. In order to obtain more Rp1fragments from this subfamily, primer combinationFor2/Low4 was used to PCR amplify the LRR-encoding re-gion (�2 kb) of Rp1 homologues from Triticum and its rel-ative A. tauschii. PCR products were cloned into TOPOvector and individual clones were sequenced as describedabove.

A single Rp1 homologue is present in the genome of B.distachyon. Primers flanking the Rp1 homologue were de-signed, and full-length Rp1 homologues were amplifiedfrom 25 genotypes of B. distachyon (supplementary tablesS1 and S2, Supplementary Material online). The amplifiedPCR products were treated with Exo-SAP (Fermentas) andthen sequenced directly using primer Brach-R. If the se-quencing chromatogram was clean, the PCR fragmentswere fully sequenced through by primer walking. However,if the chromatogram had two peaks, the PCR productswere cloned and individual clones were sequenced as de-scribe above.

Rp1 homologues were amplified from three genotypes ofsorghum using a pair of primers Trp-F/Trp-R, which areconserved in all Rp1 homologues cloned from sorghumand maize. PCR amplification, cloning, and sequencingwere identical to the method described above to amplifynearly full-length Rp1 homologues in Zea.

Southern Hybridization Using Digoxigenin LabelingSystemTo estimate the copy number of Rp1 genes in Oryza, DNAgel blot analysis was conducted according to the protocolfrom Digoxigenin (DIG)-High Prime DNA Labeling and De-tection Starter Kit I (Roche, Mannheim, Germany). BACplasmid DNA of 14 genotypes ofOryza species was digestedwith EcoR I endonuclease (Fermentas). The 250-bp probewas a mixture of PCR fragments amplified from seven di-verse rice Rp1 genes amplified using primer combinationProbe-F1/Probe-R1 (supplementary table S2, Supplemen-tary Material online). The mixed fragments were labeledwith digoxigenin-11-dUTP using DIG-High Prime.

Mapping of Rp1 Homologues in Wheat GenotypeChinese Spring Using Deletion LinesRp1 fragments amplified from wheat cultivar ChineseSpring (CS) were mapped using its aneuploid and deletionstocks (Endo and Gill 1996). A total of 146 aneuploidand deletion stocks including 21 nullisomic–tetrasomic,24 ditelosomic, and 101 deletion lines were used (Penget al. 2004). A set consisting of five Southern blots, eachwith 30 lanes, was hybridized with the Rp1 probe contain-ing an even mixture of 23 Rp1 fragments from CS, in thehybridization reaction. The conventional Southern hybrid-ization approach was adopted with some minor modi-fications (http://wheat.pw.usda.gov/NSF/project/mapping_data.html). If a band in the Southern profile is presentin the wild-type CS but absent only in a specific seriesof CS cytogenetic stocks, the corresponding Rp1 homo-logue is located to a specific region or bin in a chromosome(Endo and Gill 1996; Peng et al. 2004). To map individualRp1 fragments from CS, primers specific to each fragmentwere designed and used to screen a set of ditelosomic anddeletion lines related towheat group 4 chromosomes becausemost of the CS Rp1 homologues were mapped to chromo-some 4AL via Southern hybridization. Similarly, amplificationof PCR products from wild-type CS but not from one or sev-eral deletion lines would map the corresponding Rp1 homo-logue to a certain bin or specific chromosome region.

Sequence AnalysesThe nucleotide sequences of Rp1 homologues were alignedusing ClustalX (Thompson et al. 1997) and refined usingGeneDoc (http://www.nrbsc.org/gfx/genedoc/). Neighbor-joining (NJ) trees using Kimura’s two-parameter model wereconstructed and bootstrap values (1,000 replications) werecalculatedusingMEGA4.0 (Tamuraet al. 2007). Sequenceex-changesweredetectedusingGeneconv (Sawyer 1989),whichdetects identical tracts in twootherwise divergent fragments,

Evolution of the Rp1 Family in Poaceae · doi:10.1093/molbev/msq216 MBE

315

with no mismatch allowed between converted tracts(gscal 5 0). If one gene was detected as having gene con-versionswithseveraldifferentgenes in thesameregion,onlythe eventwith the largest conversion tractwas reported.Allgene conversions detected by Geneconv were examinedand confirmed visually. Chi-square test was used to testif there was significant difference between the frequencyof gene conversions among genes within a species andthe frequency of gene conversions among genes from dif-ferent species. Theexpectedgeneconversionnumber is cal-culated based on the total number of pairs of sequencesfrom the same species and the total number of pairs ofsequences from two different species.

To determine the number of ancestral genes (foundergenes) that generated the current Rp1 homologues inZea, the number of distinct fragments (vary at least fournucleotides) in a window of 70 bp of all Rp1 homologuesfrom Zea was counted. Ten windows (each represents anLRR-encoding unit) of 87 genes from Zea species were ex-amined. The largest number of distinct fragments in a win-dow represents the minimum number of ancestral genesfor the Rp1 gene family in Zea.

Results

Duplication and Deletion of the Rp1 Locus inDifferent Species of PoaceaeThe Rp1 locus in maize is located 2–3 Mb from the tip oftelomere on the short arm of chromosome 10 according tothe updated genome sequences of maize (Schnable et al.2009). The Rp1 locus in sorghum, called rph1 locus, islocated on chromosome 8 (Ramakrishna et al. 2002). Thislocus (called locus A in this artice for convenience) and itsflanking region show good synteny between maize and sor-ghum (fig. 1A). The genes flanking the Rp1 locus in maizewere used to identify its syntenic region in the genomes ofrice and B. distachyon, and two and one syntenic regionswere identified, respectively. Surprisingly, no Rp1 homo-logues are present at this locus in the genomes of riceor B. distachyon (fig. 1A).

Blast search using the Rp1-D sequence as query identi-fied four and one highly similar homologues from the ge-nomes of rice and B. distachyon, respectively. The four Rp1homologues in rice cultivar Nipponbare are clustered onchromosome 1, whereas the single-copy Rp1 homologuein B. distachyon is located on chromosome 2. The Rp1 locusin rice and B. distachyon (referred as locus B) shows goodsynteny but they are not syntenic to the Rp1 locus (locus A)in maize (fig. 1B).

