OsCERK2/OsRLK10, a homolog of OsCERK1, has a potential role for chitin-triggered immunity and arbuscular mycorrhizal symbiosis in rice

Abstract

In rice, the lysin motif (LysM) receptor-like kinase OsCERK1, originally identified as the essential molecule for chitin-triggered immunity, plays a key role in arbuscular mycorrhizal (AM) symbiosis. As we previously reported, although AM colonization was largely repressed at 2 weeks after inoculation (WAI), arbuscules were observed at 5 WAI in oscerk1 mutant. Conversely, most mutant plants that defect the common symbiosis signaling pathway exhibited no arbuscule formation. Concerning the reason for this characteristic phenotype of oscerk1, we speculated that OsRLK10, which is a putative paralog of OsCERK1, may have a redundant function in AM symbiosis. The protein sequences of these two genes are highly conserved and it is estimated that the gene duplication occurred 150 million years ago. Here we demonstrated that OsCERK2/OsRLK10 induced AM colonization and chitin-triggered reactive oxygen species production in oscerk1 knockout mutant as similar to OsCERK1. The oscerk2 mutant showed a slight but significant reduction of AM colonization at 5 WAI, indicating the contribution of OsCERK2 for AM symbiosis. However, the oscerk2;oscerk1 double-knockout mutant produced arbuscules at 5 WAI as similar to the oscerk1 mutant, indicating that the redundancy of OsCERK1 and OsCERK2 did not explain the mycorrhizal colonization in oscerk1 at 5 WAI. These results indicated that OsCERK2 has a potential to regulate both chitin-triggered immunity and AM symbiosis and at least partially contributes to AM symbiosis in rice though the contribution of OsCERK2 appears to be weaker than that of OsCERK1.

Introduction

Plants are exposed to various microorganisms in nature. To prevent fatal outcomes resulting from pathogenic infections, plants have developed a sophisticated immune system. Plants can also establish symbiotic relationships with beneficial microorganisms. Arbuscular mycorrhizal (AM) fungi can establish symbiosis with most terrestrial plants (>80%) and can promote phosphate absorption in host plants (Wang and Qiu 2006). Although AM symbiosis is the most widespread symbiotic association, and estimated to have originated 400 million years ago (MYA), its signaling pathway remains elusive (Remy et al. 1994).

Previous studies using legumes revealed that a part of the signaling pathway of AM symbiosis is also required for nitrogen-fixing root nodule (RN) symbiosis. This pathway is known as the common symbiosis signaling pathway (CSSP or CSP), and is considered the core signaling pathway of symbiosis (Parniske 2008). In both AM and RN symbiosis, CSSP is activated by the perception of signal molecules known as Nod factors in RN symbiosis and mycorrhizal lipochitooligosaccharides (Myc-LCOs) or tetramer of chitin (GlcNAc)₄ in AM symbiosis by LysM-receptor-like kinases (LysM-RLKs) in host plants (Genre et al. 2013; Maillet et al. 2011). The commonality of pathways between RN symbiosis and AM symbiosis led to the generally-accepted idea that nitrogen-fixing RN symbiosis is a co-opted part of the genetic program of AM symbiosis (Parniske 2008).

Interestingly, these symbioses are thought to have a shared origin with immunity. Most terrestrial plants recognize a specific pattern of the microbe/pathogen and trigger a defense response, which is so called microbe/pathogen-associated molecular pattern (MAMP/PAMP)-triggered immunity (Albert et al. 2020). In rice, the LysM receptor-like kinase, OsCERK1 is involved in the response to fungal chitin and bacterial peptidoglycan and lipopolysaccharide (Desaki et al. 2018; Kouzai et al. 2014). Furthermore, the oscerk1 mutant plants demonstrated the severe delay of AM colonization and defects to the Ca²⁺ spiking response against (GlcNAc)₄ and the germinated spore exudates (Carotenuto et al. 2017; Miyata et al. 2014; Zhang et al. 2015). These facts indicate that OsCERK1 is involved in both the initiation process of AM symbiosis and MAMP-triggered immunity. In addition, this finding suggests that immunity and symbiosis have close evolutional relationships. This bifunctional role of the OsCERK1 ortholog receptor was reported in various plants: SlLYK12 in tomato and PanLYK3 in Parasponia can induce both chitin-triggered immunity and AM or RN symbiosis (Liao et al. 2018; Rutten et al. 2020). Although the switching mechanism between symbiosis and immunity control could be universal across terrestrial plants, it remains elusive. OsCERK1 should be the key factor in understanding this precise mechanism.

