Correspondence: Teruo Sone, Research Faculty of Agriculture, Hokkaido University, Kita-9 Nishi-9 Kita-ku, Sapporo 060-8589, Japan. Tel.: +81 11 706 2502; fax: +81 11 706 4961; e-mail: firstname.lastname@example.org
AVR-Pia, an avirulence gene in the genome of the rice blast fungus Magnaporthe oryzae, triggers a hypersensitive reaction in rice cultivars harbouring the resistance gene Pia. The copy number of AVR-Pia was revealed to vary from one to three among M. oryzae isolates avirulent to Pia rice, and three copies of the gene were located on a single chromosome in strain Ina168, from which the gene was originally cloned. The spontaneous avr-Pia mutant originated from Ina168, named Ina168m95-1, which lacks the AVR-Pia gene, and was therefore used to elucidate the molecular mechanism of the deletion of all three copies of AVR-Pia. Screening and analysis of cosmid clones indicated that two copies of the DNA-type transposon Occan (Occan9E12 and Occan3A3) were located on the same chromosome, and three copies of AVR-Pia were located in between the two Occan elements. Ina168m95-1 contains a conserved Occan element, named Occanm95-1, between sequences homologous to the 5′-flanking region of Occan3A3 and the 3′-flanking region of Occan9E12. In addition, sequence polymorphisms indicated a homologous recombination between Occan3A3 and Occan9E12, which resulted in Occanm95-1. Based on these observations, we propose the hypothesis that homologous recombination in the two Occan elements leads to the deletion of AVR-Pia in Ina168m95-1.
Rice blast, which is caused by infection with the rice blast fungus Magnaporthe oryzae, is the most devastating disease of rice. Magnaporthe oryzae is an ascomycete, the imperfect stage of which is called Pyricularia oryzae (Dean et al., 2005). One of the major controlling strategies for the disease is utilization of appropriate resistant cultivars that have major resistance (R) gene(s) against rice blast. The effect of an R gene is based on direct or indirect interaction with the corresponding avirulence (AVR) gene in the pathogen genome, known as the gene-for-gene theory (Flor, 1971).
Mutations in AVR genes lead to the breakdown of resistance mechanisms controlled by R genes. Many cases of such breakdown were reported for Japanese blast resistant rice cultivars during the 1970s (Kiyosawa, 1982). The breakdown of a resistance gene motivated the introduction of a new R gene, only to be followed by the breakdown of this new gene. This kind of cat-and-mouse game between rice breeder and pathogen enriched the pathogen variety, thereby broadening the spectrum against host R genes in the field. For the sustainable use of the R genes, knowledge of the mutational mechanisms of AVR genes in the pathogen is necessary.
Thus far, a number of AVR genes in M. oryzae have been cloned and their mutations analysed. AVR-Pita1 is known for its instability, resulting in diverse mutational mechanisms at this locus, including deletion and translocation of the gene, transposon insertion into the gene and promoter sequence, and point mutations (Orbach et al., 2000; Kang et al., 2001; Khang et al., 2008; Takahashi et al., 2010; Chuma et al., 2011). Insertion of a mixed interspersed nuclear element (MINE) was found in the ACE1 avirulence gene (Fudal et al., 2005). Insertion of a Pot3 element and point mutations was shown to be responsible for the loss of function of AVRPiz-t (Li et al., 2009). AVR-Pia, AVR-Pii and AVR-Pik have all been cloned recently (Miki et al., 2009; Yoshida et al., 2009). AVR-Pia and AVR-Pii showed an all-or-nothing manner of conservation among field isolates, indicating the occurrence of gene deletion. As shown above, gene deletion is one of the major mechanisms of AVR gene mutation; however, the molecular details of this mechanism have not yet been clarified. Spontaneous deletion mutant strains, rather than field isolates that are missing the corresponding AVR gene, are necessary to understand these details. Comparison of the AVR gene locus between a mutant and its parental strain should provide a clear image of the deletion process. In the cloning strategy for AVR-Pia by Miki et al. (2009), a spontaneous virulent mutant strain, Ina168m95-1, originating from field isolate Ina168, was utilized. In that study, the deletion of AVR-Pia caused the loss of avirulence towards Pia rice. However, frequent appearances of repetitive sequences in the flanking region of the AVR gene inhibited the identification of the deleted region.
