The Lesson Learned from the Unique Evolutionary Story of Avirulence Gene AvrPii of Magnaporthe oryzae

Blast, caused by Magnaporthe oryzae, is one of the most destructive diseases affecting rice production. Understanding population dynamics of the pathogen’s avirulence genes is pre-required for breeding and then deploying new cultivars carrying promising resistance genes. The divergence and population structure of AvrPii was dissected in the populations of southern (Guangdong, Hunan, and Guizhou) and northern (Jilin, Liaoning, and Heilongjiang) China, via population genetic and evolutionary approaches. The evolutionary divergence between a known haplotype AvrPii-J and a novel one AvrPii-C was demonstrated by haplotype-specific amplicon-based sequencing and genetic transformation. The different avirulent performances of a set of seven haplotype-chimeric mutants suggested that the integrity of the full-length gene structures is crucial to express functionality of individual haplotypes. All the four combinations of phenotypes/genotypes were detected in the three southern populations, and only two in the northern three, suggesting that genic diversity in the southern region was higher than those in the northern one. The population structure of the AvrPii family was shaped by balancing, purifying, and positive selection pressures in the Chinese populations. The AvrPii-J was recognized as the wild type that emerged before rice domestication. Considering higher frequencies of avirulent isolates were detected in Hunan, Guizhou, and Liaoning, the cognate resistance gene Pii could be continuously used as a basic and critical resistance resource in such regions. The unique population structures of the AvrPii family found in China have significant implications for understanding how the AvrPii family has kept an artful balance and purity among its members (haplotypes) those keenly interact with Pii under gene-for-gene relationships. The lesson learned from case studies on the AvrPii family is that much attention should be paid to haplotype divergence of target gene.


Introduction
Rice (Oryza sativa L.), domesticated into subspecies Xian (indica) and Geng (japonica), is one of the top three cereal crops in China in terms of area and production [1,2]. Rice blast, caused by the fungal pathogen M. oryzae (Mo; syn. Pyricularia oryzae), is the most concerned

Discovery of a New Haplotype AvrPii-C
Ten isolates (CHL22, 346, 584, 590, 624, EHL0314, 0317, 0319, 0329, 0445) that were avirulent on a rice monogenic line carrying Pii, IRBLi-F5, were selected from various Mo populations in China and were subjected to amplification of both coding sequence (CDS) and the full-length (FL) sequence of AvrPii via the regular polymerase chain reaction (PCR) system, with the key primer pairs AvrPii-J/C-CDS_F/R, and AvrPii-J-FL_F/R, respectively. Both primer pairs were designed based on the known reference sequence of AvrPii aforementioned. Because half of the ten isolates tested did not produce any amplicon with the FL primer pair, the alternative approach called a thermal asymmetric interlaced PCR (Tail-PCR) system was employed for creating specific sequences of both 5 and 3 regions of the representative isolate, CHL346; both might be distinct from the reference sequence ( Figure S1). After three rounds of Tail-PCR, the certain amplicons derived from 5 and 3 regions were subjected to sequencing. Then, the key primer pair, AvrPii-C-FL_F/R, was devised based on both sequences for re-amplifying the target gene of the ten isolates via the regular PCR system. The full-length sequences of the ten isolates were compared using the Multalin website (http://multalin.toulouse.inra.fr/multalin/). Information on the primers and PCR programs were shown in Table S1.

Functional Validation of Both Haplotypes
To construct transgenic vectors of both haplotypes, the full-length sequences of AvrPii-C (as that was firstly discovered from the Chinese isolates in the present study) and AvrPii-J (as it was firstly discovered from the Japanese isolate Ina168) [13], respectively, were reamplified using Q5 High-Fidelity DNA polymerase (Tsingke, Nanjing, China) based on the genomic DNA of isolates CHL346 and CHL22. The amplicons were inserted into pGEM-T (Promega, Dalian, China) to generate two constructs pGEM-T_AvrPii-C and _AvrPii-J. Following their amplification from the pGEM-T templates with the key primer pairs, AvrPii-C-GT_F/R (GT, genetic transformation) and AvrPii-J-GT_F/R, the transgene sequences were introduced into the binary vector pBHT2-AscI with an Asc I recognition site (Table S1) [7,8]. Before their transformation into a virulent recipient isolate CHL357, each construct was validated by sequencing. At least five independent hygromycin-resistant transformants per construct were carried forward for testing their functionality onto the Pii carrier, IRBLi-F5, and their specificity onto four monogenic lines carrying individual resistance genes, i.e., IRBL5-M (Pi5), IRBLk-Ka (Pik), IRBLb-B (Pib), and IRBLta-CP1 (Pita) [16]. The phenotype of each isolate was scored as A (0 reaction: avirulent), MA (1 to 2 reaction: moderately avirulent), MV (3 to 4 reaction: moderately virulent), and V (5 reaction: virulent) through the routine spraying inoculation procedure as described previously [2,5,17].

