A genetically linked pair of NLR immune receptors shows contrasting patterns of evolution
Edited by David Baulcombe, University of Cambridge, Cambridge, United Kingdom; received September 14, 2021; accepted April 19, 2022
Significance
Plants have evolved sophisticated defense mechanisms to fend off pathogens. Plant nucleotide-binding leucine-rich-repeat receptor (NLR) proteins play crucial roles in detecting pathogen molecules inside plant cells and mounting defense responses. Here, we identified the Pias gene from rice, which encodes the NLR pair Pias-1 “helper” and Pias-2 “sensor.” These proteins function together to detect the pathogen molecule AVR-Pias of Magnaporthe oryzae and defend against rice blast disease. Pias is allelic to the previously reported Pia gene. A comparison of Pias/Pia alleles among Oryza species showed that the Pias/Pia helper is evolutionarily and functionally conserved, whereas the Pias/Pia sensor shows highly dynamic evolution, with various host domains integrated into similar positions, allowing it to detect a wide variety of pathogen molecules.
Abstract
Throughout their evolution, plant nucleotide-binding leucine-rich-repeat receptors (NLRs) have acquired widely divergent unconventional integrated domains that enhance their ability to detect pathogen effectors. However, the functional dynamics that drive the evolution of NLRs with integrated domains (NLR-IDs) remain poorly understood. Here, we reconstructed the evolutionary history of an NLR locus prone to unconventional domain integration and experimentally tested hypotheses about the evolution of NLR-IDs. We show that the rice (Oryza sativa) NLR Pias recognizes the effector AVR-Pias of the blast fungal pathogen Magnaporthe oryzae. Pias consists of a functionally specialized NLR pair, the helper Pias-1 and the sensor Pias-2, that is allelic to the previously characterized Pia pair of NLRs: the helper RGA4 and the sensor RGA5. Remarkably, Pias-2 carries a C-terminal DUF761 domain at a similar position to the heavy metal–associated (HMA) domain of RGA5. Phylogenomic analysis showed that Pias-2/RGA5 sensor NLRs have undergone recurrent genomic recombination within the genus Oryza, resulting in up to six sequence-divergent domain integrations. Allelic NLRs with divergent functions have been maintained transspecies in different Oryza lineages to detect sequence-divergent pathogen effectors. By contrast, Pias-1 has retained its NLR helper activity throughout evolution and is capable of functioning together with the divergent sensor-NLR RGA5 to respond to AVR-Pia. These results suggest that opposite selective forces have driven the evolution of paired NLRs: highly dynamic domain integration events maintained by balancing selection for sensor NLRs, in sharp contrast to purifying selection and functional conservation of immune signaling for helper NLRs.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Plants are continually attacked by a multitude of microbial pathogens. Pathogens secrete effector molecules to enable the invasion of their hosts (1). To counter this, plants have evolved a surveillance system that detects pathogen effectors inside the plant cell, leading to effector-triggered immunity (ETI) (2). Nucleotide-binding leucine-rich-repeat receptors (NLRs) play pivotal roles in ETI, which frequently leads to hypersensitive response (HR)–mediated cell death (3, 4). NLR genes underwent lineage-specific expansions in most plant genomes (∼150 in Arabidopsis thaliana and ∼500 in rice [Oryza sativa]) and are among the most variable genes in plants, pointing to strong selection pressure from pathogens (5, 6). NLR proteins are characterized by a conserved nucleotide-binding (NB) domain and a leucine-rich-repeat (LRR) domain. NLRs are divided into two major groups depending on the type of N-terminal domain: NLRs with the N-terminal coiled-coil (CC) domain are called CNLs (CC-NLRs), and those with the N-terminal Toll-like domain are called TNLs (TIR-NLRs). CNLs are widespread in the plant kingdom. TIRs are grouped into the canonical TIR and TIR2 subclasses. Though TIR2-NB proteins are found in monocot plants, TNLs with canonical TIR domains have been detected in dicot but not in monocot plants (7, 8). Adenosine diphosphate/adenosine triphosphate exchange at the NB domain (9, 10) and oligomerization of NLRs (11–16) trigger ETI signaling. The Arabidopsis CNL ZAR1 forms a pentamer “resistosome” complex after binding to a host protein and a pathogen effector. This leads to the protrusion of the N-terminal alpha helix, which perturbs the plasma membrane to trigger the HR (10, 15), a process potentially mediated by Ca2+ influx (17). The Arabidopsis TNL RPP1 forms a tetramer resistosome upon binding to its cognate effector ATR1, followed by an increase in its NAD+ase activity (11), which triggers cell death (18, 19). These recent breakthroughs have started to reveal the biochemical links between NLR molecular structure and HR induction (20).
NLRs can function as singletons, in pairs, or in networks (4). Arabidopsis ZAR1 (10), RPP1 (21), and many other NLRs function as singleton NLRs and recognize avirulence effectors (AVRs) directly or indirectly. Two NLR proteins encoded by genetically linked genes function together as paired NLRs. One of the paired NLRs frequently has a noncanonical domain called the integrated domain (ID). IDs are thought to have been derived from other host proteins (8, 22). Examples of paired NLRs include RPS4/RRS1 in Arabidopsis (23, 24) and RGA4/RGA5 (25, 26), Pik1/Pik2 (27, 28), and Pii1/Pii2 in rice (29, 30). Multiple NLRs encoded by unlinked genes may function together; these network NLRs (31) include NRC2, NRC3, and NRC4 in Solanaceae plants. When NLRs function as a pair or network, one NLR is involved in the recognition of the AVR effector (the “sensor NLR”), whereas the other plays a role in signaling (the “helper NLR”). Recent studies have expanded our understanding of the genetic architecture of NLR pairs and networks; however, how they function together and how they evolved remain elusive.
A phylogenetic study of 4,184 NLRs in the genomes of 7 Poaceae species including rice, wheat, and other grass species grouped the NLRs into 24 major clades, including 3 clades (MIC1, MIC2, and MIC3) containing the majority of NLRs with IDs (32). NLRs in clade MIC1 (including RGA5) are characterized by a wide variety of IDs integrated into similar positions in the NLR after the LRR domain. The study reported a conserved 43–amino acid motif between the LRR and ID named the CID (conservation and association with IDs) motif. The authors proposed a possible evolutionary mechanism of ID generation and ID shuffling mediated by the CID motif. However, this study examined only a few genomes from each species, with a focus on interspecies diversity within the Poaceae, and provided limited information about the recent diversification of NLRs within a single genus. In addition, experimental validation of the hypotheses underpinning the functional diversity of NLRs with IDs remains limited.
Rice blast disease caused by blast fungus (Magnaporthe oryzae [syn. Pyricularia oryzae]) is a major disease of rice that threatens world food security (33). The best way to control this disease is to deploy cultivars with resistance (R) genes. To date, 40 R genes against blast have been reported, and 25 have been cloned, most of which encode NLRs (34, 35). However, the molecular interactions between rice NLRs and blast AVRs are understood for only a small number of cases, including Pita and AVR-Pita (36, 37), Pia and AVR-Pia/AVR1-CO39 (26), Pik and AVR-Pik (28, 38), Pii and AVR-Pii (29, 30, 39), and Pizt and AVR-Pizt (40, 41).
The rice Pia pair, consisting of the NLR helper RGA4 and the NLR sensor RGA5, is one of the most well-characterized NLR pairs (25, 26). Overexpression of RGA4 in Nicotiana benthamiana and rice protoplasts caused HR-like cell death in the respective plant species, suggesting that the helper RGA4 is responsible for HR signaling. RGA4-mediated cell death was suppressed by coexpression of RGA5, indicating that RGA5 negatively regulates RGA4-mediated defense signaling. Finally, coexpression of RGA5, RGA4, and AVR-Pia triggered cell death via the direct binding of AVR-Pia to the heavy metal–associated (HMA) ID of the sensor RGA5 (26). This study, together with a study of the Arabidopsis RPS4/RRS1 pair (23, 24), supported a negative regulation model for paired NLRs in which helper NLRs function in HR signaling and sensor NLRs function in the suppression of helper-NLR activity. Upon binding or modification of the sensor-NLR IDs by pathogen AVRs, this suppression is released, allowing HR signaling to proceed. However, a recent study of rice Pikp paired NLRs suggests a helper–sensor cooperation model. Expression of the helper NLR alone did not cause cell death in N. benthamiana and coexpression of the helper and sensor was required for triggering effector-dependent HR-like cell death (42).
Here, we describe the rice NLR pair Pias, which recognizes the M. oryzae effector AVR-Pias. Pias encodes the helper Pias-1 and the sensor Pias-2 and is allelic to the previously characterized Pia gene, encoding the NLR pair helper RGA4 and sensor RGA5. Pias-2 carries a C-terminal DUF761 domain at a similar position to the HMA domain of RGA5. We show that the Pias/Pia helper-NLR lineage is evolutionarily and functionally conserved, while its sensor-NLR lineage shows highly dynamic evolution with various host domains integrated into similar positions, possibly allowing it to detect a wide variety of pathogen molecules.
Results
Isolation of Rice Pias NLR Genes and the Matching Gene AVR-Pias from M. oryzae.
