Pathogen infection causes tragic damage to host plants and serious reduction of crop yield. Resulting from their co-evolution, pathogens and plants have developed efficient mechanisms to evade detection and activate disease resistance, respectively (Huang et al., 2014). Pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) and effector-triggered immunity (ETI) have emerged as two key immunity layers in plants. Whereas PTI is associated with the perception of PAMPs by pattern recognition receptors (PRRs), ETI is the second tier of immunity where the plant recognizes secreted effector proteins which, in the absence of resistance proteins, might dampen PTI and benefit the pathogen (Jones and Dangl, 2006). ETI is often associated with rapid development of programmed cell death at the site of infection termed the hypersensitive response (HR) (Maekawa et al., 2011), a strong resistance mechanism that prevents microbial proliferation. The specific recognition of pathogen-derived avirulence (Avr) effectors by corresponding disease resistance (R) proteins activates ETI in plants. R proteins typically carry nucleotide-binding (NB) and leucine-rich repeat (LRR) (NLR) domains with varying N-terminal toll/interleukin 1 or coiled-coil domains. Recently, it was discovered that some NLRs carry effector-sensing decoy domains at their C-terminal regions (Cui et al., 2015).

In order to understand the mechanism of disease resistance to the rice blast fungus Magnaporthe oryzae, many PRRs, NLRs and resistance-associated quantitative trait loci were identified and functionally investigated using monogenic rice lines (Inukai et al., 1996; Jia and Liu, 2011; Mackill, 1992; Tsunematsu et al., 2000). In particular, most of the cloned genes encode NLR proteins (Singh et al., 2015; Wang et al., 2014). In addition, various mechanisms by which rice NLRs recognize corresponding Avr effectors have been proposed (Wu et al., 2014; Zhang et al., 2013). These include Pi-ta-dependent direct recognition of the M. oryzae Avr effector, AVR-Pita (Jia et al., 2000) and indirect interaction of Piz-t and AVR-Piz-t (Park et al., 2016). More recently, the paired NLRs, RGA4 and RGA5, were shown to function together in recognition of the M. oryzae Avr effectors AVR1-CO39 and AVR-Pia (Césari et al., 2014). Although the mechanistic basis differs between each other, these recognition events activate downstream defense signaling and efficiently restrict pathogen proliferation in rice.

Despite the effective disease resistance against blast fungus conferred by NLRs in rice, the rapid onset of genetic variability and pathogenicity of blast fungus has led to the disarmament of resistance conferred by specific R genes. Hence, plants have evolved a repertoire of R genes to counter pathogen arsenals, as evidenced by sequence polymorphisms in various cultivars (Chisholm et al., 2006; Jacob et al., 2013; McDonald and Linde, 2002; Schulze-Lefert and Panstruga, 2011). In several cases, allelic sequence diversity of NLR genes have caused functional variation, especially in regards to Avr recognition. For instance, the Pi2/Pi9 locus contains six NLR-coding genes, of which three cloned genes, Pi2, Pi9, and Piz-t, confer different resistance specificities. Further study indicated that the difference is the result of a few amino acid changes (Wu et al., 2012). Among 20 sequence polymorphisms detected in 24 Pita natural variants, only the 918th amino acid is critical for AVR-Pita recognition (Bryan et al., 2000; Wang et al., 2016).

Pi5 and Pii genes were identified in a number of rice cultivars. The similarity of resistance specificity conferred by Pii and Pi5 was reported previously (Ebron et al., 2004; Selisana et al., 2017). Consistently, in polymerase chain reaction (PCR) analysis of 24 different monogenic lines, we found that Pi5-specific markers are present only in Pi5-, Pi3-, and Pii-carrying lines (Yi et al., 2004). In a separate study, Pi3 was shown to be identical to Pi5 based on DNA gel-blot and genomic sequence analyses (Jeon et al., 2003). The Pi5 gene was identified and cloned from RIL260, a known rice cultivar resistant to M. oryzae isolates collected from Korea and the Philippines, by a map-based cloning method (Lee et al., 2009). Interestingly, it was shown that Pi5-mediated resistance requires two adjacently located NLR genes, Pi5-1 and Pi5-2, which was demonstrated by coexpressing their genomic DNA fragments in the susceptible cultivar Dongjin. Therefore, we hypothesized that Pii-mediated resistance requires a pair of NLR genes that are allelic to those of Pi5. Here in this work, we cloned the Pii resistance genes and analyzed the allelism relationship by complementation test.