In this study, further characterization of the AVR-Pia locus in Ina168 and its spontaneous deletion mutant was attempted by sequencing analysis, to obtain a key for the elucidation of the deletion mechanisms of AVR-Pia gene.
Materials and methods
Fungal and bacterial strains and DNA library
Magnaporthe oryzae isolate Ina168, the mutant Ina168m95-1 (Kito et al., 2003) and seven field isolates (Table 1) were used in this study. These strains were stored at −20 °C as dried mycelia on pieces of filter paper and were inoculated onto appropriate medium before use. Escherichia coli strains TOP10 and DH10B (Invitrogen, Carlsbad, CA) were used for recombinant DNA experiments and for the maintenance of the cosmid library, respectively. A genomic library of strain Ina168 based on cosmid pMOcosX (Kito et al., 2003) was used. Cosmids used in this study were screened by the sib selection method (Miki et al., 2009).
Table 1. AVR-Pia genotype of Magnaporthe oryzae strains used in this study
DNA extraction, restriction analysis and Southern hybridization
Fungal DNA was extracted using the method described by Sone et al. (1997). Plasmid and cosmid DNA was extracted using Quantum Prep Plasmid Miniprep or Maxiprep Kits (Bio-Rad, Hercules, CA). Restriction enzyme digestion of DNA was performed according to the instructions of the enzyme manufacturers. Capillary blotting for Southern hybridization was performed using Hybond-N+ (GE Healthcare, Buckinghamshire, UK), as described in the manufacturer's instructions. Labelling and detection of DNA probes were performed with the AlkPhos Direct Nucleotide Labeling and Detection System (GE Healthcare). Computer analyses of DNA sequences, including ORF analysis, were performed using Genetyx-MAC software (Genetyx, Tokyo, Japan).
The sequence of cosmid 9E12 was analysed by a combination of two attempts of pyrosequencing at Hokkaido System Science Co. Ltd. (Hokkaido, Japan). First, we used the Genome Analyzer II (Illumina, San Diego, CA) with Velvet assembler software. The method could not provide a contig longer than 10 kb, so we performed additional analysis using the Genome Sequencer FLX (Roche) with GS Denovo Assembler V2.0 software. Data from both sequencing attempts were then assembled using atgc software (Genetyx). The sequence of Occan9E12 and its flanking region in cosmid 9E12 was also sequenced directly by the Sanger method using a BigDye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster City, CA) and analysed using an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems). Cosmid 3A3 and the inverse PCR fragment were sequenced using the Sanger method as described above. Sequences determined in this study were deposited in the database under accession numbers AB727651–AB727653.
All the PCR primers used in this study are listed in Table 2. Inverse PCR for the amplification of the DNA region containing Occanm95-1 was performed with AccuPrime Taq Polymerase (Invitrogen). All other PCR amplifications were performed with GoTaq Polymerase (Promega, Madison, MI), following the manufacturer's instructions.
Table 2. Oligonucleotide PCR primers used in this study
Pulsed-field gel electrophoresis
Chromosomal-sized DNA samples were prepared using the method described by Sone et al. (1997). Pulsed-field gel electrophoresis was performed using a CHEF-DRIII apparatus (Bio-Rad) following the method described by Kito et al. (2003). Gels were subjected to Southern blot analysis after ethidium bromide staining.