As part of a large genetic screen to identify novel rice blast resistance genes, we crossed japonica-type rice cultivar Hitomebore to 20 different rice cultivars representing the worldwide genetic diversity of rice, resulting in the generation of recombinant inbred lines (RILs) of the F7 to F9 generations. We inoculated the parental rice lines with a panel of M. oryzae isolates and recorded their resistance or susceptibility to each isolate. Since Hitomebore was susceptible to M. oryzae isolate 2012-1 but indica-type accession WRC17 [cultivar Keiboba (43)] was resistant to this isolate, we set out to isolate the resistance genes in WRC17 rice against the 2012-1 pathogen. The 58 RILs derived from a cross between Hitomebore and WRC17 segregated into 52 resistant and 6 susceptible lines (Fig. 1A), indicating that WRC17 likely contains more than one locus conferring resistance against 2012-1.
Fig. 1.
To identify the resistance genes, we performed whole-genome sequencing and RNA sequencing (RNA-seq) of WRC17 and conducted an association study using a bioinformatics pipeline we named “RaIDeN” (https://github.com/YuSugihara/RaIDeN) using the RILs segregating for these phenotypes (Fig. 1 and SI Appendix, Fig. S1). We sequenced the genomes WRC17, Hitomebore, and six RILs showing susceptibility to the isolate 2012-1 on an Illumina DNA sequencer (SI Appendix, Table S1) and subjected the short reads of WRC17 to de novo assembly with DISCOVAR (https://www.broadinstitute.org/software/discovar/blog/), resulting in the WRC17 reference genome sequence (SI Appendix, Table S2). We also performed RNA-seq of WRC17 leaves that had been inoculated with M. oryzae (2012-1). The RNA-seq reads were mapped to the WRC17 reference genome, revealing 22,561 genes expressed from the WRC17 genome. The short reads obtained from Hitomebore and six susceptible (S-) RILs were aligned to the genome sequences of the 22,561 expressed genes. We reasoned that Hitomebore and the six S-RILs share the same DNA sequences in the candidate genes responsible for their resistance that are different from the sequences in resistant WRC17 (SI Appendix, Fig. S1). Two types of DNA polymorphisms were considered: 1) presence/absence of the genes and 2) single-nucleotide polymorphisms (SNPs) in the genes. We identified 14 genes that were present in WRC17 but absent from Hitomebore and the six S-RILs, in addition to 839 genes with shared SNPs among Hitomebore and the S-RILs that were different from the sequences of WRC17.
From this group of 853 genes (Dataset S1), we selected resistance gene analogs (RGAs) using “RGAugury” (44), which predicts genes encoding putative NLRs, receptor-like kinases (RLKs), and receptor-like proteins (RLPs). This analysis identified 38 RGAs as the candidate genes (SI Appendix, Table S3). Since most rice resistance genes against M. oryzae reported to date are NLRs, we focused on 18 NLR (11 CNL and 7 NL) genes as the candidate resistance genes of WRC17 against M. oryzae isolate 2012-1. Among the 18 NLR genes, only one (NL-04) showed presence/absence polymorphisms, and the rest contained SNPs associated with the phenotypes. We developed DNA markers in the candidate NLR genes and studied their association with phenotypes using 58 RILs segregating for resistance and susceptibility to the isolate 2012-1 (SI Appendix, Fig. S2A). This analysis showed that 16 NLRs were tightly linked to each other and that the two other NLRs were linked to each other, suggesting that the two loci (designated Pi-W17-1 and Pi-W17-2) are involved in the resistance of WRC17 against 2012-1 (Fig. 1B). Genotyping of the 52 RILs using the markers located in the two loci, Pi-W17-1 (CNL-04) and Pi-W17-2 (NL-07 and W17.11718), suggested that the lines became resistant when either of the two loci contained the WRC17-type allele (Fig. 1A and SI Appendix, Fig. S2A).
The genomic position of Pi-W17-1 corresponds to that of the previously reported Pia locus (25), and the position of Pi-W17-2 corresponds to that of the Pita (=NL-07) and Ptr (=W17.11718) loci (36, 45, 46). Among the progeny showing resistance against 2012-1, we selected 19 RILs containing Pi-W17-1 but not Pi-W17-2. These lines shared 10 candidate NLR genes within the Pi-W17-1 region (Fig. 1B and SI Appendix, Fig. S2B). We performed RNA interference (RNAi)–mediated gene silencing of eight of these genes (encoding proteins over 900 amino acids long) (SI Appendix, Fig. S2C) using the RIL HW-RIL7, which contains Pi-W17-1 but lacks Pi-W17-2. When CNL-04 or CNL-05 was silenced, its resistance against 2012-1 became compromised (SI Appendix, Fig. S2C and D). We confirmed this result by generating CNL-04 and CNL-05 knockout mutant lines by CRISPR-Cas9–mediated genome editing (Fig. 1C and SI Appendix, Fig. S3). These data suggest that Pi-W17-1–mediated resistance requires both the neighboring NLRs CNL-04 and CNL-05. Indeed, the position of CNL-04 corresponds to that of RGA4, whereas the position of CNL-05 corresponds to that of RGA5 of Pia (Fig. 1D) (25). Interestingly, CNL-04 is similar to RGA4 in terms of both structure and DNA sequence (96.6% DNA sequence identity), whereas CNL-05 has a distinct structure and a DNA sequence that diverged from RGA5 (59.8% DNA sequence identity) (SI Appendix, Fig. S4). Pia RGA5 encodes a protein with an HMA domain in its C terminus (25), whereas CNL-05 encodes a protein with a 19–amino acid motif corresponding to domain of unknown function 761 (DUF761) near its C terminus. The physical distance between CNL-04 and CNL-05 is 8.7 kb, which is longer than that between RGA4 and RGA5 (3.7 kb) (Fig. 1D). In view of the substantial differences between CNL-05 and RGA5, we decided to name this WRC17 allele Pias, CNL-04 as Pias-1, and CNL-05 as Pias-2 (SI Appendix, Fig. S5).
To isolate the AVR-Pias avirulence gene cognate of rice NLR Pias, we performed an association study of expressed genes encoding candidate effector proteins (see SI Appendix, Fig. S6 and Tables S4 and S5 for details). This analysis identified three genes (G9141, G9435, and G9532) as candidates of AVR-Pias. We selected M. oryzae isolate Ao92-06-2, which is compatible with HW-RIL7, transformed it with each of the candidate genes, and tested their interactions with HW-RIL7 (SI Appendix, Fig. S7). Transformation with one of the candidate genes, G9532, rendered Ao92-06-2 incompatible with HW-RIL7 (SI Appendix, Fig. S7A), suggesting that G9532 is AVR-Pias. To validate this result, we generated a knockout mutant of G9532 in the 2012-1 background, which became compatible with HW-RIL7 (Fig. 1E and SI Appendix, Fig. S8). These results indicate that G9532 is AVR-Pias, which is recognized by Pias. Both Pias-1 and Pias-2 are required for the recognition of AVR-Pias, as knockout of either NLR gene abrogated resistance (SI Appendix, Fig. S7B). AVR-Pias is a 91–amino acid protein with a secretion signal peptide (Fig. 1F). This protein contains a 12–amino acid Toxin18-like motif (a feature of proteins belonging to the conotoxin O superfamily) in its C terminus.
Various Host Domains Are Integrated into Pias/Pia Sensor NLRs.
Pias and Pia are allelic to each other, and both are composed of a pair of NLRs. The helper NLRs (Pias-1 and RGA4) are conserved, whereas the sensor NLRs (Pias-2 and RGA5) are divergent, with different integrated domains. To explore the diversity and evolution of Pias/Pia NLRs in the entire Oryza genus, we obtained the genomic DNA sequences of the Pias/Pia locus from 171 accessions representing 11 Oryza species as well as 4 non-Oryza species of Poaceae, including Setaria italica (foxtail millet), Panicum hallii (Hall’s panicgrass), Hordeum vulgare (barley), and Aegilops tauschii (Tausch’s goatgrass) (Dataset S2). To validate the gene structures, we used RGA4/RGA5 and Pias-1/Pias-2 genes as well as gene models supported by RNA-seq for 10 Oryza accessions as queries to infer the gene models of 167 Oryza accessions and 4 non-Oryza species using Exonerate software (www.ebi.ac.uk/~guy/exonerate) (Datasets S3 and S4). The details of the gene model prediction pipeline and the results of RNA-seq alignment to the Pias-2/RGA5 genes are given in SI Appendix, Fig. S9. Remarkably, Pias-2/RGA5 sensor proteins contain a wide variety of IDs, with up to nine different domains, including HMA, DUF761, DUF677, Zinc_ribbon_12, PKc_MAPKK (PKc_M), PKc_like, and WRKY, inserted into a similar position after the LRR domain (Fig. 2A). Around the junction of the ID-containing fragment and the LRR domain, we identified a conserved 145– to 146–amino acid motif (Fig. 2B), which spans the four LRRs (LRR-9 to LRR-12). This junction region partially overlaps but is different from the previously reported CID motif (SI Appendix, Fig. S10) (32). Pias-2 contained a 441–amino acid fragment with six LRRs and the DUF761 ID downstream of this junction sequence (SI Appendix, Fig. S5).