Genomic DNA PCR and reverse transcription PCR (RT-PCR) analysis

Twelve PCR primer sets (Supplementary Table S1) were designed across the Pi5-1 and Pi5-2 genomic DNA sequences to amplify the counterparts in Pii-carrying cultivar Fujisaka5. PCR amplification was performed in a final volume of 40 μl. The Taq DNA polymerase (BIOFACT, Korea) was used according to the manufacturer’s instructions. The PCR product purified was sequenced. The whole sequence of Pii was assembled from the fragments sequenced separately and aligned to Pi5 sequences using the ClustalW2 program.

To analyze gene expression, total RNA from leaf blades of four-week-old transgenic plants was extracted using RNAiso Plus according to the manufacturer’s protocol (Takara Bio, Japan). Reverse transcription was performed according to the manufacturer’s protocol using ReverTra Ace^® qPCR RT Master Mix with gDNA Remover (Toyobo, Japan). The cDNA obtained was used as template to measure expression by quantitative PCR (qPCR) with gene-specific primers (Supplementary Table S1). Three biological replications were carried out for each sample.

Vector construction and generation of transgenic rice plants

Full-length cDNAs of Pi5-1 and Pi5-2, and Pii-1 and Pii-2 were isolated by RT-PCR from leaves of RIL260 carrying Pi5 and Fujisaka5 carrying Pii, respectively. The cDNAs subcloned into pJET1.2 (Thermo Scientific, USA) were transferred to the plant overexpression vectors pPZP2Ha3 (Fuse et al., 2001), pGWB514 and pGWB518 (Nakagawa et al., 2007), and a modified pCAMBIA3301 under the control of CaMV35S promoter (35S). To construct the RNAi vectors, unique sequences of Pi5-1 and Pi5-2 were amplified by PCR with gene-specific primers (Supplementary Table S1) and inserted into the pANDA vector with the antisense sequence upstream of the sense one as described (Miki and Shimamoto, 2004). To create OsExo70F3 mutants, a designed guide RNA (5′-TCCCATCACCCCGCACAGCC-3′) was cloned into an entry vector, pOs-sgRNA. The resulting sgRNA expression cassette was transferred into pH-Ubi-cas9-7 to construct the Exo70F3 CRISPR/Cas9 vector (Miao et al., 2013).

To produce transgenic rice plants, the Agrobacterium tumefaciens LBA4404 strains harboring the constructed vectors were grown for three days at 28°C, and transgenic calli were obtained via the Agrobacterium-mediated co-cultivation method as described previously (Jeon et al., 2000). Pi5-1, Pi5-2 and Pii-1 cDNA constructs were introduced to a susceptible cultivar Dongjin using hygromycin selection. Pii-2 and Pii-2Δ cDNA constructs were introduced into Pii-1 cDNA transgenic rice using phosphinothricin selection as a second selectable marker. The Exo70F3 CRISPR/Cas9 construct was transformed to transgenic Dongjin plants coexpressing Pi5-1 and Pi5-2 genomic DNAs using phosphinothricin selection (Lee et al., 2009). Transgenic rice plants were selected on a medium containing 50 mg•L⁻¹ hygromycin or 10 mg•L⁻¹ phosphinothricin and 250 mg•L⁻¹ cefotaxime.

To identify OsExo70F3 mutants, PCR amplification with the primers (Supplementary Table S1) specific to the target PAM region was performed with genomic DNAs of primary transgenic plants. The PCR product purified was sequenced to determine mutations created by the CRISPR/Cas9 system (Miao et al., 2013). Full-length cDNAs of OsExo70F3 were amplified from all OsExo70F3 mutants by RT-PCR (Supplementary Table S1) and sequenced to confirm the mutations.

Pathogen inoculation

For blast fungus inoculation, M. oryzae isolate PO6-6 was cultured on V8 juice agar plates (80 ml V8 Juice [Campbell’s Soup Company, USA] L⁻¹, 15 g agar L⁻¹, pH 6.8) for two weeks under fluorescent light. Subsequently, spores were collected and suspended in water to reach a concentration of 5 × 10⁶ ml⁻¹ for spot inoculation and 2 × 10⁵ ml⁻¹ for spray inoculation. Fully expanded healthy leaf regions, without any cell death lesions, of six-week-old and four-week-old plants, respectively, were inoculated with the spore suspensions prepared for spot inoculation and spray inoculation (Kanzaki et al., 2002). Leaf samples were collected and photographed to evaluate blast lesions when disease symptoms were well developed in susceptible controls. Three replications were carried out for each sample.