Apart from the multifunctionality of OsCERK1, the characteristic aspect of the oscerk1 phenotype in AM symbiosis is worth studying. Though the oscerk1 knockout mutant plants severely defect AM colonization, especially in the early stage at approximately 2 weeks after inoculation (WAI), AM fungi start to colonize in the later stage at approximately 5 WAI. This indicates that OsCERK1 is related to the initiation process of AM symbiosis but is not essential for AM colonization. In contrast, previous studies reported that various CSSP mutant plants including oscastor, ospollux, osccamk, and oscyclops did not display any arbuscules even in 6 WAI (Banba et al. 2008; Chen et al. 2007; Gutjahr et al. 2008; Nakagawa and Imaizumi-Anraku 2015; Parniske 2008; Yano et al. 2008). The phenotypes were not evaluated in the same condition, but there is a possible phenotype difference between OsCERK1 and the other CSSP mutants. The reason for this difference can be explained by the existence of other LysM receptors, which have a redundant function. In rice, there are 10 LysM-RLKs, among which OsRLK10 exhibited the highest similarity to OsCERK1 (Buendia et al. 2018; Shimizu et al. 2010). Therefore, a redundant function of OsRLK10 and OsCERK1 has been predicted but the function of OsRLK10 has not yet been elucidated.

Here we characterized OsRLK10, focusing on AM symbiosis and chitin-triggered immunity. The phenotype of OsRLK10ox;oscerk1, which is the plants overexpressed OsRLK10 in oscerk1, demonstrated that OsRLK10 has similar capabilities of AM symbiosis and chitin-triggered immunity, thus, we renamed OsRLK10 to OsCERK2. However, contrary to our expectation, both double knock-out mutant of oscerk2 and oscerk1 (oscerk2;oscerk1) and oscerk1 produced arbuscules in 5 WAI. In contrast, osccamk mutant plants did not exhibit any colonization at all in the same experimental conditions. The oscerk2 mutant displayed a slight reduction in AM colonization but the phenotype is much weaker than that of oscerk1 mutant plants, suggesting that OsCERK1 contributes AM symbiosis rather than OsCERK2.

Materials and methods

Plant materials and transformation

Oryza sativa cv. Nipponbare WT (Pii) and WT (+/+) were used as the wild type. Mutants oscerk1 #53 (Kouzai et al. 2014) and osccamk (Banba et al. 2008) were used in this study. The vectors for generating OsCERK2ox;oscerk1 and OsCERK2-DVox;oscerk1 were constructed using a Gateway pENTR/D-TOPO cloning kit (Invitrogen, http://www.invitrogen.com/) and Gateway LR clonase enzyme mix (Invitrogen). A destination vector, pSTARA R-5, was purchased from Kumiai Chemicals Industry Co. Ltd.

The transcriptional level of OsCERK2 in bispyribac resistant regenerated plants was analyzed by quantitative RT-PCR. Total RNA was prepared from each cell line using a Plant Total RNA Mini Kit (Favorgen Biotech, Taiwan) and subjected to cDNA synthesis using a QuantiTect reverse transcription kit (Qiagen). Quantitative RT-PCR was performed using TaqMan gene expression assay reagent using a 7500 Fast Real-Time PCR system (Applied Biosystems). Oryza sativa ubiquitin (OsUBQ) was used as an internal control to normalize the quantity of mRNA. All primer sets used in the study are listed in Supplementary Table S1.

Phylogenetic tree

Amino acid sequences of OsCERK1 and OsCERK2 homolog in O. brachyantha, O. punctata, O. barthii, O. glumipatula, O. longistaminata, O. rufipogon, O. meridionalis and O. nivara, Leersia perrieri, Brachypodium distachyon, Triticum aestivum, Solanum lycopersicum and, Arabidopsis thaliana were obtained from Ensembl Plants (http://plants.ensembl.org/index.html). Amino acid sequence alignment was performed using ClustalW Ver.2.1 in the DNA Data bank of Japan. Figure tree ver.1.4.0 was used to construct phylogenetic trees.

AM colonization assay

Rice seeds were sterilized with 70% ethanol and immersed in water for 5 days at 30°C. The germinated seeds were inoculated with Rhizophagus irregularis spores and incubated in the special soil mixture as described previously (Miyata et al. 2014) for 2 or 5 weeks. The roots of these plants were stained with trypan blue and the number of infection units per plant was counted. At 5 WAI, the root length colonization ratio was calculated as described by Gutjahr et al. (2008).

Chitin-induced ROS assay

Chitoheptose and chitooctaose were kindly supplied by Yaizu Suisankagaku Industrial Co. (http://www.yskf.jp/yskfen/index.html), and re-acetylated before use. Suspension culture of O. sativa L. cv. Nipponbare, oscerk1, OsCERK2ox;oscerk1 and OsCERK2-DV ox;oscerk1 was maintained using modified N6 medium, and sub-cultured as previously reported (Yamada et al. 1993).

