The resistance (R) gene Pi37, present in the rice cultivar St. No. 1, was isolated by an in silico map-based cloning procedure. The equivalent genetic region in Nipponbare contains four nucleotide binding site–leucine-rich repeat (NBS–LRR) type loci. These four candidates for Pi37 (Pi37-1, -2, -3, and -4) were amplified separately from St. No. 1 via long-range PCR, and cloned into a binary vector. Each construct was individually transformed into the highly blast susceptible cultivar Q1063. The subsequent complementation analysis revealed Pi37-3 to be the functional gene, while -1, -2, and -4 are probably pseudogenes. Pi37 encodes a 1290 peptide NBS–LRR product, and the presence of substitutions at two sites in the NBS region (V239A and I247M) is associated with the resistance phenotype. Semiquantitative expression analysis showed that in St. No. 1, Pi37 was constitutively expressed and only slightly induced by blast infection. Transient expression experiments indicated that the Pi37 product is restricted to the cytoplasm. Pi37-3 is thought to have evolved recently from -2, which in turn was derived from an ancestral -1 sequence. Pi37-4 is likely the most recently evolved member of the cluster and probably represents a duplication of -3. The four Pi37 paralogs are more closely related to maize rp1 than to any of the currently isolated rice blast R genes Pita, Pib, Pi9, Pi2, Piz-t, and Pi36.
BLAST, caused by the filamentous ascomycete Mangnaporthe grisea (Hebert) Barr, is one of the most devastating of rice diseases (Ou 1985). The rice/M. grisea combination has been developed into a well-established host–pathogen model (Velent 1990; Jia et al. 2000; Soderlund et al. 2006), particularly as many of the interactions between host resistance (R) and pathogen avirulence (Avr) genes can be satisfactorily explained by the classical gene-for-gene hypothesis (Flor 1971; Jia et al. 2000). Over 50 major rice blast R genes have been described in the literature (Chen et al. 2005; Liu et al. 2005), and 7 of these (Pib, Pita, Pi9, Pid2, Pi2, Piz-t, and Pi36) have now been isolated (Wang et al. 1999; Bryan et al. 2000; Chen et al. 2006; Qu et al. 2006; Liu et al. 2007). Six of the 7 belong to the nucleotide binding site–leucine-rich repeat (NBS–LRR) class of R gene, as they encode a protein carrying both a nucleotide binding site and a leucine-rich repeat domain. The exception, Pid2, encodes a receptor-like kinase protein.
NBS–LRR R genes are the commonest type of resistance gene (Hammond-Kosack and Jones 1997; Bai et al. 2002). The NBS domain contains three short amino-sequence motifs, a kinase-1a or P-loop (phosphate-binding loop), kinase-2, and kinase-3, and is thought to be involved in signal transduction (Traut 1994; Dangl and Jones 2001). The LRR region plays a critical role in the determination of resistance specificity (Parker et al. 1997; Meyers et al. 1998). The xxLxLxx motif within the LRR domain is predicted to form a β-strand/β-turn structure, allowing the variable residues to interact with the pathogen Avr gene product (Hammond-Kosack and Jones 1997; Jones and Jones 1997); it is these residues that are most subject to diversity selection (Parniske et al. 1997; Ellis et al. 2000; Sun et al. 2001; Zhou et al. 2006).
The analysis of R genes isolated from various host species has revealed that most are set within a complex locus composed of multiple copies of closely related genes. Prominent examples are the maize rp1 cluster (Saxena and Hooker 1968; Collins et al. 1999), the barley Mla cluster (Wei et al. 2002), the wheat Pm3 cluster (Yahiaoui et al. 2004), the flax L gene cluster (Shepherd and Mayo 1972; Islam et al. 1989), and the rice Xa26 and Pi9 clusters (Sun et al. 2004; Qu et al. 2006). The rice and Arabidopsis thaliana genome sequences have shown that the majority of NBS–LRR genes occur within tandem arrays (Bai et al. 2002; Meyers et al. 2003). This characteristic clustering of R genes has been proposed to facilitate the evolution of novel resistance specificities via recombination or gene conversion (Hulbert 1997), with some well-characterized examples at the flax L locus and the maize rp1 locus (Ellis et al. 1999; Smith and Hulbert 2005). The identification and isolation of both host R and pathogen Avr genes will serve to clarify many of the molecular mechanisms underlying specific host–pathogen recognition in plants, and a detailed understanding of gene organization within R gene clusters will help in the interpretation of the evolution of these complex loci.