Copy number of AVR-Pia in M. oryzae field isolates
Miki et al. (2009) reported that AVR-Pia is conserved only in field isolates that are avirulent towards Pia rice cultivars, owing to a single 6.6 kb band in Southern hybridization of HindIII-digested DNA using AVR-Pia as a probe (Fig. 1b). Further characterization of the region flanking AVR-Pia was attempted by applying various restriction enzymes in the Southern analysis of strain Ina168 (Fig. 1c). None of the enzymes, except for EcoRI, cuts inside the AVR-Pia gene; therefore, the enzymes were expected to produce a single band. EcoRI, however, produces a single cut in AVR-Pia and was expected to produce two bands. Southern analysis showed the expected signals for DNA digested with HindIII, DraI, BcnI, HincII, MflI, ApaI or BamHI (Fig. 1c), but additional bands were detected in DNA samples digested with BglII, KpnI, SalI, PstI, SmaI or EcoRI. Three bands detected in BglII-digested DNA indicated the conservation of three copies of AVR-Pia in the Ina168 genome (Fig. 1c). For further confirmation, two additional probes, V3 and V5, were designed on each of the two EcoRI-digested fragments of AVR-Pia (Fig. 1a) and used to separately probe the EcoRI-digested Ina168 DNA (Fig. 2a). Probing with V5 resulted in two hybridized bands, and probing with V3 resulted in three bands. The latter result confirms the conservation of three AVR-Pia genes. The stronger signal in the lower band from V5 hybridization suggests the hybridization of twice the amount of DNA compared with the higher band. Therefore, the appearance of four bands in the hybridization pattern of AVR-Pia in EcoRI-digested Ina168 DNA indicates the presence of six overlapping hybridization bands. Similar interpretation of the hybridization result of EcoRI-digested DNA of various strains with AVR-Pia indicated that the copy number of this gene varies from one to three (Fig. 2b).
The chromosomal location of AVR-Pia genes was investigated using pulsed-field gel electrophoresis (Fig. 2c). The AVR-Pia DNA probe hybridized with the largest chromosome in contour-clamped homogenous electrical field (CHEF)-separated chromosomes from Ina168. The strong signal detected in the isolate Shin83-34 was probably due to the higher concentration of DNA contained in the sample. This result indicates that all three copies of AVR-Pia are located on the same chromosome. Taken together with the fact that no copy of AVR-Pia was conserved in the mutant Ina168m95-1 (Fig. 2b), it was suggested that the mutant Ina168m95-1 had lost all three copies of AVR-Pia.
Adjacent region of AVR-Pia locus
Miki et al. (2009) reported that the insert sequence in the cosmid 46F3, which contains AVR-Pia, was completely lost in the mutant Ina168m95-1. To obtain the adjacent DNA sequence of cosmid 46F3, an additional cosmid clone 9E12 was sequenced. Cosmid 9E12 was assumed to not contain AVR-Pia (Miki et al., 2009), but it was expected to contain the sequence adjacent to that of cosmid 46F3. The sequence of cosmid 9E12 overlapped with cosmid 46F3 but revealed 12.1 kb of novel sequence flanking the transposon Occan (Kito et al., 2003) in cosmid 46F3 (named Occan9E12) (Fig. 3a). BLAST search indicated that this sequence mainly consists of nonrepetitive, low-copy sequence, with the exception of one retrotransposon MINE (Fudal et al., 2005). Conservation of this region in the mutant Ina168m95-1 was verified by PCR and Southern blot analyses. Each of the 21 sets of primers amplified DNA fragments with the same size from both the parental strain and the mutant strain (data not shown). We performed Southern blot analysis using a DNA fragment from the 12.1 kb 3′-flanking region of Occan9E12 as a probe (Probe A) against DNA of the parent and mutant strains digested with restriction enzymes that cut in that region (Fig. 3b). With all restriction enzymes, we detected signals in both Ina168 and Ina168m95-1, and restriction fragment length polymorphisms (RFLPs) between the two strains were also detected (Fig. 3b). This result not only indicates the conservation of the 3′-flanking region of Occan9E12 in the mutant but also demonstrates that the locations of the restriction sites, other than one in the 3′-flanking region of Occan9E12, are different from those in cosmid 46F3. The restriction fragments from the mutant Ina168m95-1 might cover the boundary of the AVR-Pia deletion event.
