Conserved fungal effector suppresses PAMP-triggered immunity by targeting plant immune kinases

Significance

Multiple effectors of bacterial pathogens target immune kinases such as BAK1 and BIK1, but it is unclear whether this strategy is employed by fungal pathogens. We reveal here that a fungal effector named NIS1 is broadly conserved in filamentous fungi in the Ascomycota and Basidiomycota, thus being regarded as a core effector, and has the ability to suppress PAMP-triggered immunity. Importantly, NIS1 targets BAK1 and BIK1, interfering with their essential functions for immune activation upon pathogen recognition. Multifaceted analyses including the knockout of NIS1 revealed that it plays a critical role in fungal infection. These findings demonstrate that to infect host plants, filamentous fungi deploy a core effector that attacks conserved immune kinases critical for the ancestral defense system.

Abstract

Plant pathogens have optimized their own effector sets to adapt to their hosts. However, certain effectors, regarded as core effectors, are conserved among various pathogens, and may therefore play an important and common role in pathogen virulence. We report here that the widely distributed fungal effector NIS1 targets host immune components that transmit signaling from pattern recognition receptors (PRRs) in plants. NIS1 from two Colletotrichum spp. suppressed the hypersensitive response and oxidative burst, both of which are induced by pathogen-derived molecules, in Nicotiana benthamiana. Magnaporthe oryzae NIS1 also suppressed the two defense responses, although this pathogen likely acquired the NIS1 gene via horizontal transfer from Basidiomycota. Interestingly, the root endophyte Colletotrichum tofieldiae also possesses a NIS1 homolog that can suppress the oxidative burst in N. benthamiana. We show that NIS1 of multiple pathogens commonly interacts with the PRR-associated kinases BAK1 and BIK1, thereby inhibiting their kinase activities and the BIK1-NADPH oxidase interaction. Furthermore, mutations in the NIS1-targeting proteins, i.e., BAK1 and BIK1, in Arabidopsis thaliana also resulted in reduced immunity to Colletotrichum fungi. Finally, M. oryzae lacking NIS1 displayed significantly reduced virulence on rice and barley, its hosts. Our study therefore reveals that a broad range of filamentous fungi maintain and utilize the core effector NIS1 to establish infection in their host plants and perhaps also beneficial interactions, by targeting conserved and central PRR-associated kinases that are also known to be targeted by bacterial effectors.

Plants have evolved two layers of antimicrobial defenses: pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) and effector-triggered immunity (ETI). PTI is mediated by membrane-embedded receptor-like proteins (RLPs), receptor-like kinases (RLKs) and receptor-like cytoplasmic kinases (RLCKs), whereas ETI generally occurs when cytoplasmic resistance (R) proteins detect specific pathogen effectors (1).

The immune kinases BAK1/SERK3 and BIK1 have been studied extensively in Arabidopsis thaliana as the central regulatory RLK and RLCK, respectively, working with multiple pattern recognition receptors for PAMP sensing and signaling (2–6). FLS2, another well-characterized RLK that recognizes the bacterial PAMP flagellin (flg22), recruits BAK1 upon ligand perception to initiate PTI signaling (2, 3, 7). In contrast, BIK1 forms a complex with FLS2 in the steady state (4, 5). Upon flg22 elicitation, BAK1 associates with FLS2 and phosphorylates BIK1; activated BIK1 then phosphorylates BAK1 and FLS2 before dissociating from the FLS2-BAK1 complex to transmit the signal to the downstream pathway (4, 5, 8). Phosphorylated BIK1 also activates the NADPH oxidase RBOHD through phosphorylation events to trigger a reactive oxygen species (ROS) burst, which is one of the earliest PTI responses (9, 10).

Many types of bacterial pathogen effectors target these RLK- and RLCK-type kinases. For example, Pseudomonas syringae AvrPto, AvrPtoB, HopF2, and HopB1 target BAK1, and Xanthomonas oryzae Xoo2875 targets OsBAK1 (the BAK1 homolog in rice) (11–14). AvrPto and AvrPtoB bind to BAK1 and interfere with formation of the FLS2-BAK1 complex (11). On the other hand, P. syringae AvrPphB and Xanthomonas campestris AvrAC target BIK1 (5, 15). AvrPphB, a cysteine protease, degrades PBS1-like kinases, including BIK1 (5), while the uridylyl transferase AvrAC conceals important phosphorylation sites in the activation loop of BIK1 (15). These findings strongly suggested that inhibiting these RLK- and RLCK-type kinases is advantageous to bacterial pathogens. However, it remained unclear whether this strategy is also employed by fungal and oomycete pathogens.

Our knowledge of the molecular functions of fungal and oomycete effectors is now expanding. The reported functions of these effectors are highly diverse and include inhibition of host-secreted lytic enzymes (16, 17), modulation of the plant ubiquitination system (18), autophagy (19), and blocking of the exposure of the fungal PAMP chitin to its corresponding receptor(s) in plants (20, 21).

Here we report that a so-called “core” effector named necrosis-inducing secreted protein 1 (NIS1), which is highly conserved in filamentous fungal plant pathogens, has the ability to target RLKs such as BAK1 and the RLCK BIK1 and thereby to impair PTI signaling. Core effectors can be defined by their wide distribution among strains of a particular pathogen. For example, high-throughput genome sequencing of 65 strains of Xanthomonas axonopodis pv. manihotis, the causal agent of cassava bacterial blight, revealed that all strains shared nine common effectors, indicating that these nine are core effectors in this bacterial pathogen (22). On the other hand, extensive genome analyses on fungal pathogens also reveal the presence of abundant homologous effector proteins across different species, which are likewise widely defined as core effectors. NIS1 was first identified in cucumber anthracnose fungus, Colletotrichum orbiculare, as an effector that causes cell death in Nicotiana benthamiana (Nb) after Agrobacterium-mediated transient expression (23). Notably, this effector was found to be broadly conserved in filamentous fungal pathogens, and we therefore considered NIS1 as a putative core effector. However, deletion of NIS1 had no effect on C. orbiculare virulence on its natural host, cucumber. Since C. orbiculare infects Nb in addition to Cucurbitaceae (24), we also inoculated the nis1∆ strain of C. orbiculare onto Nb; again, however, the mutant exhibited WT-like virulence on Nb. Therefore, the molecular and physiological function of this effector remained elusive.

In this study, we show that C. orbiculare NIS1, as well as NIS1 homologs of the crucifer anthracnose fungus Colletotrichum higginsianum and the rice blast fungus Magnaporthe oryzae, commonly suppress both hypersensitive response (HR) cell death triggered by the oomycete PAMP INF1 and ROS generation triggered by flg22 and the fungal PAMP chitin in Nb. These findings indicate that NIS1 suppresses multiple PTI responses in Nb. Our study demonstrates that NIS1 targets the RLK BAK1 and its orthologs in other plants, and also the RLCK BIK1, both of which are critical for PTI. The study also revealed that both BAK1 and BIK1 targeted by NIS1 are critical for plant immunity in Arabidipsis toward Colletotrichum fungi. We found that transient expression of NIS1 in Nb enhanced susceptibility to C. orbiculare. Furthermore, targeted gene disruption of NIS1 in M. oryzae resulted in a severe reduction of virulence on both barley and rice susceptible cultivars, indicating the importance of the conserved effector NIS1 for fungal virulence.

