Background

In the long history of parasitism, adaptive mutations have occurred between hosts and pathogens, and selection pressure has traditionally been considered the main force driving this coevolution. To date, two hypotheses have been proposed regarding these dynamics: arms race and trench warfare evolution between host resistance genes (R) and pathogen avirulence genes (AVR) [1]. The arms race hypothesis is considered the principal hypothesis, in which both host R genes and pathogen AVR genes are under directional selection and the alleles are derived by mutation. In brief, pathogens evolve a virulence gene to overcome host defense, while the hosts evolve a new resistance allele to defeat the virulence genes of the pathogen. In contrast, in the trench warfare hypothesis, the evolution of both host R genes and pathogen AVR genes is nondirectional.

Rice blast is one of the most destructive diseases in rice-growing regions and is caused by the filamentous ascomycetous fungus Magnaporthe oryzae (synonym of Pyricularia oryzae). Employing resistant rice varieties with major resistance (R) genes is considered the most important strategy for controlling this disease and crop loss that is also environmentally friendly and economical. To date, ≤26 R genes in rice have been cloned: Pb1, Pia, Pib, Pid2, Pid3, Pik, Pikh/Pi54, Pikm, Pikp, Pish, Pit, Pita, Pizt, Pi1, Pi2, Pi5, Pi9, pi21, Pi25, Pi36, Pi37, Pi56, Pi63, PiCO39 (http://www.ricedata.cn/gene/gene_pi.htm), Pi64 [2] and Pigm [3].

Rice resistance genes can recognize the corresponding AVR of M. oryzae and initiate their immune reaction. To date, 12 AVR genes in M. oryzae have been cloned: AVR-Pi54 [4], AVR-Pi9 [5], AVR-Pib [6], AVR-Pia [7], AVR-Pii [7], AVR-Pik/km/kp [7], AVR-Pizt [8], ACE1 [9], AVR-Pita [10], AVR1-CO39 [11], PWL1 [12], and PWL2 [13]. The AVR-Pik/km/kp gene of M. oryzae determines the effectiveness of the R gene Pik/km/kp. AVR-Pik/km/kp encodes a putative secreted protein with 113 amino acids and two conserved motifs: motif-1, [LI]xAR[SE][DSE], and motif-2, [RK]CxxCxxxxxxxxxxxxH (similar to the C2H2 zinc finger motif) [7]. The AVR-Pik/km/kp gene was cloned from an isolate of Ina168 but found to be absent in the assembled sequence of isolate 70–15, which is recognized by the host Pik resistance protein and triggers the defense response [7]. Five AVR-Pik alleles (AVR-Pik-A, AVR-Pik-B, AVR-Pik-C, AVR-Pik-D, and AVR-Pik-E) were found [7], and AVR-Pik-D (20.5%) and AVR-Pik-E (1.4%) were detected among 77 isolates [14]. Four AVR-Pik alleles (AVR-Pik-A, AVR-Pik-C, AVR-Pik-D, and AVR-Pik-E) were found among 39 isolates worldwide (three isolates from Europe, six isolates from America, seven isolates from Africa and 23 isolates from Asia), and AVR-Pik-D was the most frequent allele (15 out 39), while the AVR-Pik-A, AVR-Pik-C, and AVR-Pik-E alleles had similar frequencies (7–9 out of 39) [15]. AVR-Pik/km/kp has evolved via gene gain/loss [7], while substitution mutations were observed in the coding regions of AVR-Pik/km/kp in M. oryzae populations, and 16 single nucleotide polymorphisms (SNPs) were found in regions without signal domains in Chinese rice blast isolates [16].

The Pik locus is located on the long arm of chromosome 11 and is known to have a resistance function [17,18,19,20]. At the Pik locus, five rice blast R genes (Pik, Pik-m, Pik-p, Pik-h and Pik-s) are involved, among which 4 R genes (Pik, Pik-m, Pik-p and Pik-h) have been isolated [18, 21,22,23,24] and Pik is regarded as the youngest allele [22]. Pik, Pik-m, Pik-p and Pik-h were cloned and found to encode a putative CC-NBS-LRR protein [18, 23, 25, 26]. The CC domain of Pik-1 physically binds to the AVR-Pik effector of M. oryzae to trigger Pik-specific resistance [15, 23]. The rice resistance gene Pik-s is still not cloned. Monogenic lines containing 24 rice blast resistance genes, including Pik, Pik-m, Pik-p, Pik-h and Pik-s, were developed and will be used to characterize the pathogenicity of rice blast fungus [27].

