Abstract
A subset of plant nucleotide-binding domain and leucine-rich repeat-containing (NLR) proteins carry extraneous integrated domains that have been proposed to mediate pathogen effector recognition. The current view is that these unconventional domains function by directly binding or serving as substrates for pathogen effectors, yet only a few domains have been functionally characterized to date. Here we report that the integrated NOI domain of the rice NLR protein Pii-2, together with its partner Pii-1, mediates immunity to the rice blast fungus Magnaporthe oryzae by indirect recognition of the AVR-Pii effector. We discovered that the Pii-2 NOI domain does not physically interact with the effector itself but instead binds the host protein OsExo70-F3, which is a target of AVR-Pii. We further identified mutations within the NOI core motif (PxFGxW) of Pii-2 that abolish both OsExo70-F3 binding and Pii-mediated resistance to M. oryzae expressing AVR-Pii. This led us to propose a novel conceptual model in which an NLR-integrated domain functions to detect host proteins targeted by pathogen effectors, in a framework that extends classical indirect recognition models.
Plants have evolved immune receptors to defend against pathogen infection. These include cytosolic nucleotide-binding domain and leucine-rich repeat-containing (NLR) proteins that recognize pathogen avirulence (AVR) effectors and activate a complex immune response that includes hypersensitive cell death. Plant NLRs have a modular architecture comprising either an N-terminal coiled-coil (CC) or a Toll/interleukin-1 receptor (TIR) homology domain, followed by nucleotide binding (NB) and leucine rich repeat (LRR) domains. To date, several reports have shown that genetically-linked NLR pairs (e.g., RPS4/RPS1, RGA4/RGA5 and Pikp-1/Pikp-2) are required to act together for the recognition of cog<nate AVRs. Of these pairs, one NLR is a sensor NLR that perceives pathogen effectors while the other is a helper NLR that is necessary to activate immune signaling (ref. 1,2). A number of sensor NLRs, perhaps as many as 10%, have recently been reported to contain additional unconventional domains called integrated domains (IDs) that appear to have originated from host targets of effectors (ref. 3,4). The current view is that NLR-integrated domains function as baits or decoys that bind pathogen effectors or serve as substrates of effectors, in order to detect invading pathogens. One example is Arabidopsis RRS1, an NLR that carries a C-terminal domain with similarity to WRKY transcriptional factors. RRS1 is required for the perception of bacterial effectors PopP2 and AvrRps4, which are bacterial effectors that target Arabidopsis WRKY proteins to enhance virulence. Therefore, Le Roux et al. (2015) concluded that a WRKY domain targeted by pathogen effectors has integrated into RRS1 to facilitate recognition of the bacterial effectors (ref. 5,6). Other examples include the rice NLR pairs Pia (RGA4/RGA5) and Pikp (Pikp-1/Pikp-2), which recognize the effectors AVR-Pia and AVR-PikD of the rice blast fungus, Magnaporthe oryzae (syn. Pyricularia oryzae), respectively. Remarkably, both RGA5 and Pikp-1 include a Heavy Metal-Associated (HMA) domain integrated in different positions in these NLRs. Binding of the integrated HMA domain to the corresponding AVR effectors triggers the hypersensitive response and immunity to the blast fungus (ref. 7,8). These findings let to the proposal that unconventional integrated domains of NLRs have evolved to directly bind pathogen effectors or serve as a substrate for the effectors’s activity (ref. 9). As yet, the degree to which IDs associate with host proteins is largely unknown.
The Pii locus in rice encodes a pair of CC-NLR proteins, Pii-1 and Pii-2, that detect the M. oryzae effector AVR-Pii and mount an effective immune response against races of this pathogen that express AVR-Pii (ref. 10,11). Despite repeated attempts using various protein-protein interaction methods, we did not detect direct interaction between AVR-Pii and either one of Pii-1 and Pii-2 (ref. 12). Instead, we discovered that AVR-Pii binds two members of the large EXO70 protein family of rice, OsExo70F-2 and OsExo70F-3 (ref. 12). EXO70 is a subunit of the exocyst, an evolutionarily conserved vesicle tethering complex that functions in the last stage of exocytosis (ref. 13,14). Although we do not understand the degree to which AVR-Pii binding to EXO70 contributes to virulence, we found that OsExo70-F3 is genetically required for Pii-mediated immunity to M. oryzae expressing AVR-Pii (ref. 12). This led us to propose that Pii response to AVR-Pii follows the indirect recognition model, in which the NLR protein recognizes modifications of a host protein targeted by the effector (ref. 12).