The incubation process of rice callus is described previously (Hayafune et al. 2014). Cells harvested 5 days after transfer to the new medium were used in the experiments. After pre-incubation for 30 min, 1 nM of (GlcNAc)₈ or sterilized H₂O was treated and incubated for 30 min, 60 min, 90 min, 120 min, and 180 min respectively at 25°C. The ROS species generated in the reaction mixture was determined using the luminol-dependent chemiluminescence assay (Schwacke and Hager 1992). The experiment was performed in triplicate.

Expression of Halo-tagged OsCERK2 in N. benthamiana

For the construction of the expression vector for Halo tagged OsCERK2 (OsCERK2-Halo), HaloTag was inserted into the C-terminal region of OsCERK2 (ED–TM–JM) in pENTR/D-TOPO using the same methodology of OsCERK1-HaloTag in our previous report (Shinya et al. 2012). Briefly, the DNA fragment encoding the HaloTag was amplified from the pFN18K vector (Promega) with the primer set Entrylast-F and OsCERK2-948R for OsCERK2 (Supplementary Table S1). The PCR product was ligated into the OsCERK2(ED–TM–JM) in pENTR/D-TOPO, which was amplified using inverse PCR with the primer set OsCERK2-Halo infu-F and OsCERK2-Halo infu-R for OsCERK2 (Supplementary Table S1) and the corresponding full-length cDNA as a template, by using the In-Fusion Cloning Kit (TAKARA). PCR products encoding the Halo-Tag and OsCERK2(ED–TM–JM) sequences were treated with the Cloning Enhancer (TAKARA) before the infusion cloning reaction. The resulting OsCERK2 (ED–TM–JM)-HaloTag (OsCERK2-Halo) subcloned into pENTR/D-TOPO was first digested with EcoRV or NruI to disable the kanamycin resistance gene in OsCERK2-Halo and then transferred to a binary vector pEAQ-DEST1 (Sainsbury et al. 2009) using the LR reaction (Invitrogen). The expression vectors thus obtained were transformed into Agrobacterium tumefaciens LBA4404 using electroporation. To express the Halo-tagged OsCERK2 in Nicotiana benthamiana, the transformed A. tumefaciens was pressure-infiltrated into N. benthamiana leaves as reported previously (Sainsbury et al. 2009). The leaves were collected 6 days after infiltration.

Measurement of chitin-binding activity of OsCERK2 using affinity labeling with biotinylated-GN8 and colloidal chitin

Affinity labeling of GN8-Bio, the conjugate of biocytin hydrazide and (GlcNAc)₈, was performed as described previously (Shinya et al. 2010). Microsomal membrane fractions (MF) were prepared from the OsCERK2-Halo, OsCERK1-Halo, and CEBiP expressing N. benthamiana leaves or rice suspension-cultured cell according to our previous report (Shinya et al. 2012). MF (10 µg of CEBiP and OsCERK1-Halo, 30 µg of OsCERK2) was mixed with 0.4 µM of GN8-Bio with and without competing sugars (40 µM) and adjusted to 30 µl with a binding buffer. After incubation for 1 h on ice, 3 µl of 3% EGS (ethylene glycol bis[succinimidylsuccinate]) solution was added to the mixture and left to stand for 30 min. The reaction was stopped by adding of 1 M Tris, mixed with SDS-PAGE sample buffer, boiled for 10 min, and used for SDS-PAGE. After blotting onto the PVDF membrane, detection of biotinylated proteins was performed using a rabbit antibody against biotin (Rockland or BETHYL), Halo Tag (Promega), or CEBiP as a primary antibody and horseradish peroxidase-conjugated goat anti-rabbit IgG (MP Biomedicals) as a secondary antibody.

For the binding assay with colloidal chitin, microsomal membrane fraction was prepared from the OsCERK2-Halo-expressing leaves. To solubilize the membrane proteins, the microsomal membrane was suspended with phosphate-buffered saline (PBS) containing 0.5% Triton X-100, 1 mM phenylmethylsulfonylfluoride and 2 mM dithiothreitol, and kept overnight on a shaker at 450 r.p.m. at 4°C. The suspension was ultracentrifuged at 100,000×g for 30 min at 4°C. The solubilized protein fraction was mixed with colloidal chitin, kept on ice for 2 h, and centrifuged at 16,000×g at 4°C. The supernatant contained most of the proteins not bound to the colloidal chitin and was named the “flow-through” fraction. The precipitate was washed three times with PBS and then eluted with SDS-PAGE loading buffer, giving the “bound” fraction. Halo-tagged proteins in each fraction were detected using western blotting with an anti-HaloTag antibody (Promega) (Shinya et al. 2012). Colloidal chitin was prepared as described previously (Hon et al. 1995; Huynh et al. 1992).