Cloning and analysis of the boundary of the deleted region in the mutant
Conservation of the Occan9E12 3′-flanking region in the mutant Ina168m95-1 is reminiscent of that a boundary of the deleted region in which three copies of AVR-Pia were contained is located in Occan9E12 or in the adjacent regions. To analyse this boundary, the unknown DNA region in the SacI fragment containing the Occan9E12 3′ region was cloned from Ina168m95-1 by inverse PCR using a pair of primers designed within the region. Inverse PCR resulted in a 5.0 kb DNA fragment, and sequence analysis of the fragment revealed an Occan element (Occanm95-1) and a novel 2.6 kb DNA fragment located adjacent to Occanm95-1 (Fig. 3a). This result indicates that the 5′-flanking sequence of Occan9E12 was lost in the mutant, while the 3′-flanking region was conserved in the mutant. This result indicates that the Occan9E12 element is located at one of the boundaries of the deletion of AVR-Pia in Ina168m95-1 (Fig. 3a).
The 2.6 kb novel sequence in the inverse PCR fragment was then used for screening additional cosmid clones, which were expected to contain another boundary of the deletion, from the genomic library of Ina168. A cosmid clone, named 3A3, was identified to contain a sequence homologous to the 2.6 kb sequence (Fig. 3a). Interestingly, sequence analysis of the cosmid 3A3 revealed another Occan sequence (Occan3A3) adjacent to the 2.6 kb sequence in the same direction as Occan9E12 and an additional 6.1 kb sequence in the opposite side of the Occan3A3 element. This additional sequence contains a retrotransposon MGL sequence with a Pot2 transposon inserted inside the element; it showed no homology with the Occan9E12 3′-flanking sequence in Ina168. The conservation of cosmid 3A3 DNA fragments in the mutant Ina168m95-1 was investigated by Southern blot analysis (Fig. 3c). Two probes (probes B and C) from the 5′ side of Occan3A3 and one probe, D, from the 3′ side of Occan3A3 were hybridized with SacI-digested DNA from Ina168 and the mutant. Probe B hybridized with a 3.5 kb band in both strains. Probe C hybridized with two bands in both strains, but only one band was the same in molecular weight. The other band that hybridized with Probe C in Ina168m95-1 was the same size, 10 kb, as the band that hybridized with Probe A (Fig. 3b). This indicates that this 10 kb band corresponds to DNA with the SacI fragment containing Occanm95-1, and accordingly, the higher molecular weight band in Ina168 corresponds to the SacI fragment containing Occan3A3. Probe D hybridized with the same two bands that hybridized with Probe C in Ina168, whereas the band corresponding to the Occanm95-1-containing SacI fragment did not hybridize with Probe D in the mutant Ina168m95-1. These results indicate that Ina168m95-1 conserved the 5′ region of Occan3A3 but not the 3′ region. Occan3A3 was presumed to be located at the other boundary of the deletion event in Ina168m95-1.
Occan9E12, Occan3A3 and AVR-Pia are located on the same chromosome in Ina168
To confirm that the two boundaries identified above are located on the same chromosome in Ina168, a Southern blot analysis using low-copy DNAs adjacent to each of the two Occan elements, with AVR-Pia as probes, was performed against CHEF-separated chromosomal DNA from Ina168 and Ina168m95-1 (Fig. 4a). All probes hybridized with the largest chromosomal band in Ina168. Two probes, except AVR-Pia, hybridized with the same band in Ina168m95-1. These results suggest that the two Occan elements are located on the same chromosome as the three copies of AVR-Pia in Ina168, and only the AVR-Pia genes were lost in Ina168m95-1.