Pikm and Pikp exhibit a high level of resistance to blast fungus from Fujian Province and can be used in parents for resistance breeding in Fujian Province [28]. Pikm, Piks, and Pikp are moderately resistant in Sichuan and Guizhou Provinces, China [29]. Pikm, Piks, and Pik are moderately resistant, while Pikh exhibits high resistance in Guangdong Province, China [30], and 35.4% of 82 rice germplasm resources were found to carry Pikh by molecular analysis [31]. Eighty of 229 rice cultivars and breeding materials carry the Pik locus in Fujian Province, based on PCR detection [32]. Different resistance spectra of Pik, Pikm, Pikp, Pikh and Piks at the Pik locus were detected in 282 blast isolates collected from Yunnan Province, China [33]. The R genes of the Pik locus exhibit high resistance to Chinese rice blast fungus.

Further understanding the molecular evolution of the AVR gene has potential implications for the development of resistance breeding, the rational use of resistance genes in production, and the deployment of more effective strategies to control the disease. Regarding the long-term interactions between the pathogen and its host, the host employs resistance genes to prevent infection by the pathogen; however, the pathogen attempts to overcome them, and the coevolution of the pathogen and its host becomes discernible at the genome level [34, 35]. The pathogen utilizes mutation to adapt to novel host alleles and the environment, while its genome structure is highly variable and impacted by host selection [15, 36, 37]. Analyzing DNA sequence variation of AVR-Pik/km/kp alleles of M. oryzae in field isolates will help to understand the effectiveness and durability of the resistance gene Pik alleles in China.

The goal of the present study was to analyze the DNA sequence variation of AVR-Pik/km/kp alleles in field isolates of M. oryzae to understand the variation and coevolutionary mechanism of M. oryzae AVR-Pik/km/kp alleles and rice Pik alleles in Yunnan Province.

Results

Efficacy of Pik genes and detection frequency of AVR-Pik alleles

Based on the disease reactions, the efficacy of the Pik genes Pik, Pikm, Pikp, Pikh and Piks were examined. Some 223, 256, 154, 276 and 83 of the 366 isolates (collected from different rice growing regions of Yunnan and selected as representative isolates) tested were avirulent to the Pik, Pikm, Pikp, Pikh and Piks gene-containing rice monogenic lines IRBLk-K, IRBLkm-Ts, IRBLkp-K60, IRBLkh-K3 and IRBLks-F5, respectively (Table 1). The frequency of avirulence to Pik, Pikm, Pikp, Pikh and Piks was 60.9, 69.9, 42.1, 75.4 and 22.7%, respectively, while the remaining 143, 110, 212, 90 and 283 isolates were virulent to the corresponding R gene (Table 1). Of 366 isolates, AVR-Pik/km/kp alleles of 278 were amplified by AVR-Pik/km/kp (AVR-Pik allele)-specific primers (pex31F/pex31R) (Table 1), and the mean percentage of the AVR-Pik/km/kp allele was 76.0%. The highest percentage of AVR-Pik/km/kp was 90.3% in the M. oryzae population collected from northeastern Yunnan, whereas the lowest percentage was 66.7% from northwestern Yunnan (Table 1). The percentages of AVR-Pik/km/kp were 77.8, 90.3, 66.7, 72.7, 89.3 and 68.3% in central, northeastern, northwestern, southeastern, southwestern and western Yunnan, respectively. Similarly, the percentages of AVR-Pik/km/kp were 74.5 and 77.0% in Xian/Indica (XI) and Geng/Japonica (GJ) rice-growing regions in Yunnan. These findings suggest that Pik loci have different effective uses in preventing blast infections in most rice production areas in Yunnan.

Table 1 Distribution of AVR-Pik genes and avirulent isolates of M. oryzae collected from Yunnan, China, in IRBLk-K, IRBLkm-Ts, IRBLkp-K60, and IRBLkh-K3
Full size table