Construction and Characterization of Haplotype-Chimeric Mutant
Because multiple variations were scattered in the whole sequences of both haplotypes ( Figure S2), the haplotype-chimeric mutants were constructed via a domain-swapping approach ( Figure S3) [8]. To create the two haplotype-chimeric mutants in 5 and 3 regions, two chimeric templates were amplified with respective primer pairs, with the outside one being haplotype-specific (pure color arrow) and the inside one being haplotype-chimeric (chimeric color arrow); then the haplotype-chimeric mutants were generated by overlap-PCR with the pre-amplified fragment, both serving as megaprimers, for overlap extension. To create three haplotype-chimeric mutants targeted in the CDS region, self-ligation-PCR was conducted with a set of laddering chimeric primers using the TaKaRa MutanBEST Kit (TaKaRa, Dalian, China), according to the manufacturer's protocol (Table S1 and Figure S3).
Seven haplotype-chimeric mutants (Mai1-7) with two donor and one recipient isolates were phenotyped as aforementioned. Then the repeated and typical phenotype of each mutant was quantified based on Mo growth in the leaves of IRBLi-F5 by estimating the ratio between the copy number of a Mo sequence (Po2) and the rice Ubi gene (GenBank accession D12629) with qPCR system as described previously [8]. To establish relationships (if any) between phenotypes and protein structures of the haplotype-chimeric mutants, two models were employed to draw their 3D-structures (I-TASSER, https://zhanggroup.org/I-TASSER/; SWISS-MODEL, https://swissmodel.expasy.org/interactive) ( Figure S4).
Interactions of the wild types and three CDS derived haplotype-chimeric mutants with the paired Pii proteins were tested in the yeast two-hybrid (Y2H) system (Clontech, Dalian, China), in which the mature (20-70) and the full-length (FL) AvrPii proteins with the five versions, i.e., FL, CC (coiled coil), NBS (nucleotide binding site), LRR (leucine rich repeat), CtNL (carbon terminal non-LRR sequence), of the paired Pii protein (coded as #1, #2) were in reciprocal bait/prey combinations ( Figure S5). The detailed procedure was adopted from the previous studies [7,18].

Shape of Population Structure
Six Mo populations ((Guangdong (GD), Hunan (HN), and Guizhou (GZ) from southern; and Liaoning (LN), Jilin (JL), and Heilongjiang (HLJ) from northern China)), each consisting of 60 rice-derived isolates, were selected based on regionally representative rice cultivars and sites sampled (Table S2). An additional set of 17 weed-derived Mo isolates was used as outgroup for checking the origins of haplotypes. A total of 377 isolates were phenotyped on the Pii carrier as aforementioned. The isolates were genotyped with four markers, each two, a shorter fragment (MK1) tending to CDS, and a longer one (MK2) to 5 region for the individual haplotypes, by the key PCR primer pairs, AvrPii-C-MK1_F/R, AvrPii-C-MK2_F/R, AvrPii-J-MK1_F/R, and AvrPii-J-MK2_F/R. Each marker was given a genotype code of either 0 (absence of amplicon) or 1 (presence of amplicon). Then, the genotype of each isolate was simply scored as 0 (absence of AvrPii), 1-C (AvrPii-C), 1-J (AvrPii-J), and 1-C_J (AvrPii-C and AvrPii-J), when genotypes of each two markers for the individual haplotype were consistent. Taken together, the parameter of each combination of phenotype/genotype was counted out for each population for shaping its population structure.