Our data show that lineages of pathogens as different as bacteria and fungi share an effector-mediated strategy to interfere with immune kinases that transmit signaling from pathogen-recognizing receptors, suggesting that this effector innovation in phytopathogens has been essential to cope with PTI, which is probably universal in plants. The finding that NIS1 is conserved in a broad range of filamentous fungi in both Ascomycota and Basidiomycota also tells us that the effector NIS1 was already present in the common ancestor of these phyla and has been evolutionarily maintained to facilitate plant infection.

Results

The Conserved Filamentous Fungal Effector NIS1 Suppresses INF1-Induced Cell Death and Interacts with BAK1.

We previously identified the effector NIS1 (hereafter called CoNIS1) from the hemibiotroph C. orbiculare as a secreted protein that induces cell death in Nb (23). We found 219 of putative CoNIS1 homologs in UniProtKB (SI Appendix, Table S1). CoNIS1 homologs are conserved in a broad range of fungi belonging to Sordariomycetes, Dothideomycetes, Eurotiomycetes, Orbiliomycetes, and Leotiomycetes in the Ascomycota, and to Exobasidiomycetes, Agaricomycetes, and Tremellomycetes in the Basidiomycota, which include numerous pathogenic fungi (Fig. 1 and SI Appendix, Fig. S1). This suggests that NIS1 is an evolutionarily ancient effector, and its broad conservation implies the importance of NIS1 as a possible core effector in fungal pathogens. Intriguingly, we also found that the root endophyte C. tofieldiae possesses a NIS1 homolog (Fig. 1 and SI Appendix, Fig. S1), indicating that the conservation of NIS1 is not limited to pathogens but extends to beneficial species (25).

Fig. 1.

NIS1 is widely distributed among fungal lineages of Ascomycota and Basidiomycota. An unscaled phylogenetic tree of selected NIS1 homologs reconstructed by treefixDTL (62) is shown. Out of 219 putative NIS1 homologs, we selected 171 homologs by removing redundant information using the following selection criteria: (i) When multiple homologs with identical sequences were present, one homolog was arbitrarily chosen and retained; and (ii) when multiple homologs from a different strain of the same species were present, homologs from an arbitrarily chosen strain was retained. Leaf labels represent abbreviated names of the species to which the NIS1 homologs belong. When multiple NIS1 homologs are present in a species, unique numbers were assigned to each leaf label. Leaf labels and nodes are color coded according to class. The track outside the tree represents fungal phylum. NIS1 homologs from C. orbiculare (CoNIS1), C. higginsianum (ChNIS1), C. tofieldiae (CtNIS1), and Magnaporthe oryzae (MoNIS1) are shown in bold. Potential transfer from Basidiomycota (Tremellomycetes) to Ascomycota (Sordariomycetes) is indicated by the gray arrowhead. For full species and gene names, see SI Appendix, Tables S1 and S2.

We next investigated whether CoNIS1 can suppress plant immune responses. The well-known oomycete PAMP elicitor INF1 strongly induces cell death in a restricted number of Nicotiana and Solanum species within solanaceous plants, including Nb (26, 27). Because C. orbiculare infects Nb (24), and because INF1-induced cell death is reportedly suppressed not only by oomycete effectors but also by effectors of organisms in different kingdoms (28, 29), we first examined possible effects of CoNIS1 on the INF1-elicited immune response in Nb. Transient expression of INF1 by agroinfiltration induces rapid HR cell death in Nb within 2 d postinoculation (dpi) (30); this is much faster than CoNIS1-induced cell death, which is first observed after 5 dpi. Taking advantage of the different timing between CoNIS1- and INF1-induced necrosis, we tested whether CoNIS1 can suppress INF1-induced cell death. We found that INF1-induced cell death was clearly suppressed at 3 dpi of Agrobacterium harboring a plasmid expressing INF1 by preexpression of CoNIS1 (Fig. 2A and SI Appendix, Fig. S2). CoNIS1 eventually caused cell death at 6 dpi (SI Appendix, Fig. S3). Moreover, the NIS1 homolog of crucifer anthracnose fungus C. higginsianum (ChNIS1) also suppressed INF1-induced cell death (Fig. 2A and SI Appendix, Fig. S2). We previously found that the NIS1 homolog of the rice blast fungus M. oryzae (MoNIS1) did not induce necrotic lesions in Nb (23), in contrast to CoNIS1 and ChNIS1, indicating that MoNIS1 is functionally distinct from the Colletotrichum NIS1s. Interestingly, the phylogenetic analysis revealed that MoNIS1 is distantly related to CoNIS1 and ChNIS1 but more closely related to NIS1 homologs in the Basidiomycota (Fig. 1). Comparisons of NIS1 phylogeny and species phylogeny suggest that the common ancestor of Magnaporthales, Diaporthales, and Xylariales gained a NIS1 gene from Tremellomycetes (SI Appendix, Fig. S4).

Fig. 2.

NIS1 suppresses INF1-induced cell death and directly interacts with BAK1. (A) Suppression of INF1-induced cell death by the NIS1 homologs of C. orbiculare, C. higginsianum, and M. oryzae. Infiltration sites of A. tumefaciens harboring a plasmid expressing each NIS1 homolog or A. tumefaciens harboring the empty plasmid pBICP35 as a vector control (vector) (dashed white circles) were challenged with A. tumefaciens harboring a plasmid expressing INF1 (solid black circles) in Nb. The photograph was taken at 3 dpi. Similar results were obtained in two additional experiments. (B) Co-IP of BAK1 and NbSERK3 with NIS1 homologs. BAK1-GFP, NbSERK3-GFP, or GFP (control) was transiently coexpressed with CoNIS1-HA or MoNIS1-HA in Nb. Total protein extracts were subjected to IP with anti-GFP magnetic beads followed by immunoblotting (IB) with anti-HA and anti-GFP antibodies. Asterisks indicate truncated or modified forms of the expressed proteins that were detected by anti-GFP antibody. Similar results were obtained in two additional experiments. (C) Interaction of NIS1 homologs lacking a signal peptide with the cytoplasmic region of Arabidopsis BAK1 (BAK1-cyto) in yeast two-hybrid assays. The interactions were examined by assessing growth on the selective medium QDO/X/A (SD/−Ade/−His/−Leu/−Trp/x-a-Gal/Aba). DDO (SD/−Leu/−Trp) medium was used for a mating control. Similar results were obtained from one additional experiment. (D) A part of MoNIS1-mCherry signal was colocalized with BAK1-GFP1 located in the plasma membrane in Nb. BAK1-GFP1 and a functional MoNIS1-mCherry were transiently expressed in Nb via agroinfiltration. At 2 dpi, the fluorescence signal was examined under a confocal microspope. (Scale bar = 10 µm.)