Molecular cloning of Pii cDNAs

In order to verify the pair of NLRs as orthologues of Pi5-1 and Pi5-2 genomic sequence, we used twelve sets of primers, designed based on Pi5-1 and Pi5-2 genome sequences (Fig. 1A and Supplementary Table S1), and amplified the respective PCR fragments from genomic DNA of a rice cultivar Fujisaka5 known to show Pii-mediated resistance (Fig. 1B). All primer pairs did not amplify PCR products in the susceptible cultivar, Dongjin, except for the Pi5-1 F5/R5 primer set which could amplify a genomic sequence similar to a part of Pi5-1 C-terminal region from Dongjin (which is also conserved in the genome of susceptible cultivar Nipponbare). DNA sequence analysis of PCR products showed that the Pi5 counterparts in Fujisaka5 are highly similar to Pi5 genes (Lee et al., 2009). Meanwhile, an independent study deployed the Mutmap-Gap method to characterize the Pii locus, suggesting the presence of a Pi5 allele in cultivar Hitomebore (Takagi et al., 2013). The resistance gene Pii was isolated in the mutant lines that have lost Pii function. Nucleotide alignment analysis denoted the identity of the previously identified Pii gene (Takagi et al., 2013) with our Pi5-1 counterpart in Fujisaka5, herein referred to as Pii-1 (GenBank accession No. MH490982), whereas the other gene is referred to as Pii-2 (GenBank accession No. MH490983) in our study. In order to verify cDNA sequences of Pii-1 and Pii-2 in Fujisaka5, Pii cDNAs were cloned by RT-PCR using cDNA prepared from the leaves infected with M. oryzae PO6-6, and subjected to sequence analysis. Six and 11 polymorphic amino acids were confirmed in Pii-1 and Pii-2 from cultivar Fujisaka5 as compared to Pi5-1 and Pi5-2, respectively (Fig. 1C). Additionally, Pii-2 contains an extended C-terminus of 20 amino acids, in contrast to Pi5-2. DNA sequence analysis suggested that Pii-2 and Pi5-2 might have different splicing patterns due to significant variation in the last intron (Supplementary Fig. S1).

To test if Pii-1 and Pii-2 are required for AVR-Pii recognition in Fujisaka5, RNA interference (RNAi)-mediated knockdown of Pii-1 or Pii-2 transcripts were induced by introducing RNAi constructs of Pi5-1 or Pi5-2 into a Pii monogenic line (IRBLi-F5) that carries both Pii-1 and Pii-2 derived from a cross between Fujisaka5 and susceptible background cultivar Lijiangxintuanheigu (Tsunematsu et al., 2000) (Fig. 2A). Expression of Pii-1 or Pii-2 was significantly decreased in Pii-1/Pi5-1 or Pii-2/Pi5-2 RNAi lines, respectively, as evaluated by quantitative RT-PCR in the selected transgenic lines (Fig. 2B). To test whether the reduction in Pii-1 or Pii-2 expression correlates with reduced disease resistance/AVR-Pii recognition, we challenged wild-type IRBLi-F5 and RNAi transgenic lines with M. oryzae PO6-6, which carries AVR-Pii and AVR-Pi5 (Kobayashi et al., 2007), using a spray inoculation method. Distinct lesions developed at seven days post-infection (dpi) in transgenic plants (Fig. 2C), suggesting that the resistance by Pii was abolished due to the reduced expression of Pii-1 or Pii-2. This result indicates that both Pii-1 and Pii-2 are required for AVR-Pii-triggered resistance to M. oryzae.

