We collected rice blast-infected leaf samples from rice blast hotspots in 13 nurseries of the Sichuan Basin, including DZ, YS, TJ, YA, QW, ST, QS, ZZ, PX, NX, NB, AZ, and JY (Figure 1A). Then, single-spore isolation method was used to establish single-spore isolates following a previous report (Jia, 2009). A total of 383 single-spore isolates were obtained from the infected leaves of the susceptible variety Lijiangxin Tuan Heigu (LTH). Isolates were cultured on PDA medium (200 g/l potato, 20 g/l dextrose, and 13 g/l agar) in a growth chamber at 25°C in a 12-h light/12-h dark photoperiod for 7 days and were stored at −20°C on filter paper.

The fungal isolates were cultured on PDA medium in plates covered with cellophane membrane at 25°C in a 12-h light/12-h dark photoperiod for 7 days. Then, mycelia were harvested and frozen in liquid nitrogen for grinding to fine powder. Total genomic DNA was extracted via cetyltrimethylammonium bromide (CTAB) method [10 mM Tris-HCl, pH 8.0, 1 mM ethylenediaminetetraacetic acid (EDTA), 100 mM NaCl, 2% sodium dodecyl sulfate (SDS)] for all the fungal isolates (Saghai-Maroof et al., 1984). PCR was performed with gene-specific primers (Supplementary Table 1) for amplifying eight avirulence genes, namely, AVR-Pib, AVR-Pik, AVR-Pia, AVR-Pi9, AVR-Pi54, AVR-Pita1, AVR-Pizt, and AVR-Pii from isolated DNA samples. Each PCR reaction contains 2.5 μl of 10X EasyTaq buffer, 0.25 μl of 5 U/μl EasyTaq DNA polymerase (TransGen Biotech Corporation Ltd., Beijing, China), 0.5 μl of 10 mM deoxynucleotide triphosphates (dNTPs), 0.5 μl of each 10 μM primer, 1 μl of 20–50 ng/μl fungal genomic DNA, and 19.5 μl of distilled water. Amplifications were performed in the Bio-Rad Thermal Cycler (C1000, Bio-Rad Laboratories, Life Science Research, California, USA) with the following PCR program: 95°C for 3 min, 30 cycles at 95°C for 30 s, 52 to 58°C for 30 s, 72°C for 30 s, and a final extension at 72°C for 8 min. PCR products were examined by electrophoresis in 1% agarose gel using 120 V for 30 min and the gel was stained with GelRed Nucleic Acid Stain (Biotium, USA). The sizes of PCR amplicons were estimated by Trans2000 DNA Ladder (TransGen Biotech Corporation Ltd., Beijing, China). The PCR amplicons were sequenced by TsingKe Biotech Corporation Ltd. (Beijing, China). The obtained sequences were aligned with the sequences of each AVR gene to detect natural variations.

Rice blast fungal isolates on PDA medium were cultured on OTA media (50 g/l oatmeal, 200 ml/l tomato juice, and 15 g/l agar) at 25°C with 12-h light followed by 12-h dark treatment for producing mycelium. 7 days later, the mycelia were scraped with a cell spreader and the plates were exposed to continuous light at 28°C for 3 to 5 days to promote sporulation. Then, the conidia were washed with 2 ml double-distilled water (ddH₂O) per petri dish and the suspension was adjusted the concentration to 1 × 10⁵ conidia/ml.

The indicated lines and susceptible control LTH were used for pathogenicity assays in laboratory. The rice seeds were immersed in water for 2 days at 37°C in darkness for germination and then grown in the greenhouse with 28/24 ± 1°C day/night temperature, 70% relative humidity, and 14-h/10-h light/dark period. At three-leaf to five-leaf seedling stage, the second youngest leaves were detached and placed on the ddH₂O with 6-benzylaminopurinehydrochloride (6-BA) in square petri dish after punch wounded. Then, the punch-wounded leaves were drop inoculated with 10 μl of spore suspension at the wound sites as a previous report (Kong et al., 2012). The inoculated leaves in petri dish were incubated in dark at 26°C in the greenhouse with 90% relative humidity for 24 h (Fang et al., 2018). Lesion size was examined at 4 to 7 days after inoculation to assess the pathogenicity of isolates.

The DNA sequences were assembled and aligned with DNAMAN9 (Lynnon Biosoft) and the reference sequences of AVR genes were obtained from the National Center for Biotechnology Information (NCBI) with GenBank accessions given in Supplementary Table 1.