Fig. 2.
We studied the frequency of different IDs in the 11 Oryza species used to reconstruct a phylogenetic tree based on whole-genome sequences (Fig. 2C and Datasets S2 and S4). The IDs of the cultivated rice O. sativa (44 samples) are shared by the HMA and DUF761 domains with similar frequencies. O. rufipogon (13 samples), the wild progenitor of O. sativa, contains DUF677 and PKc_M in addition to HMA and DUF761. Oryza species belonging to the A-genome group (O. sativa, O. rufipogon, O. barthii, O. glumaepatula, and O. meridionalis) contain a higher proportion of the DUF761 ID, suggesting its importance in their defense. However, outside the A-genome species, the DUF761 ID is absent. Instead, PKc_M (O. brachyantha and O. punctata), DUF677 (O. officinalis), and HMA (O. australiensis) are dominant, indicating that the integration of the DUF761 domain–containing fragment into Pias-2/RGA5 likely occurred in the immediate ancestor of the A-genome species that diverged from other Oryza lineages over 2.3 million y ago (TimeTree; www.timetree.org/).
A comparison of the genome sequences around Pias/Pia sensor-NLR genes between O. punctata and O. sativa cultivar Nipponbare revealed an interesting conserved region (Fig. 2D and E). Both of their coding regions contain the conserved junction sequences, whereas the IDs downstream of this sequence are different: PKc_M for O. punctata W1582 and DUF761 for O. sativa Nipponbare. However, in the 30.8-kb region downstream of the O. sativa Nipponbare Pias-2 gene, two regions (blocks a and b) share high DNA sequence similarity with the junction and ID regions of the O. punctata W1582 Pias-2 homolog (Fig. 2D and E and SI Appendix, Fig. S11): Block a corresponds to the junction region to exon-6, and block b corresponds to exon-7 to -10 of the O. punctata W1582 Pias-2 homolog. Perhaps this conserved sequence downstream of the O. sativa Nipponbare Pias sensor is a footprint of the replacement of the ID from PKc_M to the DUF761-containing fragment via homologous recombination at the junction region (Fig. 2F). A survey of 51 A-genome Oryza accessions with the DUF761 ID revealed that the O. punctata W1582 junction–ID–like sequence is widely conserved in O. sativa, O. rufipogon, O. meridionalis, and an accession of O. barthii (accession W1702) but not in the majority of O. barthii and O. glumaepatula accessions (SI Appendix, Fig. S12 and Table S6). These results indicate that this possible recombination occurred only once and that the footprint was probably lost in the latter two species. We did not detect similar footprints in other sensor Pias/Pia NLRs with non-DUF761 IDs.
Contrasting Patterns of Evolution of Pias/Pia Sensors and Helpers.
To explore the evolutionary patterns of the Pias/Pia NLR locus, we reconstructed phylogenetic trees of Pias/Pia NLR pairs separately for the helper NLR (Pias-1 and RGA4) and the sensor NLR (Pias-2 and RGA5) using the amino acid sequences of the helper NLRs (full-length protein) and sensor NLRs (full-length protein except the ID domain) of 22 Oryza accessions and 4 accessions from other Poaceae genera. Note that Pias/Pia helper and sensor NLRs belong to distantly related NLR clades (32). Overall, the helper-NLR tree (Fig. 3A) is consistent with the species tree (Fig. 2C) with a few exceptions. Species in the AA-genome group (O. sativa subsp. japonica, O. sativa subsp. indica, O. rufipogon, O. barthii, O. glumaepatula, and O. meridionalis) clustered together. Only the placement of O. australiensis (accession W0008) is not congruent between the Pias-1/RGA4 helper-NLR tree and the species tree. On the other hand, the sensor-NLR tree is quite distinct from the helper-NLR tree and is not congruent to the species tree. Remarkably, the sensor NLRs show a higher level of divergence than the helper NLRs (Fig. 3A). The sensor phylogeny includes two major clades (C1 and C2) separated by deep branches. RGA5 of O. sativa cv. Sasanishiki (25) belongs to the C1 clade, whereas Pias-2 of O. sativa WRC17 belongs to the C2 clade. Four species of the AA-genome group (O. sativa subsp. japonica, O. sativa subsp. indica, O. rufipogon, O. barthii, and O. meridionalis) and O. punctata (BB genome) have sensor-NLR alleles from both the C1 and C2 clades, whereas O. australiensis (EE genome), O. granulata (GG genome), and O. brachyantha (FF genome) have C1 alleles, and O. glumaepatula (AA genome) and O. officinalis (CC genome) have C2 alleles. Therefore, phylogenetic analysis of the sensor NLR of the Pias/Pia locus pointed to a transspecies polymorphism, which is reminiscent of the major histocompatibility (MHC) gene polymorphism in vertebrates (47).
Fig. 3.
To determine whether the paired NLRs have accumulated mutations at different rates, we calculated the nucleotide diversity and Tajima’s D (48). The nucleotide diversities of the Pias-2/RGA5 sensor were markedly higher than those of the Pias-1/RGA4 helper for each of the CC, nucleotide binding site (NBS), and LRR(-LII) domains (Fig. 3B). Also, the three domains of the Pias-1/RGA4 helper and Pias-2/RGA5 sensor had contrasting negative and positive Tajima’s D, respectively (Fig. 3C), indicating purifying selection in Pias-1/RGA4 and balancing selection in Pias-2/RGA5, especially in the LRR domain. These results suggest that the genetically linked helper and sensor NLRs of the Pias/Pia locus in the genus Oryza have undergone contrasting modes of evolution.
To determine the types of nucleotide substitutions in the Pias-1/RGA4 helper and Pias-2/RGA5 sensor, we calculated dN (nonsynonymous mutations) and dS (synonymous mutations) (Fig. 3D). Consistent with higher nucleotide diversity and Tajima’s D, we noted overall higher dN and dS in Pias-2/RGA5, notably in the NBS and LRR domains (Fig. 3D). We also observed that the pairwise dN and dS of Pias-1/RGA4 were constrained by the low genetic divergence in many of the examined pairs. Nonetheless, dS was higher in Pias-1/RGA4 for the CC domain than in Pias-2/RGA5 (Fig. 3D and SI Appendix, Fig. S13). These results suggest that different evolutionary patterns affect the different domains of Pias-1/RGA4 and Pias-2/RGA5.
The Pias-1 Helper Functions with the RGA5 Sensor to Recognize AVR-Pia.
Pia comprises the helper-NLR RGA4 and the sensor-NLR RGA5. The expression of RGA4 triggers HR cell death in rice as well as N. benthamiana, and RGA5 suppresses RGA4-mediated cell death (49). Upon the binding of AVR-Pia to the HMA ID of RGA5, this suppression is released, and HR-like cell death is triggered (49). To address the functional conservation of helper/sensor NLRs of the Pias/Pia locus, we examined the functions of helper NLRs by transiently overexpressing five Pias/Pia helper NLRs in N. benthamiana using agroinfiltration (SI Appendix, Fig. S14). Only RGA4 supported strong cell death, and other alleles including Pias-1 caused weaker cell death, indicating that the role of RGA4 as a strong HR inducer is not typical among the tested helper NLRs (SI Appendix, Fig. S14). Next, we tested the effects of coexpressing Pias-1 and Pias-2, which surprisingly resulted in stronger cell death than that caused by Pias-1 expression alone (SI Appendix, Fig. S15). Coexpression of AVR-Pias with Pias-1 and Pias-2 did not alter the level of cell death (SI Appendix, Fig. S15). A coimmunoprecipitation analysis showed that Pias-1 and Pias-2 interact (SI Appendix, Fig. S16), and a yeast two-hybrid assay showed that the CC domains of Pias-1 and Pias-2 form homo- and heterodimers (SI Appendix, Fig. S16). These results suggest that Pias-1 and Pias-2 physically interact like RGA4/RGA5, but their mode of action is different from that of the RGA4/RGA5 pair.
We next examined the effects of exchanging various helper and sensor Pias/Pia NLRs by performing a transient expression assay in N. benthamiana leaves (Fig. 4A and SI Appendix, Fig. S17). Coexpression of the sensor RGA5 with the two heterologous helpers (Pias-1 and RGA4-Ogr from O. granulata) suppressed HR-like cell death, indicating that RGA5 can suppress the cell death induced by helper NLRs other than RGA4 (Fig. 4A and SI Appendix, Fig. S17). Furthermore, this RGA5-mediated suppression of the Pias-1–triggered HR was released via coexpression of AVR-Pia, which resulted in cell death (Fig. 4A and SI Appendix, Fig. S17). A similar result was obtained using the RGA4-Ogr helper (Fig. 4A and SI Appendix, Fig. S17). These results indicate that the sensor RGA5 properly functions with helpers other than RGA4 to recognize AVR-Pia and mount the HR in N. benthamiana.
Fig. 4.