A novel AVR-Pikh gene was identified to be associated with AVR-Pik/km/kp alleles

The AVR-Pik/km/kp gene is an effector gene with 342 nucleotides encoding a putative secreted protein possessing one signal peptide of 57 nucleotides in the first exon in the open reading frame (ORF) [7]. A total of 10 AVR-Pik haplotypes, including the five original AVR-Pik alleles AVR-Pik_D (GenBank Accession No. AB498875) (H01), AVR-Pik_A (AB498876) (H02), AVR-Pik_B (AB498877) (H03), AVR-Pik_C (AB498878) (H04), and AVR-Pik_E (AB498879) (H05), were identified based on the DNA sequence assemblies of 201 isolates (Table 2). The remaining 77 isolates were sequenced, but they had double peaks and were removed for further analysis. Five novel AVR-Pik/km/kp haplotypes (H06-H10) were identified. Alignment of DNA sequence assemblies of the AVR-Pik/km/kp gene from 201 isolates revealed six polymorphic sites in the exon region, and none of them were in the signal peptide region (Table 2). Six sites in the exon region resulted in amino acid substitutions (Table 3). Moreover, the AVR-Pik/km/kp allele sequence assemblies among the 201 isolates were predicted to produce 10 functional proteins (Table 3). Among these 10 proteins, amino acid variations were predicted to occur at five positions. All variations occurred throughout the protein, except for the putative secreted proteins possessing the [RK]CxxCxxxxxxxxxxxxH] motif (Table 3; Additional file 1: Figure S1). Amino acid variations at M78K were found in six isolates, all of which were virulent in the monogenic lines IRBLk-K (with Pik), IRBLkm-Ts (with Pikm), IRBLkp-K60 (with Pikp), IRBLkh-K3 (with Pikh) and IRBLks-F5 (with Piks) (Table 3). This finding suggests that amino acid 78 M is critical for the avirulence function of AVR-Pik/km/kp/kh loci. The isolates of the H01, H07 and H09 haplotypes harbored the avirulence genes AVR-Pik/km/kp/kh, the isolates of H05 and H08 harbored AVR-Pik/km/kh, the isolates of H06 harbored AVR-Pikm/kh, and the isolates of H02 and H03 harbored AVR-Pikh because these isolates were avirulent to the corresponding R gene(s) (Table 3). The isolates of H04 and H10 had overcome the resistance of all Pik alleles at the loci (Table 3). Thus, the novel avirulence gene AVR-Pikh was identified, and the evolution of AVR-Pik alleles of M. oryzae was involved. The 10 haplotypes did not harbor AVR-Piks because the isolates were virulent to the monogenic line IRBLks-F5 (harboring Pi-ks) (Table 3). Some 75 isolates contained AVR-Pik/km/kp/kh (frequency of 36.4%), 55 isolates contained AVR-Pik/km/kh (frequency of 26.7%), four isolates contained AVR-Pikm/kh (frequency of 1.9%), and 50 isolates contained AVR-Pikh (frequency of 24.9%). Some 17 isolates did not contain these avirulence genes (Additional file 1: Table S1). In summary, five novel AVR-Pik loci were identified, and 91.5% of the total isolates contained AVR-Pikh, which is widely distributed in southwestern China.

Table 2 Haplotypes of AVR-Pik loci in rice blast fungus in Yunnan, China
Full size table
Table 3 Variation in the AVR-Pik loci proteins in rice blast fungus in Yunnan, China
Full size table

Stepwise evolution and haplotype diversity of AVR-Pik loci in M. oryzae

Among the 10 AVR-Pik haplotypes, the haplotypes H01, H02, H03, H04 and H05 were identical to the original AVR-Pik alleles of AVR-Pik_D (GenBank Accession No. AB498875), AVR-Pik_A (AB498876), AVR-Pik_B (AB498877), AVR-Pik_C (AB498878), and AVR-Pik_E (AB498879) (Table 2), respectively. Seven haplotypes were detected in 88, 37 and 39 M. oryzae isolates from western, central and northeastern Yunnan, respectively. Six haplotypes were detected in 17 M. oryzae isolates from southeastern Yunnan, three haplotypes were detected in 10 M. oryzae isolates from southwestern Yunnan, and only one haplotype was detected in 10 M. oryzae isolates from northwestern Yunnan (Table 4). Ten and eight haplotypes were found in the GJ and XI rice-growing regions, and the diversity index (DI) was 0.79 and 0.75 for these regions, respectively. Similarly, the DI was 0.78, 0.68, 0.65, 0.62, 0.54, and 0 for northeastern, central, western, southeastern, southwestern, and northwestern Yunnan, respectively (Table 4). In summary, the DI of AVR-Pik alleles was ordered in Yunnan Province as follows: northeastern>central>western>southeastern>southwestern>northwestern. The DI of AVR-Pik alleles in the GJ rice-growing region was similar to that in the XI rice-growing region.