Remarkably, MoNIS1 also suppressed INF1-induced cell death (Fig. 2A and SI Appendix, Fig. S2), despite its evolutionary remoteness from NIS1 homologs of other Ascomycete fungi (Fig. 1 and SI Appendix, Fig. S4). Therefore, the fungal effector NIS1, widely conserved in fungal pathogens, commonly suppresses INF1-induced PTI in Nb. It is noteworthy that MoNIS1 failed to induce necrotic lesions in Nb (23), thus uncoupling the suppression of INF1-induced cell death from the activity of NIS1 itself to induce cell death in Nb.

BAK1, a member of the SERK family, is a central regulator required for recognition of multiple PAMPs in Nb and Arabidopsis (2, 3, 6). Importantly, silencing of the BAK1 homolog of Nb, NbSerk3, canceled INF1-induced cell death in Nb (3, 31). It was also reported that a receptor for INF1, named ELR, identified in the wild potato Solanum microdontum, associates with BAK1 homologs of potato (32), further supporting the role of the BAK1 homolog in INF1-induced cell death in Nb.

Therefore, to assess the possibility that NIS1 targets BAK1, we looked for an interaction of BAK1 with CoNIS1 and MoNIS1 in a coimmunoprecipitation (co-IP) analysis in Nb. Green fluorescent protein (GFP)-tagged BAK1 was expressed together with each NIS1 homolog in Nb, and subsequent co-IP analysis using anti-GFP beads revealed that both CoNIS1 and MoNIS1 coimmunoprecipitated with BAK1-GFP (Fig. 2B), indicating their interaction with BAK1 in planta. CoNIS1 also associated with NbSERK3 (Fig. 2B). A yeast two-hybrid assay revealed that both CoNIS1 and MoNIS1 lacking the signal peptide (SP) interact with the cytoplasmic region of BAK1 (Fig. 2C). We then transiently expressed a functional mCherry-fused MoNIS1 and BAK1-GFP in Nb (SI Appendix, Fig. S5). Part of the mCherry signal was detected in the cyotosol, and the remainder colocalized with the signal of BAK1-GFP located in the cell’s plasma membrane in Nb (Fig. 2D). The result is consistent with our previous report that mCherry-fused CoNIS1 secreted by M. oryzae translocated into rice cells (33), as well as accumulating in a biotrophic interfacial complex (BIC) (34, 35). Collectively, these findings suggest that NIS1 acts as a cytoplasmic effector and has the ability to interact with the cytoplasmic region of BAK1. CoNIS1 and MoNIS1 also interacted with BKK1 (SI Appendix, Fig. S6), a close paralog of BAK1 in Arabidopsis that can form complexes with PAMP receptors on ligand binding and transmit PTI signals (6).

To gain further insights into the NIS1-BAK1 interaction, we performed a deletion analysis of CoNIS1 (Fig. 3A). We found that CoNIS1 lacking the C-terminal 30 amino acids (CoNIS1∆C30) suppressed INF1-induced cell death (Fig. 3A and SI Appendix, Fig. S2) and still interacted with BAK1 (Fig. 3B). In contrast, CoNIS1 lacking the C-terminal 60 amino acids (CoNIS1∆C60) failed to suppress INF1-induced cell death and displayed a severe reduction in BAK1 association (Fig. 3 A and B and SI Appendix, Fig. S2). These findings illustrate a correlation between suppression of INF1-induced cell death and interaction with BAK1.

Fig. 3.

The ability of NIS1 to suppress INF1-induced cell death correlates with NIS1-BAK1 interaction. (A) Suppression assay of INF1-induced cell death by C-terminal truncation mutants of CoNIS1. Infiltration sites of A. tumefaciens harboring a plasmid expressing CoNIS1-HA, CoNIS1∆C30-HA, or CoNIS1∆C60-HA, or A. tumefaciens harboring the empty plasmid pBICP35 as a vector control (vector) (dashed white circles), were challenged with A. tumefaciens harboring a plasmid expressing INF1 (solid black circles) in Nb. The photograph was taken at 3 dpi. Similar results were obtained in two additional experiments. (B) Co-IP of Arabidopsis BAK1 and the HA-tagged C-terminal truncation series of CoNIS1. BAK1-GFP was transiently coexpressed with CoNIS1-HA, CoNIS1∆C30-HA, or CoNIS1∆C60-HA in Nb. Total protein extracts were subjected to IP with anti-GFP magnetic beads followed by IB with anti-HA and anti-GFP antibodies. Similar results were obtained in four additional experiments. (C) Suppression assay of INF1-induced cell death by CoNIS1∆C30 with an Ala point mutation series. Infiltration sites of A. tumefaciens harboring a plasmid expressing CoNIS1∆C30-HA, CoNIS1∆C30-S123A-HA, CoNIS1∆C30-Y125A-HA, CoNIS1∆C30-V127A-HA, or CoNIS1∆C30-G132A-HA, or A. tumefaciens harboring the empty plasmid pBICP35 as a vector control (vector) (dashed white circles), were challenged with A. tumefaciens harboring a plasmid expressing INF1 (solid black circles) in Nb. The photograph was taken at 3 dpi. Protein extracts were analyzed by IB with anti-HA antibody (Upper). CBB staining was used as a loading control (Lower). Similar results were obtained in one additional experiment. (D) Y125 is dispensable for the interaction of CoNIS1∆C30 with Arabidopsis BAK1. BAK1-GFP was transiently coexpressed with CoNIS1∆C30-HA, CoNIS1∆C30-Y125A-HA, or CoNIS1∆C60-HA in Nb. Total protein extracts were subjected to IP with anti-GFP magnetic beads followed by IB with anti-HA and anti-GFP antibodies. Similar results were obtained from additional two independent experiments.

The clear difference between CoNIS1∆C30 and CoNIS1∆C60 in the suppression assay for INF1-induced cell death implied that the C-terminal 31- to 60-amino acid region is important for the cell death suppression. To assess this, we performed alanine-scanning mutational analyses instead of further deletion. We selected amino acids located at the end of NIS1∆C30 while considering the properties and conservation of each residue: Ser at position 123 to Ala (S123A), Tyr at 125 to Ala (Y125A), Val at 127 to Ala (V127A), and Gly at 132 to Ala (G132A) (SI Appendix, Fig. S1). The suppression assay for each of these mutants revealed that introducing the Y125A mutation abolished the ability of CoNIS1∆C30 to suppress INF1-induced cell death although it had no effect on the protein’s stability (Fig. 3C). Interestingly, CoNIS1∆C30 carrying Y125A still interacted with BAK1 (Fig. 3D), implying that Y125A specifically reduces the interaction with NbSERK3 and its paralogs but not Arabidopsis BAK1. Alternatively, NIS1 might target unidentified components critical for INF1-induced cell death in addition to BAK1-type immune kinases. We also revealed that full-length CoNIS1 carrying Y125A still caused cell death in Nb (SI Appendix, Fig. S7), further supporting the uncoupling of the cell death-inducing activity from the ability to suppress INF1-induced cell death. Tyr-125 is moderately conserved in NIS1s of other fungal pathogens (41%) and its substituents are usually other aromatic amino acids, Phe (38%, including MoNIS1) or Trp (6%) (SI Appendix, Fig. S1).