Resequencing and Evolution Analysis
All the isolates with amplicons including weed-derived Mo isolates were re-amplified by the key primer pairs, AvrPii-C-RS_F/R (RS, resequencing), and/or AvrPii-J-RS_F/R. Particularly, an isolate belonging to genotype 1-C/J might gain two amplicons by both key primer pairs. Then all the amplicons carrying the FL sequences were subjected to resequencing, directly. The AvrPii allelic sequences were aligned with the known reference sequence of isolate Ina168 using DnaSP v6.12.03 software (http://www.ub.edu/dnasp/), and their CDSs were subjected to evolutionary analysis with the following parameters. Segregating site (S) was derived from pair comparison among isolates tested. Nucleotide diversity (π) was estimated by Nei's function with the Jukes and Cantor correction [19,20]. Selection on the family was evaluated by the Fu and Li's D* and F* parameters [21], and the average rate of non-synonymous (Ka) and synonymous (Ks) substitutions (Ka/Ks) was assessed as described by Nei and Gojobori [22]. Then, three kinds of key selection forces were briefly defined: the positive selection (also defined as Darwinian selection or directional selection) is featured with key evolutionary parameters, D < 0 and/or Ka/Ks (dN/dS) > 1, that leads to fixation of target allele that increases the fitness of individuals; the balancing selection (disruptive selection or diversified selection) is featured with key parameter, D > 0, that favors diversity; and the purifying selection (negative or stabilizing selection) is featured with key parameter, Ka/Ks ≤ 1, that eliminates deleterious mutations [23][24][25].
For deeply dissecting evolutionary features of the AvrPii family, six Mo natural populations were recombined into four comparable groups, i.e., all AvrPii group, haplotype AvrPii-J group, haplotype AvrPii-C group, and a regional group. The target genes were further divided into the entire coding region, non-signal peptide region (mature protein), and signal peptide region, if necessary. Additionally, 17 weed-derived isolates served as an outgroup for evolutionary comparison, if any.

Haplotypic Features of AvrPii
All the ten avirulent isolates selected from various Chinese Mo populations were amplified the expected CDS products with the key primer pair, AvrPii-J/C-CDS_F/R, whereas only the first five isolates were with the expected FL amplicons by the key primer pair, AvrPii-J-FL_F/R, indicating that sequences of the second primer pair might be less matched with those of the second five isolates ( Figure 1A,B). The key primer pair, AvrPii-C-FL_F/R, which was devised based on the specific sequences generated from both 5 and 3 regions of the representative isolates CHL346 via Tail-PCR system ( Figure S1), was employed to re-amplify the ten isolates. The reverse result was observed, i.e., the first five isolates were without any amplicon, except for #1, CHL22, and the second five one with the expected products. The reasonable explanation for genotype of the exceptional isolate CHL22 is that it carries two copies of AvrPii, i.e., AvrPii-J and AvrPii-C, which resembles some isolates shown in Table S2.
matched with those of the second five isolates ( Figure 1A,B). The key primer pair, AvrPii-C-FL_F/R, which was devised based on the specific sequences generated from both 5′ and 3′ regions of the representative isolates CHL346 via Tail-PCR system ( Figure S1), was employed to re-amplify the ten isolates. The reverse result was observed, i.e., the first five isolates were without any amplicon, except for #1, CHL22, and the second five one with the expected products. The reasonable explanation for genotype of the exceptional isolate CHL22 is that it carries two copies of AvrPii, i.e., AvrPii-J and AvrPii-C, which resembles some isolates shown in Table S2. Thus, the key primer pairs, AvrPii-J-RS_F/R, and AvrPii-C-RS_F/R, respectively, were used for amplifying the first and second five isolates for the haplotypic ampliconbased sequencing. As expected, there were extensive variations in both 5′ and 3′ regions, and intermediate ones in CDS region between the two haplotypes (two typical sequence difference as shown in Figures S1 and S2). The target gene from the first five isolates was Thus, the key primer pairs, AvrPii-J-RS_F/R, and AvrPii-C-RS_F/R, respectively, were used for amplifying the first and second five isolates for the haplotypic ampliconbased sequencing. As expected, there were extensive variations in both 5 and 3 regions, and intermediate ones in CDS region between the two haplotypes (two typical sequence difference as shown in Figures S1 and S2). The target gene from the first five isolates was temporally defined as AvrPii-J, and that from the second five isolates as AvrPii-C. Transgenic progenies derived from both haplotypes expressed a consistent functionality and specificity on five monogenic rice lines carrying Pii and other four individual resistance genes ( Figure 1C). Noticeably, both haplotypes expressed moderately avirulent to the Pii carrier.
Altogether, the new haplotype, AvrPii-C (GenBank MK830673), is a novel and functional one in the AvrPii family.