The rice cultivar St. No. 1 confers partial resistance to Japanese and complete resistance to Chinese isolates of blast (Ezuka et al. 1969a,b; Yunoki et al. 1970; Chen et al. 2005). Much of this resistance is due to the presence of Pi37 (Chen et al. 2005). In this article, we describe the in silico map-based cloning of this gene, which is located in a gene cluster set in a recombination-suppressed region.
MATERIALS AND METHODS
Candidate gene cloning:
The gene annotation programs FGENESH (http://www.softberry.com) and RiceGAAS (http://ricegaas.dna.affrc.go.jp/rgadb/) were used to identify candidates for Pi37 within the Nipponbare genomic sequence defined by the flanking markers RM543 and FPSM1 (Figure 1). These candidate sequences, including both their promoter and terminator, were amplified from the genomic DNA of St. No. 1 by the long-range PCR (LR-PCR) procedure described elsewhere (Liu et al. 2007). Necessary primer sequences and restriction enzymes are listed in Table 1. PCR products were purified by agarose gel electrophoresis and inserted into the SalI site of the binary vector pCAMBIA1300 to form constructs R37L1CAM, R37L2CAM, R37L3CAM, and R37L4CAM. All clones were validated by sequencing.
Constructs containing a single candidate gene were transformed into Agrobacterium tumefaciens strain EHA105 by electroporation (GenePluser Xcell TM, Bio-Rad, Hercules, CA). Stability of the constructs was checked as previously described (Qu et al. 2003), and the constructs were then individually transformed into the highly blast-susceptible rice cultivar Q1063, following the methods elaborated by Hiei et al. (1994). The reaction to blast infection of the primary transgenics (T0) and their progeny (T1 and T2) was tested by artificial inoculation with isolate CHL1159 (Pan et al. 2003; Chen et al. 2005). The Pi37 donor St. No. 1 and the susceptible recipient Q1063 were used as controls for the efficacy of the pathological experiment. Transgene copy number in a number of blast resistant T0 plants was assessed by Southern hybridization, as described previously (Liu et al. 2007). The presence of the transgene was also verified by cleaved amplified polymorphic sequence (CAPS) markers, using the primer pair 37CDSF and 37CDSR and digested by EcoRI. Patterns of transgene segregation and the association between the presence of the transgene and resistance to blast infection were studied in the T1 and T2 generations derived from single transgene copy T0 individuals (supplemental Table S1 at http://www.genetics.org/supplemental/).
Rapid amplification of cDNA ends (RACE) was conducted using the GeneRacer kit (Invitrogen, Groningen, The Netherlands), following the manufacturer's instructions. Total leaf RNA was extracted 24 hr after infection from both St. No. 1 and the highly blast-susceptible Lijiangxintuanheigu (LTH). Primers for the first round of amplification of the 5′ RACE were GS1 and the GeneRacer 5′ primer. A 50-fold dilution of the resulting PCR product served as template for the second round of amplification, using as primers GS2 and GeneRacer 5′. The 3′ RACE employed GS3 in combination with the GeneRacer 3′ primer. A mediate RT–PCR fragment was amplified by primers GS4 and GS5—this overlaps the 5′ RACE and 3′ RACE fragments (Figure 3A and supplemental Table S1 at http://www.genetics.org/supplemental/). The RACE products were inserted into pGEM-T (Promega, Madison, WI) for sequencing. Allelic variants for the coding sequences were derived from St. No. 1, LTH, Q1063, and Nipponbare (Table 2). Sequence similarities were calculated using the Matcher program (http://bioweb.pasteur.fr/seqanal/interfaces/matcher.html), while TSSP and POLYAH (http://www.softberry.com/berry.html) were used to identify the promoter and polyadenylation regions. Protein sequence homology was derived from a BLASTP analysis (Altschul et al. 1997). Multiple sequence alignments and phylogenetic analysis were conducted using MEGA 3.1 (http://www.megasoftware.net/mega.html). Theoretical isoelectric points (pI) and protein molecular weights were computed by the DNAstar software package.