NIS1 Inhibits PAMP-Induced ROS Generation and Binds to the RLCK BIK1.

We investigated whether NIS1 can cancel different types of cell death in Nb. Transient expression of Avr3a, an avirulence protein of Phytophthora infestans, together with its cognate R protein R3a triggers HR cell death in Nb (30). Interestingly, MoNIS1 suppressed R3a/Avr3a-dependent HR cell death (SI Appendix, Fig. S8). We could not test this ability for CoNIS1 because of its own capacity to induce cell death in Nb. R3a/Avr3a-dependent HR cell death is not canceled in NbSerk3-silenced plants (36), implying that at least MoNIS1 has other targets in addition to BAK1. To assess whether NIS1 inhibits PTI responses that do not accompany an HR response, we measured ROS produced after flg22 treatment in Nb transiently expressing CoNIS1 or each NIS1 homolog (Fig. 4A). CoNIS1, ChNIS1, and MoNIS1 all blocked the ROS burst, demonstrating that this effector from multiple pathogens suppresses flg22-triggered ROS generation in Nb. CoNIS1∆C30, but not CoNIS1∆C60, also blocked the ROS burst (Fig. 4A). We further found that CoNIS1 and MoNIS1 suppressed ROS generation triggered by the fungal PAMP chitin in Nb, although the suppression activity of MoNIS1 was weaker than that of CoNIS1 (Fig. 4B). The finding that NIS1 can suppress more than one PAMP-induced ROS production pathway is consistent with the wide conservation of this effector among fungal pathogens.

Fig. 4.

NIS1 prevents BIK1-RBOHD interaction by targeting BIK1 and inhibits PAMP-triggered ROS generation. (A) Suppression of flg22-triggered ROS production by NIS1 in Nb. Total ROS production [represented as relative luminescence units (RLU)] by Nb transiently expressing CoNIS1-HA, MoNIS1-HA, ChNIS1-HA, CoNIS1∆C30-HA, or CoNIS1∆C60-HA, or harboring empty vector as control, was measured after treatment with 500 nM flg22. Results are average + SE (n = 12). Similar results were obtained in two additional experiments. (B) Suppression of chitin-triggered ROS production by NIS1 in Nb. Total ROS production (represented as RLU) by Nb transiently expressing CoNIS1-HA or MoNIS1-HA after treatment with 1 mg/mL chitin. Results are average + SE (n = 12). Similar results were obtained in two additional experiments. (C) Co-IP of Arabidopsis BIK1 with CoNIS1. BIK1-GFP or GFP (control) was transiently coexpressed with CoNIS1-HA, CoNIS1∆C30-HA, or CoNIS1∆C60-HA in Nb. Total protein extracts were subjected to IP with anti-GFP magnetic beads followed by IB with anti-HA and anti-GFP antibodies. Similar results were obtained in three additional experiments. (D) Co-IP of Arabidopsis BIK1 with MoNIS1. The asterisk marks a truncated form of the expressed protein that was detected by anti-GFP antibody. Similar results were obtained in three additional experiments. (E) Prevention of BIK1-RBOHD association by NIS1. Co-IP was performed in Nb transiently expressing BIK1-GFP and 3xFLAG-RBOHD with CoNIS1-HA or CoNIS1∆C60-HA. A. tumefaciens harboring the empty plasmid pBICP35 was infiltrated (vector) as a control for the expression of CoNIS1-HA and CoNIS1∆C60-HA. Total protein extracts were subjected to IP with anti-GFP magnetic beads followed by IB with anti-FLAG, anti-HA, and anti-GFP antibodies. Similar results were obtained from one additional experiment. (F) CoNIS1∆C30-Y125A suppressed flg22-triggered ROS production in Nb. Similar results were obtained in one additional experiment. (G) NIS1 of the root endophyte C. tofieldiae suppressed flg22-triggered ROS production in Nb. Similar results were obtained in two additional experiment.

BIK1 plays a critical role in ROS generation triggered by multiple PAMPs (flg22, elf18, and chitin) in Arabidopsis by directly interacting with and phosphorylating the NADPH oxidase RBOHD (9, 10). As shown above, NIS1 interacts with the cytoplasmic domain of BAK1/SERK3 and BKK1 (Fig. 2C and SI Appendix, Fig. S6), which is composed almost entirely of a serine/threonine protein kinase domain. The kinase domain of BIK1 shares 43% identity and 80% similarity with the cytoplasmic kinase domain of BAK1. Thus, we checked the potential interaction between NIS1 and BIK1 by co-IP assay in Nb. This revealed that CoNIS1 can indeed interact with BIK1 (Fig. 4C). Analysis of the truncated forms of CoNIS1 indicated that CoNIS1∆C30, but not CoNIS1∆C60, associates with BIK1 (Fig. 4C), suggesting that the binding modes of CoNIS1 to BIK1 and BAK1/SERK3 (Fig. 3B) are similar. MoNIS1 also interacts with BIK1 (Fig. 4D). These results indicate that NIS1 is able to associate with a RLCK, BIK1, as well as with RLKs including BAK1/SERK3.

As noted above, the phosphorylation of RBOHD by BIK1 results in a PAMP-dependent ROS burst (9, 10). Given that Arabidopsis BIK1 and RBOHD interact with each other in Nb (10), we asked whether CoNIS1 binding to BIK1 affects the BIK1-RBOHD interaction in Nb. Consistent with a previous report (10), we detected a BIK1-RBOHD interaction by co-IP assay using GFP-tagged BIK1 (Fig. 4E). Importantly, RBOHD protein coimmunoprecipitation with BIK1 almost completely disappeared in the presence of full-length CoNIS1 (Fig. 4E). On the other hand, the CoNIS1∆C60 mutant, which lacks the ability to bind to BIK1, did not affect the amount of coimmunoprecipitating RBOHD. These results suggest that NIS1 blocks the interaction between BIK1 and RBOHD by targeting BIK1. The Y125A mutation abolished the ability of CoNIS1∆C30 to suppress INF1-induced cell death. We then asked whether CoNIS1∆C30-Y125A suppresses flg22-triggered ROS generation in Nb. Notably, CoNIS1∆C30-Y125A clearly suppressed ROS generation (Fig. 4F), which is likely consistent with the finding that CoNIS1∆C30-Y125A maintains the ability to interact with one of its targets, namely BAK1 (Fig. 3D).