Finally, we generated transgenic rice plants (HW-RIL7:35S-RGA5) expressing the RGA5 transgene driven by the 35S cauliflower mosaic virus (CaMV35S) promoter in the Pias (Pias-1 and Pias-2) background. We challenged these lines with two M. oryzae isolates: Ao92-62-2 harboring the AVR-Pia transgene and Ao92-62-2 harboring the AVR-Pias transgene (Fig. 4B and SI Appendix, Fig. S18). Remarkably, HW-RIL7:35S-RGA5 exhibited resistance against M. oryzae containing AVR-Pia (Fig. 4B). Although the HW-RIL7:35S-RGA5 lines consistently showed resistance against M. oryzae containing AVR-Pia, their resistance against M. oryzae containing AVR-Pias varied. We crossed an rga4 mutant in the Sasanishiki background [Sas1493 (25)] with HW-RIL7 and obtained F1-B plants. These plants, harboring intact RGA5 as well as Pias-1 and Pias-2, now recognized and triggered resistance against M. oryzae containing AVR-Pia but not AVR-Pias (Fig. 4B). Similar results were obtained for F1-A plants generated by crossing Sasanishiki (wild type; WT) with HW-RIL7 (Fig. 4B). These results suggest that Pia function is dominant over Pias function in terms of the recognition of AVR-Pia and AVR-Pias.
These results suggest that the Pias-1 helper functions together with RGA5 in rice to recognize and mount resistance against M. oryzae containing AVR-Pia, suggesting that the helper function has been conserved over the long history of Pias/Pia evolution.
Discussion
Our study investigated the evolution of a pair of genetically linked NLRs in the genus Oryza and provided experimental evidence that the two paired NLRs have evolved in dramatically contrasting fashions. This study points to the evolution of a modular architecture of paired NLRs. Division of roles between a conserved helper NLR for signaling and a divergent sensor NLR with a cassette-like receptor domain for pathogen sensing may have given plants the ability to efficiently fend off rapidly evolving microbe pathogens.
We identified and functionally characterized the rice R gene Pias. This gene encodes the paired NLRs Pias-1 helper and Pias-2 sensor, which recognize the M. oryzae effector AVR-Pias. Pias is allelic to the well-studied R gene Pia, encoding the NLRs RGA4 helper and RGA5 sensor (25), which recognizes the effectors AVR-Pia (50) and AVR1-CO39 (26). The allelic sensor-NLRs Pias-2 and RGA5 carry different domains at their C termini. In RGA5, the integrated domain HMA directly binds to and recognizes two M. oryzae effectors, AVR-Pia and AVR1-CO39 (26). We have not yet detected direct binding between the DUF761 ID of Pias-2 and AVR-Pias despite several attempts. Perhaps the recognition of AVR-Pias by Pias-2 requires other host components; indeed, the recognition of AVR-Pii by Pii-2 requires the rice protein OsExo70-F3, which binds to both AVR-Pii and Pii-2 (30, 39).
What is the origin of the integrated DUF761 domain of Pias-2? The rice sensor-NLRs RGA5 and Pik-1 contain HMA domains as IDs. The ID HMA shares high amino acid sequence similarity with rice small heavy metal–associated domain proteins [sHMAs (51, 52)]. We revealed that the M. oryzae effector AVR-Pik binds to and stabilizes sHMA proteins, likely to promote pathogen infection (51, 53). To identify the proteins that provide the DUF761 domain to the Pias-2 ID, we performed BLAST searches against the rice protein database using the short Pias-2 DUF761 sequence (19 amino acids) as a query (SI Appendix, Fig. S19). This identified 15 proteins with a similarity threshold of E <10. Most of the 15 proteins contained DUF761 at their C termini. Only a few functional studies of DUF761-containing proteins have been performed. A study in cotton (Gossypium hirsuta) showed that GhCFE1A, containing DUF761 and DUF4408 domains, binds to actin proteins and localizes to the endoplasmic reticulum (ER) upon overexpression in N. benthamiana (54). Knockdown of GhCFE1A did not cause any phenotypic changes, while its overexpression led to delayed cotton fiber cell elongation. These results suggest that GhCFE1A is a linker protein that mediates the formation of the ER network and actin cytoskeleton. Arabidopsis A70, encoding a DUF761-containing protein, is specifically induced in the incompatible interaction with Pseudomonas syringae, but not in the compatible interaction (55). Knockout of the Arabidopsis DUF761-containing protein gene DUF761-1 did not have any phenotypic effects, whereas overexpressing DUF761-1 altered plant morphology and resulted in a constitutive defense response, leading to enhanced resistance against P. syringae (56). These findings suggest that DUF761-containing proteins function in defense, presumably mediated by the actin–ER network. Previous studies of the NLRome of Arabidopsis (6) and NLR-ID of various members of the plant kingdom (8) showed that DUF761 is one of the most common domains integrated into NLRs as the ID. In Arabidopsis, DUF761 was integrated into TIR-NLR, and the NLRs with DUF761-ID are in almost all cases paired with helper NLRs (6). These findings suggest that the DUF761 domain is a major target of pathogen effectors and has been frequently integrated into NLRs. Future studies should investigate how AVR-Pias is recognized by Pias and how AVR-Pias interferes with host cellular processes by its possible interaction with DUF761-containing proteins.
Divergent Sensor NLRs in the Pias/Pia Lineage.
The Pias-2/RGA5 sensor-NLR lineage is extremely divergent among Oryza species, with up to six different ID motifs integrated at their C termini (Fig. 2). These hugely divergent IDs may mediate the detection of a diversity of effectors from the blast fungus and possibly other pathogens. We hypothesize that the diversity of the Pias-2/RGA5 lineage has been maintained by natural selection to maintain various IDs that recognize the invasion of pathogens by directly binding to effectors or guarding host factors that are modified by effectors.
Within Oryza species, there are two major clades, C1 and C2. Notably, alleles from both clades are maintained within the species O. sativa, O. rufipogon, O. barthii, O. meridionalis, and O. punctata. The observed transspecies allelic divergence and their roles in detecting pathogen molecules are similar to those of the MHC locus of vertebrates (47). Individuals with higher heterozygosity at the MHC locus might have higher fitness (overdominance) due to their ability to bind to a larger number of pathogen peptides (57, 58). Perhaps in the ancestral outcrossing Oryza species, heterozygous plants with a larger repertoire of NLRs with different IDs had a selective advantage against a multitude of pathogens. It is also possible that frequency-dependent selection helped maintain this polymorphism. When the frequency of a pathogen effector in a population increases, the frequency of an allele for a cognate sensor NLR will increase, resulting in a reduction in the frequency of pathogen alleles in the population. In turn, the frequency of another effector gene will increase in the pathogen, and the frequency of the cognate sensor NLR will increase. According to this Red Queen model, the allele frequencies of both effector and sensor NLRs oscillate and may be maintained for a long time by balancing selection (57, 59, 60). In summary, the highly divergent evolution of Pias/Pia sensor NLRs with variable IDs seems driven by the fitness gain obtained by an enhanced recognition capability of pathogen effectors.
Genetic Mechanism of ID Switching in Pias/Pia NLRs.
The Pias/Pia sensor NLRs contain various ID sequences at the similar position downstream of the LRR domain. This suggests the presence of a mechanism for “cassette”-like exchange of IDs between sensor NLRs. In the upstream region of the ID, we identified a highly conserved stretch of 145–amino acid sequences. A recent study reported a 43–amino acid CID motif conserved in the region between the LRR and ID of the MIC1 NLR clade by studying the NLRs of seven Poaceae species (32). The conserved region identified here encompasses the CID motif (SI Appendix, Fig. S10) but extends in the N-terminal direction by ∼100 amino acids including four LRRs. The authors hypothesized that the CID could serve as a recombination point of integration of genomic sequences. Despite the difference in the conserved motif, our data basically support the hypothesis that this region serves as the recombination point for the integration of endogenous sequences matching various protein domains. In support of this idea, the downstream sequence of a Pias sensor NLR with DUF761 contains a DNA sequence similar to the junction and ID regions of the O. punctata Pias sensor homolog. Perhaps the PKc_M ID of the original O. punctata Pias was replaced by the DUF761 ID–containing fragment, and this switch caused the translocation of PKc_M to the region downstream of the sensor-NLR gene (Fig. 2F). In view of the high sequence conservation between the O. punctata PKc_M ID and the downstream sequence, it is also possible that the downstream sequence was functional until the recent past, the Pias sensor contained the dual IDs DUF761 and PKc_M in the same molecule, or Pias switched between these two IDs, possibly via alternative splicing. Future studies should address the mechanism of the junction sequence-mediated recombination.
Function of Pias/Pia Paired NLRs.