Table 4 Distribution of AVR-Pik haplotypes in different rice-growing regions
Full size table

Six nucleotide variations in the exons of AVR-Pik alleles were observed (Additional file 1: Figure S1 and Table S2), and a haplotype network based on sequence variation was developed (Fig. 1). Four microevolutionary clusters of AVR-Pik, AVR-Pikm, AVR-Pikp, and AVR-Pikh were observed among 201 field isolates (Fig. 1). The five original AVR-Pik alleles AVR-Pik_D (H01), AVR-Pik_A (H02), AVR-Pik_B (H03), AVR-Pik_C (H04), and AVR-Pik_E (H05) were involved in the networks. The isolates of H01, H05, H07, H08 and H09 were avirulent to IRBLk-K (with Pik), whereas the isolates of H02, H03, H04, H06 and H10 were virulent to Pik (Table 3; Fig. 1). The isolates of H01, H05, H06, H07, H08 and H09 were avirulent to IRBLkm-Ts (with Pikm), whereas the isolates of H02, H03, H04, and H10 were virulent to Pikm (Table 3; Fig. 1). The isolates of H01, H07 and H09 were avirulent to IRBLkp-K60 (with Pikp), whereas the isolates of H02, H03, H04, H05, H06, H08 and H10 were virulent to Pikp (Table 3; Fig. 1). The isolates of H01, H02, H03, H05, H06, H07, H08 and H09 were avirulent to IRBLkh-K3 (with Pikh), whereas the isolates of H04 and H10 were virulent to Pikh (Table 3; Fig. 1). These findings suggest that there were four distinct stepwise-evolved patterns (AVR-Pik, AVR-Pikm, AVR-Pikp, and AVR-Pikh) in rice-growing regions of Yunnan.

Fig. 1
figure 1

The haplotype network for the 10 AVR-Pik alleles. The original AVR-Pik allele is designated as the H01 haplotype in the network. Haplotypes are separated by mutational events. All haplotypes are displayed as circles. The size of the circles corresponds to the haplotype frequency. Haplotypes H01 to H05 are the same as AB498875, AB498876, AB498877, AB498878 and AB498879 (GenBank Accession No.) of AVR-Pik and were obtained from GenBank. Green indicates avirulence to the corresponding R gene, and yellow indicates virulence to the corresponding R gene

Full size image

A possible scenario for M. oryzae AVR-Pik allele-rice Pik allele interactions and coevolution was constructed (Fig. 2). The AVR-Pik homolog H01 (AVR-Pik-D) was derived from an ancestral M. oryzae gene. The Pik allele, Piks, cannot recognize the three alleles AVR-Pik-D (H01), H07 and H09; thus, the other Pik allele, Pikp, evolved that can recognize these three alleles, while the altered alleles H05 (AVR-Pik-E) and H08 evolved to virulence from avirulence via nucleotide substitution to avoid recognition by Pikp (Table 2; Fig. 2). For this situation, another Pik allele, Pik, evolved that can recognize five alleles, namely, AVR-Pik-D (H01), H07, H09, AVR-Pik-E (H05) and H08. Then, yet another AVR-Pik allele, H06, was derived that cannot be recognized by Pikp and Pik. Next, the rice R gene Pikm was utilized that recognizes AVR-Pik-D (H01), H07, H09, AVR-Pik-E (H05), H08 and H06. Then, two more AVR-Pik alleles, namely, AVR-Pik-A (H02) and AVR-Pik-B (H03), were derived that cannot be recognized by Pikp, Pik and Pikm. Next, the rice R gene Pikh was utilized that recognizes AVR-Pik-D (H01), H07, H09, AVR-Pik-E (H05), H08, H06, AVR-Pik-A (H02) and AVR-Pik-B (H03). Then, another two AVR-Pik alleles, namely, AVR-Pik-C (H04) and H10, evolved that cannot be recognized by any of the five Pik alleles (Table 2; Fig. 2). These patterns show the stepwise evolution of AVR-Pik and Pik interaction and coevolution. Interestingly, the AVR-Pik allele H07 was derived from H01, which can be recognized by Pikp, Pik and Pikm. Thus, the altered allele H06 from H07 can avoid recognition by Pikp and Pik; next, the altered allele H08 from H06 can avoid recognition by Pikp, while the altered allele H04 from H08 avoids recognition by any of the five Pik alleles. Similarly, the H09 allele was derived from H07, which can be recognized by Pikp, Pik, Pikm and Pikh; thus, the altered allele H02 allele from H09 can avoid recognition by Pikp, Pik, and Pikm (Table 2; Fig. 2). The H05 allele can be recognized by Pik, Pikm and Pikh, while the altered allele H04 from H05 can avoid recognition by Pikp, Pik, and Pikm (Table 2; Fig. 2). These results suggest that the avoidance evolution of AVR-Pik loci of M. oryzae was involved in the interaction and coevolution with the Pik loci of M. oryzae in nature.