Gene expression analysis:
For semiquantitative RT–PCR analysis, total RNA was isolated with the TRIzol reagent (Invitrogen, Carlsbad, CA) from 250 mg of seedling (three- to four-leaf stage) leaf of St. No. 1 (Pi37) and LTH collected 0, 24, and 48 hr after inoculation with isolate CHL1159. RT–PCR was carried out in two steps: briefly, ∼1 μg total RNA was reverse transcribed by SuperScript III RT (Invitrogen), and a 1-μl aliquot of the RT reaction used as template for the subsequent PCR. Primers GS4 and GS5 were used as Pi37 gene-specific primers (supplemental Table S1 at http://www.genetics.org/supplemental/; Figure 3A). Primers Actin1F and Actin1R were used as an internal control (supplemental Table S1). The RT–PCR was initiated with 1 cycle at 94°/3 min, followed by cycling at 94°/30 sec, 62°/60 sec, and 72°/90 sec. A sample was removed from the thermocycler every 3 cycles between the 23rd and the 35th cycles. Equal volumes of these PCRs were electrophoresed through a 1.5% agarose gel for product quantification. The RT–PCR products were also ligated into pGEM-T for sequence validation.
Subcellular location of Pi37:
The deduced Pi37 peptide sequence was subjected to subcellular location prediction using WoLF POSRT (http://wolfpsort.org/). The domain containing a subcellular signal was amplified by primers Gfp371F and Gfp371R (supplemental Table S1 at http://www.genetics.org/supplemental/), containing the NotI and NocI sites (underlined in supplemental Table S1). After digestion, the PCR fragment was ligated in frame to the C terminus of the eGFP coding region of pUC18 and expressed under the control of the CaMV 35S promoter. The constructs (1 μg) were coated on 1.1-mm diameter gold beads and shot into onion epidermal cell layers by a pneumatic particle gun (PDS-1000/He, Bio-Rad). Bombardment conditions were 128 in. Hg vacuum, 1300 psi He, target distance 6 cm). The cells were then cultured on MS medium for 24 hr at 22° and observed by a confocal microscopy (TCS SP2, Leica, Wetzlar, Germany) with a filter set providing 455–490 nm excitation and emission above 507 nm.
Identification of candidate genes for Pi37:
The location of Pi37 has been defined by recombinational analysis as lying between the two microsatellite loci RM543 and FPSM1 (separated by, respectively, 10 and 1 recombinants) and cosegregating with RM302, FPSM4, RM212, FSTS4, S15628, FSTS1, and FSTS3 (Figure 1A; Chen et al. 2005). In Nipponbare, this 374-kb region is covered by the four BAC clones P0490D09, P0413G02, P0010B10, and B1100D10 (Figure 1B), and contains 118 predicted genes. BLAST analysis of these genes identified 4 as having an NBS–LRR structure, clustering within a 55-kb interval (from 22,313 to 75,167 bp) on B1100D10 (Figure 1C). On the basis that the majority of R genes are in the NBS–LRR class, these 4 genes, designated Pi37-1, -2, -3, and -4 (Figure 1C), were considered to be the likeliest candidates for Pi37.
Isolation of candidate genes:
Four primer pairs were designed from the Nipponbare sequence, including the necessary restriction sites and cloning protection base (Table 1). LR–PCR products of expected size 7.3, 9.3, 7.3, and 7.0 kb were successfully amplified, and these were ligated to form, respectively, the constructs R37L1CAM, R37L2CAM, R37L3CAM, and R37L4CAM. To guard against potential PCR-based artifacts, two independent LR–PCRs were conducted for each of the four candidate genes, and two to three clones per construct were sequenced. Sequence comparisons showed that Pi37-1 and -4 in St. No. 1 share identical sequence with their equivalents in Nipponbare, while St. No. 1 Pi37-2 is 99.7 and -3, 99.8% homologous to their Nipponbare equivalents (supplemental Table S2 at http://www.genetics.org/supplemental/). Thus Pi37-2 and -3, but not -1 or -4, remained as potential candidates for Pi37.