Pia has been extensively studied (49) and serves as a paradigm for paired NLRs together with the paired NLRs RPS4/RRS1 (23). In these two cases, the helper NLR is regarded as a cell-death inducer and the sensor NLR as a suppressor that maintains the complex in an inactive state when the pathogen is absent. Once the sensor has been modified by direct binding (RGA5) or modification (RRS1) of the ID by pathogen effectors, the suppression of the helper is released and HR-like cell death occurs. However, Pias NLRs do not function according to this model. In the N. benthamiana assay, Pias-1 functioned as a weak cell-death inducer and Pias-2 did not suppress Pias-1–mediated cell death. Indeed, Pias-1 and Pias-2 together triggered stronger cell death than that caused by Pias-1 alone. A recent functional study of rice Pikp, another paired NLR, also showed that the helper NLR alone does not cause cell death in the N. benthamiana system and that the helper and sensor cooperate to trigger HR (42). Therefore, the hypothesized functional roles of the helper as a cell-death inducer and the sensor as a cell-death suppressor as well as a detector of effector molecules may not be as prevalent as assumed. The system of negative regulation of a cell-death inducer by a suppressor encoded by genetically separate factors carries tremendous risks given that a loss-of-function mutation in the suppressor gene kills the carrier cells and incurs a genetic load. Therefore, such an extreme negative regulation system is unlikely to be maintained over a long period of evolution. We predict that cooperative NLRs in pairs or networks are more prevalent (31). It is also possible that the Pias/Pia-paired NLR system is regulated by additional components in rice cells that are absent from N. benthamiana. It would be interesting to experimentally determine what proportions of paired NLRs function in negative regulation and in cooperation using N. benthamiana transient expression assays. Further studies are needed to decipher the full regulatory network of Pias/Pia-paired NLR-mediated resistance.
Helper NLRs in the Pias/Pia Lineage Are Functionally Conserved.
Phylogenetic reconstruction of the Pias/Pia NLR locus revealed that the helper-NLR Pias-1/RGA4 is conserved, whereas the sensor-NLR Pias-2/RGA5 is highly divergent (Fig. 3). An Arabidopsis NLRome study showed that some NLR pairs coevolved, with the phylogenetic trees of helper and sensor NLRs corresponding (6). Our findings for Pias/Pia NLRs do not align with these observations. A functional study in N. benthamiana showed that Pias-1–mediated cell death was suppressed by RGA5 and that Pias-1 together with RGA5 functions in the recognition of AVR-Pia, leading to cell death (Fig. 4A). Similarly, an RGA4 homolog of O. granulata that is phylogenetically distant from O. sativa also functions with RGA5 to recognize AVR-Pia (Fig. 4A). Moreover, an HW-RIL7 rice line harboring Pias as well as an RGA5 transgene recognized AVR-Pia (Fig. 4B). These results suggest that the function of the Pias-1/RGA4 helper lineage is conserved, which is in line with its conserved amino acid sequences. It is possible that the separation of the roles of NLRs between the conserved helper and divergent sensor allowed for higher flexibility of pathogen recognition compared with singleton NLRs. This flexibility would allow the plant to cope with the rapid evolution of pathogens, which exhibit larger population sizes and shorter generation times than the host plants. The functional understanding of the modular structure of paired NLRs revealed here provides a basis for engineering NLRs to detect various effectors and to confer resistance to crops against pathogens.
Materials and Methods
Rice Pathogenicity Assays.
Rice leaf blade punch inoculation was performed using the M. oryzae isolates. A conidial suspension (3 × 105 conidia per milliliter) was punch-inoculated onto a rice leaf 1 mo after seed sowing. The inoculated plants were placed in a dew chamber at 27 °C for 24 h in the dark and transferred to a growth chamber with a 16-h light/8-h dark photoperiod. Disease lesions were scanned 10 d postinoculation, and lesion size was measured using ImageJ software (61).
RNA-Seq of Rice and Barley Leaves Infected with M. oryzae 2012-1 Isolate.
Total RNA was extracted from rice- and barley-infected leaves using an SV Total RNA Isolation System (Promega). One microgram of total RNA was used to prepare each sequencing library with an RNA Sample Prep Kit v2 (Illumina). The two types of libraries, created from infected barley and rice leaves, were sequenced by paired-end (PE) and single-end (SE) sequencing using the NextSeq 500 platform.
DNA-Seq for the RaIDeN Pipeline.
Genomic DNA was extracted from WRC17, Hitomebore, and S-RIL leaves using a NucleoSpin Plant II Kit (Macherey Nagel). Libraries for PE short reads were constructed using a TruSeq DNA LT Sample Prep Kit (Illumina). The PE library of WRC17 and other libraries were sequenced on the Illumina MiSeq and HiSeq 4000 platforms.
Generation of Pi-W17-1 with Knocked-Down Expression of Candidate NLR Genes in Rice and qRT-PCR.
The eight types of gene knockdown (RNAi) constructs (pANDA-Pi-W17-1 candidate NLRs) were generated by PCR amplification of a specific fragment of each Pi-W17-1 candidate NLR gene from WRC17 WT complementary DNA (cDNA). The sequences were cloned into the Gateway vector pENTR/D-TOPO (Invitrogen) and transferred into recombination sites of the pANDA vector (62) using LR Clonase (Invitrogen). The resulting vectors with eight types of pANDA-Pi-W17-1 candidate NLR genes were introduced into Agrobacterium tumefaciens (strain EHA105) and used for A. tumefaciens–mediated transformation of HW-RIL7 following the method of Okuyama et al. (25). Total RNA was extracted from leaves using an SV Total RNA Isolation System (Promega) and used for qRT-PCR. cDNA was synthesized from 500 ng total RNA using a PrimeScript RT Reagent Kit (Takara Bio). qRT-PCR was performed using a StepOne Real-Time PCR Instrument (Applied Biosystems) with KAPA SYBR FAST PCR Master Mix (Kapa Biosystems). Melting curve analysis (from 60 to 95 °C) was included at the end of the cycles to ensure the consistency of the amplified products. The comparative Ct (ΔΔCt) method was used to calculate the expression of CNL-04 (CNL-05) relative to the rice ACTIN gene (LOC_Os03g50885) as an internal control. The data presented are the average and SDs from three experimental replications. The primers used to generate the RNAi construct and for qRT-PCR are listed in Dataset S5.
Generation of Rice Mutants of CNL-04 and CNL-05 by CRISPR-Cas9–Mediated Genome Editing.
Rice knockout mutants of CNL-04 and CNL-05 were generated using the CRISPR-Cas9 system developed by Mikami et al. (63). Sense and antisense target sequences were designed using the web-based service CRISPRdirect (crispr.dbcls.jp), annealed, and cloned into the pU6::ccdB::gRNA cloning vector following digestion with BbsI as the target sequence. The target sequence with the OsU6 promoter was cloned into the pZH::gYSA::MMCas9 vector following digestion with AscI and PacI. The resulting vectors (pZH::gYSA::MMCas9-CNL-04 and -CNL-05) were introduced into A. tumefaciens (strain EHA105) and used for A. tumefaciens–mediated transformation of HW-RIL7 following the method of Okuyama et al. (25). The resulting regenerated T0 plants were sequenced, and the mutation type was confirmed using primers listed in Dataset S5.
Genetic Complementation of the Candidate AVR-Pias.
Three candidate gene constructs (pCB1531-AVR-Pias candidate) were generated by PCR amplification of the coding sequences of the AVR-Pias candidate genes from cDNA prepared from M. oryzae 2012-1–infected barley leaves. The sequences were digested with XbaI and BamHI and cloned into pCB1531-pex22p-EGFP following the method of Yoshida et al. (50) that had been linearized by digestion with XbaI and BamHI. The resulting vectors were used to transform Ao92-06-2 (lacking AVR-Pias) following the method of Sweigard et al. (64). M. oryzae isolate 2012-1 mutated in G9532 was generated using the CRISPR-Cas9 system developed by Arazoe et al. (65). Sense and antisense target sequences were designed using the web-based service CRISPRdirect (crispr.dbcls.jp), annealed, and cloned into the pCRISPR-Cas-U6-1 cloning vector following the method of Arazoe et al. (65). To generate the targeting vector TV-G9532, the 5′ flanking region of G9532 was amplified and cloned into pCB1636 (64) containing a hygromycin resistance gene that had been linearized by inverse PCR using primers pCB1636iv2fwd and pCB1636iv2rev as described by Shimizu et al. (66) using In-Fusion cloning (Clontech). Subsequently, the 3′ flanking region was amplified and cloned into a plasmid containing the 5′ flanking region that had been linearized by inverse PCR using primers pCB1636iv1fwd and pCB1636iv1rev as described by Shimizu et al. (66) using In-Fusion cloning (Clontech). The resulting vectors were used to transform the 2012-1 isolate (containing AVR-Pias) following the method of Sweigard et al. (64). The primers used for construct generation are listed in Dataset S5.
Genome Sequences Used for the Study.