As mentioned above, we found that a NIS1 homolog is conserved in a root endophyte, C. tofieldiae (Fig. 1 and SI Appendix, Fig. S1) (25). Interestingly, we noticed that C. tofieldiae expresses the NIS1 homolog abundantly during Arabidopsis colonization at 6, 10, 16, and 24 dpi by referring to deposited RNA sequencing data of Hiruma et al. (25). This raises the possibility that C. tofieldiae utilizes NIS1 for its beneficial interaction with Arabidopsis. Therefore, we asked whether the C. tofieldiae NIS1 homolog is able to suppress plant immune responses in the same way as the NIS1s of the pathogens by performing the suppression assay for flg22-triggered ROS generation in Nb. Remarkably, C. tofieldiae NIS1 also suppressed flg22-triggered ROS generation (Fig. 4G).

NIS1 Inhibits the Kinase Activity of Both BAK1 and BIK1.

We found that NIS1 has no detectable effect on the stability of BAK1 or BIK1 in planta, suggesting that the fungal effector NIS1 possesses no activity to degrade or cleave these RLK and RLCK members (Fig. 5A). In PTI, phosphorylation relaying by RLKs (FLS2, BAK1, etc.) and RLCKs (BIK1, etc.) is critical for the initiation of downstream signaling (37). To investigate whether NIS1 affects the kinase activities of BAK1 and/or BIK1, we performed an in vitro kinase assay with these proteins, purified using a maltose binding protein (MBP) tag, in the presence or absence of CoNIS1 protein. We prepared a semipurified fraction containing CoNIS1 lacking the SP secreted from tobacco BY2 cells expressing this NIS1. For BAK1, we prepared the cytoplasmic domain (CD) of BAK1 instead of the full-length version. The assay revealed that CoNIS1 strongly inhibited autophosphorylation of MBP-BAK1CD (Fig. 5B). We also found that CoNIS1 inhibited autophosphorylation of MBP-BIK1 (Fig. 5C). These results suggest that NIS1 binds to BAK1 and BIK1 and then inhibits their kinase activity.

Fig. 5.

NIS1 inhibits the kinase activity of both BAK1 and BIK1. (A) Effects of NIS1 homologs (CoNIS1 and MoNIS1) on protein stability of BAK1 (Left) and BIK1 (Right). CoNIS1-HA or MoNIS1-HA were transiently coexpressed with BAK1-GFP or BIK1-GFP in Nb. A. tumefaciens harboring the empty plasmid pBICP35 was infiltrated (vector) as a control for the expression of CoNIS1-HA and MoNIS1-HA. Protein extracts were analyzed by immunoblotting with anti-HA and anti-GFP antibodies. (B) NIS1 inhibits autophosphorylation of BAK1. Purified MBP-fused BAK1 cytoplasmic domain (MBP-BAK1CD) were incubated with His-tagged CoNIS1 proteins purified from the culture medium of BY-2 cells secreting CoNIS1 (CoNIS1, 500 ng). We incubated MBP-fused kinase proteins with the equivalent fraction purified from the culture medium of BY-2 cells that do not secrete CoNIS1 as a negative control (NC). The His-tagged CoNIS1 protein fraction was also diluted fivefold with kinase reaction buffer and used for the assay (CoNIS1, 100 ng). In vitro kinase assay using [γ-³²P] ATP was performed and autophosphorylation was visualized by autoradiography. CBB was used as a loading control. Similar results were obtained from one additional experiment. (C) NIS1 inhibits autophosphorylation of BIK1. Purified MBP-fused BIK1 (MBP-BIK1) were subjected to the analysis described in B. Similar results were obtained from three additional experiments.

Arabidopsis BAK1 and BIK1 Targeted by NIS1 Are Critical for Immunity to Colletotrichum Fungi.

Because NIS1 binds to RLKs (BAK1/SERK3 and BKK1) and a RLCK (BIK1) and inhibits their kinase activity, we then investigated the impact of the loss of these RLK/RLCKs on Arabidopsis immunity to Colletotrichum fungi. We first inoculated the adapted C. higginsianum onto the Arabidopsis bak1-5 mutant. The bak1-5 mutation is a semidominant allele of BAK1 with a specific phenotype related to PAMP responsiveness (38). Quantitative assay for the size of developed lesions revealed that C. higginsianum displayed enhanced virulence to the bak1-5 mutant (Fig. 6A). We then inoculated C. higginsianum onto the bik1 mutant and the bik1 pbl1 mutant because PBL1 is highly similar to, and displays partial functional redundancy with BIK1 (5). C. higginsianum also showed enhanced virulence to bik1 and bik1 pbl1, similar to bak1-5 (Fig. 6A). Microscopic analysis revealed that the appressorium-mediated entry rate in bak1-5 was higher than in the wild-type (WT) plant (Fig. 6B). We also inoculated nonadapted Colletotrichum tropicale present in the Colletotrichum gloeosporioides complex (previously reported as C. gloeosporioides) onto these mutants together with a mutant defective in PEN2 (PENETRATION 2), which is critical for nonhost resistance against C. tropicale (39, 40). We found that the nonadapted C. tropicale developed clear lesions in the pen2, bak1-5, bik1, and bik1 pbl1 mutants but not in the wild-type plant (Fig. 6C). C. tropicale entry rates also increased in both bak1-5 and bik1 pbl1 compared with the wild-type plant (Fig. 6D). Collectively, these results suggest that the RLK and RLCK targeted by NIS1 play an important role in Arabidopsis immunity toward both adapted and nonadapted Colletotrichum fungi, including preinvasive resistance that is critical for pathogen entry control.

Fig. 6.

BAK1 and BIK1 targeted by NIS1 are involved in Arabidopsis immunity toward Colletotrichum fungi. (A) Quantification of lesion size of adapted C. higginsianum on Arabidopsis mutants. A conidial suspension of C. higginsianum (2.5 × 10⁵ conidia per milliliter) was inoculated onto leaves of Arabidopsis WT (Col-0), bak1-5, bik1-1, and bik1-1 pbl1-1. Lesion sizes were determined 5 dpi. At least 50 lesions in each experiment were measured. Means and SD were calculated from three independent experiments. The statistical significance of differences between means was determined by two-tailed t test. *P < 0.05 (comparison with Col-0). (B) Quantitative analysis of C. higginsianum entry. Conidia of C. higginsianum (5 × 10⁵ conidia per milliliter) were inoculated onto each plant line, as above, and inoculated plants were incubated for 3 d. One hundred appressoria produced by C. higginsianum on each plant were observed. Means and SD were calculated from five independent plant samples. The statistical significance of differences between means was determined by two-tailed t test. *P < 0.01 (comparison with Col-0). (C) Inoculation assay of nonadapted C. tropicale on Arabidopsis mutants. A conidial suspension of C. tropicale with 0.1% glucose (5 × 10⁵ conidia per milliliter) was inoculated onto Arabidopsis WT (Col-0), pen2-1, bak1-5, bik1-1, and bik1-1 pbl1-1. Inoculated plants were incubated for 4 d. (D) Quantitative analysis of C. tropicale entry. Conidia of C. tropicale with 0.05% glucose were inoculated onto each plant line, as above, and inoculated plants were incubated for 14 h. At least 100 conidia were observed in each experiment. Means and SD were calculated from three independent experiments. The statistical significance of differences between means was determined by two-tailed t test. *P < 0.01 (comparison with Col-0).