Sorghum, which is in the same subfamily (Panicoideae)as maize (fig. 1C), has two Rp1 loci. One of them is locatedon chromosome 8 and syntenic to locus A (fig. 1A),whereas the other is located on chromosome 3 and syn-tenic to locus B (fig. 1B). Barley, which is in the same sub-family (Pooideae) as B. distachyon (fig. 1C), also has bothloci A and B, in addition to three other Rp1 loci (Rostokset al. 2002). The Rp1 locus on barley chromosome 7(5H)was believed to be syntenic to the maize Rp1 locus (Rostoks

et al. 2002), whereas the Rp1 locus on barley chromosome 5is syntenic to the Rp1 locus in rice because its flankingmarkers ABC261 and BCD1930 are also flanking the Rp1locus in rice. We concluded that there were at least twoancient Rp1 loci in Poaceae, and the two ancient Rp1 locimight have experienced constant duplications and/or de-letions in different lineages of the Poaceae family.

Striking Variation of Rp1 Copy Number in DifferentSpeciesThe sequenced genomes of rice and B. distachyon have fourand one Rp1 homologues, respectively. We used Southernblot and/or sequencing strategy to investigate the copynumber of Rp1 homologues at the locus B of Oryza speciesand in the genome of B. distachyon. Our data showed thatthe locus B of the Oryza species included in this study had

FIG. 1. The two ancient Rp1 loci in Poaceae species. (A) The Rp1locus A is present in maize and sorghum but absent in rice andBrachypodium. (B) The Rp1 locus B is present in rice, Brachypodium,and Sorghum but absent in maize. The genes shaded in gray are Rp1homologues. (C) The phylogenetic relationships and the divergenttime of the major genera included in this study (Gale and Devos1998; Hilton and Gaut 1998; Bennetzen 2000; Freeling 2001; Huanget al. 2002; Vogel et al. 2010).

Luo et al. · doi:10.1093/molbev/msq216 MBE

316

fewer than five copies of Rp1 homologues (fig. 2) andthat there was only one or two Rp1 homologues in eachB. distachyon genotype (see below).

PCR cloning method was used to investigate the copynumber of Rp1 homologues in wild maize, which mayhave a large number of Rp1 homologues in a genome(Smith et al. 2004). To validate this approach, four indepen-dent PCRs were carried out for genotype Ames21889 ofZ. may ssp. parviglumis using a conserved primer combina-tion For2/Low4 (supplementary table S2, SupplementaryMaterial online). Sequencing 264 clones derived fromthe four PCRs gave 70 distinct fragments. Of the 70 Rp1fragments, 23 were present in at least two PCRs, confirmingtheir presence in the genome. The other 47 Rp1 fragmentswere found in only one of the four PCRs. To exclude thepossibility of PCR artifacts, PCR primers specific to 12 of the47 fragments were designed and the amplified PCR prod-ucts were sequenced. The results showed no evidence ofPCR artifacts, in terms of PCR recombination. The validatedPCR method was applied to estimate the Rp1 copy numberin other four genotypes of Zea species. A total of 48, 65, 50,and 49 clones were sequenced for genotypes PI566691 andPI384062 of Z. may ssp. parviglumis, genotype PI441933 ofZ. luxurians, and genotype PI462368 of Z. diploperennis, re-spectively. That generated 13, 31, 42, and 17 diverse frag-ments, respectively. The total number of Rp1 homologuesin Z. luxurians PI441933 might be unprecedentedly largebecause 42 distinct fragments were discovered after se-quencing 50 randomly chosen clones. The high copy num-ber of Rp1 homologues in Zea forms a striking contrast tothe low copy number in O. sativa and B. distachyon.

Large Number of Diverse Rp1 Homologues in ZeaSpecies due to Duplications Followed by FrequentSequence ExchangesTo investigate the mechanism generating the large num-ber of diverse Rp1 homologues in Zea, we obtained 63

nearly full-length (at least 3,659 bp) Rp1 homologues fromthe five genotypes of Zea species (supplementary table S3,Supplementary Material online). These 63 Rp1 homo-logues were combined with 24 maize Rp1 sequencesretrieved from GenBank for further analysis. An NJ dis-tance tree was constructed using Rp1 homologues fromsorghum and Pi37 from rice as outgroups (fig. 3). The87 genes from the four Zea species formed a clade exclu-sively, with a bootstrap number of 100. The genes fromZea exhibited 85.3–99.9% (with an average of 92.3%)nucleotide identities with each other. The average pair-wise nucleotide identity between genes within a species(85.6–99.9%, with an average of 92.1%) is similar to thosebetween genes in different species (85.3–99.5%, with anaverage of 92.5%). t-Test shows that the average nucleo-tide identities between genes from the same species haveno significant difference from those between genes fromdifferent Zea species.

All the Rp1 homologues from Zea species, includingthe functional resistance gene Rp1-D, showed a chimericstructure (fig. 4A). Eighty-four sequence exchanges weredetected among the 87 maize Rp1 homologues using soft-ware Geneconv (P, 0.01). The exchange tracts varied from145 to 2,316 bp, with an average of 629 bp. The exchangetracts were mainly located in the middle of genes, and onlyeight of the 84 exchange tracts extended to one end of thegenes. Of the 84 detected sequence exchanges, 45 occurredbetween genes within a species and 39 between genes fromdifferent species. Chi-square test showed that the sequenceexchanges within a species was significantly more thanexpected (v2 5 9.26, degrees of freedom 5 1, P , 0.01).

To investigate the minimum number of ancestral genesthat generated the 87 diverse Zea Rp1 homologues, individ-ual LRR-encoding units of the 87 genes were analyzed(see Materials and Methods). The number of distinct frag-ments in each LRR-encoding unit of the 87 genes variedfrom 4 to 14. Therefore, the 87 Rp1 homologues obtainedfrom Zea might have been generated by as many as 14 an-cestral genes through frequent sequence exchanges amongthem.

To further investigate the main genetic events ac-counted for the evolution of Rp1 locus in maize, theRp1 locus in maize line B73 was reassembled using the par-tially sequenced BAC clones and published sequences(Ramakrishna et al. 2002; Schnable et al. 2009). Three con-tigs of 750-kb sequences, each containing at least one Rp1homologue, were obtained. Additional sequences of ap-proximately 280 kb were distributed in three unassembledregions and contained no Rp1 homologues. Therefore, theRp1 locus from maize line B73 spans over 1 Mb. It harborsat least 11 full-length and 3 truncated Rp1 homologues.Dot plot analysis of the 750-kb assembled sequences dis-covered several duplications of greater than 10 kb, and fiveof them contained an Rp1 homologue (data not shown).The mechanisms for the duplications containing Rp1 ho-mologues remain unclear. Nevertheless, none of the largeduplications were tandem, ruling out the possibility of re-cent unequal crossover.