Genome sequences of 171 accessions of Poaceae plants were used, including 167 Oryza accessions as well as one accession each from A. tauschii, H. vulgare, P. hallii, and S. italica. These also included sequences of O. sativa WRC17 (this study) and O. sativa Sasanishiki (25). Genome sequences of 66 accessions were obtained from public databases: 52 accessions of O. sativa and O. rufipogon (67), O. sativa subsp. indica ShuHui498 (68), O. sativa subsp. japonica Nipponbare (69), O. rufipogon W1943 (National Center for Gene Research, Chinese Academy of Sciences, China), 5 accessions from O. barthii, O. glumaepatula, O. meridionalis, O. punctata, and O. brachyantha (The Oryza Map Alignment Project), O. officinalis W0002 (National Institute of Genetics, Japan), O. granulata W0067B (70), A. tauschii AL8/78 (71), H. vulgare Morex (72), P. hallii FIL2 (Department of Energy Joint Genome Institute), and S. italica Yugu18 (73) (see Dataset S2 for details). For 101 accessions of the wild Oryza species (O. barthii, O. glumaepatula, O. meridionalis, O. punctata, O. officinalis, O. brachyantha, and O. granulata), next-generation sequencing (NGS) reads (fastq format) of whole-genome sequences were retrieved from the NGS National Center for Biotechnology Information (NCBI) database (Dataset S2) and used for de novo assembly by MaSuRCA (74). For the two wild Oryza accessions, O. glumaepatula W2184 and O. punctata W1582, DNA sequencing was performed using Oxford Nanopore Technology (ONT) using genomic DNA extracted from their leaves. Sequencing was performed using the PromethION System with a FLO-PRO002 flow cell (ONT). Base calling of ONT reads was performed on FAST5 files using Guppy (ONT). Subsequently, low-quality reads were filtered out, and de novo assembly was performed using NECAT software (https://github.com/xiaochuanle/NECAT/). To further improve the accuracy of the assembly, Racon software (https://github.com/lbcb-sci/racon) was applied twice, and Medaka (https://github.com/nanoporetech/medaka) was used to correct misassembly. One round of consensus correction was performed using BWA (75) and HyPo (https://github.com/kensung-lab/hypo) on Illumina short reads for the accessions.
Gene Annotation and Detection of Pias/Pia Orthologs.
To infer the protein-coding regions of Pias-1/RGA4 and Pias-2/RGA5 genes, and to obtain information about their ID sequences, we used the pipeline shown in SI Appendix, Fig. S9. We retrieved Pias-1/RGA4 and Pias-2/RGA5 gene models from the genome sequences and RNA-seq data publicly available for seven Oryza samples (O. barthii IRGC105608, O. glumaepatula GEN1233_2, O. meridionalis W2112, O. punctata IRGC105690, O. australiensis W0008, O. brachyantha IRGC101232, and O. granulata W0067B). For O. rufipogon W1943 and O. officinalis W0002, genome sequences were publicly available, but we performed new RNA-seq analyses to improve gene prediction. For the sample O. punctata W1582, we performed genome sequencing and RNA-seq analyses. We also used the RGA4/RGA5 gene model of O. sativa cv. Sasanishiki (25) and the Pias-1/Pias-2 gene model of O. sativa cv. Keiboba (this study). The gene models of Pias-1/RGA4 and Pias-2/RGA5 of these 12 Oryza samples were used as queries to annotate IDs in the genome assembly of 167 Oryza samples using Exonerate (www.ebi.ac.uk/~guy/exonerate). However, 10 samples of O. glumaepatula and 6 samples of O. brachyantha did not match known domains. Therefore, we incorporated RNA-seq data for each sample for the two species, resulting in the annotation of the Zinc_ribbon_12 (O. glumaepatula) and HMA (O. brachyantha) IDs. In the next round, we used the gene models of 12 samples used in the first round of Exonerate as well as two new samples (O. glumaepatula W2184 and O. brachyantha W0655) as queries to infer IDs in the assembled genomes of 167 Oryza samples (Datasets S2, S3, and S4). This resulted in the identification of five known domains and one unknown domain in the Oryza ID sequences.
Amino Acid Sequences and Accession Numbers of Pias/Pia Orthologs.
Amino acid sequences of Pias/Pia homologs used in this study were retrieved from the Eukaryotic Genome Annotation of the NCBI database; accession numbers for Pias-1/RGA4 homologs and Pias-2/RGA5 homologs are as follows: XP_015617251.1 and XP_015617810.1 for O. sativa subsp. japonica Nipponbare, OsR498G1119642600.01 and OsR498G1119642700.01 for O. sativa subsp. indica ShuHui498, XP_020148260.1 and XP_020148256.1 for A. tauschii AL8/78, HORVU.MOREX.r2.4HG0288000.1 and HORVU.MOREX.r2.4HG0288010.1 for H. vulgare Morex, XP_025826635.1 and XP_025827327.1 for P. hallii FIL2, and XP_004979045.1 and XP_004979046.2 for S. italica Yugu18.
Phylogenetic Analysis of Pias-1/RGA4 and Pias-2/RGA5 Orthologs.
The protein sequences of the Pias-1/RGA4 and Pias-2/RGA5 orthologs were aligned using webPRANK (https://www.ebi.ac.uk/goldman-srv/webprank/) (76). We used the whole amino acid sequences of the Pias-1/RGA4 orthologs but only the partial amino acid sequences (CC-NB-LRR-junction domains) of the Pias-2/RGA5 orthologs due to the very low sequence similarity after the junction domains. A maximum-likelihood tree was constructed with IQ-TREE v2.0.3 (77) using 1,000 ultrafast bootstrap replicates (78). The models were automatically selected by ModelFinder (79) in IQ-TREE (77). ModelFinder (79) selected “JTT + G4” for Pias-1/RGA4 orthologs and “JTT + R2” for Pias-2/RGA5 orthologs. Finally, the phylogenetic trees were drawn with FigTree v1.2.3 (tree.bio.ed.ac.uk/software/figtree/).
Analysis of DNA Polymorphisms, dN and dS.
The coding DNA sequences (CDSs) of the Pias-1/RGA4 and Pias-2/RGA5 orthologs were aligned using the codon-based aligner MACSE v2.05 (80). We applied MACSE v2.05 not only to the entire CDS of Pias-1/RGA4 and CC-NBS-LRR-junction of Pias-2/RGA5 but also to each domain (CC, NBS, and LRR[-junction sequence]) using default parameters. We evaluated DNA polymorphisms of the Pias-1/RGA4 and Pias-2/RGA5 orthologs, calculating the nucleotide diversity (π) and Tajima’s D (48) using MEGA X v10.2.4 (81). Then, for each alignment, the maximum-likelihood trees were constructed using IQ-TREE v2.0.3 with 1,000 ultrafast bootstrap replicates (78) and ModelFinder (79). Based on these alignments and trees, we calculated the pairwise dN and dS using the YN00 program (82) in PAML v4.8 (83).
Expression Constructs Used in the Cell-Death Assay.
Expression constructs for five types of helper NLRs, Pias-1 (O. sativa subsp. indica, WRC17), RGA4 (O. sativa subsp. japonica, Sasanishiki), RGA4-Oru (O. rufipogon accession W1943), RGA4-Oau (O. australiensis accession W0008), and RGA4-Ogr (O. granulata accession W0067B) (pCambia1300S-“helper NLR”:HA), were generated by PCR amplification of the coding sequences from cDNA generated from leaf material and cloned into the binary vector pCambia1300S (www.cambia.org) that had been linearized by digestion with PstI and SpeI by In-Fusion cloning. Expression constructs for two types of sensor NLRs, Pias-2 (O. sativa subsp. indica, WRC17) and RGA5 (O. sativa subsp. japonica, Sasanishiki) (pCambia1300S-FLAG: “sensor NLR”), were generated by PCR amplification of the coding sequences from cDNA generated from leaf material and cloned into the binary vector pCambia1300S (www.cambia.org) that had been linearized by digestion with SalI and PstI by In-Fusion cloning. The resulting vectors were introduced into A. tumefaciens (strain GV3101). The primers used to generate the expression constructs are listed in Dataset S5.
Cell-Death Assay in N. benthamiana.
Transient expression of Pias/Pia allelic NLR and AVR-Pia was performed by infiltrating 4- to 5-wk-old N. benthamiana plants with A. tumefaciens carrying the expression vector. A. tumefaciens suspensions in infiltration buffer (10 mM 2-[N-morpholino]ethanesulfonic acid), 10 mM MgCl2, and 150 µM acetosyringone, pH 5.6) were adjusted to the densities shown in SI Appendix, Table S7. The autofluorescence value under ultraviolet (UV) light was scored using a luminescent image analyzer (ImageQuant LAS-4000, Cytiva).
Data Availability
The DNA-seq and RNA-seq data from this study are listed in Dataset S6 and have been deposited in the DNA Databank of Japan (BioProject accession nos. PRJDB9440, PRJDB12353, PRJDB12884, PRJDB12891, and PRJDB12902). The nucleotide sequences of Pias-1, Pias-2, and AVR-Pias have been deposited in the DNA Databank of Japan (Pias-1: LC672059; Pias-2: LC672060; and AVR-Pias: LC672061).
All study data are included in the article and/or supporting information.
Acknowledgments
This study was supported by Japan Society for the Promotion of Science KAKENHI 15H05779, 20H05681, and 16H06279 (PAGS) (to R.T.) and 21K14834 (to M.S.), Royal Society UK–Japan International Exchange Grants JPJSBP120215702 and IEC\R3\203081 (to R.T., S.K., and M.J.B.), the Biotechnology and Biological Sciences Research Council (UKRI-BBSRC, UK Grant BB/P012574), European Research Council (ERC BLASTOFF Project 743165), John Innes Foundation, and Gatsby Charitable Foundation. Computations were partially performed on the supercomputer at the Research Organization of Information and Systems, National Institute of Genetics. We also thank the National Institute of Genetics (National BioResource Project), Japan, and the Genebank at the National Agriculture and Food Research Organization, Japan, for providing the seeds of wild rice and WRC17, respectively.