Here we report that Pii recognition of AVR-Pii follows a novel conceptual model of NLR perception of plant pathogens. We found that the 280 C-terminal amino acids of Pii-2 (Pii-2-CT) bind the rice protein OsExo70-F3 that is targeted by the AVR-Pii effector (Fig. 1, Supplementary Fig. 1, 2 and 3). We further delineated the interaction region to the 103 amino acids at the C-terminus of the protein (Pii-2-CTC) (Supplementary Fig. 1, 2 and 3). Interestingly, this region of Pii-2 contains a short sequence with similarity to a core motif of nitrate-induced (NOI) domain (PxFGxW) that is also conserved in the Pi5-2 which is allelic to Pii-2 (Supplementary Fig. 3 and ref. 3,15). The NOI domain also defines the Arabidopsis RPM1-INTERACTING PROTEIN4 (RIN4), which is known to negatively regulate plant defense and is targeted by bacterial effectors AvrRpt2, AvrB and AvrRpm1 (ref. 16). In fact, the protease effector AvrRpt2 cleaves RIN4 protein at the NOI motif PxFGxW. To further investigate the importance of the Pii-2 NOI motif, we introduced multiple mutations in this region and tested the resulting mutants for binding to OsExo70-F3 (Fig.1, Supplementary Fig. 4). Mutations in the NOI conserved residues, Pro(P), Phe(F) and Trp(W) abolished binding to OsExo70-F3 (Fig. 1, Supplementary Fig. 4 and 5). This indicates that the Pii-2 NOI core motif is required for OsExo70-F3 binding.
(A) A schematic diagram of Pii-1 and Pii-2 NLRs. Coiled-coil (CC), nucleotide-binding (NB) and leucine rich repeat (LRR) domains are shown in light grey. OsExo70-F3 binds the C-terminus of the Pii-2 (Pii-2-CT) shown in dark grey (Pii-2-CT) that contains the core motif (PxFGxW) of NOI domain. (B) Alignment of amino acid sequences around the NOI core motif of Arabidopsis RIN4, rice Pi5-2 and Pii-2 proteins. Residues of the NOI core motif are shown in green. Asterisks indicate conserved amino acids among Arabidopsis RIN4, Pi5-2 and Pii-2 proteins, which are substituted for Ala (A) in the Pii-2-CT-m5 mutant. (C) Yeast two hybrid (Y2H) assay showing the interaction of OsExo70-F3 with the wild-type Pii-2-CT but not with the Pii-2-CT-m5 mutant. Protein interactions were tested in the yeast grown on Quadruple dropout (QDO) + 3AT medium (Trp−Leu−Ade−His− Xα gal+ +10 mM 3AT: right panels). Viability of all transformed yeast cells were demonstrated by growing them on double dropout (DDO) medium (Trp−Leu−: left panels). Yeast cells were spotted onto the media as tenfold dilution series (x1, x10−1, x10−2). Empty vectors (labeled as “empty”) were used as negative controls.
Next, we investigated the role of the NOI motif in Pii-mediated immunity. We determined the effect of mutations within the NOI core motif on Pii-mediated resistance to M. oryzae using genetic complementation of a pii-2 loss-of-function rice line with Pii-2 transgenes carrying the mutations in the NOI motif described above. To obtain a pii-2 loss-of-function mutant rice, we screened 5,600 ethyl methanesulfonate (EMS) mutant lines of rice cultivar Hitomebore, which carries the Pii gene (ref. 17). This led to the identification of two pii-2 loss-of-function mutants, Hit5882 and Hit13701, carrying an amino acid substitution (V371D) and a nonsense mutation (Q247stop) in the Pii-2 NB-domain, respectively (Supplementary Fig. 3; ref. 11). A full-length cDNA of wild-type Pii-2 under the control of the 35S promoter (Hit5882+WT-Pii-2) transformed in Hit5882 complemented Pii-mediated resistance (Fig. 2). In contrast, transformation of Hit5882 with a full-length cDNA of Pii-2 carrying the NOI-m5 mutation failed to confer Pii-dependent resistance in all transgenic lines tested (Hit5882+Pii-2-NOI-m5; Fig. 2). The Pii-2-NOI-m5 mutant also failed to complement the second Pii-loss-of-function rice line Hit13701 (Supplementary Fig. 6). Overall, our data genetically link Pii-2 NOI motif binding to OsExo70-F3 with Pii-mediated resistance to M. oryzae expressing the AVR-Pii effector. This suggests that NOI binding to EXO70 is crucial for Pii function.