Complementation analysis of the candidate genes:
To carefully confirm the candidates deduced from the reference sequence information, all four constructs, which carry each candidate gene, were individually transformed into the highly susceptible cv. Q1063. A total of 6, 68, 132, and 39 independent primary transformants were generated using, respectively, R37L1CAM, R37L2CAM, R37L3CAM, and R37L4CAM. When infected with blast isolate CHL1159, all the transgenic plants involving R37L1CAM, R37L2CAM, or R37L4CAM were highly susceptible, but 24 out of 132 R37L3CAM transformants were resistant. Three of these (LZ75, LZ76, and LZ85) were genotyped by Southern blotting. One copy of Pi37-3 was present in LZ76 and two in both LZ75 and LZ85 (supplemental Figure S1 at http://www.genetics.org/supplemental/). Monogenic inheritance of the Pi37-3 transgene was displayed among the progeny of LZ76-4, which produced a segregation ratio between resistant and susceptible not significantly different from 3:1 (28 resistant and 6 susceptible progeny, χ2 = 0.84, P < 0.30). These progeny could also be used to demonstrate that perfect cosegegration between reaction to blast infection and the presence of the allele-specific marker generated by primer pairs 37CDSF and 37CDSR (Figure 2B). Thus Pi37-3 is a strong candidate for Pi37.
Molecular characterization of Pi37:
The full-length St. No. 1 Pi37 cDNA was obtained by RT–PCR and RACE–PCR and was compared to the genomic sequence. The gene contains a 197-bp 5′ and a 603-bp 3′ untranslated region (UTR) and two introns. The first intron is 3943-bp long and is positioned within the 5′ UTR, ending 23 bp upstream of the ATG start codon. The second intron is 124-bp long, and is positioned within the 3′ UTR, starting 39 bp downstream of the TGA stop codon. The transcript length of Pi37 is 3873 bp (Figure 3A). The deduced Pi37 open reading frame encodes a 1291-residue polypeptide with an estimated molecular weight of 147 kDa, and a pI of 5.98. The N-terminal section contains three typical NBS family motifs (Van Der Biezen and Jones 1998), specifically GGAGKS (beginning at residue 222), LLVLDDV (beginning at residue 297), and GSRVLVTSRR (beginning at residue 327). These correspond to, respectively, the kinase 1a (P-loop), the kinase 2, and the kinase 3a consensus motifs (Traut 1994; Grant et al. 1995). The C-terminal region of the protein is composed of 25 irregular LRRs between residues 590 and 1290 (Figure 3B).
A full-length cDNA was also generated and sequenced from the susceptible LTH. Sequence comparison analysis revealed no structural differences between the alleles (data not shown), but there were nine single-nucleic-acid substitutions present, which generated six amino acid differences between the alleles (Table 2). From additional cDNA sequences obtained from Q1063 and Nipponbare, and from Pi37-4 (see below), it was possible to conclude that all the susceptible genotype Pi37 gene products share two residue differences relative to the sequence of the resistant-type product. These are V239A (a valine at position 239 replacing an alanine) and I247M (an isoleucine at position 247 replacing a methionine) (Table 2; Figure 3B).
Expression analysis of Pi37:
After 23 RT–PCR cycles, the amplicon was barely detectable, but it became readily detectable from 26 cycles onward (Figure 4). The expression level of the resistant allele appears to be higher than that of the susceptible one, and there was some evidence for induction of expression during the 48 hr following inoculation. Overall, Pi37 appears to be constitutively expressed, although its expression level was somewhat promoted in the presence of the pathogen. The C terminus of the predicted amino acid sequence of the Pi37 product includes a cytoplasmic subcellular localization signal domain. Transient expression in onion epidermal cells was employed to confirm this prediction in vivo. This experiment showed that the eGFP–Pi37 C-terminus fusion protein was distributed throughout the cytoplasm, but not in the nucleus (Figure 5). (Note that the GFP-only control signal was evenly distributed throughout the cell.)
Evolutionary analysis of the Pi37 gene cluster:
A BLAST analysis of the Nipponbare sequence identified two duplications (33,911–54,539 and 54,540–75,167) within the overall cluster (Figure 1C). The duplicated segment is composed of two elements—a gene member (Pi37-3 or -4) and an intergenic sequence (Interseq1 or Interseq2), with a nucleotide identity, respectively, 98.8 and 99.1% (a part of these data is given in supplemental Table S2 at http://www.genetics.org/supplemental/). A YhyA-like transposon is present in each intergenic sequence (data not shown). The sequence identity between the four Pi37 candidates ranged from 78.4 (-1 vs. -2) to 98.8% (-3 vs. -4) (supplemental Table S2). This suggests that Pi37-3 was recently evolved from -2, which in turn was derived from -1, while -4 is probably the youngest gene, emerging as a duplication of -3.