Functional evaluation of Pi5 and Pii cDNAs in transgenic rice

To further investigate whether the Pii-1/Pii-2 pair is required, we generated a series of binary constructs carrying Pi5 or Pii variants whose expression is controlled by the 35S promoter (Fig. 3A). Pi5 constructs were transformed separately to a susceptible rice cultivar Dongjin, in which the Pi5 allelic sequence is absent (Figs. 3B and 3C). While transgenic lines expressing Pi5-1 or Pi5-2 cDNA displayed wild-type morphology, transgenic plants expressing both Pi5-1 and Pi5-2 cDNAs developed by crossing the transgenic lines carrying Pi5-1 and Pi5-2 displayed an autoimmune lesion mimic (LM) phenotype. The phenotype was characterized by cell death areas arising from the leaf tip and apparently seen in the third leaf at five weeks of age under paddy field conditions in the summer (Supplementary Fig. S2). To validate the cloned Pi5-1 and Pi5-2 cDNAs in disease resistance to blast fungus, the transgenic plants expressing Pi5-1 and/or Pi5-2 cDNAs were spot inoculated with M. oryzae PO6-6 (Figs. 3B and 3C). A susceptible control, Dongjin, as well as transgenic lines expressing Pi5-1 or Pi5-2 cDNA displayed long brown lesion at nine dpi, characteristic of a compatible interaction. Importantly, the transgenic lines carrying genomic sequences (Lee et al., 2009) or cDNAs of both Pi5-1 and Pi5-2 were resistant to M. oryzae PO6-6, demonstrating that the cloned cDNAs of Pi5-1 and Pi5-2 are functional.

In order to verify if the transgenic plants expressing Pii-1 and/or Pii-2 show similar disease resistance/susceptibility phenotypes to Pi5-expressing lines, susceptible Dongjin plants were first transformed with Pii-1 cDNA. The possibility of two spliced forms of Pii-2 due to the significant difference in the last intron of Pii-2 and Pi5-2 (Supplementary Fig. S1) led us to investigate the function of Pii-2 cDNAs both with a longer C-terminus and without it (Pii-2Δ). The Pii-2Δ variant lacks the last exon and therefore has a C-terminal region identical to Pi5-2 (Fig. 3A). Pii-2 or Pii-2Δ cDNA was then introduced to Pii-1 cDNA-expressing transgenic Dongjin. Each of the five transgenic plants with different expression of each transgene was evaluated for susceptibility (Supplementary Fig. S3). While transgenic lines expressing only Pii-1 cDNA were susceptible, all lines expressing Pii-2 or Pii-2Δ cDNA additionally were resistant to M. oryzae PO6-6 (Figs. 3B and 3C). In addition, the transgenic plants expressing Pii-2 or Pii-2Δ without Pii-1 were susceptible (Figs. 3B and 3C). This result showed that Pii-mediated resistance is similar to Pi5, consistent with our hypothesis that Pii and Pi5 display similar recognition specificity. Interestingly, we also observed the LM phenotype in some transgenic plants expressing Pii-2 or Pii-2Δ cDNA in the presence of Pii-1 cDNA at the late stage of development (Supplementary Fig. S3), which is consistent with the observation from Pi5 transgenic rice plants. To confirm that this resistance is Pii- and Pi5-mediated, we challenged all the transgenic plants with a compatible M. oryzae KI215 isolate lacking AVR-Pii and AVR-Pi5 (Lee et al., 2009). As expected, the lesions developed at nine dpi in all the transgenic plants were comparable to the levels of the susceptible cultivar Dongjin (Fig. 4).

Genetic swapping between Pii and Pi5 paired NLR counterparts

The high similarity of the Pii- and Pi5-paired R genes further prompted us to test the combination of Pii and Pi5 genes. Therefore, we generated transgenic lines expressing Pii-1 and Pi5-2 by crossing a Pii-1 cDNA overexpressing line to the line expressing Pi5-2 genomic DNA. The resulting plants thus expressed both Pii-1 and Pi5-2 (Fig. 3C). In addition, we similarly generated transgenic plants expressing Pii-2 or Pii-2Δ in the presence of Pi5-1 cDNA by introducing Pii-2 or Pii-2Δ cDNAs to Pi5-1 cDNA expressing line. Surprisingly, the crossed plants showed clear incompatible interaction with M. oryzae PO6-6 and compatible with M. oryzae KI215 (Figs. 3B, 3C, and 4). In the line with that, transgenic plants expressing Pii-2 or Pii-2Δ in Pi5-1 cDNA carrying background also exhibited the obvious resistance to M. oryzae PO6-6 and susceptibility to M. oryzae KI215 (Figs. 3B, 3C, and 4). These data indicate that Pii and Pi5 genes function interchangeably in triggering ETI.