NIS1 Contributes to Fungal Virulence.

We previously reported that targeted deletion of CoNIS1 did not affect the virulence of C. orbiculare on Nb or cucumber, its natural host (23). This suggested either a functional redundancy of another effector(s) with CoNIS1 in C. orbiculare, or that CoNIS1 does not contribute to the virulence of C. orbiculare. To clarify this, we tested whether Nb expressing NIS1 displays enhanced susceptibility to C. orbiculare. We transiently expressed MoNIS1 in Nb leaves under the control of the cauliflower mosaic virus 35S promoter and then inoculated C. orbiculare onto the treated leaves. We expressed MoNIS1 instead of CoNIS1 because CoNIS1, but not MoNIS1, induced necrotic lesions, which would interfere with the evaluation of lesion development by the pathogen. The result revealed that MoNIS1 expression significantly enhanced lesion development by C. orbiculare (Fig. 7 A and B). This finding suggests that the conserved effector NIS1 has a role in the virulence of C. orbiculare against plants including Nb.

Fig. 7.

NIS1 contributes to virulence of fungal pathogens. (A) Transient expression of MoNIS1 promotes lesion formation by C. orbiculare on Nb. Conidial suspensions (5 × 10⁵ conidia per milliliter) of C. orbiculare wild-type strain were drop inoculated onto infiltration sites of A. tumefaciens harboring the plasmid expressing MoNIS1 or A. tumefaciens harboring the empty plasmid pBICP35 as a vector control (vector). Dashed lines represent the boundaries of infiltrated areas. The photograph was taken at 6 dpi. (B) Quantification of lesion number on Nb leaves transiently expressing MoNIS-HA after C. orbiculare inoculation. Conidial suspensions (5 × 10⁵ conidia per milliliter) of C. orbiculare wild-type strain were spray inoculated onto infiltration sites of A. tumefaciens harboring a plasmid expressing MoNIS1 or of A. tumefaciens harboring the empty plasmid pBICP35 (vector). Lesion number for each infiltrated area was counted at 4 dpi. Center lines show the medians. Box limits indicate the 25th and 75th percentiles. Whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles. Crosses represent sample means. Data points are plotted as open circles; n = 8 biological replicates. The statistical significance of differences between means was determined by two-tailed t test. **P < 0.01. (C) Inoculation assay of the monis1∆ mutant on barley (H. vulgare ‘Nigrate’). Conidial suspensions of WT (Δku70), monis1Δ, and monis1Δ expressing CoNIS1 genomic sequence were inoculated onto barley leaves, and incubated for 5 d. (D) Inoculation assay of the monis1∆ mutant on rice (O. sativa ‘Moukoto’). Conidial suspensions of WT (Δku70) and monis1Δ were inoculated onto rice leaves and incubated for 7 d. (E) Co-IP of rice OsBAK1 and NIS1 homologs. OsBAK1-GFP was transiently coexpressed with CoNIS1-HA or MoNIS1-HA in Nb. Similar results were obtained from one additional experiment.

We also generated a MoNIS1 knockout (monis1∆) mutant of M. oryzae by targeted gene disruption. The monis1∆ mutant exhibited normal colony growth, the same as the WT strain (SI Appendix, Fig. S9 A and B). Remarkably, inoculation assays of the mutant on barley and rice revealed that it possessed severely reduced virulence on both hosts compared with WT (Fig. 7 C and D); this phenotype was complemented by reintroduction of MoNIS1 (SI Appendix, Fig. S9C). The effector MoNIS1 therefore plays a critical role in host infection by M. oryzae. Strikingly, the virulence of the monis1∆ mutant was restored by introducing a genomic region encompassing CoNIS1 (Fig. 7C). It has been reported that OsBAK1 is involved in rice immunity against M. oryzae (41), and we found that both CoNIS1 and MoNIS1 indeed interact with OsBAK1 transiently expressed in Nb (Fig. 7E).

Discussion

The work presented here has revealed that the fungal core effector NIS1 targets BAK1/SERK3 and BIK1, key kinases in plant PTI signaling that have only previously been reported to be targeted by bacterial effectors (5, 11, 13, 15). It was previously reported that the fungal effector Pep1 is essential for host penetration by Ustilago maydis (42). Pep1 is conserved in Ustilaginaceae, where it is regarded as a core effector (43), but not outside this family. In contrast, the effector NIS1 is conserved in fungi belonging to both Ascomycota and Basidiomycota. Thus, we consider that NIS1 is a core effector, acquired anciently by fungal pathogens. This idea is supported by our finding that NIS1 targets PAMP recognition and signaling machineries that are probably universally conserved in higher plants. Indeed, it was recently reported that the moss Physcomitrella patens recognizes chitin and mounts defense responses via the RLK CERK1 (44), indicating that RLK-dependent PAMP recognition is an ancestral system in plants.

Since the sequence of NIS1 has no similarity to bacterial effectors that target RLKs and RLCKs, our findings also suggest independent effector innovation in bacterial and fungal pathogens to attack RLK and RLCK class kinases that are critical for PTI. Independent effector innovation at the interkingdom level has also been detected in multiple plant proteins such as Rcr3 protease, which is targeted by protease inhibitors from pathogenic fungi, oomycetes, and nematodes (17, 45, 46). BAK1-related RLKs and BIK1-related RLCKs are well conserved and critical for PTI; thus, we consider that once a fungus had innovated NIS1, it was essential to maintain it as a core effector during evolution. In contrast to NIS1, bacterial effectors that attack RLKs and RLCKs, such as AvrPto, are not broadly conserved among bacterial pathogens. It remains unclear why this type of effector is not widely conserved among bacterial pathogens. As noted above, genome sequencing of 65 strains of X. axonopodis pv. manihotis identified nine core effectors that are shared by all tested strains of this bacterial pathogen (22). However, these effectors are not strongly conserved in X. campestris. Furthermore, there is no overlap between X. campestris and P. syringae core effectomes (47). In contrast, many putative core effectors are conserved in fungal species: for example, ∼200 core effector-like genes are conserved in four different Colletotrichum species (48), suggesting that infection strategies based on core effectors, including NIS1, are more important in fungal pathogens than in bacterial pathogens.