(A) Results of rice leaf blade spot inoculation assay of M. oryzae. Two M. oryzae isolates, with (transgenic Sasa2+AVR-Pii; top) or without (Sasa2; bottom) AVR-Pii were spot inoculated onto the leaves of rice lines Moukoto (Pii-), Hitomebore (Pii+), Hit5882, a pii-2 Hitomebore mutant, and transgenic Hit5882 lines expressing a full-length cDNA of the wild-type Pii-2 (Hit5882+WT-Pii-2) and those expressing the mutant version of Pii-2 with m5 mutation in the NOI core motif (Hit5882+Pii-2-NOI-m5). Representative data of two line each of Hit5882+WT-Pii-2 (W1 and W2) and Hit5882+Pii-2-NOI-m5 (m2 and m3) are shown. (B) Bar graphs showing disease lesion lengths after inoculation of an M. oryzae isolate Sasa2+AVR-Pii onto rice lines from panel A, labelled on the bottom. Here, five lines each from Hit5882+WT-Pii-2 (W1 to W5) and Hit5882+Pii-2-NOI-m5 (m1 to m5) were tested. The graph illustrates average lesion lengths and standard deviations from six inoculated spots per line. (C) Expression of Pii-2 transgenes in the tested rice plants. Messenger RNA from Pii-2 transgene but not from intrinsic gene was specifically detected by semi-quantitative RT-PCR using a combination of primers derived from Pii-2 gene and the binary vector sequences. Rice Actin was used as control.
The current view is that integrated domains of NLR immune receptors function by directly binding or serving as substrates for pathogen effectors. Our findings suggest that NLR-integrated domains may function by detecting host proteins targeted by pathogen effectors in a novel model in which an integrated domain of an NLR mediates indirect recognition of a pathogen effector (ref. 18,19) (Fig. 3). We hypothesize that the rice NLR Pii-2 has integrated a NOI domain originating from a host NOI-EXO70 complex targeted by AVR-Pii. This NOI integration has enabled Pii to monitor OsExo70-F3 via physical interaction, which allowed indirect detection of AVR-Pii (Fig. 3). Recent reports showing RIN4-EXO70 interactions (ref. 20,21) support this hypothesis.
(A) A widely accepted model of NLR-ID function. NLR directly interacts with pathogen effector via its integrated domain, and subsequently activates immune response. An Arabidopsis sensor NLR, RRS1 (top), has the WRKY domain at the C-terminus, and bacterial effectors AvrRps4/PopP2 directly bind to the RRS1 WRKY ID (ref. 5,6). The rice sensor NLRs, Pikp-1 (middle) and RGA5 (bottom), carry integrated HMA domain. Pathogen AVR effectors AVR-PikD and AVR-Pia, respectively, directly bind to their cognate HMA (ref. 7,8). In the current model (ref. 1, 2), these sensor NLRs trigger ETI together with their NLR helpers. (B) A novel model of ID function in indirect detection of pathogen effectors. Pii-2 is an NLR with an ID, NOI core motif (PxFGxW), residing in its C-terminal region. Pii-2-ID directly binds OsExo70-F3, which is necessary for AVR-Pii recognition. ΘsExo70-F3 also binds AVR-Pii. We propose that Pii-2 monitors OsExo70-F3, the status of which is altered upon binding of AVR-Pii. Thus, Pii-2 ID guards a host protein, OsExo70-F3, that is targeted by a pathogen effector, AVR-Pii.