Supporting Information
Appendix 01 (PDF)
- Download
- 28.55 MB
Dataset S01 (XLSX)
- Download
- 484.60 KB
Dataset S02 (XLSX)
- Download
- 22.71 KB
Dataset S03 (XLSX)
- Download
- 29.13 KB
Dataset S04 (XLSX)
- Download
- 25.78 KB
Dataset S05 (XLSX)
- Download
- 14.36 KB
Dataset S06 (XLSX)
- Download
- 12.15 KB
References
1
S. A. Hogenhout, R. A. Van der Hoorn, R. Terauchi, S. Kamoun, Emerging concepts in effector biology of plant-associated organisms. Mol. Plant Microbe Interact. 22, 115–122 (2009).
2
J. D. Jones, J. L. Dangl, The plant immune system. Nature 444, 323–329 (2006).
3
J. Kourelis, R. A. L. van der Hoorn, Defended to the nines: 25 years of resistance gene cloning identifies nine mechanisms for R protein function. Plant Cell 30, 285–299 (2018).
4
H. Adachi, L. Derevnina, S. Kamoun, NLR singletons, pairs, and networks: Evolution, assembly, and regulation of the intracellular immunoreceptor circuitry of plants. Curr. Opin. Plant Biol. 50, 121–131 (2019).
5
R. M. Clark et al., Common sequence polymorphisms shaping genetic diversity in Arabidopsis thaliana. Science 317, 338–342 (2007).
6
A. L. Van de Weyer et al., A species-wide inventory of NLR genes and alleles in Arabidopsis thaliana. Cell 178, 1260–1272.e14 (2019).
7
Q. Pan, J. Wendel, R. Fluhr, Divergent evolution of plant NBS-LRR resistance gene homologues in dicot and cereal genomes. J. Mol. Evol. 50, 203–213 (2000).
8
P. F. Sarris, V. Cevik, G. Dagdas, J. D. Jones, K. V. Krasileva, Comparative analysis of plant immune receptor architectures uncovers host proteins likely targeted by pathogens. BMC Biol. 14, 8 (2016).
9
M. Bernoux et al., Comparative analysis of the flax immune receptors L6 and L7 suggests an equilibrium-based switch activation model. Plant Cell 28, 146–159 (2016).
10
J. Wang et al., Ligand-triggered allosteric ADP release primes a plant NLR complex. Science 364, eaav5868 (2019).
11
S. Ma et al., Direct pathogen-induced assembly of an NLR immune receptor complex to form a holoenzyme. Science 370, eabe3069 (2020).
12
R. Martin et al., Structure of the activated ROQ1 resistosome directly recognizing the pathogen effector XopQ. Science 370, eabd9993 (2020).
13
H. Sharif et al., Structural mechanism for NEK7-licensed activation of NLRP3 inflammasome. Nature 570, 338–343 (2019).
14
J. L. Tenthorey et al., The structural basis of flagellin detection by NAIP5: A strategy to limit pathogen immune evasion. Science 358, 888–893 (2017).
15
J. Wang et al., Reconstitution and structure of a plant NLR resistosome conferring immunity. Science 364, eaav5870 (2019).
16
L. Zhang et al., Cryo-EM structure of the activated NAIP2-NLRC4 inflammasome reveals nucleated polymerization. Science 350, 404–409 (2015).
17
G. Bi et al., The ZAR1 resistosome is a calcium-permeable channel triggering plant immune signaling. Cell 184, 3528–3541.e12 (2021).
18
L. Wan et al., TIR domains of plant immune receptors are NAD+-cleaving enzymes that promote cell death. Science 365, 799–803 (2019).
19
S. Horsefield et al., NAD+ cleavage activity by animal and plant TIR domains in cell death pathways. Science 365, 793–799 (2019).
20
H. Adachi, S. Kamoun, A. Maqbool, A resistosome-activated ‘death switch.’ Nat. Plants 5, 457–458 (2019).
21
K. V. Krasileva, D. Dahlbeck, B. J. Staskawicz, Activation of an Arabidopsis resistance protein is specified by the in planta association of its leucine-rich repeat domain with the cognate oomycete effector. Plant Cell 22, 2444–2458 (2010).
22
T. Kroj, E. Chanclud, C. Michel-Romiti, X. Grand, J. B. Morel, Integration of decoy domains derived from protein targets of pathogen effectors into plant immune receptors is widespread. New Phytol. 210, 618–626 (2016).
23
S. J. Williams et al., Structural basis for assembly and function of a heterodimeric plant immune receptor. Science 344, 299–303 (2014).
24
C. Le Roux et al., A receptor pair with an integrated decoy converts pathogen disabling of transcription factors to immunity. Cell 161, 1074–1088 (2015).
25
Y. Okuyama et al., A multifaceted genomics approach allows the isolation of the rice Pia-blast resistance gene consisting of two adjacent NBS-LRR protein genes. Plant J. 66, 467–479 (2011).
26
S. Cesari et al., The rice resistance protein pair RGA4/RGA5 recognizes the Magnaporthe oryzae effectors AVR-Pia and AVR1-CO39 by direct binding. Plant Cell 25, 1463–1481 (2013).
27
I. Ashikawa et al., Two adjacent nucleotide-binding site-leucine-rich repeat class genes are required to confer Pikm-specific rice blast resistance. Genetics 180, 2267–2276 (2008).
28
A. Maqbool et al., Structural basis of pathogen recognition by an integrated HMA domain in a plant NLR immune receptor. eLife 4, e08709 (2015).
29
H. Takagi et al., MutMap-Gap: Whole-genome resequencing of mutant F2 progeny bulk combined with de novo assembly of gap regions identifies the rice blast resistance gene Pii. New Phytol. 200, 276–283 (2013).
30
K. Fujisaki et al., An unconventional NOI/RIN4 domain of a rice NLR protein binds host EXO70 protein to confer fungal immunity. bioRxiv [Preprint] (2017). https://www.biorxiv.org/content/10.1101/239400v1 (Accessed 24 December 2017).
31
C. H. Wu et al., NLR network mediates immunity to diverse plant pathogens. Proc. Natl. Acad. Sci. U.S.A. 114, 8113–8118 (2017).
32
P. C. Bailey et al., Dominant integration locus drives continuous diversification of plant immune receptors with exogenous domain fusions. Genome Biol. 19, 23 (2018).
33
E. Pennisi, Armed and dangerous. Science 327, 804–805 (2010).
34
J. Liu et al., Recent progress and understanding of the molecular mechanisms of the rice-Magnaporthe oryzae interaction. Mol. Plant Pathol. 11, 419–427 (2010).
35
Y. Wu et al., Combination patterns of major R genes determine the level of resistance to the M. oryzae in rice (Oryza sativa L.). PLoS One 10, e0126130 (2015).
36
M. J. Orbach, L. Farrall, J. A. Sweigard, F. G. Chumley, B. Valent, A telomeric avirulence gene determines efficacy for the rice blast resistance gene Pi-ta. Plant Cell 12, 2019–2032 (2000).
37
Y. Jia, S. A. McAdams, G. T. Bryan, H. P. Hershey, B. Valent, Direct interaction of resistance gene and avirulence gene products confers rice blast resistance. EMBO J. 19, 4004–4014 (2000).
38
H. Kanzaki et al., Arms race co-evolution of Magnaporthe oryzae AVR-Pik and rice Pik genes driven by their physical interactions. Plant J. 72, 894–907 (2012).
39
K. Fujisaki et al., Rice Exo70 interacts with a fungal effector, AVR-Pii, and is required for AVR-Pii-triggered immunity. Plant J. 83, 875–887 (2015).
40
C. H. Park et al., The E3 ligase APIP10 connects the effector AvrPiz-t to the NLR receptor Piz-t in rice. PLoS Pathog. 12, e1005529 (2016).
41
R. Wang et al., Immunity to rice blast disease by suppression of effector-triggered necrosis. Curr. Biol. 26, 2399–2411 (2016).
42
R. Zdrzałek, S. Kamoun, R. Terauchi, H. Saitoh, M. J. Banfield, The rice NLR pair Pikp-1/Pikp-2 initiates cell death through receptor cooperation rather than negative regulation. PLoS One 15, e0238616 (2020).
43
Y. Kojima, K. Ebana, S. Fukuoka, T. Nagamine, M. Kawase, Development of an RFLP-based rice diversity research set of germplasm. Breed. Sci. 55, 431–440 (2005).
44
P. Li et al., RGAugury: A pipeline for genome-wide prediction of resistance gene analogs (RGAs) in plants. BMC Genomics 17, 852 (2016).
45
H. Zhao et al., The rice blast resistance gene Ptr encodes an atypical protein required for broad-spectrum disease resistance. Nat. Commun. 9, 2039 (2018).
46
X. Meng et al., The broad-spectrum rice blast resistance (R) gene Pita2 encodes a novel R protein unique from Pita. Rice (N. Y.) 13, 19 (2020).