We isolated the single spores from diseased leaves of susceptible variety LTH in 13 nurseries of the Sichuan Basin (Figure 1A). Finally, 383 isolates were obtained and subjected to investigate the variations of AVR genes. We analyzed the nucleotide sequence polymorphisms of eight known AVRs, namely, AVR-Pi9, AVR-Pi54, AVR-Pita1, AVR-Pib, AVR-Pia, AVR-Pizt, AVR-Pii, and AVR-Pik. First, we investigated their presence based on PCR amplification and classified the 383 isolates into the 16 race groups based on the presence of AVR genes. We found that five AVR genes, namely, AVR-Pik, AVR-Pi54, AVR-Pizt, AVR-Pi9, and AVR-Pib, were present in all the nurseries and the isolates containing the five AVR genes formed the major race groups of M. oryzae in the Sichuan Basin (Figures 1B,C). Nevertheless, AVR-Pita1 was present in the isolates from ST, TJ, and DZ, but almost (the present ratio < 0.1) or completely absent in five rice blast nurseries, including ZZ, AZ, PX, JY, and NB. AVR-Pii was present in the isolates of ZZ, but almost or completely absent in six rice blast nurseries, namely, ST, TJ, NX, PX, JY, and NB (Figure 1B). AVR-Pia was only present in a few isolates obtained from ZZ, DZ, and YS.

We speculated that the isolates that did not contain AVR-Pia may become virulent to the monogenic IRBL line carrying cognate R gene Pia. We inoculated these isolates to IRBLa-A. As our expectation, IRBLa-A showed susceptibility to all the isolates (Figure 2), indicating that loss of AVR-Pia breaks down the resistance of Pia. Similarly, those strains that did not contain AVR-Pita1, AVR-Pii, and AVR-Pia may breakdown their cognate R genes.

We classified the race group of M. oryzae based on their harbored AVR genes. The 383 isolates were divided into the 16 race groups (Figure 1C). The groups C and D are the two largest groups consisting of 156 and 110 isolates, respectively. The groups G and A consist of 64 and 18 isolates, respectively. The rest groups each consist of fewer than 10 isolates. To estimate the diversity of M. oryzae in each prefecture, we calculated the number of the race groups in each nursery (Figure 1B). There are 11 groups existed in YS, suggesting that it has the most abundant diversity of M. oryzae. The rest nurseries have one to seven groups. Among them, the seven groups were identified in YA, while only the one and two groups were identified in ST and ZZ, respectively. There are similar amounts of detected isolates in these nurseries, such as 20 in YA, 19 in ST, and 19 in ZZ (Figure 1A). These data indicate that the diversity of M. oryzae shows distinct geographic feature in the Sichuan Basin.

Although AVR-Pib has high presence in most isolates (Figure 1B), its cognate Pib gene has been lost resistance in the Sichuan Basin (Zhang et al., 2017). Therefore, we investigated the sequence diversity of AVR-Pib. The genomic sequence covering the coding sequence (CDS) region was amplified via PCR using AVR-Pib-specific primers (Figure 3A; Supplementary Table 1). The PCR products can be divided into four different amplification patterns (APs) to denote no amplicon (AP0) and different sizes of amplicons (AP1-PA3) (Figures 3A,B). These APs include 10 variations that are designed haplotype 1 (H1) to haplotype 10 (H10) (Figure 3C). AP1 is the biggest PCR amplicon with over 2,000 bps, which was detected in 235 isolates and is larger than the expected size of AVR-Pib (GeneBank: KM887844.1), indicating that there is an insertion. Consistently, sequencing analysis identified an 1,861 bp insertion that shared 99.84% nucleotide sequence identity to the transposon Pot3 (GeneBank: AF333034.1) and flanked with two base pairs (5′-TA-3′) target site duplication (TSD). The insertion locates at four novel sites in 5′-untranslated region (5′-UTR) of AVR-Pib, including at −54, −231, −235, and −241 bp upstream of the start codon, which belong to H2, H3, H4, and H5 (Figure 3C). We developed an internal primer pairs, AvrPibF2/AvrPib, which could distinguish H2 from H3, H4, and H5 (Figures 3A–C). AP2 was detected in seven isolates from YA with near 1,000 bp, which defined as H1. Sequence alignment revealed that AP2 had a 482-bp insertion at the position of −85 bp from the start codon of AVR-Pib. This insertion showed 97.1% nucleotide sequence identity to the non-LTR retrotransposon Mg-SINE (GeneBank: U35313.1) and was flanked with a 12-bp sequence (CTTTTGCTTCGA) (Figure 3C). AP3 matches with the expected size of AVR-Pib (Figure 3B). Sequencing analysis identified several indels in the CDS and 5′-UTR of AVR-Pib (Figure 3C). A deletion occurs at positions 37 and 38 of the CDS region (H6) and several insertions occur in the 5′-UTR, including insertion of AAC at −124, C at −216 or −220, TAACT or TAAGT at −231, and TAACGT at −235 (H7-H9) (Figure 3C). Moreover, the sequencing chromatogram uncovered that some AP3 has double peaks, which decoded into two types of sequences with insertions at −231, −216, and −124 (H10). One sequence has insertion of TAACGT at −231, C at −216, and AAC at −124 and another sequence has insertion of TAACT at −231, C at −216, and AAC at −124 (Figure 3C). Besides, we failed to obtain amplicons from 15 isolates, which were from YS (5 isolates), TJ (2 isolates), YA (6 isolates), and QW (2 isolates) (Supplementary Table 2). These data indicate that the diverse ways were employed by AVR-Pib for escaping from R protein detection, such as insertion of Pot3, Mg-SINE, indels, and gene loss.