A wider BLAST search showed that the Pi37 sequence shares 59% identity with rp1, which confers resistance to rust in maize (Sun et al. 2001). The other characterized rice blast R gene (Pib, Pita, Pi9, Pi2, Piz-t, and Pi36) product sequences were also included in the phylogenetic analysis. Altogether, these fifteen R proteins can be grouped into five clades: rp1-dp1/rp1-dp3/rp1-dp8/rp1-kp3/rp1-dp7/rp1-kp1 (I), Pi37-1/Pi37-2/Pi37-3/Pi37-4 (II), Pita/Pi36 (III), Pi9/Pi2/Piz-t (IV), and Pib (V) (Figure 6). This analysis proposes that the four Pi37 paralogs belong to three clades and are more closely related to the maize rp1 complex than to any of the currently characterized rice blast R genes.
The function of Pi37:
Blast resistance in rice is commonly categorized into qualitative (complete) and quantitative (partial) (Yunoki et al. 1970). The latter describes an incomplete form of generally race nonspecific resistance under multigenic control (Simmonds 1991), although race-specific effects are also known [e.g., Pif (Yunoki et al. 1970), Pb1 (Fujii et al. 2000), pi21 (Fukuoka and Okuno 2001), and Pi34 (Zenbayashi-Sawata et al. 2006)]. The cultivar St. No. 1 shows complete resistance to many Chinese blast isolates, and it is now understood that this resistance is based not only on Pif, but also on Pi37 and perhaps other, as yet uncharacterized genes (Chen et al. 2005). Pi37 belongs to the NBS–LRR class of R genes, and its functionality depends on the identity of two residues (V239A and I247M) in the NBS region, located between the kinase 1a (P-loop) and kinase 2. The NBS region of certain R genes has been shown to be involved in intramolecular interactions with other domains of the protein, or in interactions with other proteins (Mestre and Baulcombe 2006). Since the two critical Pi37 substitutions ensure the complete resistance of St. No. 1, they are presumably involved in AvrPi37 recognition, an avirulence factor carried by Chinese isolates, but presumably lacking in Japanese isolates. Further gene recombination and in vitro binding analysis will be necessary to provide a detailed molecular explanation for the major effect of such a small sequence difference (Jia et al. 2000; Ellis et al. 2007).
The Pi37 protein shares more sequence homology with the products derived from the rp1 complex than with any of the other R genes characterized to date. This maize complex gene confers race-specific resistance to Puccinia sorghi (Saxena and Hooker 1968; Collins et al. 1999). The order and arrangement of alleles at this complex has been associated with the generation of novel race specificities (Richter et al. 1995), and intralocus recombinants have also been identified which confer apparently race nonspecific partial resistance. The complete dominance of the race-specific rp1 alleles is thought to be conferred by a highly efficient molecular recognition of the elicitor (Hulbert 1997). A change from partial to complete resistance may therefore be generated by quite a minor sequence change in the part of the R gene which affects the interaction with its corresponding Avr gene. Many of the important grass species carry sequences homologous to rp1 (Ayliffe et al. 2000). However, only a few of these orthologs have a demonstrated function. Although Pi37 shares homology with rp1 and confers race-specific resistance in rice to the blast fungus, it remains to be seen whether it has any effect in a heterologous situation (Elliott et al. 2002; Christensen et al. 2004; Huang et al. 2007).
The structure of Pi37:
Comparative studies have shown that intron positions are highly conserved over long evolutionary periods (Roy et al. 2003; Roy and Penny 2006). This applies to the conserved NBS region in cereal NBS–LRR genes (Bai et al. 2002). The six characterized rice blast NBS–LRR R genes Pita, Pib, Pi9, Pi2, Piz-t, and Pi36 all carry introns in their coding region. Pita has an intron in the beginning of its kinase-2 motif, one of the commonest positions for introns in cereal NBS–LRR genes (Bryan et al. 2000; Bai et al. 2002). Pib has a complex structure in which two tandemly repeated partial NBS regions carry two introns between the RNBS-B and the GLPL motif (Wang et al. 1999; Bai et al. 2002). Pi9, Pi2, and Piz-t all contain two introns, one upstream of the NBS domain and one downstream of the LRR region (Qu et al. 2006). The Pi36 coding region is interrupted by four introns in the NBS and LRR domains (Liu et al. 2007). Of the four paralogs in the Pi37 cluster, only Pi37-1 has any introns (this article, data not shown). This copy is the most outlying of the four paralogs (supplemental Table S1 at http://www.genetics.org/supplemental/). Pi37 thus appears be the first representative of a cereal NBS–LRR gene lacking an intron.