Performance of Haplotype-Chimeric Mutant
Compared to moderately avirulent performance of transgenic isolates derived from both wild haplotypes, reactions of each two haplotype-chimeric mutants derived from both 5 (Mai1-2) and 3 (Mai6-7) were quantified as moderately virulent, compared to the moderate avirulent symptoms caused by both wild types, where no fungal growth could be detected by qPCR (Figure 2). It suggests that both 5 and 3 regions were specific to the individual wild haplotypes because replacement of either 5 or 3 regions led to decline expression of their avirulence against the Pii carrier ( Figure S3; Table S3), while the performance of three haplotype-chimeric mutants, which were derived from CDS regions (Mai3-5), were almost changed from moderately avirulent to virulent. It indicates that the integrity of CDS was crucial to express functionality of the individual haplotypes. Correspondingly, the protein structures of CDS-derived chimeric mutants were distinct from both wild type ones, i.e., Mai1-2 and 6-7 ( Figure S4).
Altogether, the new haplotype, AvrPii-C (GenBank MK830673), is a novel and functional one in the AvrPii family.

Performance of Haplotype-Chimeric Mutant
Compared to moderately avirulent performance of transgenic isolates derived from both wild haplotypes, reactions of each two haplotype-chimeric mutants derived from both 5′ (Mai1-2) and 3′ (Mai6-7) were quantified as moderately virulent, compared to the moderate avirulent symptoms caused by both wild types, where no fungal growth could be detected by qPCR (Figure 2). It suggests that both 5′ and 3′ regions were specific to the individual wild haplotypes because replacement of either 5′ or 3′ regions led to decline expression of their avirulence against the Pii carrier ( Figure S3; Table S3), while the performance of three haplotype-chimeric mutants, which were derived from CDS regions (Mai3-5), were almost changed from moderately avirulent to virulent. It indicates that the integrity of CDS was crucial to express functionality of the individual haplotypes. Correspondingly, the protein structures of CDS-derived chimeric mutants were distinct from both wild type ones, i.e., Mai1-2 and 6-7 ( Figure S4).   Figure S5A,B).
The results were consistent with those reported in the previous studies [3,26]. There were three positive reactions only in Prey/Bait combinations between five full-length proteins AvrPii (Prey) and a series of domains of the paired proteins of Pii (Bait), i.e., a weaker interaction between AvrPii-J/C-3 FL and Pii-1 CC , and a stronger and weaker interactions, respectively, between AvrPii-C WT FL and Pii-1 CC and Pii-2 LRR (Figure S5C,D). These results indicate that the younger prey of the AvrPii-C, particularly the interval from #131 to #213 (Figure 2; Table S3), could be interacted with the cognate bait of the Pii, during their long-term co-evolution process.