Fig. 2
figure 2

Possible scenario for M. oryzae AVR-Pik allele-rice Pik allele interactions and coevolution. Chronological order is given on the left (time order). The AVR-Pik homolog H01 (AVR-Pik-D) was derived from an ancestral M. oryzae gene. AVR-Pik-D (H01), H07 and H09 are recognized by Pikp; thus, the altered alleles AVR-Pik-E (H05) and H08 evolved. In response to this situation, another Pik allele, Pik, evolved that can recognize five alleles, namely, AVR-Pik-D (H01), H07, H09, AVR-Pik-E (H05) and H08. Then, yet another AVR-Pik allele, H06, was derived that cannot be recognized by Pikp and Pik. Next, the rice R gene Pikm was utilized that recognizes AVR-Pik-D (H01), H07, H09, AVR-Pik-E (H05), H08 and H06. Then, two more AVR-Pik alleles, namely, AVR-Pik-A (H02) and AVR-Pik-B (H03), were derived that cannot be recognized by Pikp, Pik and Pikm. Next, the rice R gene Pikh was utilized that recognizes AVR-Pik-D (H01), H07, H09, AVR-Pik-E (H05), H08, H06, AVR-Pik-A (H02) and AVR-Pik-B (H03). Then, two other AVR-Pik alleles, namely, AVR-Pik-C (H04) and H10, evolved that cannot be recognized by any of the five Pik alleles

Full size image

Selection pressure on AVR-Pik in M. oryzae

To determine the natural selection pressure on AVR-Pik in M. oryzae in Yunnan, Tajima’s neutrality of AVR-Pik in M. oryzae was tested based on 201 AVR-Pik DNA sequences, and Tajima’s D was found to be 1.19854 (Additional file 1: Table S2). The result suggests that AVR-Pik might be under strong population expansion or either in positive selection. The results of three positive-selection models were highly consistent (Fig. 3). The sliding window shows the distribution of the Ka/Ks values across all 113 amino acids under the M8, M8a, and M7 models (Fig. 3). The results show that the Ka/Ks value of the 46th, 47th, 48th, 67th and 78th sites was > 1, suggesting that these sites were potentially subjected to purifying selection. Positively selected sites were observed only in the mature protein region among the 201 M. oryzae isolates with AVR-Pik (Fig. 3). These results showed that the amino acid sequence was conserved in the signal peptide compared with the divergent mature protein region of AVR-Pik in M. oryzae.

Fig. 3
figure 3

Sliding window of positively selected sites in the AVR-Pik alleles under the M8, M8a, and M7 models. The Y-axis indicates the ratio of the rate of nonsynonymous substitutions (Ka) to the rate of synonymous substitutions (Ks) (Ka/Ks); the X-axis indicates the position of the AVR-Pik amino acids in the site. The signal region of the variant structure is purple, and the black area represents the mature protein region on the label at the top of the figure

Full size image

To confirm the resistance of Pik alleles in the field, we assayed seedling and panicle blast disease with monogenic lines carrying Pik, Pikm, Pikp, and Pikh, which were developed by the Japan International Research Center for Agricultural Sciences (JIRCAS) and International Rice Research Institute (IRRI) in fields in Mangshi, Lufeng and Yiliang Counties in 2015 (Additional file 1: Table S3). The result suggests that IRBLkm-Ts (with Pikm), IRBLkp-K60 (with Pikp), and IRBLkh-K3 (with Pikh) were resistant, while IRBLks-F5 (with Piks) and IRBLk-Ka (with Pik) were susceptible in Mangshi County (Additional file 1: Table S3). These results suggest that M. oryzae isolates in the population holds AVR-Pikm/kp/kh genes. IRBLkh-K3 (with Pikh) was resistant in Lufeng and Yiliang, and the monogenic lines IRBLks-F5 (with Piks), IRBLk-Ka (with Pik), IRBLkm-Ts (with Pikm) and IRBLkp-K60 (with Pikp) were susceptible in Lufeng and Yiliang Counties, suggesting that the M. oryzae isolates in the population harbor AVR-Pikh. These results are consistent with the results of PCR detection and pathogenicity assays.