CRISPR/Cas9-mediated mutation of Exo70F3 in Pi5 transgenic rice

Exo70F3 was previously reported as an interactor of AVR-Pii and critical for Pii-mediated resistance. Loss of OsExo70F3 by RNAi abolished Pii-mediated resistance (Fujisaki et al., 2015). Therefore, we hypothesized that knockout mutations of OsExo70F3 would cause loss of Pi5-mediated resistance. In order to test this hypothesis, we generated OsExo70F3 loss-of-function mutants with the CRISPR/Cas9-mediated gene editing method in the transgenic Dongjin plants expressing Pi5-1 and Pi5-2 genomic DNAs described in Figure 3B. The resulting knockout mutants, exo70f3-1, exo70f3-2 and exo70f3-3, carried distinct modifications in the coding region of OsExo70F3 (Fig. 5A). Strikingly, a premature stop codon caused by the shift of reading frame in the coding region (exo70f3-1 and exo70f3-2) and loss of a large fragment (exo70f3-3) (Supplementary Fig. S4) did not affect Pi5-mediated resistance, evidenced by restricted lesion development (Figs. 5B and 5C).

The requirement of Pi5-1 and Pi5-2 in Pi5-mediated resistance was validated by introducing both genomic DNA fragments to susceptible cultivar Dongjin (Lee et al., 2009). In order to dissect their functions, the protein coding sequences of both genes were validated. In this study, we generated transgenic plants expressing Pi5-1 and Pi5-2 cDNAs by crossing each of the overexpression lines. The obvious resistance to M. oryzae PO6-6 was demonstrated in transgenic lines expressing both cDNAs, indicating that they are sufficient to confer Pi5-mediated resistance.

The similarity of resistance specificity conferred by Pii and Pi5 (Ebron et al., 2004; Selisana et al., 2017) led us to hypothesize that Pii-mediated resistance requires a pair of NLR genes that are allelic to those of Pi5. In this study, we provide several lines of evidence to prove the allelism between Pii and Pi5. Firstly, two counterpart genes, Pii-1 and Pii-2, which share high similarity to Pi5-1 and Pi5-2, respectively, were present in Fujisaka5 carrying Pii. Secondly, the paired NLRs of Pii-1 and Pii-2 were necessary for Pii-mediated resistance as confirmed in RNAi plants of Pii-1 and Pii-2 (yielding susceptibility to M. oryzae PO6-6) and in transgenic plants coexpressing both Pii-1 and Pii-2 cDNAs (yielding resistance to). Lastly, the replacement of Pii-1 by Pi5-1 or Pii-2 by Pi5-2 retained an incompatible interaction to M. oryzae PO6-6.

Cell death is often an indicator of defense induced by ETI. Therefore, LM phenotypes have been studied in-depth to understand the molecular aspect of defense (Wu et al., 2008). Cell death phenotypes in tobacco are observed when expressing PigmR, whereas HR-like symptoms demand the presence of both AVR-Pia and Pia in rice (Deng et al., 2017; Satoh et al., 2014). Previously, the introduction of both Pi5-1 and Pi5-2 genomic DNAs into Dongjin did not cause an obvious morphological phenotype (Lee et al., 2009). In contrast, LM phenotype appeared in transgenic plants expressing both Pi5 cDNAs, as well as transgenic plants expressing both Pii cDNAs, but was absent in single gene transgenic lines (Supplementary Figs. S2 and S3). A similar observation was made for the paired Arabidopsis NLRs, RPS4 and RRS1, where RPS4 auto-immunity depends partially on RRS1 (Heidrich et al., 2013). This suggests that overexpression of both paired NLR genes leads to constitutive defense activation that functions as a cell death trigger.