Population Structure of the AvrPii Family
As for Pii-based phenotyping, frequencies of avirulent isolates ranged from 8.3% (GD) to 53.3% (HN), indicating that the AvrPii family was less avirulent and diverse in both southern and northern regions in China (Figure 3; Table S2). As for the two haplotypespecific marker-based genotyping, the results derived from the shorter fragment markers (MK1) were completely repeated by that from the longer one (MK2), thereby simply giving only four genotypes, 0, 1-C, 1-J, and 1-C_J, among six Mo populations derived from rice ( Figures 1A and 3A; Table S2). Intriguingly, combinations of phenotypes/genotypes were also only four, V/0, A/1-C, A/1-J, and A/1-C_J, because all the Pii-virulent isolates were restricted to genotype 0, whereas the Pii-avirulent ones were with the remaining three. Furthermore, all the four combinations were detected in the three southern populations, and only two A/1-C and V/0 in all the northern ones, suggesting that genic diversity in the southern region was higher than those in the northern one ( Figure 3B). Together, the AvrPii family was rarer and less diverse in all the six Mo populations, particularly in the northern region. qPCR approach. The plant and fungal DNA, respectively, were inferred from the abundance of the rice Ubi and the Mo Pot2 amplicons; the data shown was given in the form mean ± standard deviation (n = 3). Similar results were obtained from two biological replicates each with three technical repeats.
Furthermore, there was not any positive interaction in Bait/Prey reciprocal combinations between four mature proteins (AvrPii-J WT 20-70, AvrPii-J/C-220-70, AvrPii-J/C-320-70, and AvrPii-C WT 20-70) and a series of domains of the paired proteins of Pii ( Figure S5A,B). The results were consistent with those reported in the previous studies [3,26]. There were three positive reactions only in Prey/Bait combinations between five full-length proteins AvrPii (Prey) and a series of domains of the paired proteins of Pii (Bait), i.e., a weaker interaction between AvrPii-J/C-3FL and Pii-1CC, and a stronger and weaker interactions, respectively, between AvrPii-C WT FL and Pii-1CC and Pii-2LRR ( Figure S5C,D). These results indicate that the younger prey of the AvrPii-C, particularly the interval from #131 to #213 (Figure 2; Table S3), could be interacted with the cognate bait of the Pii, during their long-term coevolution process.

Population Structure of the AvrPii Family
As for Pii-based phenotyping, frequencies of avirulent isolates ranged from 8.3% (GD) to 53.3% (HN), indicating that the AvrPii family was less avirulent and diverse in both southern and northern regions in China (Figure 3; Table S2). As for the two haplotype-specific marker-based genotyping, the results derived from the shorter fragment markers (MK1) were completely repeated by that from the longer one (MK2), thereby simply giving only four genotypes, 0, 1-C, 1-J, and 1-C_J, among six Mo populations derived from rice ( Figures 1A and 3A; Table S2). Intriguingly, combinations of phenotypes/genotypes were also only four, V/0, A/1-C, A/1-J, and A/1-C_J, because all the Piivirulent isolates were restricted to genotype 0, whereas the Pii-avirulent ones were with the remaining three. Furthermore, all the four combinations were detected in the three southern populations, and only two A/1-C and V/0 in all the northern ones, suggesting that genic diversity in the southern region was higher than those in the northern one (Figure 3B). Together, the AvrPii family was rarer and less diverse in all the six Mo populations, particularly in the northern region. Regarding outgroup, all 17 weed-derived isolates could not infect rice, and only seven were identified as A/1-J, indicating that the AvrPii-J, which emerged before rice domestication, is older than AvrPii-C ( Figure 3B; Table S2).

Evolutionary and Regional Divergence of the AvrPii Family
The genetic diversity of the AvrPii family among the six populations was explored in some detail by re-sequencing all 110 amplicons derived from 96 isolates, of which 14 carry both haplotypes ( Figure 3B; Table S2). The CDS harbored 20 SNPs, of which eleven were non-synonymous. The non-signal domain harbored 18 of the SNP sites, while the other two were within the signal peptide domain; as a result, the level of nucleotide diversity in the former domain (0.045) was higher than in the latter (0.013) ( Table 1). As to the individual haplotypes, only one segregating site was found in AvrPii-C group and none in AvrPii-J one, indicating that there were extremely less diverse in both haplotypes. Regarding the regional groups, both haplotypes were identified in the southern group, thereby raising 20 segregating sites for reaching a higher level (0.045) of the AvrPii diversity. On the contrary, only one haplotype, AvrPii-C, was placed in the northern one without any segregating site. Again, it was clearly suggested that diversity of the AvrPii in the southern group was largely higher than that in the northern one.   Figure 3, particularly, 14 isolates carried both AvrPii-C and AvrPii-J, those were courted out two times (also see details in Table S2). b Number of segregating sites. c Nei's nucleotide diversity based on silent site. d Fu and Li's D* and F*, and α , β represents statistical significance at p < 0.05, 0.02 level, respectively. e Nonsynonymous. f Synonymous.
In all AvrPii group, the first set of selection-related parameters D* and F* in both entire (D* = 1.74; F* = 2.65) and non-signal peptide region (D* = 1.68; F* = 2.59) were positive and higher than those in signal peptide region (D* = 0.68; F* = 1.11), indicating that the target locus has been experienced a stronger balancing selection in both regions and a slighter one in signal peptide region (Table 1). In the haplotype groups, only slight balancing selection pressure (D* = 0.51; F* = 0.79) was detected within the haplotype AvrPii-C group, and none in the haplotype AvrPii-J one. As to the regional group, there has been a stronger balancing selection pressure (D* = 1.72; F* = 3.04) on the AvrPii family in the southern group, but none in the northern one. As to the second set of selection-related parameters (Ka, Ks, and Ka/Ks), Ka/Ks in both all AvrPii and the southern groups were less than 0.57, indicating that the AvrPii family has been imposed by a stronger purifying selection (negative selection) in both groups. While a slight positive selection (Ka > Ks) was detected within the haplotype AvrPii-C group. Intriguingly, neither positive nor balancing selection pressures was detected in haplotype AvrPii-J, the regional north, and out groups (Table 1).