Then, we tested whether the isolates harboring AVR-Pib variations were virulent to IRBLb-B containing cognate Pi-b. To this end, we performed pathogenicity assays and found that isolates with short insertions are avirulent to IRBLb-B, including H7, H8, H9, and H10 (Figures 3C,D), indicating that the short sequence insertions in 5′-UTR may not impact the function of AVR-Pib. In contrast, the isolates virulent to IRBLb-B carry the other four types of variations, including Mg-SINE insertion (H1), Pot3 insertion (H2-H5), deletion in CDS (H3), and absence of AVR-Pib (Figures 3C,D), indicating that these variations caused function loss of AVR-Pib to breakdown of the Pib resistance.

To further investigate the perniciousness of invalid AVR-Pib in the Sichuan Basin, we calculated the ratios of AVR-Pib variants in all the isolates (Figure 4). Our data indicate that 26.63% isolates contain the valid AVR-Pib, i.e., wild type (WT) and short sequence insertions in 5′-UTR of AVR-Pib. The invalid AVR-Pib exists in 73.37% isolates, consisting with the conclusion that its cognate Pib gene has been lost resistance in the Sichuan Basin (Zhang et al., 2017). There are four variations of invalid AVR-Pib, Pot3 insertion, 2 bp deletion in CDS (CDS 37–38^Del), absence, and Mg-SINE insertion, which, respectively, exist in 61.36, 6.27, 3.91, and 1.83% isolates, suggesting that Pot3 insertion is the major way to invalidate AVR-Pib. In addition, the isolates carrying Pot3 insertion extensively exist in 11 nurseries, except YA and QW, and preponderantly in seven nurseries, namely, QS, ZZ, AZ, JY, ST, PX, and DZ. Variants with 2 bp deletion in CDS and Mg-SINE insertion of AVR-Pib were only detected in part of isolates from YS and YA, respectively. Some isolates lost AVR-Pib gene in YS, YA, TJ, and QW. These varieties of invalid AVR-Pib were carried by most isolates in YS and YA. However, the valid AVR-Pib existed in over half of isolates in QW, NB, NX, and TJ (Figure 4), hinting that the Pib gene may still confer resistance in these four regions, especially in QW and NX.

Previously, insertions were reported in the 5′-UTR of AVR-Pizt, resulting in the function loss of AVR-Pizt and breakdown of the Pizt resistance (Wang et al., 2020). The fragment was amplified via PCR with primers AvrPiztF and AvrPiztR (Figure 5A). There are two sizes of the amplicons obtained in 97.91% of the isolates, but failed in a few isolates from YS, QS, ZZ, and NX (Figure 5B; Supplementary Table 2). Sequencing analysis indicated that there is an insertion of 201 bp at −181 bp upstream of the start codon of AVR-Pizt (Figure 5A). The insertion shared 96% sequence identity to solo-LTR of retrotransposon Inago2 (GeneBank: AB334125.1). The insertion was detected in 59 isolates obtained from the nurseries of YS and DZ (Supplementary Table 2). According to the pathogenicity assays, three representative isolates, namely, YS046, YS125, and DZ25, harboring the insertion are virulent to the monogenic rice blast resistance line IRBLzt-T carrying Pizt (Figure 5C), indicating that the insertion disrupts the avirulent function of AVR-Pizt, leading to breakdown of the Pizt-mediated resistance in the Sichuan Basin.