The expression of CoNIS1 and ChNIS1, but not MoNIS1, caused necrotic lesion formation in Nb (23), whereas CoNIS1, ChNIS1, and MoNIS1 commonly suppressed INF1-induced cell death and PAMP-triggered ROS generation in Nb (Figs. 2A and 4A). CoNIS1 lacking its C-terminal 30 amino acids (CoNIS1∆C30) strongly reduced necrotic lesion development (23) in comparison with full-length CoNIS1, and this region is missing in MoNIS1 (SI Appendix, Fig. S1). Thus, the corresponding region is dispensable for plant immune suppression but is important to induce necrotic lesions, possibly via plant recognition and activation of cell death as a counterdefense. In this context, it is notable that C. orbiculare deploys an effector, DN3, that can reduce CoNIS1-triggered plant cell death (23). MoNIS1 is phylogenetically distant from CoNIS1 and is likely to originate from an NIS1 homolog of the Basidiomycota. Considering that Valsa mali, a Sordariomycete fungus phylogenetically close to M. oryzae, carries two copies of NIS1 homologs, one of which is closely related to MoNIS1 while the other belongs to the same clade as CoNIS1 (Fig. 1 and SI Appendix, Fig. S4), M. oryzae or a close ancestor species appears to have lost its CoNIS1-like NIS1 homolog. The loss of this NIS1 homolog might have enabled the pathogen to avoid recognition by a particular host plant while maintaining the essential ability to suppress host immunity by expressing the other NIS1.

Interestingly, it was recently reported that the expression of the Fusarium virguliforme NIS1 ortholog induced sudden death syndrome foliar symptoms in soybean, mimicking the development of soybean foliar symptoms caused by F. virguliforme itself in the field (49). Thus, NIS1-triggered cell death via plant recognition may be deployed by a pathogen for enhancing virulence in certain plant–fungal pathogen interactions. It is also noteworthy that the root endophyte fungus C. tofieldiae possesses a NIS1 homolog (Fig. 1 and SI Appendix, Fig. S1), and this fungus exhibits high expression of the NIS1 homolog during Arabidopsis colonization (25). We revealed that the C. tofieldiae NIS1 homolog suppresses flg22-triggered ROS generation in Nb in the same way as the pathogenic fungal NIS1s (Fig. 4 A and G). This suggests that C. tofieldiae uses NIS1 to colonize Arabidopsis, i.e., the endophyte fungus needs effectors, including NIS1, to establish its beneficial interaction with Arabidopsis.

Expression of CoNIS1 and its orthologs in Nb almost completely abolishes flg22-triggered ROS generation (Fig. 4A). We found that NIS1 interacts with BIK1, resulting in the inhibition of the BIK1-RBOHD interaction (Fig. 4 C and E). BIK1 induces flg22-triggered ROS generation by phosphorylating RBOHD in Arabidopsis. Arabidopsis RBOHD shares high similarity with NbRbohB (50) and complements the loss of the flg22-induced ROS burst in NbRbohB-silenced Nb plants (9), indicating that NbRbohB is a functional ortholog of Arabidopsis RBOHD. Although a functional ortholog of BIK1 has not yet been identified in Nb, we propose that NIS1 targets such an ortholog, inhibiting the PAMP-triggered ROS generation that largely depends on NbRbohB (51).

BIK1 is a key component that links PTI signals from transmembrane PAMP receptor complexes to intracellular immune responses, although it is unclear how BIK1 regulates downstream signaling except for the direct activation of RBOHD (9, 10). The rbohb-silenced Nb displayed reduced ROS production and decreased resistance to P. infestans (52). However, rbohb silencing did not reduce the immunity of Nb to C. orbiculare (53), whereas the expression of NIS1 in Nb enhanced susceptibility to C. orbiculare (Fig. 7 A and B). This indicates that inhibition of PAMP-triggered ROS generation is insufficient to reduce Nb immunity to C. orbiculare. Consistent with this, both Arabidopsis rbohD and rbohF mutants maintain immunity to C. tropicale (39), but in contrast the Arabidopsis bik1 mutant as well as the bak1-5 mutant display reduced immunity to the same pathogen (Fig. 6 C and D). It is necessary to investigate further the plant defense pathways that are effective against Colletotrichum fungi, which partially depend on BAK1 and its related RLKs as well as BIK1 and may be dampened by NIS1. We also showed that MoNIS1 suppressed R3a/Avr3a-dependent HR cell death in Nb (SI Appendix, Fig. S8), which is not suppressed in serk3-silenced Nb (36). Thus, NIS1 may have additional targets, probably belonging to a kinase family that regulates R3a/Avr3a-triggered ETI.

Deletion of the NIS1 ortholog MoNIS1 severely compromised the virulence of M. oryzae on rice and barley (Fig. 7 C and D). OsBAK1 is reportedly involved in rice immunity to M. oryzae (41). Taken together with the interaction of CoNIS1 and MoNIS1 with OsBAK1 (Fig. 7E), this result implies that M. oryzae needs to suppress OsBAK1 function via MoNIS1 for successful infection. Also, it was reported that a rice RLCK named OsRLCK185 is involved in rice immunity and targeted by the bacterial effector Xoo1488 (54). MoNIS1 may target RLCKs, including OsRLCK185, of rice and/or barley to enhance the virulence of M. oryzae toward these species.

Materials and Methods

Plasmids were constructed as described in detail in SI Appendix, SI Materials and Methods and Table S3. Methods for plant growth, yeast two-hybrid assay, ROS measurement, in vitro kinase activity assay, and colocalization assay are also described in detail in SI Appendix, SI Materials and Methods.

Fungal Strains and Media.

The C. orbiculare (syn. Colletotrichum lagenarium) wild-type strain 104-T (MAFF240422) is stored at the Laboratory of Plant Pathology, Kyoto University. C. tropicale (previously reported as C. gloeosporioides) S9275 (MAFF840071) was provided by Shigenobu Yoshida, National Institute for Agro-Environmental Sciences, Ibaraki, Japan. C. higginsianum MAFF305635 was obtained from National Agriculture and Food Research Organization, GenBank, Ibaraki, Japan. Cultures of all fungal isolates of Colletotrichum were maintained on 3.9% (wt/vol) PDA medium (BD Difco) at 24 °C in the dark. For conidiation, a C. tropicale culture was placed in a cycle of 16 h black light and 8 h dark. All strains of M. oryzae used in this study are stored at the Iwate Biotechnology Research Center. The Δku70 mutant of M. oryzae strain Guy11 (55), provided by N. J. Talbot, University of Exeter, Devon, UK, was used as the wild-type strain. Cultures of M. oryzae were maintained on oatmeal agar medium at 25 °C in the dark. For conidiation, a M. oryzae culture was grown under a black light (FS20S/BLB 20W, Toshiba) for 4 d at 22 °C, after aerial hyphae of the colonies had been washed away with sterilized distilled water.

Transient Expression in Nb.

Nb plants (5–6 wk old) were used for agroinfiltration assays. Constructs derived from pBICP35 were used to transform Agrobacterium tumefaciens GV3101 (56) by electroporation. Transformant cells were cultured and harvested by centrifugation and suspended in MMA induction buffer (1 L MMA: 5 g MS salts, 1.95 g Mes, 20 g sucrose, 200 mM acetosyringone, pH 5.6). The suspensions were infiltrated into Nb leaves using a syringe. For cell death suppression assay, at 1 d after infiltration with A. tumefaciens carrying a plasmid harboring each NIS1 homolog, the infiltration site was challenged with a recombinant A. tumefaciens strain carrying a plasmid harboring INF1 or a mixture of two A. tumefaciens strains carrying plasmids harboring either Avr3a^KI (Δ23–147) or R3a. All suspensions were incubated for 1 h before infiltration. INF1-induced cell death was observed at 3 dpi. Protein extracts were analyzed by immunoblot using anti-HA (3F10, Roche), anti-GFP [GFP(B-2), Santa Cruz Biotechnology] and anti-FLAG (F-3165, Sigma-Aldrich) antibodies. HRP-linked anti-rat (7077, Cell Signaling Technology) and anti-mouse (7076, Cell Signaling Technology) IgG were used as secondary antibodies.