The Lesson Learned from Case Study on Population Structures of the AvrPii Family
Having identified the novel haplotype, AvrPii-C, dynamic population structures of the AvrPii family were finely dissected among six Mo populations, each three from southern and northern regions of China (Figures 1-3). Whereas the apparent contradiction between the higher detected number of avirulent isolates and the lower number of detected avirulence genes of AvrPii were observed in various Mo populations previously [3,11,14,15], compared to the lower rate of presence of AvrPii being detected, the higher rate of Pii-avirulent isolates was partly due to Pii independent that was conveyed by another gene pair of Avr/R gene in such combinations [3]. It was, however, certainly conceivable that failing in detecting a yet unknown haplotype AvrPii-C in such Pii-avirulent isolates as reported in the current study was the major reason for creating the numerical gaps and/or the lower rates of presence of AvrPii. When revisiting all PCR primer sequences used in the previous studies [3,11,14,15], it was revealed that all such primers were designed from the known reference sequence of the AvrPii-J (AB498874 that resembles to CHL22 as shown in Figure S1) [2], those were not matching enough for amplifying the AvrPii-C as shown in Figure 1. Because the proportion of the AvrPii-C was almost five times higher than that of the AvrPii-J in the six Mo populations tested in the present research (Figure 3), in addition to the fact that higher resistance frequencies of the Pii carrier IRBLi-F5 were detected in Japan [27], Bangladesh [28], Kenya [29], and Vietnam [30], it was conceivable that the numerical gaps and/or extremely lower proportions of presence of AvrPii in various populations were largely due to missing the novel haplotype, AvrPii-C [3,11,14,15].
Considering haplotype divergence is one of the most important mechanisms underpinning Avr gene evolution in various plant pathosystems [3,7,15,[31][32][33][34][35][36][37][38], the lesson learned from case studies on population structures of the AvrPii family is that much attention should be focused on haplotype divergence of the target gene when there were certain numerical gaps or extremely lower proportions of presence of the target gene as reported in the previous investigations [3,11,14,15]. Briefly, the population structure of individual target gene(s) is better dissected by both phenotypic and genotypic datasets so that the reasons for the numerical gap and/or extremely lower proportion of target gene could be easily found. If done, the specific sequence for unknown haplotype, which is distinct from the known one, should be created via the Tail-PCR system in a case of fewer-than-enough reference sequences available, and/or through gene similarity analysis using reference sequence of the known one for inquiring whether there was/were candidate distinct haplotype(s) in public database (as shown in Figures S1 and S2). Then primer sets for the novel haplotype, if any, should be re-devised based the haplotype-specific sequence for genotyping and resequencing (as shown in Figure 1) [3]. Furthermore, the novel haplotype(s) might be functionally characterized via transformation test, if possible and necessary.