Our work also points to the NOI-EXO70 complex as a major target of plant pathogen effectors. These domains have therefore been acquired as sensor domains by NLR immune receptors, and both NOI and EXO70 are commonly seen integrated into NLRs (ref. 22, 23). Whether integrated NOI and EXO70 domains can mediate direct recognition of effectors remains to be determined. Nonetheless, our finding that NLR-integrated domains can indirectly detect AVR effectors expands the mechanistic view of pathogen detection by immune receptors—that the interactors of NLR-integrated domains can be either pathogen effectors, or host proteins that are targeted by effectors. This work shows the first example of indirect pathogen recognition via an integrated NLR, highlighting how the arms race between plants and pathogens has driven the emergence of a wealth of molecular interactions and mechanisms in the fight for survival.
Methods
Plasmids and Plants
The primers and their sequences used for plasmid constructions are listed in Supplementary Table 1. For Y2H assay, OsExo70-F3 in pGADT7 (prey vector), Pii-1-CC, Pii-1-NB, Pii-1-LRR+CT, Pii-2-CC, Pii-2-NB and Pii-2-LRR+CT in pGBKT7 (bait vector) have been previously constructed (ref. 17). To construct pGBKT7-Pii-2-CT and pGBKT7-Pii-2-CTC, Pii-2-CT and Pii-2-CTC fragments were amplified by PCR using primer sets: KF825f/KF826r and KF827f/KF826r, respectively, and inserted into pGBKT7 vector by utilizing EcoRI and BamHI sites. To prepare bait vectors of the Pii-2-CT mutant, Pii-2-CT-m5, the N-terminal half and the C-terminal half of Pii-2-CT fragment were amplified by using primer sets (KF825f/KF831r and KF830f/KF826r, respectively), and resulting PCR products were mixed and used as DNA template for second PCR using a primer set (KF825f /KF826r) to amplify Pii-2-CT-m5 fragment. Similarly, to construct other pGBKT7-Pii-2-CT mutants, primer sets (1st PCR: KF825f/KF833r for Pii-2-CT-VA-N and KF832f/KF826r for Pii-2-CT-VA-C; KF825f/KF835r for Pii-2-CT-PA-N and KF834f/KF826r for Pii-2-CT-PA-C; KF825f/KF837r for Pii-2-CT-FA-N and KF836f/KF826r for Pii-2-CT-FA-C; KF825f/KF839r for Pii-2-CT-GA-N and KF838f/KF826r for Pii-2-CT-GA-C; KF825f/KF841r for Pii-2-CT-WA-N and KF840f/KF826r for Pii-2-CT-WA-C) were used for PCR to amplify the N- and C-terminal half fragments of Pii-2-CT mutants. The N- and C-terminal half fragments of each Pii-2-CT mutants were mixed and used as DNA template for second PCR using a primer set (KF825f /KF826r). After the second PCR, the resulting 0.85 kb PCR products were inserted into pGBKT7 vector by utilizing EcoRI and BamHI sites.
For preparing the expression vectors in rice, full-length coding sequences of Pii-2 was amplified from cDNA by PCR using primer set, KF828f/KF829r, and was introduced into pCAMBIA1300S vector (Cambia, www.cambia.org) by utilizing KpnI and XbaI sites (pCAMBIA-Pii-2). To construct expression vector for Pii-2 mutant (Pii-2-NOI-m5), two Pii-2 fragments were amplified by using primer sets (KF828f/KF831r and KF830f/KF829r, respectively), and resulting PCR products were mixed and used as DNA template for the second PCR using a primer set (KF828f /KF829r) to amplify Pii-2-NOI-m5 fragment. The Pii-2-NOI-m5 fragment was inserted into pCAMBIA1300S vector (Cambia, www.cambia.org) by utilizing KpnI and XbaI sites (pCAMBIA-Pii-2-NOI-m5).
To prepare transgenic rice plants expressing wild-type Pii-2 and mutant Pii-2 (Pii-2-NOI-m5), pCAMBIA-Pii-2 and pCAMBIA-Pii-2-NOI-m5 were introduced into Agrobacterium tumefaciens, which were used for transformation of Hit5882 and Hit13701 that are pii-2 deficient mutants of the rice cultivar Hitomebore (ref. 11).