The sequence variation in NB-LRR genes is the result of a long co-evolution between rice and rice blast. Therefore, the significant difference in the last intron of Pii-2 and the presence of longer Pii-2 mRNA led us to investigate the function of Pii-2 truncated form (Pii-2Δ) which is absent in Fujisaka5 cDNA library. However, transgenic plants carrying either Pii-2 or Pii-2Δ additionally to Pii-1 could confer resistance to AVR-Pii and AVR-Pi5 carrying isolate PO6-6. The integration of an effector target into plant NLRs has been reported as a novel conserved mechanism called the integrated decoy model (Césari et al., 2014). RIN4 acts as molecular switch and contains two nitrate–induced (NOI) domains that function in suppressing PTI (Ray et al., 2019). In this regard, a conserved NOI core motif (PxFGxW) of the AvrRpt2-cleavage site was observed in the Pi5-2 C-terminus (Nishimura et al., 2015) that is also retained in both Pii-2 variants. Notably, a genome-wide association study using the rice diversity panel1 identified a new resistance allele at the Pi5 locus of a resistance cultivar 301279 (Kang et al., 2016). The Pi5-2 allelic gene newly identified was found to have a short C-terminus; therefore, lacking the NOI motif but retaining resistance function based on RNAi experiments. This suggests that the NOI motif of Pii-2 and Pi5-2 may not be critical for resistance, although further studies are required to confirm this.

The sequence similarity, resistance spectrum, and location on the chromosome raised the question about the identity of AVR-Pii and AVR-Pi5 effectors. AVR-Pii was previously isolated via an association genetics approach among 23 M. oryzae isolates (Yoshida et al., 2009). Subsequently, a comparative study of 77 M. oryzae isolates from the International Rice Research Institute (IRRI) has shown a correlation between the presence of AVR-Pii and activation of Pi5-mediated resistance, except for a few isolates that lack AVR-Pii but are avirulent on the Pi5 line (Selisana et al., 2017), suggesting that AVR-Pii is recognized by the Pi5-1/Pi5-2 pair or that an additional unknown Avr effector (AVR-Pi5) is present. In a recent study, a considerable inconsistency was found between the presence of AVR-Pii sequence in a large collection of M. oryzae isolates and Pii-mediated resistance in rice (i.e., resistant but without AVR-Pii or susceptible with AVR-Pii) (Lu et al., 2019), implying that another unknown AVR-Pii effector is present or that an additional factor is required for AVR-Pii function. Since AVR-Pi5 has not been identified independently, we hypothesized that testing cross-functionality of Pii and Pi5 alleles might be useful to determine if the Pii and Pi5 genes have identical function. Accordingly, we generated the combination of Pii-1 and Pi5-2 along with Pi5-1 and Pii-2 to examine the cooperation of two genes. The crossed line of Pii-1 cDNA-overexpressing plant and Pi5-2 genomic DNA-carrying as well as transgenic plants expressing Pi5-1 cDNA and Pii-2 or Pii-2Δ were resistant to M. oryzae PO6-6 which possesses AVR-Pii and AVR-Pi5, suggesting that Pii and Pi5 components can function cooperatively in conferring disease resistance. However, whether the resistance is Pii or Pi5-mediated is unknown. Therefore, the six amino acid difference in Pii-1/Pi5-1 and eleven in Pii-2/Pi5-2 might not significantly alter Avr recognition. Nevertheless, if the AVR-Pi5 is distinct from AVR-Pii, the crosses will be useful to understand the interesting mechanism of Avr recognition of Pii and Pi5.

The Exo70 protein in the exocyst complex is important in tethering and fusion of the vesicles and plasma membrane at the site of polarized exocytosis (Munson and Novick, 2006). In plants, OsExo70F3 is known to interact with AVR-Pii and is critical for Pii-mediated resistance (Fujisaki et al., 2015). Therefore, in our study, the involvement of OsExo70F3 in Pi5 action was evaluated by examining susceptibility of Exo70F3 mutations in the Pi5 background. Interestingly, Pi5-mediated resistance does not require Exo70F3 whereas Pii does, illuminating a difference in defense mechanisms among two R genes at the same locus. Differential requirement of OsExo70F3 for Pi5 and Pii-mediated resistance is puzzling because both NLR pairs seem to recognize the same Avr effector, AVR-Pii. One would hypothesize that AVR-Pii may target multiple Exo70 variants and an unidentified Exo70 variant might be required for Pi5. Alternatively, Pi5-1 and Pi5-2 may recognize an unknown Avr effector. However, so far, most of the analyzed M. oryzae isolates that are avirulent to Pi5 carry AVR-Pii. It is unlikely, although not impossible, that Pi5-1 and Pi5-2 recognize an unknown Avr effector. Therefore, together with the identification of AVR-Pi5, functional verification of exocytosis components will pave the way to uncover the hidden mechanism of Pi5-resistance.