Discussion

In this study, we found five new haplotypes in the AVR-Pik DNA sequences among field isolates of M. oryzae from various rice-producing regions in Yunnan. Numerous virulent isolates of the Pik gene containing rice varieties were identified in field isolates collected in Yunnan, suggesting that Pik was eradicated in some rice production areas due to the extensive development of Pik in China. The Pik alleles have been deployed and display high rice blast resistance in China [20, 22, 32, 39]. Complete deletions have occurred in AVR-Pik sequences among field isolates of M. oryzae from various rice-producing countries [15, 16], which agrees with our results. Numerous isolates inspected from commercial rice fields containing AVR-Pik suggest that Pik has been effective in preventing rice blast disease. In Yunnan, rice cultivars with Pikh, Pikp, Pikm, Piks, and Pik were resistant to 81.7, 62.8, 51.9, 43.4 and 39.4% of isolates (282 isolates), respectively [33]. The corresponding values of 146 isolates from Guangdong Province were 88.4, 39.0, 0, 1.4 and 57.5%, respectively [40]. These results suggest that some Pik alleles have limited effects in these rice production areas. Continued analysis of AVR-Pik alleles in these isolates will help us understand the evolutionary mechanism of AVR-Pik and predict the stability and effectiveness of Pik allele-mediated resistance under natural conditions.

Effective variations in DNA sequences have been observed in the telomere regions of several AVR genes (AVR-Pita1, AVR-Pia, and AVR-Pii) [7, 41, 42]. The transposable element (TE) insertion in the last exon of the ACE1 gene [9] and Pot3 inserted in AVR-Pizt and AVR-Pita1 all resulted in new virulent alleles. Based on the DNA sequence analysis [8, 43, 44], four variations, namely, a point mutation, segmental deletion, complete absence (6.7%) and TE insertion, were found in AVR-Pib, all of which result in loss of the avirulence function [6]. Three distinct expression profiles were found among seven of 16 functional nucleotide polymorphisms in the AVR-Pib genes [6]. These findings showed that M. oryzae uses transposons to change the expression of AVR genes to overcome R genes. In the present study, the AVR-Pik gene was present in most blast populations (76.0%) in Yunnan (Table 1), which was similar to rice blast isolates in Hunan Province [45]. We found significantly more nucleotide variation in the protein-coding region of AVR-Pik alleles, resulting in changes in amino acids and suggesting that there is intense selection pressure on AVR-Pik alleles in Yunnan.

DNA sequence variation was found in exon regions of AVR-Pik, and a total of 10 haplotypes were identified based on the six variant nucleotides among 201 isolates collected from Yunnan (Table 2). Five novel variant amino acids of the AVR-Pik loci variants in the 201 isolates were identified in the present study, which leads towards finding of five new haplotypes. Based on the virulence analysis of the strains harboring this variation, haplotypes H01, H02, H05 and H07 are more frequent in the field isolates. This result suggests that the loss of these haplotypes may have a larger fitness penalty than the loss of other alleles in the M. oryzae population. These new alleles allowed us to construct a more holonomic network among different alleles of AVR-Pik, and some novel haplotypes were found. We also identified the putative secreted proteins possessing the [LI]xAR[SE][DSE] and [RK]CxxCxxxxxxxxxxxxH] motifs in 201 isolates with AVR-Pik alleles (Table 3), which was consistent with the results of Yoshida et al. [7]. Some 126, 59, 94 and 15 isolates are variations at the amino acid positions H46N, P47A, G48D, and A67D, respectively, and four and six isolates are variations at the amino acid positions M78I and M78K, respectively (Table 3). These results showed that the 46th, 47th, 48th, 67th and 78th amino acid positions were the most variable amino acid sites among proteins of AVR-Pik/km/kp/kh.

During the long coevolution of plants and pathogens, the pathogen AVR genes have been recognized by the cognate plant R genes and triggered effective defense responses. The divergences of the AVR genes of the pathogen were shaped by host R genes and changing environmental conditions. We observed that the DI of AVR-Pik was similar in the XI and GJ regions (Table 4), and variations in AVR-Pik were different between the XI- and GJ-growing regions (Table 4). These results suggest that adaptive variations have occurred in commercial rice fields in Yunnan.