Yeast two-hybrid assay
Y2H assay was performed as described previously (ref. 24). Ten-times dilution series [OD600 = 3.0 (x1), 0.3 (x10−1) and 0.03 (x10−2)] of yeast cells were prepared and spotted onto quadruple dropout medium (QDO); basal medium lacking Trp, Leu, Ade and His but containing 5-Bromo-4-Chloro-3-indolyl a-D-galactopyranoside (X-a-gal) (Clontech). To detect interactions, both QDO medium with or without 10 mM 3-amino-1,2,4-triazole (3AT) (Sigma) was used. Yeast cells were also spotted onto double dropout medium (DDO); basal medium lacking Trp, Leu to test cell viability. To check protein accumulation in yeast cells, yeast cells were propagated in liquid DDO at 30°C overnight. Forty mg yeast cells were collected and resuspended with 160 μl GTN + DC buffer [10% glycerol, 25mM Tris-HCl (pH 7.5), 150mM NaCl, 1 mM DTT and 1 tablet of complete EDTA-free (Roche, Basel Switzerland)]. Then, 160 μl of 0.6 N NaOH was added, mixed gently, and incubated at room temperature for 10 min. Next, 160 μl of gel sample buffer [40%(w/v) glycerol, 240 mM Tris-HCl pH 6.8, 8% (w/v) SDS, 0.04% (w/v) bromophenol blue, 400 mM DTT] was added and incubated at 95 °C for 5 min. After 20,000 x g centrifugation for 5 min, the supernatant was subjected to SDS-PAGE. Proteins expressed from bait and prey vectors were immunologically detected by using anti-Myc antibody (MBL, Nagoya, Japan) and anti-HA 3F10 (Roche), respectively.
Assays for fungal pathogenicity and gene expression
Wild-type M. oryzae isolate Sasa2 (no AVR-Pii) is stored at Iwate Biotechnology Research Center, and a transgenic M. oryzae strain (Sasa2+AVR-Pii-s) was previously established as Sasa2(+22p:pex33) (ref. 17). Rice leaf blade spot inoculation with conidial suspension (5 × 105 conidia/ml) was performed as described previously (ref. 24). Disease lesions were photographed at 10 days after inoculation, and vertical length of the lesions was measured. To check Pii-2 expression in transgenic rice plants, semi-quantitative RT-PCR was performed as described previously (ref. 12). Messenger RNA from Pii-2 transgene but not from intrinsic gene was specifically detected by using a primer set (5’-CTTT CGCGAGCT CGGTACCAACAAT G-3’ and 5’-ATCCACATCATTTTGTAACAATAG-3’). As a control, rice Actin was detected by using a primer set (5’-CTGAAGAGCATCCTGTATTG -3’ and 5’-GAACCTTTCTGCTCCGATGG -3’).
Contributions
K.F. S.K. and R.T. initiated, conceived and coordinated the project; K.F. identified and characterized NOI core motif in Pii-2; E.K. generated transgenic rice plants; K.F., Y.A. and H.U. prepared plamid vectors and performed fungal infection assay and RT-PCR. K.F. and K.I. performed yeast two hybrid assay. H.S., M.B., S.K. and R.T. supervised the project; K.F., A.B., S.K., and R.T. wrote the manuscript
Competing interests
The authors declare no competing financial interests.
Corresponding author
Co-correspondence to Ryohei Terauchi and Koki Fujisaki.
Acknowledgements
We thank members of our lab for discussions and general assistance. We also thank Matt Moscow and Helen Brabham, The Sainsbury Laboratory, UK for sharing their unpublished results. This work was supported by Science and Technology Research Promotion Program for Agriculture, Forestry, Fisheries and Food Industry, Japan, Grant-in-aid for MEXT (Scientific Research on Innovative Areas 23113009), JSPS KAKENHI (Grant No. 26850029) and JSPS KAKENHI 15H05779, the Biotechnology and Biological Sciences Research Council (BBSRC, UK, grants BB/J00453 and BB/P012574).