Single-nucleotide polymorphisms in Occan elements suggest a homologous recombination between Occan9E12 and Occan3A3
Based on the results obtained in this study, one model of the deletion event occurring in Ina168m95-1 was proposed: a homologous recombination event between two Occan elements, Occan9E12 and Occan3A3, deleted three copies of AVR-Pia located between these elements (Fig. 4c). To verify this model, single-nucleotide polymorphisms (SNPs) in DNA sequences of the three Occan elements identified in this study were retrieved. Four SNPs among three Occan elements were identified (Fig. 4b). Occan9E12 and Occan3A3 differed at four positions, and Occanm95-1 showed a chimeric structure characteristic of a recombination event between the 5′ side of Occan3A3 and the 3′ side of Occan9E12. This result strongly supports the model proposed in Fig. 4c.
Several reports have revealed deletion as a common mechanism for mutating AVR genes in M. oryzae, but none of these reports investigated details of the deletion event at the molecular level. This study presents the first example of a molecular understanding of the deletion event, which is driven by a homologous recombination between two copies of DNA-type transposons. Several different transposable elements, which are dispersed throughout the genome, have been identified in M. oryzae (Dean et al., 2005; Sanchez et al., 2011). Some of them are strongly related to the AVR gene locus (Khang et al., 2008; Chuma et al., 2011), and in some cases, the insertion of such elements into the ORF or promoter region was found to be responsible for the functional loss of AVR genes (Kang et al., 2001; Fudal et al., 2005; Li et al., 2009). Results obtained in this study revealed an additional role for transposable elements as the target for homologous recombination.
Chuma et al. (2011) proposed a model for the adaptation of the rice blast fungus against resistant cultivars, in which deletion and recovery of AVR genes represents the main mechanism of adaptation. Homologous recombination can be a key process for such model, because recombination between repetitive sequences dispersed throughout the genome may cause various DNA rearrangements, including deletions, translocations and even horizontal transfers between strains. The size of the deleted DNA in the mutant Ina168m95-1 is yet to be determined, but it is likely to be over 44 kb, because the full-length cosmid 46F3 (38 kb) insert and the 6 kb sequence in the cosmid 3A3 are lost in the mutant. This indicates that the size of the DNA that is rearranged in the process can accommodate multiple AVR genes or a large AVR gene cluster for the production of secondary metabolites such as ACE1 (Böhnert et al., 2004).
The copy number of AVR-Pia indicates the duplication or deletion of AVR-Pia in Oryza isolates. Every copy in Ina168 and other isolates is located in a 6.6 kb HindIII fragment. Both termini of the HindIII sites of the fragment are located within the repetitive region (Miki et al., 2009); therefore, duplication should have occurred not only in the AVR-Pia gene and its putative promoter but together with its flanking sequences, including repetitive sequences. The contribution of recombination on the duplication still needs to be clarified by further studies of the AVR-Pia loci in a greater number of Oryza isolates of M. oryzae.
In conclusion, this study provided a detailed account of the homologous recombination between repetitive sequences in the genome of M. oryzae as a mechanism for gene deletion. Homologous recombination is induced by DNA double strand breaks, which are found to occur frequently in the M. oryzae genome (Ndindeng et al., 2010). Analyses of genes that participate in the molecular pathways of homologous recombination, such as Rhm52 and Rhm54 (Elegado et al., 2006), will be important in further studies for understanding the mutational mechanism of AVR genes in this important plant pathogen. In addition, such studies will provide us a novel strategy for the control of mutations of AVR genes of the rice blast fungus.
This work was supported by the Program for Promotion of Basic and Applied Researches for Innovations in Bio-oriented Industry (BRAIN) and Grants-in-Aid for Scientific Research (C-20580043 and B-23380024) by Japan Society for the Promotion of Science (JSPS).