FIG. 2. Southern blot of Oryza species using Rp1 fragments. Thereare only one to four bands for each BAC clone spanning the entireRp1 locus.

Evolution of the Rp1 Family in Poaceae · doi:10.1093/molbev/msq216 MBE

317

Concerted Evolution of Rp1 Homologues in GeneraZea and Sorghum but Not in Different Zea SpeciesThe frequent gene conversions did not cause concertedevolution for the Rp1 gene family in the genus Zea. Thehomologues from the four Zea species mingled in theZea clade on the NJ tree, with no independent clade forindividual Zea species (fig. 3). Such relationships were con-sistent with the observation that nucleotide identities be-tween genes within a species were similar to those betweengenes from different Zea species (table 1).

To investigate if frequent sequence exchanges homog-enized the Rp1 gene family in high phylogenetic levels,we extended our study to S. bicolor. Eight Rp1 homologueswere obtained from three genotypes of S. bicolor. The pair-wise nucleotide identity between genes within a genusvaries from 82.8% to 99.9%, which is much higher thanthe pairwise nucleotide identity between genes from differ-ent genus (80.6–83.8%). The high nucleotide identities be-tween genes within a genus were not all caused by recentgene duplications because 296 trans-specific polymorphicsites are present between genes from Zea and Sorghum.The large number of trans-specific polymorphisms may in-dicate that sequence exchanges also played important roleon the homogenization of genes within a genus. The genesfrom different genus, Zea or Sorghum, formed independentclades in the distance tree (fig. 3). The formation of inde-pendent clades for Zea and Sorghum indicated that con-certed evolution occurred for the Rp1 gene family in thegenus level though it was not observed in different Zeaspecies (fig. 3).

Low Rp1 Haplotype Diversity in Cultivated RiceThe Rp1 locus in rice cultivar Nipponbare contains fourgenes, Os01g57270, Os01g57280, Os01g57310, andOs01g57340. Comparison of the four genes and their inter-genic sequences showed that Os01g57310 is a chimeragenerated through a recent unequal crossover betweenOs01g57280 and Os01g57340 (fig. 4B). The 5# (ca. 3.2 kb)of the chimera is identical to Os01g57340 and its 3# (ca.600 bp) is identical to Os01g57280. The unequal crossoverresulted in a duplication of 20,628 bp, and the duplicatedsequence has only four single-nucleotide polymorphisms(SNPs) with its progenitor sequence. Rice cultivar St. No. 1has a similar Rp1 locus to that in Nipponbare: the four pairs ofRp1 homologues in the two cultivars exhibit 100%, 99.7%,99.8%, and 100% nucleotide identities, respectively (Linet al. 2007). The chimera in St. No. 1 is the functional genePi37, which encodes resistance against rice Blast (Lin et al.2007). The resistance gene Pi37 has four SNPs fromOs01g57310, two of which result in change of amino acids.Because the chimera Os01g57310 is identical to its progenitorsequences (see above), the gene Pi37 apparently evolvedfrom Os01g57310 through four point mutations. Partialsequencing the BAC clones spanning the Rp1 locus in ricecultivars ZS97 and ZH11 suggested that these two cultivarsalso have the same Rp1 haplotype as Nipponbare. Screening24 rice cultivars using nine primer combinations indicated

FIG. 3. An NJ distance tree for Rp1 homologues from Zea andSorghum. Pi37 from rice was used as an outgroup. The numbers onnodes are bootstrap values, and values lower than 60 are not shown.Genes with prefix ‘‘Rp’’ are from Zea may ssp. may, with prefix‘‘Zmp’’ are from Z. may ssp. parviglumis, with prefix ‘‘Zl’’ are fromZ. luxurians, and with prefix ‘‘Zd’’ are from Z. diploperennis.

Luo et al. · doi:10.1093/molbev/msq216 MBE

318

that five (including ZS97 and ZH11) cultivars have theNipponbare-like haplotype. The Nipponbare-like haplotypein these cultivars was further supported by their Rp1 homo-logues, which were almost identical to their correspondinggenes in Nipponbare. The only variations of Rp1 sequencesfrom rice cultivars of the Nipponbare-like haplotype are a fewSNPs and a diverse region in genes OjLXJ-2 and OjGYDD-2resulted from a gene conversion.

Compared with the Rp1 haplotype in Nipponbare,rice cultivar 93-11 has a very different Rp1 haplotype.The region spanning Rp1 locus in 93-11was resequenced,resulting in 170,565-bp sequence with two gaps of un-known size. Three full-length and one truncated Rp1 ho-mologues were discovered. The centromeric part of theRp1 locus in rice cultivar 93-11 is similar to that in Nippon-bare, including two Rp1 homologues and their intergenicregion. However, the telomeric part of the Rp1 locusvaries dramatically in these two cultivars, with no similarityin the intergenic region. Screening 24 rice cultivars usingnine PCR primer combinations located at the Rp1 locusshowed that no other rice cultivars have the 93-11–likeRp1 haplotype. In addition to the Nipponbare-like and93-11–like haplotypes, the screening of 24 rice cultivarsusing the 9 primer combinations identified another 2 hap-lotypes, with 4 and 14 cultivars, respectively. The identifi-cation of 4 Rp1 haplotypes among the 24 rice cultivarsbased on the results of 9 PCRs is consistent with thesequencing results of Rp1 homologues in these cultivars(see below).

Differentiation of Type I and Type II Rp1Homologues in OryzaWe obtained 56 full- or nearly full-length Rp1 homologuesfrom 24 rice cultivars and 9 wild Oryza species using severalstrategies (supplementary tables S1 and S2, SupplementaryMaterial online). They were combined with five rice Rp1homologues from GenBank for further analysis. The riceRp1 homologues exhibit 75.6–100% nucleotide identitywith each other. A distance tree was constructed to showtheir sequence relationship (fig. 5). The Oryza Rp1 homo-logues form several distinguishable clades in the tree. Oneclade, marked as Type II in figure 5, contains eight genes,which have only three SNPs in their 3,767 bp sequences.The eight genes from this clade have at most 79.2% nucle-otide identity with other Rp1 homologues. No sequenceexchanges were observed between the eight genes withother paralogues. We conclude that this clade of genes rep-resents a lineage of Type II Rp1 homologues (Kuang et al.2004). This lineage of Type II genes is present in 6 of the 24rice cultivars screened. It is also present in O. rufipogon butnot in other wild Oryza genotypes included in this study.The Type II gene in rice cultivars Nipponbare and 93-11 is inthe same orientation as other genes in the cluster, suggest-ing that gene orientation does not account for the differ-entiation of Type I and Type II Rp1 homologues in Oryza.