AVR-Pita1 locates close to the telomere region on the third chromosome. Some literatures reported that AVR-Pita1 is easy to be lost or exists as different avirulent haplotypes (Zhou et al., 2007; Singh et al., 2014; Damchuay et al., 2020). In this study, we used gene-specific primers to amplify fragments containing the whole CDS region of AVR-Pita1 (Supplementary Table 1). The expected size fragment was successfully amplified from 191 (49.9%) isolates. Analyzing the sequence of fragment, we identified five variations for AVR-Pita1 (Supplementary Table 3). All these variations had been reported to be still avirulent (Li et al., 2014; Damchuay et al., 2020). However, over half of isolates (192, 50.1%) failed to produce an amplificon, indicating that AVR-Pita1 is losing and going to breakdown the resistance function of Pita in the Sichuan Basin.

To investigate the natural variation of AVR-Pik in the field isolates of the Sichuan Basin, we analyzed the sequence of CDS region of AVR-Pik. The AVR-Pik gene was successfully amplified from all the isolates, indicating that AVR-Pik exists widely in the isolates from the Sichuan Basin. Sequencing analysis identified five AVR-Pik alleles, namely, AVR-PikA, AVR-PikB, AVR-PikC, AVR-PikD, and AVR-PikE (Figure 6A). Among them, single AVR-PikA, AVR-PikB, AVR-PikC, or AVR-PikD was characterized in less than 10% isolates. Especially, AVR-PikC is present only in 0.26% isolates. In contrast, AVR-PikE is abundant and exists in 35.25% isolates. The rest isolates (53.00%) were characterized as harboring two AVR-Pik alleles, namely, AVR-PikD + AVR-PikA and AVR-PikD + AVR-PikE, which were detected in 42.56 and 10.44% isolates, respectively (Figure 6B). Therefore, AVR-PikE and AVR-PikD + AVR-PikA are dominant AVR-Pik alleles in the blast fungal populations in the Sichuan Basin. In addition, we identified a new deletion of CTTT, which happened at −42 to −39 upstream of the start codon of AVR-PikD in the AVR-PikD + AVR-PikE harbored isolates (Figure 6A).

Meanwhile, we analyzed the distribution of different AVR-Pik alleles in the 13 nurseries (Figure 6C). Single AVR-PikA was found in isolates from YA (25%) and NX (90%). AVR-PikB was detected in isolates from NB (100%), PX (81.82%), and AZ (58.33%). AVR-PikC was detected only in one isolate from YS. Single AVR-PikD was detected in YA (5%) and QW (20%). Single AVR-PikE was detected in five nurseries that accounted for 63.37, 70, 80, 86.96, and 94.74% of the isolates from YS, YA, QW, QS, and ZZ, respectively. Single AVR-PikE was also detected in two isolates from DZ and five isolates from AZ. Double alleles of AVR-PikD + AVR-PikA were detected in DZ (97.62%), YS (0.99%), TJ (100%), ST (100%), PX (18.18%), and JY (100%), whereas another double alleles of AVR-PikD + AVR-PikE were found in YS (34.65%), QS (13.04%), ZZ (5.26%), and NX (10%). Taken together, AVR-PikA was the dominant allele of AVR-Pik in the nursery of NX. AVR-PikB was dominant in the PX, NB, and AZ. AVR-PikE was dominant in the YS, YA, QW, QS, and ZZ. AVR-PikD + AVR-PikA were dominant in the JY, TJ, ST, and DZ. These data indicate that AVR-Pik alleles existed in all the isolates with distinct geographical feature and, thus, cultivars carrying Pikm can be exploited in the Sichuan Basin.

Specific primers for MAT1-1 and MAT1-2 (Supplementary Table 1) were used to investigate the mating type composition of isolates in the M. oryzae population. In total, 171 isolates are MAT1-1 and 174 isolates are MAT1-2 (Figure 7A), indicating that the distribution of MAT1-1 and MAT1-2 is relatively balanced in the Sichuan Basin. In addition, we also identified hermaphroditic type in 38 isolates, which have both the mating types simultaneously. Single mating type was found in the isolates from 6 of 13 nurseries, namely, ST, PX, NB, JY, QS, and ZZ. Two mating types were detected in the rest seven nurseries and the hermaphroditic isolates were found in TJ, DZ, and YA (Figure 7B). Even though the sexual status is hardly observed under natural conditions, sexual reproduction potential is still present in regions where the isolates have both the mating types, hinting that the parasexual recombination may be an important way to promote variation of AVR genes in the Sichuan Basin.