The Lesson Learned from Evolutionary Story of the AvrPii Family
Accumulating evidence from population genetic study on various Avr genes demonstrated that the gene-for-gene principle has played a pivotal role in featuring their population structures [3,[7][8][9]11,34,39,40]. As per the definitions of three kinds of the key selections, stronger balancing and purifying selection forces were exerted on the Chinese populations and only positive selection pressure was exerted on the Thai population, when datasets available in both Lu et al., 2019 [3] and Sirisathaworn et al., 2017 [11] were re-assayed by the same approaches employed in the current study (Table S4). As pathogen Avr genes have generally evolved in response to positive selection [7][8][9]11,[39][40][41], the unique population structures of the AvrPii family found in China have significant implications for understanding how the AvrPii family has kept an artful balance and purity among its mem-bers (haplotypes), those keenly interact with the cognate R gene, Pii, under gene-for-gene relationships.
Furthermore, several intriguing points, which were extracted from five comparable groups in the current study, should be marked. First, it could be clearly deemed that the haplotype AvrPii-C was a mutant emerged after rice domestication, and the haplotype AvrPii-J as wild type has emerged before rice domestication (Figure 3; Table S2). Second, the population structures between haplotypes AvrPii-C and AvrPii-J groups were slightly different, the former one received weaker balancing and positive selection forces, thereby scattering in all the six Mo populations across China, and the latter one has received a weaker purifying and even neutral selection pressure, thereby being maintained and restricted in the three southern populations (Table 1; Figure 3). Third, the population structures between southern and northern groups were distinct. The former one has experienced both stronger balancing and purifying selection pressure, thereby keeping all the four combinations of phenotypes/genotypes (A/1-J, A/1-C, A/1-C_J, and V/0), and the latter one has received a weaker purifying and even neutral selection pressure, thereby keeping maintained only two (A/1-C and V/0), which led to show certainly higher genetic diversity in the southern region. Fourth, higher frequencies of avirulent isolates were detected in southern HN (53.3%) and GZ (35.0%) and in northern LN (40.0%), but over three-fold lower ones in their neighboring regions, i.e., the southern GD (8.3%), and the northern JL (13.3%) and HLJ (10.0%) ( Figure 3). That, in turn, indicated that the cognate R gene, Pii, could be continuously used as basic and critical resistance resource in HN, GZ, and LN populations, even though it was recognized as one of the extremely rarer and weaker R genes in the Chinese rice breeding programs [2,5,12,[42][43][44]. Again, it provided further weight to the idea that the AvrPii family was a smarter one to keep a sophisticated balance and purity not only its members but also its living places.
It is worthy learning more marvelous lessons from evolution stories of the AvrPii family featured by its balanced, purified, and directed haplotypes in global Mo populations including trans-species ones.

Conclusions
The study has identified a sequence-distinct and functional haplotype AvrPii-C, compared to the known one, AvrPii-J. It was believed that failing in amplifying of AvrPii-C was the major reason for causing the numerical gaps between higher proportions of the Pii-avirulent isolates and/or extremely lower proportions of presence of AvrPii in various populations previously reported. The unique population structure of the AvrPii family was shaped by its balanced, purified, and directed haplotypes in the Chinese populations. The AvrPii-C was recognized as a younger one that emerged after rice domestication. The cognate resistance gene Pii could be continuously used as a basic and critical resistance resource in Hunan, Guizhou, and Liaoning provinces, China.

Patent
Qinghua Pan, Xing Wang, Cheng Li, Jianqiang Wen, Yu Zhang, Ling Wang. Identification and applications of the avirulence gene AvrPii-C of Magnaporthe oryzae (ZL201811273480.3, Granted on 14 August 2020) Supplementary Materials: The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/genes14051065/s1, Figure S1: Discovery of a novel haplotype AvrPii-C in Chinese populations of M. oryzae; Figure S2: Alignment of the full-length sequences of both haplotypes, AvrPii-J and AvrPii-C; Figure S3: Constructure of haplotype-chimeric mutants by PCRbased domain swapping from AvrPii-J to AvrPii-C; Figure S4: The 3D-structures of a set of five AvrPii proteins; Figure S5: Interactions between AvrPii and Pii proteins in the Y2H system; Table S1. PCR primers and programs used in the current study; Table S2. Genotypes and phenotypes of 377 isolates of M. oryzae characterized in the current study [45]; Table S3. The full-length sequences of seven haplotype-chimeric mutants of the AvrPii family; Table S4. Comparison of evolutionary parameters of the AvrPii family among three Mo populations, of which parameters of the two populations previously reported were re-computed by the same approaches used in the current study.