The Type II genes are the most divergent from the othergenes among all Rp1 homologues from Oryza species, fol-lowed by those obtained from O. australiensis, O. brachyan-tha, and O. coarctata (fig. 5). All other Rp1 genes cloned

FIG. 4. The chimeric structures of functional resistance genes. (A) Rp1-D from maize. (B) Pi37 from rice. Pi37 is a chimera between Os1g57280and Os1g57340 generated through unequal crossover. The shaded region shows identical sequences. The numbers on top show positions inRp1-D or Pi37.

Table 1. Nucleotide Identities between Rp1 Homologues from Different Species/Genera.

Oryza(%)

Brachypodiumdistachyon (%)

Triticum(%)

Sorghubicolor (%)

Zea(%)

Zea may ssp.may (%)

Z. may ssp.parviglumis (%)

Z. luxurians(%)

Z. diploperennis(%)

Oryza 89.3 74.7 72.8 69.7 69.4 69.4 69.3 69.3 69.4B. distachyon 96.5 76.8 71.8 71.4 71.5 71.3 71.4 71.4Triticum 81.0 69.7 69.4 69.4 69.3 69.5 69.3S. bicolor 91.3 81.9 81.8 81.9 81.8 81.9Z. may ssp. may 92.0 91.7 93.1 92.9Z. may ssp.parviglumis

91.8 92.9 92.8

Z. luxurians 95.7 94.9Z. diploperennis 94.8

Evolution of the Rp1 Family in Poaceae · doi:10.1093/molbev/msq216 MBE

319

from Oryza species exhibit 86.8–100% nucleotide identityand form a clade in figure 5. They have chimeric structures,generated through sequence exchanges among paralogues(fig. 4B). At least 15 sequence exchanges were detectedamong the genes in this clade. Some of the sequence ex-changes occurred before speciation of O. sativa, O. nivara,O. alata, and O. rufipogon. We concluded that these genesevolved in a pattern of Type I R-genes. Therefore, there wasdifferentiation of Type I and Type II R-genes in the Rp1cluster of Oryza.

Single-Copy Rp1 Locus in B. distachyon GenomeUnlike other species, B. distachyon genome has a singlecopy at the Rp1 locus. To better understand the evolutionof single-copy R-gene locus, full-length Rp1 homologues

were amplified from 25 genotypes using primer combina-tion Brach-F and Brach-R (supplementary table S2, Supple-mentary Material online). Sequencing the PCR productsusing primer Brach-R generated clean chromatogramsfor 16 of the 25 genotypes, indicating the presence of a sin-gle gene in these genotypes. The 16 PCR products givingclean chromatograms were sequenced directly. The othernine genotypes showed nearly identical chromatogramsand have double peaks, suggesting multiple fragments inthe PCR products. The PCR products amplified fromone of such genotypes (PB-292) were cloned and se-quenced. Two distinct sequences were obtained. Therefore,a total of 19 Rp1 homologues were obtained from B. dis-tachyon, including one retrieved from the sequenced ge-nome of B. distachyon. The 19 Rp1 genes are groupedinto two major clades (group G1 and G2 in fig. 6). TheG1 group has seven genes, with only seven SNPs andone 29-bp deletion. The G2 group consisted of 12 genes;9 of the 12 G2 genes are identical (G2a), but the other3 (G2b) were divergent from G2a, showing only 98.5%to 99.0% nucleotide identity with each other. The nucleo-tide identities between genes in the clades G1 and G2 varyfrom 91.9% to 93.9%. Primers specific to G1, G2a, and G2bwere designed and used to screen all genotypes of B. dis-tachyon. Consistent with previous cloning and sequencingresults, the PCR amplification showed that there are threedistinct Rp1 homologues (either alleles or paralogues) indifferent genotypes of B. distachyon, and some genotypeshave a single copy, whereas some genotypes have twocopies (fig. 7).

High Divergence of Rp1 Homologues in Wheat andBarleyThe efforts to amplify full- or nearly full-length Rp1 homo-logues fromwheat were partially successful, and only 19 Rp1homologues were cloned from three wheat cultivars. PCRand sequencing results using primer combination For2and Low4, which amplify approximately 2-kb LRR-encodingsequence from the Rp1 family, show that there are manymore Rp1 homologues in wheat genomes. To avoid

FIG. 5. An NJ distance tree for Rp1 homologues from Oryza. Rp1-Dfrom maize was used as an outgroup. The numbers on nodes arebootstrap values, and values lower than 60 are not shown. Geneswith prefix ‘‘Oi’’ are from Oryza sativa ssp. indica, with prefix ‘‘Oj’’are from O. sativa ssp. japonica, with prefix ‘‘Or’’ are from O.rufipogon, with prefix ‘‘On’’ are from O. nivara, with prefix ‘‘Og’’ arefrom O. glaberrima, with prefix ‘‘Op’’ are from O. punctata, withprefix ‘‘Om’’ are from O. minuta, with prefix ‘‘Oal’’ are from O. alta,with prefix ‘‘Oau’’ are from O. australiensis, with prefix ‘‘Oc’’ arefrom O. coarctata, and with prefix ‘‘Ob’’ arse from O. brachyantha.

FIG. 6. An NJ distance tree for Rp1 homologues cloned from 18genotypes of Brachypodium distachyon. The numbers on nodes arebootstrap values, and values lower than 60 are not shown. Theprefix of the genes represents the genotypes from which the geneswere cloned.