The genomic organization of the Pi37 gene cluster:
The Pi37 gene cluster has been located in a recombination-suppressed 374-kb region, flanked by regions showing enhanced recombination (Chen et al. 2005). This pattern is reminiscent of the maize rp1 gene complex. Suppression of recombination has also been noted in some other R gene regions, such as Mi (Van Daelen et al. 1993), Mla (Wei et al. 1999), Pita2 (Nakamura et al. 1997), Pi-CO39(t) (Chauhan et al. 2002), Pi5 (Jeon et al. 2003), and I loci (Vallejos et al. 2006). Recombination frequency is severely reduced in the hemizygous state (Ozias-Akins et al. 1998; Goel et al. 2003). The Pi37 region lies within a segment introgressed from indica into japonica rice (Yunoki et al. 1970), and the dominant mode of inheritance of the four markers (see FPSM4, FSTS4, FSTS1, and FSTS3 in Figure 1A) cosegregating with Pi37 is suggestive that hybrids between Pi37 carriers and noncarriers may well be effectively hemizygous for the introgression segment (Chen et al. 2005; Vallejos et al. 2006). Thus the absence of localized recombination between resistant and susceptible haplotypes may have driven the evolution of diversity at this R gene locus. In the “trench warfare” model for the evolution of the host–pathogen interaction (Stahl et al. 1999), natural selection maintains a dynamic equilibrium between susceptible and resistant alleles in a population but at a cost to fitness. When the relevant pathogens or even pathogen races are absent, the maintenance of specific R genes can therefore confer an evolutionary cost (Tian et al. 2002, 2003; Zhou et al. 2004). While the existence of the Pi37 gene cluster may allow for a greater rate of novel R gene generation via intragenic recombination, the maintenance of the R gene might therefore become a liability to the host. As a result, suppression of recombination at the Pi37 locus may represent an adaptive mechanism supporting balancing selection between resistant and susceptible alleles.
The Pi37 gene cluster contains four NBS–LRR genes, arranged in tandem in the same orientation. The complex shares substantial homology to the maize rp1 complex (Peng et al. 1999; Ayliffe et al. 2000). rp1 haplotypes vary in the number of homologs present, ranging from as few as 1 to as many as 50 (Monosi et al. 2004). On the basis of sequence comparisons, the rp1 homologs of maize, sorghum, and barley are thought to all have arisen independently from a single rp1-like gene present in the common grass ancestor (Ramakrishna et al. 2002). Pi37 may well have evolved from the same common rp1-like gene ancestor. The simplest predicted series of events leading the present status of the Pi37 locus starts with an ancient triplication of a single copy Pi37-1 gene to give rise to Pi37-2 and -3, and this was followed later by a second duplication event in which the Pi37-3 sequence gave rise to the present day -3 and -4. A more detailed analysis of the intergenic sequences separating the Pi37 members in various haplotypes should provide evidence to test this model and will allow for further elucidation of the evolution of this R gene cluster.
In conclusion, we have applied the in silico map-based cloning method to successfully isolate the functional gene Pi37 and have characterized the R gene cluster in which it lies. This general approach is proving to be an effective means for the genetic dissection of gene complexes in recombination-suppressed regions.
This research was supported by grants from the National 973 project (2006CB1002006), the National 863 projects (2006AA10A103; 2006AA100101), the National Natural Science Foundation (30570994), the Innovation Research Team Project from the Ministry of Education of China (IRT0448), and the Guangdong Provincial Natural Science Foundation (039254).
Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession no. DQ923494.1.
Communicating editor: A. Paterson
- Received August 15, 2007.
- Accepted September 4, 2007.
- Copyright © 2007 by the Genetics Society of America