47
S. V. Edwards, P. W. Hedrick, Evolution and ecology of MHC molecules: From genomics to sexual selection. Trends Ecol. Evol. 13, 305–311 (1998).
48
F. Tajima, Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123, 585–595 (1989).
49
S. Césari et al., The NB-LRR proteins RGA4 and RGA5 interact functionally and physically to confer disease resistance. EMBO J. 33, 1941–1959 (2014).
50
K. Yoshida et al., Association genetics reveals three novel avirulence genes from the rice blast fungal pathogen Magnaporthe oryzae. Plant Cell 21, 1573–1591 (2009).
51
K. Oikawa et al., The blast pathogen effector AVR-Pik binds and stabilizes rice heavy metal-associated (HMA) proteins to co-opt their function in immunity. bioRxiv [Preprint] (2020). https://www.biorxiv.org/content/10.1101/2020.12.01.406389v1.full (Accessed 2 December 2020).
52
A. Białas et al., Two NLR immune receptors acquired high-affinity binding to a fungal effector through convergent evolution of their integrated domain. eLife 10, e66961 (2021).
53
J. H. R. Maidment et al., Multiple variants of the fungal effector AVR-Pik bind the HMA domain of the rice protein OsHIPP19, providing a foundation to engineer plant defense. J. Biol. Chem. 296, 100371 (2021).
54
F. Lv et al., GhCFE1A, a dynamic linker between the ER network and actin cytoskeleton, plays an important role in cotton fibre cell initiation and elongation. J. Exp. Bot. 66, 1877–1889 (2015).
55
W. Truman, M. H. Bennett, I. Kubigsteltig, C. Turnbull, M. Grant, Arabidopsis systemic immunity uses conserved defense signaling pathways and is mediated by jasmonates. Proc. Natl. Acad. Sci. U.S.A. 104, 1075–1080 (2007).
56
Y. Zhang, F. Zhang, X. Huang, Characterization of an Arabidopsis thaliana DUF761-containing protein with a potential role in development and defense responses. Theor. Exp. Plant Physiol. 31, 303–316 (2019).
57
N. Takahata, M. Nei, Allelic genealogy under overdominant and frequency-dependent selection and polymorphism of major histocompatibility complex loci. Genetics 124, 967–978 (1990).
58
S. B. Piertney, M. K. Oliver, The evolutionary ecology of the major histocompatibility complex. Heredity 96, 7–21 (2006).
59
M. E. Woolhouse, J. P. Webster, E. Domingo, B. Charlesworth, B. R. Levin, Biological and biomedical implications of the co-evolution of pathogens and their hosts. Nat. Genet. 32, 569–577 (2002).
60
R. Terauchi, K. Yoshida, Towards population genomics of effector-effector target interactions. New Phytol. 187, 929–939 (2010).
61
C. A. Schneider, W. S. Rasband, K. W. Eliceiri, NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671–675 (2012).
62
D. Miki, K. Shimamoto, Simple RNAi vectors for stable and transient suppression of gene function in rice. Plant Cell Physiol. 45, 490–495 (2004).
63
M. Mikami, S. Toki, M. Endo, Comparison of CRISPR/Cas9 expression constructs for efficient targeted mutagenesis in rice. Plant Mol. Biol. 88, 561–572 (2015).
64
J. Sweigard, F. Chumley, A. Carroll, L. Farrall, B. Valent, A series of vectors for fungal transformation. Fungal Genet. Newsl. 44, 52–55 (1997).
65
T. Arazoe et al., Tailor-made CRISPR/Cas system for highly efficient targeted gene replacement in the rice blast fungus. Biotechnol. Bioeng. 112, 2543–2549 (2015).
66
M. Shimizu et al., RNA-seq of in planta-expressed Magnaporthe oryzae genes identifies MoSVP as a highly expressed gene required for pathogenicity at the initial stage of infection. Mol. Plant Pathol. 20, 1682–1695 (2019).
67
Q. Zhao et al., Pan-genome analysis highlights the extent of genomic variation in cultivated and wild rice. Nat. Genet. 50, 278–284 (2018).
68
H. Du et al., Sequencing and de novo assembly of a near complete indica rice genome. Nat. Commun. 8, 15324 (2017).
69
International Rice Genome Sequencing Project, The map-based sequence of the rice genome. Nature 436, 793–800 (2005).
70
Z. Wu et al., De novo genome assembly of Oryza granulata reveals rapid genome expansion and adaptive evolution. Commun. Biol. 1, 84 (2018).
71
A. V. Zimin et al., Hybrid assembly of the large and highly repetitive genome of Aegilops tauschii, a progenitor of bread wheat, with the MaSuRCA mega-reads algorithm. Genome Res. 27, 787–792 (2017).
72
M. Mascher et al., A chromosome conformation capture ordered sequence of the barley genome. Nature 544, 427–433 (2017).
73
J. L. Bennetzen et al., Reference genome sequence of the model plant Setaria. Nat. Biotechnol. 30, 555–561 (2012).
74
A. V. Zimin et al., The MaSuRCA genome assembler. Bioinformatics 29, 2669–2677 (2013).
75
H. Li, R. Durbin, Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics 26, 589–595 (2010).
76
A. Löytynoja, N. Goldman, webPRANK: A phylogeny-aware multiple sequence aligner with interactive alignment browser. BMC Bioinformatics 11, 579 (2010).
77
B. Q. Minh et al., IQ-TREE 2: New models and efficient methods for phylogenetic inference in the genomic era. Mol. Biol. Evol. 37, 1530–1534 (2020).
78
D. T. Hoang, O. Chernomor, A. von Haeseler, B. Q. Minh, L. S. Vinh, UFBoot2: Improving the ultrafast bootstrap approximation. Mol. Biol. Evol. 35, 518–522 (2018).
79
S. Kalyaanamoorthy, B. Q. Minh, T. K. F. Wong, A. von Haeseler, L. S. Jermiin, ModelFinder: Fast model selection for accurate phylogenetic estimates. Nat. Methods 14, 587–589 (2017).
80
V. Ranwez, E. J. P. Douzery, C. Cambon, N. Chantret, F. Delsuc, MACSE v2: Toolkit for the alignment of coding sequences accounting for frameshifts and stop codons. Mol. Biol. Evol. 35, 2582–2584 (2018).
81
S. Kumar, G. Stecher, M. Li, C. Knyaz, K. Tamura, MEGA X: Molecular evolutionary genetics analysis across computing platforms. Mol. Biol. Evol. 35, 1547–1549 (2018).
82
Z. Yang, R. Nielsen, Estimating synonymous and nonsynonymous substitution rates under realistic evolutionary models. Mol. Biol. Evol. 17, 32–43 (2000).
83
Z. Yang, PAML: A program package for phylogenetic analysis by maximum likelihood. Comput. Appl. Biosci. 13, 555–556 (1997).
Information & Authors
Information
Published in
Classifications
Copyright
Copyright © 2022 the Author(s). Published by PNAS. This open access article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).
Data Availability
The DNA-seq and RNA-seq data from this study are listed in Dataset S6 and have been deposited in the DNA Databank of Japan (BioProject accession nos. PRJDB9440, PRJDB12353, PRJDB12884, PRJDB12891, and PRJDB12902). The nucleotide sequences of Pias-1, Pias-2, and AVR-Pias have been deposited in the DNA Databank of Japan (Pias-1: LC672059; Pias-2: LC672060; and AVR-Pias: LC672061).
All study data are included in the article and/or supporting information.
Submission history
Received: September 14, 2021
Accepted: April 19, 2022
Published online: June 30, 2022
Published in issue: July 5, 2022
Keywords
Acknowledgments
This study was supported by Japan Society for the Promotion of Science KAKENHI 15H05779, 20H05681, and 16H06279 (PAGS) (to R.T.) and 21K14834 (to M.S.), Royal Society UK–Japan International Exchange Grants JPJSBP120215702 and IEC\R3\203081 (to R.T., S.K., and M.J.B.), the Biotechnology and Biological Sciences Research Council (UKRI-BBSRC, UK Grant BB/P012574), European Research Council (ERC BLASTOFF Project 743165), John Innes Foundation, and Gatsby Charitable Foundation. Computations were partially performed on the supercomputer at the Research Organization of Information and Systems, National Institute of Genetics. We also thank the National Institute of Genetics (National BioResource Project), Japan, and the Genebank at the National Agriculture and Food Research Organization, Japan, for providing the seeds of wild rice and WRC17, respectively.
Notes
This article is a PNAS Direct Submission.
Authors
Competing Interests
Competing interest statement: S.K. receives funding from the industry on NLR biology.
Metrics & Citations
Metrics
Altmetrics
Citations
Cite this article
A genetically linked pair of NLR immune receptors shows contrasting patterns of evolution, Proc. Natl. Acad. Sci. U.S.A.
119 (27) e2116896119,
https://doi.org/10.1073/pnas.2116896119
(2022).
Copied!
Copying failed.
Export the article citation data by selecting a format from the list below and clicking Export.
Cited by
Loading...
View Options
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDFLogin options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to access the full text.