Yunnan is one of the diversification centers of the cultivated Asian rice species Oryza sativa. The three wild species O. rufipogon, O. officinalis and O. meyeriana also exist in the area [46]. Over 5000 accessions of rice germplasms were collected from fields and preserved. Among them, 227 rice accessions were characterized by a set of differential rice blast isolates, and 38 and 25 of 227 rice accessions contained the rice blast resistance genes Pik and Pikm, respectively [46]. The observed Tajima’s D of 1.19854 (Additional file 1: Table S2) suggests that AVR-Pik/km/kp/kh loci may be under population expansion or purifying selection shaped by the cognate Pik loci in rice-growing regions of Yunnan. Most isolates carried AVR-Pikh and Pikh, with high resistance, in Yunnan and Guangdong Provinces. This pattern may be due to Pikh being a widely distributed resistance gene in rice accessions. These results agree with those of Zhai et al. [22].

AVR-Pik is recognized specifically by the Pik in rice, and AVR-Pik directly physically binds the N-terminal coiled-coil domain of Pik. These observations were confirmed by yeast two-hybrid and coimmunoprecipitation assays [15]. Four alleles of AVR-Pik (AVR-Pik_D, AVR-Pik_E, AVR-Pik_A, and AVR-Pik_C) in Japanese isolate populations coevolved with the rice Pik alleles Pikp, Pik and Pikm [15]. Four alleles of AVR-Pik in the Chinese M. oryzae population showed stepwise evolution with the rice Pik alleles Pikp, Pik, Pikm, and Pkh [16]. Highly variable Pik alleles were observed, and stepwise changes in both the AVR-Pik of M. oryzae and Pik of rice were found in the field [16]. These observations indicate that AVR-Pik has been strongly targeted by hosts [16]. In the present study, we found both avoidance and stepwise-evolved AVR-Pik allele-rice Pik allele interactions and coevolution (Table 3; Fig. 2), which implies the presence of a high diversity of rice varieties in Yunnan. The AVR-Pik alleles have been regularly under selection by antagonistic alleles in host populations. Similarly, the wheat-infecting lineages from Brazil and Bangladesh appeared to be genetically distinct and displayed reticulate evolution in population genomic analyses of transcriptomic SNPs [47].

A stepwise mutation process has been demonstrated for virulence acquisition in Fusarium oxysporum f. sp. ciceris and Puccinia striiformis f. sp. tritici [48,49,50]. In the present study, we found one major episode of mutation evolution of AVR-Pik alleles and seven minor mutation evolution patterns (Fig. 2). The alternative mutation pattern can seemingly convert from avirulence to virulence via occasional mutation and showed higher efficiency (Fig. 2). These results may be due to the strong positive selection pressure imposed by the corresponding Pik allele of the host and the environment. Similarly, AVRL567 can convert from avirulent to virulent by a set of stepwise mutations leading to amino acid substitution [51]. Stepwise evolution has been observed in AVR-Pik [15, 16]. The possible evolution of AVR-Pik found in the present study was more complex than expected in the rice-growing regions of Yunnan.

Conclusion

We detected five novel haplotypes in the field population by using 201 isolates, constructed a complex network of AVR-Pik alleles, and evaluated the effectiveness of Pik alleles in rice production areas of Yunnan. Our findings support the premise that functional AVR-Pik possesses diversified sequence structures and can avoid recognition by hosts via multiple site variations. Haplotype H10 originates from the frequently distributed H2 haplotype, and H4 originated from H5 and/or H8. These haplotypes can overcome all detected Pik alleles to date. Although the H4 and H10 haplotypes have low frequencies, surveillance of these two alleles in field populations is crucial because of their high risk of increasing in abundance in the background of Pik-containing rice varieties. Management must retard selection on the allele, possibly by avoiding its proliferation in agricultural practices. The prediction of blast occurrence should be based on the frequency and distribution of the allele of multiple loci, e.g., Pik and AVR-Pik, in isolate populations under field conditions.