Luo et al. · doi:10.1093/molbev/msq216 MBE

320

discrepancy, the 2-kb LRR-encoding sequences instead ofnearly full-length genes were analyzed for the Pooideaesubfamily. Using primer combination For2 and Low4, 27 dis-tinct fragments of Rp1 homologues (TaCS1–TaCS27) wereobtained from wheat cultivar CS. PCR products of 23 ran-domly chosen CS Rp1 homologues were pooled and used asprobe for Southern hybridization to screen 146 CS cytoge-netic stocks including 21 nullisomic–tetrasomic, 24 ditelo-somic, and 101 deletion lines. Among the 14 scorable bandsin the Southern image, the majority (9/14) were missing inN4AT4D, Dt4AS, 1BL1 (4AL4), 2DL9 (4AL4), 3DS6 (4AL11/4AL5), 4AL5, 4AL4, and 6AL2 (4AL4), but all 14 bands arepresent in all other lines (fig. 8A). Based on the principle ofdeletion bin mapping in wheat (Endo and Gill 1996; Penget al. 2004), we conclude that the majority of Rp1 homo-logues in CS are likely located on wheat chromosomearm 4AL and further to the terminal region that covers20% of the long arm of chromosome 4A (4AL4-0.8-1.00,fig. 8B). To confirm above Southern hybridization results,PCR primer combinations specific to 25 genes were success-fully designed and used to screen the CS cytogenetic stocks(fig. 8C); 12 Rp1 homologues were mapped to 4AL and 1wasmapped to 4AS, whereas the other 12 were not mappedeither because of PCR failure or no missing bands in anydeletion lines. Therefore, most of the Rp1 homologuesin CS were located on the terminal region (0.8–1.00) of4AL, consistent with the results of Southern hybridization(fig. 8A–C).

Expressed sequence tag–based comparative mapping in-dicated that wheat chromosome region of 4AL4-0.80-1.00is actually a 7BS segment resulted from a 4AL/7BS recipro-cal translocation (Miftahudin et al. 2004), and this 7BS re-gion (0.59–1.00) is syntenic to rice chromosome 6 (Sorrellset al. 2003; Hossain et al. 2004), which does not have an Rp1locus.

Most Rp1 homologues cloned from CS exhibit low nucle-otide identities among each other, and the nucleotide di-versity of Rp1 homologues in the CS was only 81.1%. Ifthe 12 genes successfully mapped to 4AL in CS (i.e., pa-ralogues in one cluster) are considered only, their nucleotidediversity is still 81.2%, considerably lower than the diversityofRp1 homologues located in theRp1 cluster of Zea (92.5%).

To better understand the evolution of Rp1 in Triticumand its close relatives, we extended our study to two ad-ditional wheat cultivars as well as it close relatives. Twelve,

fifteen, thirteen, three, two, and nine distinct short frag-ments of Rp1 homologues were amplified from wheatcultivars Zhengmai9023, Hua2152, T. durum (Langdon),T. monocum, A. tauschii, and T. urartu, respectively. Theywere combined with the 27 Rp1 homologues in CS, 3 genesfrom barley genotype Morex to construct an NJ distancetree (fig. 9). Another six barley homologues were excludedbecause only NBS-encoding but not LRR-encoding regionswere available for these genes. The tree shows that therewere four ancient lineages of Rp1 homologues in the Pooi-deae subfamily, and all four lineages have representativesfrom wheat. Most Rp1 homologues from cultivar CS belongto one lineage, which were mapped to either 4AL or 4AS.The three barley genes are distributed in three different lin-eages (fig. 9). The fourth lineage with no barley genes hasa nearly full-length Rp1 homologue from wheat cultivar CS(TaCS-6). Comparison of the NBS-encoding sequence ofTaCS-6 with the six barley genes with only NBS-encodingsequences available showed that MorexE (AF414175) wasmost similar to TaCS-6, with 93.5% nucleotide identity be-tween them and likely belongs to lineage 3. Therefore, bothbarley and wheat had highly divergent Rp1 homologues,with members from all four Rp1 lineages in the Pooideaesubfamily. Furthermore, the four barley genes belongingto four different lineages are located on four differentchromosomes (Rostoks et al. 2002).

Discussion

Two Ancient Rp1 Loci in the Grass FamilyThe Rp1 homologues from 21 species in the three majorsubfamilies of the grass family were extensively sampledand analyzed. The sorghum genome, which might havethe ancient Rp1 haplotype in the common ancestor ofthe grass family, has two Rp1 loci (A and B). However,the genomes of rice, B. distachyon, and maize maintain onlyone of them. Interestingly, maize harbors locus A, whereasrice and B. distachyon harbor locus B, indicating indepen-dent deletion of the Rp1 locus in Poaceae (fig. 1). In con-trast to the loss of Rp1 locus, the Rp1 locus might haveduplicated in some species. It was shown previously thatthe Rp1 homologues were present on at least five chromo-somes in barley genome, including Rp1 loci A and B (Ayliffeet al. 2000; Rostoks et al. 2002). The Rp1 homologues fromthese loci are highly divergent from each other and belongto distinct lineages (fig. 9). These lineages also contain ho-mologues from wheat, suggesting that the duplication ofRp1 loci might have occurred before speciation of Triticumand Hordeum. As mentioned above, rice and B. distachyonhave locus A only. Because B. distachyon is closely relatedwith wheat/barley and divergent from rice, it is very likelythat rice and B. distachyon had independent deletions ofRp1 locus B.

Large Variation of Rp1 Copy Number in DifferentSpeciesThe species included in this study vary not only in Rp1 lo-cus number but also in copy number. The Rp1 locus in the

FIG. 7. Brachypodium distachyon has one or two Rp1 homologues ina genome. Three primer combinations specific to three differentRp1 lineages (fig. 6) were used to screen B. distachyon genotypes.Genotype name is shown above each lane.

Evolution of the Rp1 Family in Poaceae · doi:10.1093/molbev/msq216 MBE

321

25 genotypes of B. distachyon has at most two genes. Incontrast, Z. may ssp. parviglumis genotype Ames21889has at least 70 Rp1 homologues. Zea luxurians PI441933likely has a large number of Rp1 homologues in its genomebecause 42 distinct genes were obtained after sequencing50 randomly chosen clones. Though different genotypes ofZ. may have various number of Rp1 homologues, it is ob-vious that the average of Rp1 homologues in one Zea ge-nome is considerably higher than that in B. distachyon.

The large number of Rp1 homologues in Zea might bethe results of unequal crossovers. The Rp1 locus in maizewas shown to be prone to recombination and high fre-quency of unequal crossover was observed experimentally(Hulbert 1997; Sun et al. 2001). The Rp1 locus is locatedapproximately 2–3 Mb from the telomere, and its proxim-ity to telomere might promote structure change of R-genelocus (David et al. 2009). Unequal crossover, on the otherhand, may cause deletions of genes. In contrast to the ex-pansion of the Rp1 locus in Zea, Tripsacum dactyloides,a close relative of Zea, might have had complete deletionof the Rp1 locus. We failed to get any PCR amplificationsfrom two genotypes of T. dactyloides using five primercombinations that are conserved in Zea and Sorghum (datanot shown). Frequent unequal crossovers were also foundin other resistance gene cluster (Van der Hoorn et al. 2001;Kruijt et al. 2004; Nagy and Bennetzen 2008).