Coimmunoprecipitation Assays.

Microsomal fractions were extracted from agroinfiltrated Nb leaves at 2 dpi by homogenizing with extraction buffer [50 mM Tris⋅HCl pH 7.4, 150 mM NaCl, 5% glycerol, 0.5% Triton X-100, protease inhibitor mixture (11836153001, Roche)] on ice. Crude protein extracts were centrifuged at 20,000 × g for 10 min to remove cell debris, and the supernatants were then incubated with µMACS anti-GFP magnetic microbeads (130-091-288, Miltenyi Biotec) for 1 h at 4 °C with gentle rotation. For MoNIS1-BIK1 interaction, the supernatants were incubated with GFP-trap-A (ChromoTek) for 2 h at 4 °C. After three washing steps with wash buffer (50 mM Tris⋅HCl pH 7.4, 150 mM NaCl, 5% glycerol, 0.1% Triton X-100), immunoprecipitates were analyzed by immunoblotting with anti-HA (3F10, Roche), anti-GFP [GFP(B-2), Santa Cruz Biotecnology] and anti-FLAG (F-3165, Sigma-Aldrich) antibodies. HRP-linked anti-rat (7077, Cell Signaling Technology) and anti-mouse (7076, Cell Signaling Technology) IgG were used as secondary antibodies.

Pathogen Inoculation Assays.

In an inoculation assay of C. orbiculare on Nb, 2 d after infiltration with A. tumefaciens carrying a plasmid harboring MoNIS1-HA or pBICP35 (empty vector), 10 µL of a conidial suspension (5 × 10⁵ conidia per milliliter) of C. orbiculare 104-T strain was drop inoculated (for lesion size evaluation) or spray inoculated (for lesion number quantification) onto the infiltration areas of detached leaves. After incubation for 4 d at 24 °C, the lesion number of each infiltrated area was measured. Mean and SD were calculated for “lesion number per 100 cm²” values from eight independent infiltration areas. The statistical significance of differences between means was determined by unpaired t test (two tailed). Statistical significance was defined as a P value of 0.01 or lower. An inoculation assay of adapted C. higginsianum on Arabidopsis mutants was performed as follows: A conidial suspension of C. higginsianum (2.5 × 10⁵ conidia per milliliter) was inoculated onto leaves of Arabidopsis WT (Col-0), bak1-5, bik1-1, and bik1-1 pbl1-1 plants and incubated for 5 d. The lesion diameter of inoculated leaves was determined with Adobe Photoshop. At least 50 lesions were measured in each experiment. Mean and SD were calculated from three independent experiments. The statistical significance of differences between means was determined by unpaired t test (two tailed); statistical significance was defined as a P value of 0.05 or lower. To measure entry (invasion) rates, conidia (5 × 10⁵ conidia per milliliter) of the pathogen were inoculated on cotyledons, which were then mounted in water under a coverslip, with the inoculated surface facing the objective lens. The invasion ratio (%) was calculated by the following numerical formula: (number of appressoria forming invasive hyphae)/(number of appressoria) × 100. One hundred appressoria were observed at 3 dpi on each plant. Mean and SD were calculated from five independent plant samples. The statistical significance of differences between means was determined by unpaired t test (two tailed), with statistical significance defined as a P value of 0.01 or lower. An inoculation assay of nonadapted C. tropicale on Arabidopsis mutants was performed as follows: A conidial suspension of C. tropicale (5 × 10⁵ conidia per milliliter) with 0.1% glucose was inoculated onto leaves of Arabidopsis WT (Col-0), pen2-1, bak1-5, bik1-1, and bik1-1 pbl1-1 plants and incubated for 4 d. To measure entry (invasion) rates, the pathogen was inoculated with 0.05% glucose on cotyledons and the invasion ratio (%) was calculated by the following numerical formula: (number of conidia forming invasive hyphae)/(number of germinating conidia) × 100. At least 100 conidia were observed at ∼14 hpi in each experiment. Means and SD were calculated from three independent experiments. The statistical significance of differences between means was determined by unpaired t test (two tailed). Statistical significance was defined as a P value of 0.01 or lower. For an inoculation assay of M. oryzae on barley and rice, a conidial suspension of M. oryzae (5 × 10⁴ conidia per milliliter) containing 0.01% Tween 20 was sprayed onto susceptible barley cotyledons (Hordeum vulgare ‘Nigrate’) and rice seedlings (Oryza sativa ‘Moukoto’) of the fourth leaf stage. Inoculated plants were placed in a dew chamber at 27 °C for 24 h in the dark and then transferred to a growth chamber with a photoperiod of 16 h.

MoNIS1 Gene Disruption and Complementation in M. oryzae.

To obtain protoplasts, hyphae of M. oryzae strains were incubated for 3 d in 200 mL of YG medium (5 g of yeast extract, and 20 g of glucose per liter). Protoplast preparation and transformation were performed as described previously (57). Hygromycin- or bialaphos-resistant transformants were selected on plates with 300 µg/mL of hygromycin B (Wako Pure Chemicals) or 250 µg/mL of bialaphos (Wako Pure Chemicals), respectively. Targeted gene disruption of MoNIS1 in M. oryzae was performed by homologous recombination using a gene disruption vector. The primers used for generating the vector are listed in SI Appendix, Table S3. Transformants were analyzed by PCR with the primers MoNIS1F/MoNIS1R, with HPH-specific primers, or with bialaphos resistance gene (bar)-specific primers.

Phylogenetic Analysis.

We used ETE Toolkit version 3.0.0b35 (58) to construct a maximum likelihood tree of NIS1 homologs. Multiple alignment of protein sequences of NIS1 homologs was generated with Clustal Omega version 1.2.1 (59). Alignment cleaning was performed by trimAl v1.4 (60) with “–gt 0.1” setting. A maximum likelihood tree was obtained using PhyML version 20131022 (61) with the JTT model and 100 bootstrap replicates. Species tree topologies were built with CVTree version 5.0 (62, 63) using the predicted proteomes of 141 fungal species (SI Appendix, Table S2). The final NIS1 protein tree was reconstructed using Treefix-DTL version 1.0.2 (64) with the above-generated NIS1 protein alignments (SI Appendix, Fig. S1), NIS1 ML tree (SI Appendix, Fig. S4) and species tree (SI Appendix, Fig. S4) as inputs. Trees were visualized using iTOL (ref. 65; https://itol.embl.de/).