Methods

Rice cultivars, fungal isolates, culture, and pathogenicity assays

The Pik, Pikm, Pikp, Pikh, and Piks gene-containing rice monogenic lines IRBLk-K, IRBLkm-Ts, IRBLkp-K60, IRBLkh-K3 and IRBLks-F5, respectively, and the susceptible backcrossing parent Lijiangxintuanheigu (LTH, without Pik) were used for pathogenicity assays (the seeds were originally acquired from Japan International Research Center for Agricultural Sciences (JIRCAS), and the JIRCAS undertook the formal identification of the plant material. The seeds conserved in plant germplasm resources bank of Yunnan Academy of Agricultural Sciences). A total of 366 isolates were collected, single-spore purified, and examined. All isolates were stored at − 20 °C on filter paper and grown in petri dishes containing oatmeal agar for spore production at room temperature under blue and white fluorescent lighting. Disease reactions were determined using a modified standard pathogenicity assay, as described by Jia et al. [52]. Specifically, rice seedlings at the 3- to 4-leaf stage were placed in a plastic bag and spray inoculated with a spore suspension of 1–5 × 105 spores/mL. After inoculation, the plastic bags were sealed to maintain a high relative humidity (90–100%) for 24 h before removing the plants from the bags. Subsequently, the plants were maintained in a greenhouse for an additional 6 days to allow the development of disease symptoms. The disease reactions were rated visually based on the number and extent of lesions on the second youngest leaf using the 0–5 disease scale. A value of 0–1 indicated resistant, 2 indicated moderately resistant, and 3–5 indicated susceptible. Five seedlings were used each time, the experiment was repeated once more, and the mean disease scores were used to determine resistance versus susceptibility.

DNA preparation, PCR amplification, and DNA sequencing

Fungal isolates were grown in complete liquid media at 25 °C for six to 8 days to produce mycelia under dark conditions. DNA was then isolated from mycelia using the cetyl trimethylammonium bromide (CTAB) method [53]. The primers pex31F (5′-TCGCCTTCCCATTTTTA-3′) and pex31R (5′-GCCCATGCATTATCTTAT-3′) were used to amplify the AVR-Pik allele and for sequencing using the methods of Yoshida et al. [7]. Specifically, PCRs were performed using 2× Taq PCR MasterMix (Tiangen Biotech Co. Ltd., Beijing, China). Each PCR consisted of the following components: 25 μl of Taq PCR Master Mix (containing 25 U of Taq DNA polymerase, 10X Tiangen PCR buffer, 15 mM MgCl2, and 200 μM each dNTP), 1 μl of each 10 μM primer, 2 μl of fungal genomic DNA, and 21 μl of distilled water (provided in the Tiangen kit). Reactions were performed in a Bio-Rad Thermal Cycler (C1000, Bio-Rad Laboratories, Life Science Research, CA, USA) with the following PCR program: 1 cycle at 95 °C for 3 min for initial denaturation, followed by 29 cycles at 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 30 s and a final denaturation at 72 °C for 7 min. All PCRs were repeated three times (20 μl for detection, 50 μl for sequencing). The size of the amplified fragment was estimated by DL2000 DNA Ladder (Tiangen Biotech Co. Ltd., Beijing, China). PCR products were sequenced using the same primers as mentioned above for PCR amplification. DNA was sequenced by Shanghai Life Technologies Biotechnology Co., Ltd. (Shanghai, China). The amplicon from each isolate was sequenced three times.

Resistance evaluation of Pik alleles in the field

The monogenic lines IRBLk-Ka, IRBLkm-Ts, IRBLks-F5, IRBLkp-K60, and IRBLkh-K3 (carrying Pik, Pikm, Piks, Pikp, and Pikh, respectively) were planted in fields in Mangshi, Lufeng and Yiliang Counties in Yunnan Province in 2015. The seedlings and panicles were surveyed for blast disease, and the resistance was evaluated.

Data analysis

DNA sequences of AVR-Pik were assembled by the Vector NTI V.10 software suite (Invitrogen, Carlsbad, California, USA) and aligned using DNASTAR V7.10 software (http://www.dnastar.com/). The number of DNA haplotypes and polymorphic sites (π) and the sliding window were calculated using DnaSP v5.10.01 software [54]. Haplotype network analysis was performed using TCS1.21 (http://darwin.uvigo.es/) [55]. The DI was calculated as the frequency of haplotypes or protein types in the rice blast fungus population following the method of Fontaine et al. [38]: DI = (1-∑ni = 1pi2), where pi is the frequency of haplotype i in a population. Tajima’s neutrality test was performed using MEGA V5.10. The analysis of positive selection was performed using the Selection Server program (http://selecton.tau.ac.il). Three models were used to identify the positively selected sites under the query of AVR-Pik: M8 (positive selection enabled, beta + w ≥ 1), M8a (beta + w = 1, null model), and M7 (beta, null model). The data were then imported into Microsoft Excel for statistical analysis and to draw the sliding window.