High copy number of resistance genes should be selectedfor and maintained in plant genomes, if resistance is con-sidered only. However, R-gene clusters in plant genomesdid not expand unlimitedly. It is possible that large R-genelocus may change chromosome structure and cause ge-nome instability. Furthermore, the presence of an R-genesequence in a genome may decrease its fitness in somecases (Tian et al. 2003; Nagy and Bennetzen 2008). The se-lection mechanisms for high copy number of Rp1 homo-logues in Zea but single copy in B. distachyon remainunknown.

Concerted Evolution of Disease-Resistance GenesFrequent sequence exchanges tend to homogenize a genefamily. Consequently, homologues within a species/genus/family are more similar than those from different species/genus/family, a phenomenon called concerted evolution.Homogenization of disease-resistance genes could decreasea plant’s resistance and any sequence exchanges homoge-nizing an R-gene family may be selected against. It was hy-pothesized that there was ‘‘diversifying’’ selection aftersequence exchanges (Kuang et al. 2004). Newly generatedchimeras with novel resistance function will be selected forand maintained in a population. In contrast, sequence ex-changes homogenizing two genes and reducing resistancespecificity will be selected against. Frequent sequence ex-changes followed by diversifying selections will maintainhigh diversity of the coding sequences of Type I disease-resistance genes though their intron sequences may behomogenized (Kuang et al. 2004, 2005). In fact, therehas been no report of concerted evolution of R-genes inclosely related plant species. In this study, we showed that

FIG. 8. Mapping Rp1 homologues using aneuploid and deletionstocks of wheat genotype CS. (A) Southern blot analysis of CSaneuploid and deletion line using an even mixture of 23 Rp1fragments as probe. The first five lanes show five bands. N4AT4D isa nullisomic line of chromosome 4A; Dt4AS is a ditelosomic lineof short arm of chromosome 4A; 4AL11, 4AL5, and 4AL4 are 4ALdeletion lines. The last five lanes are used as comparison and show14 bands. (B) A bin map of wheat chromosome 4AL showingphysical location of 9 Rp1 homologues. This map was derived fromthe above-mentioned Southern blot analysis. (C) Screening CSditelosomic and deletion lines related to wheat group 4 chromo-somes using PCR primers specific to individual Rp1 homologuecloned from CS. Lanes 1–23 are CS ditelosomic or deletion lines:1: Dt4AL, 2: Dt4BS, 3: Dt4DS, 4: Dt4DL, 5: 4AS-3, 6: 4AS-4, 7: 4AS-1,8: 4AL-12, 9: 4AL-13, 10: 4AL-5, 11: 3DS-6 (4AL-11), 12: 4AL-4, 13:4BS-1, 14: 4BS-8, 15: 4BS-4, 16: 4BL-1, 17: 4BL-5, 18: 4DS-2, 19: 4DS-3,20: 4DS-1, 21: 4DL-9, 22: 4DL-13, 23: 4DL-12. Lane 24 is wide-type CS.

Luo et al. · doi:10.1093/molbev/msq216 MBE

322

there were frequent sequence exchanges between Rp1 ho-mologues within each Zea species. However, such frequentsequence exchanges did not result in concerted evolution,probably due to short divergence time as well as the afore-mentioned diversifying selection after sequence exchanges(Gaut and Clegg 1993; Hilton and Gaut 1998). Similarly,there is no evidence of concerted evolution for the Rp1homologues in Oryza (fig. 5). Like in Zea species, frequentsequence exchanges have occurred between Rp1 homo-logues in S. bicolor, which had evolutionary patterns typicalof Type I R-genes. The Rp1 homologues in Zea and Sorghumformed independent clades in the distance tree (fig. 3),showing obvious concerted evolution. The concerted evo-lution between Sorghum and Zea was likely due to theirlong divergence time (Doebley et al. 1990). Tripsacum di-verged 4.5–4.8 Ma with Zea (Hilton and Gaut 1998), and itwould be interesting to know if there was concerted evo-lution between the Rp1 homologues in these two genera.Unfortunately, we failed to obtain any Rp1 homologuesfrom the two genotypes of T. dactyloides included in thisstudy.

Distinct Evolutionary Patterns of Rp1 Homologuesin Different SpeciesThe Rp1 homologues in Zea and Sorghum exhibit evolu-tionary patterns typical of Type I R-genes. Unlike theRp1 homologues in Zea and Sorghum, the homologuesin wheat and its relatives showed no sign of sequence ex-changes among paralogues. These genomes maintain manydivergent Rp1 homologues, which evolved independently.Apparently, the Rp1 homologues in wheat and barley wereduplicated in the early stage of subfamily Pooideae, andthey might have evolved following the birth-and-deathmodel (Michelmore and Meyers 1998). The Rp1 homo-logues in B. distachyon also evolved independently, but on-ly a single Rp1 homologue is present in B. distachyon. Wehypothesize that the Rp1 homologues in wheat, barley, B.distachyon as well as their close relatives might haveevolved in a pattern of Type II R-genes.

Differentiation of Type I and Type II Genes in RiceBoth Type I and Type II Rp1 homologues were found in theRp1 cluster in Oryza species. The Type II genes in Oryzaspecies were most similar to other Rp1 homologues ob-tained from Oryza species, suggesting that the differentia-tion of Type I and Type II Rp1 homologues occurred afterthe speciation of Oryza. In this study, the Type II gene wasfound in cultivated rice and wild riceO. rufipogon. The TypeII Rp1 homologue was the most divergent member in theRp1 family in Oryza, suggesting that the differentiation ofType I and Type II genes occurred right after the speciationof genus Oryza. It remains to be investigated whether the

FIG. 9. An NJ distance tree for Rp1 homologues from wheat, barley,and their close relatives. The numbers on nodes are bootstrap

values, and values lower than 60 are not shown. Genes with prefix‘‘Ta’’ are from Triticum aestivum, with prefix ‘‘At’’ are from Aegilopstauschii, with prefix ‘‘Tm’’ are from T. monocum, with prefix ‘‘Tu’’are from T. urartu, with prefix ‘‘Td’’ are from T. durum, and withprefix ‘‘Morex’’ are from barley cultivar Morex.

Evolution of the Rp1 Family in Poaceae · doi:10.1093/molbev/msq216 MBE

323

Type II gene is also present in other Oryza species. Exceptthe identified Type II Rp1 gene in Oryza, the three genesobtained from O. australiensis, O. brachyantha, andO. coarctata were the most divergent in the Rp1 family ob-tained from Oryza species. No sequence exchanges weredetected between these genes and the Type I genes inOryza. Their evolutionary patterns remain unclear for lackof enough Rp1 sequences from these species.

Most Rp1 homologues obtained from Oryza showedevolutionary patterns of Type I R-genes. It was proposedthat there are a large number of Type I R-genes in a pop-ulation because each genotypemay harbor a different set ofType I R-genes (Kuang et al. 2004). Surprisingly, only a fewdistinct Rp1 homologues were found after screening 24 ricecultivars. We hypothesize that only few Rp1 haplotypeswere introduced into the cultivated rice through domesti-cation and breeding. The number of distinct Rp1 homo-logues in a natural population of wild rice might belarge. This prediction is partially supported by the relativelyhigh diversity of the 17 Rp1 homologues obtained fromwild rice (fig. 5).

Evolution of the Single-Copy Rp1 Gene inB. distachyonOnly one or two Rp1 homologues were present in the 25 B.distachyon genotypes in our study, which include diploids,tetraploids, and hexaploid. The Rp1 homologues aregrouped into three lineages, G1, G2a, and G2b. All ten dip-loids have only one Rp1 homologue. However, 9 of the 15polyploids have 2 genes. It is possible that the three Rp1homologues in these nine polyploids are located at thesame locus of different genomes. Therefore, the Rp1 locusin B. distachyon might have a single gene. Notice, even inthis case, the three genes should be treated as paraloguesrather than alleles. First, this locus in other species such aswheat has more than one copy of Rp1 homologue, suggest-ing that the progenitor species of B. distachyonmight havemultiple copies of Rp1 paralogues at this locus. The single-copy gene at this locus might be the result of independentdeletions of all other Rp1 paralogues at this locus, respec-tively. Second, the G1 and G2a/b genes exhibit nucleotideidentity of only 93.9–94.1%, which is much lower thana typical nucleotide identity between two alleles. Third,some polymorphic sites (SNPs) between G1 and G2a/b genes were also found in different paralogues in wheat.Thus, we tend to consider the three Rp1 homologues in B.distachyon as paralogues.

The sequencing of plant genomes of model species pro-vides valuable resources for understanding the organiza-tion, structure, and evolution of R-genes. However, ourdata showed that the evolutionary patterns of a resistancegene family might vary dramatically in closely related spe-cies. Great care should be taken when applying R-gene in-formation in a model species to its relatives. Genome-widecomparative analysis of R-genes will further promote ourunderstanding of R-gene evolution, which is critical for con-servation biology, efficient cloning of R-gene, and breedingof resistant crops.

The sequences of Rp1 homologues generated in thisstudy were deposited in GenBank under accession num-bers GU733153–GU733316.

Supplementary MaterialSupplementary tables S1–S3 are available atMolecular Biologyand Evolution online (http://www.mbe.oxfordjournals.org/).

AcknowledgmentsWe thank Dr Meizhong Luo for gifts of several research ma-terials, Dr Jiming Jiang for DNA of T. dactyloides, Dr DongfaSun for wheat cultivars Zhengmai9023 and Hua2152,Dr Sibin Yu for rice cultivars, Dr Qinghua Pan for sharinginformation of Rp1 homologues in rice cultivar St. No. 1.This work was supported by the ‘‘973’’ National Key BasicResearch Program grant no. 2009CB119000, China Trans-genic grant no. 2009ZX08009-045B, National NaturalScience Foundation of China grant no. 30921002 to H.K.and Chinese Academy of Sciences—the Important Direc-tional Program of Knowledge Innovation Engineering grantno. KSCX2-YW-Z-0722 and the ‘‘973’’ National Key BasicResearch Program grant no. 2009CB118300 to J.P.

ReferencesAmmiraju JSS, Luo MZ, Goicoechea JL, et al. (27 co-authors). 2006.

The Oryza bacterial artificial chromosome library resource:construction and analysis of 12 deep-coverage large-insert BAClibraries that represent the 10 genome types of the genus Oryza.Genome Res. 16:140–147.

Ayliffe MA, Collins NC, Ellis JG, Pryor A. 2000. The maize Rp1 rustresistance gene identifies homologues in barley that have beensubjected to diversifying selection. Theor Appl Genet.100:1144–1154.

Bennetzen JL. 2000. Comparative sequence analysis of plant nucleargenomes: microcolinearity and its many exceptions. Plant Cell.12:1021–1029.

Collins N, Drake J, Ayliffe M, Sun Q, Ellis J, Hulbert S, Pryor T. 1999.Molecular characterization of the maize Rp1-D rust resistancehaplotype and its mutants. Plant Cell. 11:1365–1376.

David P, Chen NW, Pedrosa-Harand A, Thareau V, Sevignac M,Cannon SB, Debouck D, Langin T, Geffroy V. 2009. A nomadicsubtelomeric disease resistance gene cluster in common bean.Plant Physiol. 151:1048–1065.

Doebley J, Durbin M, Golenberg EM, Clegg MT, Ma DP. 1990.Evolutionary analysis of the large subunit of carboxylase (rbcL)nucleotide sequence among the grasses (Gramineae). Evolution44:1097–1108.

Endo TR, Gill BS. 1996. The deletion stocks of common wheat.J Hered. 87:295–307.

Freeling M. 2001. Grasses as a single genetic system: reassessment2001. Plant Physiol. 125:1191–1197.

Gale MD, Devos KM. 1998. Plant comparative genetics after 10years. Science 282:656–659.

Gaut BS, Clegg MT. 1993. Molecular evolution of the Adh1 locus inthe genus Zea. Proc Natl Acad Sci U S A. 90:5095–5099.

Goff SA, Ricke D, Lan TH, et al. (55 co-authors). 2002. A draftsequence of the rice genome (Oryza sativa L. ssp. japonica).Science 296:92–100.

Luo et al. · doi:10.1093/molbev/msq216 MBE

324

Hilton H, Gaut BS. 1998. Speciation and domestication in maize andits wild relatives: evidence from the globulin-1 gene. Genetics150:863–872.

Hossain KG, Kalavacharla V, Lazo GR, et al. (46 co-authors). 2004. Achromosome bin map of 2148 expressed sequence tag loci ofwheat homoeologous group 7. Genetics 168:687–699.

Huang S, Sirikhachornkit A, Su X, Faris J, Gill B, Haselkorn R,Gornicki P. 2002. Genes encoding plastid acetyl-CoA carboxylaseand 3-phosphoglycerate kinase of the Triticum/Aegilopscomplex and the evolutionary history of polyploid wheat. ProcNatl Acad Sci U S A. 99:8133–8138.

Hulbert SH. 1997. Structure and evolution of the Rp1 complexconferring rust resistance in maize. Annu Rev Phytopathol. 35:293–310.

Kruijt M, Brandwagt BF, de Wit PJ. 2004. Rearrangements in the Cf-9disease resistance gene cluster of wild tomato have resultedin three genes that mediate Avr9 responsiveness. Genetics 168:1655–1663.

Kuang H, Caldwell KS, Meyers BC, Michelmore RW. 2008. Frequentsequence exchanges between homologs of RPP8 in Arabidopsisare not necessarily associated with genomic proximity. Plant J.54:69–80.

Kuang H, Wei F, Marano MR, et al. (22 co-authors). 2005. The R1resistance gene cluster contains three groups of independentlyevolving, type I R1 homologues and shows substantial structuralvariation among haplotypes of Solanum demissum. Plant J. 44:37–51.

Kuang H, Woo SS, Meyers BC, Nevo E, Michelmore RW. 2004.Multiple genetic processes result in heterogeneous rates ofevolution within the major cluster disease resistance genes inlettuce. Plant Cell. 16:2870–2894.

Lin F, Chen S, Que Z, Wang L, Liu X, Pan Q. 2007. The blastresistance gene Pi37 encodes a nucleotide binding site leucine-rich repeat protein and is a member of a resistance gene clusteron rice chromosome 1. Genetics 177:1871–1880.

Liu J, Liu X, Dai L, Wang G. 2007. Recent progress in elucidating thestructure, function and evolution of disease resistance genes inplants. J Genet Genomics. 34(9):765–776.

Meyers BC, Kozik A, Griego A, Kuang H, Michelmore RW. 2003.Genome-wide analysis of NBS-LRR-encoding genes in Arabidop-sis. Plant Cell. 15:809–834.

Michelmore RW, Meyers BC. 1998. Clusters of resistance genes inplants evolve by divergent selection and a birth-and-deathprocess. Genome Res. 8:1113–1130.

Miftahudin, Ross K, Ma XF, et al. (50 co-authors). 2004. Analysis ofexpressed sequence tag loci on wheat chromosome group 4.Genetics 168:651–663.

Mondragon-Palomino M, Meyers BC, Michelmore RW, Gaut BS.2002. Patterns of positive selection in the complete NBS-LRRgene family of Arabidopsis thaliana. Genome Res. 12:1305–1315.

Nagy ED, Bennetzen JL. 2008. Pathogen corruption and site-directedrecombination at a plant disease resistance gene cluster.Genome Res. 18:1918–1923.

Paterson AH, Bowers JE, Bruggmann R, et al. (45 co-authors). 2009.The Sorghum bicolor genome and the diversification of grasses.Nature 457:551–556.

Peng JH, Zadeh H, Lazo GR, et al. (30 co-authors). 2004. Chromosomebin map of expressed sequence tags in homoeologous group 1 ofhexaploid wheat and homoseology with rice and Arabidopsis.Genetics 168:609–623.

Ramakrishna W, Emberton J, Ogden M, SanMiguel P, Bennetzen JL.2002. Structural analysis of the maize Rp1 complex revealsnumerous sites and unexpected mechanisms of local rearrange-ment. Plant Cell. 14:3213–3223.

Rostoks N, Zale JM, Soule J, Brueggeman R, Druka A, Kudrna D,Steffenson B, Kleinhofs A. 2002. A barley gene family homologousto the maize rust resistance gene Rp1-D. Theor Appl Genet. 104:1298–1306.

Sawyer S. 1989. Statistical tests for detecting gene conversion. MolBiol Evol. 6:526–538.

Schnable PS, Ware D, Fulton RS, et al. (167 co-authors). 2009. TheB73 maize genome: complexity, diversity, and dynamics. Science326:1112–1115.

Shen J, Araki H, Chen L, Chen JQ, Tian D. 2006. Unique evolutionarymechanism in R-genes under the presence/absence polymor-phism in Arabidopsis thaliana. Genetics 172:1243–1250.

Smith SM, Hulbert SH. 2005. Recombination events generatinga novel Rp1 race specificity. Mol Plant Microbe Interact. 18:220–228.

Smith SM, Pryor AJ, Hulbert SH. 2004. Allelic and haplotypicdiversity at the Rp1 rust resistance locus of maize. Genetics167:1939–1947.

Sorrells ME, La Rota M, Bermudez-Kandianis CE, et al. (39 co-authors). 2003. Comparative DNA sequence analysis of wheatand rice genomes. Genome Res. 13:1818–1827.

Sun Q, Collins NC, Ayliffe M, Smith SM, Drake J, Pryor T, Hulbert SH.2001. Recombination between paralogues at the Rp1 rustresistance locus in maize. Genetics 158:423–438.

Tamura K, Dudley J, Nei M, Kumar S. 2007. MEGA4: molecularevolutionary genetics analysis (MEGA) software version 4.0. MolBiol Evol. 24:1596–1599.

Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG.1997. The CLUSTAL_X windows interface: flexible strategies formultiple sequence alignment aided by quality analysis tools.Nucleic Acids Res. 25:4876–4882.

Tian D, Traw MB, Chen JQ, Kreitman M, Bergelson J. 2003. Fitnesscosts of R-gene-mediated resistance in Arabidopsis thaliana.Nature 423:74–77.

Van der Hoorn RA, Kruijt M, Roth R, Brandwagt BF, Joosten MH, DeWit PJ. 2001. Intragenic recombination generated two distinct Cfgenes that mediate AVR9 recognition in the natural populationof Lycopersicon pimpinellifolium. Proc Natl Acad Sci U S A.98:10493–10498.

Vogel JP, Garvin DF, Mockler TC, et al. (174 co-authors). 2010.Genome sequencing and analysis of the model grass Brachypo-dium distachyon. Nature 463:763–768.

Evolution of the Rp1 Family in Poaceae · doi:10.1093/molbev/msq216 MBE

325