Localization assay

A transient expression system with N. benthamiana was used for localization analysis. Subcloned OsCERK2 in pENTR/D-TOPO was inserted into the binary vector pGWB5. The expression vectors were used to transform Agrobacterium tumefaciens LBA4404 by electroporation. N. benthamiana leaves were then pressure-infiltrated with the transformed A. tumefaciens. The expressed GFP-protein was then localized using an FV1000-D confocal microscope (Olympus). GFP was excited using a 488 nm laser and fluorescence was detected at 500–600 nm.

Gene targeting

To produce the oscerk2 knockout mutant, the homologous recombination method was used. Vector construction and rice transformation for gene targeting were performed by following the methods in previous reports (Kouzai et al. 2014; Ozawa et al. 2012). Homologous recombination was confirmed using PCR using the 4 primer pairs presented in Supplementary Table S1. The regenerated plants were self-pollinated to obtain the homozygous oscerk2 line and crossed with oscerk1 to obtain the double knock-out mutant, oscerk2;oscerk1. We regenerated gene-targeted calli and obtained self-pollinated seeds (T1 generation). Genotyping was performed using PCR with the primer described in Kouzai et al. (2014) and in Supplementary Table S1.

Results

OsCERK2, paralog gene of OsCERK1 is genus Oryza specific

Full-length sequence of OsCERK2 is registered in the NCBI database as a hypothetical protein OsJ_29979, but is not found in the databases in the Rice Genome Annotation Project (RADB: http://rice.plantbiology.msu.edu) or The Rice Annotation Project (RAP: https://rapdb.dna.affrc.go.jp/index.html). The partial sequence of OsCERK2 is listed as LOC_Os09g33630 in RADB and as Os09t0511000-01 in RAP.

Based on these data, OsCERK2 (OsRLK10) is the closest homolog of OsCERK1 among the 10 LysM receptor-like kinases in rice and belongs to the same branch of OsCERK1 as described by Shimizu et al. (2010). OsCERK2 encodes a receptor-like kinase consisting of a signal peptide, an extracellular domain (ED) containing three LysM motifs, a transmembrane (TM) region, and an intracellular Ser/Thr kinase domain. The whole sequence of the OsCERK2 protein shares 62% amino acid sequence homology with OsCERK1. The sequence homology was higher in the kinase domain (73.5%) than in the ED (54.8%) (Figure 1A, Supplementary Figure S1). In particular, the position of LysM motifs and their sequences are highly conserved between OsCERK1 and OsCERK2, and the sequence similarities and identities in the LysM motifs are 82–100% and 42–60%, respectively. The high similarity of the genomic sequence of OsCERK2 and OsCERK1 (71%), together with strong conservation of the intron-exon structure (Supplementary Figure S1B), suggested that OsCERK2 is the paralog gene of OsCERK1.

Figure 1. Characterization of OsCERK2. (A) Predicted structure and similarity of OsCERK2 and OsCERK1. Signal peptide (SP), LysM motif, and transmembrane domain (TM) are described. Black dots display the internal kinase domain. The amino acid number predicted by SOSUI (http://harrier.nagahama-i-bio.ac.jp/sosui/sosuisignal/sosuisignal_submit.html) and counted at the end of signal peptides which is. (B) Phylogenic tree of OsCERK2 and OsCERK1 homologs in 8 wild rice species; Oryza. brachyantha, O. punctata, O. barthii, O. glumipatula, O. longistaminata, O. rufipogon, O. meridionalis, and O. nivara. Alongside the outgroup species, Leersia perrieri, Brachypodium distachyon, Triticum aestivum, Solanum lycopersicum, and Arabidopsis thaliana. The CERK1 clade is shown in blue and the CERK2 clade is shown in pink. The black triangle indicates OsCERK2 and the white triangle indicates OsCERK1. OsCERK1 in L. perrieri is shown with the red triangle. All the gene sequences are obtained from Ensembl Plants.

To understand the evolutional trajectory of OsCERK2, we constructed a phylogenic tree using the sequences of OsCERK2 and OsCERK1 homolog genes in 8 wild rice species; Oryza brachyantha, O. punctata, O. barthii, O. glumipatula, O. longistaminata, O. rufipogon, O. meridionalis and O. nivara. The following outgroup species were also used; Leersia perrieri, Brachypodium. distachyon, Triticum aestivum, Solanum lycopersicum, and Arabidopsis thaliana. Interestingly, the OsCERK2 clade and OsCERK1 clade were apparently separated, and the nine species belonging to the genus Oryza, including O. sativa, exhibited corresponding genes in both the OsCERK2 clade and OsCERK1 clade. However, L. perrieri, which is the nearest outgroup species of Oryza, and the other outgroup species, including B. distachyon, T. aestivum, S. lycopersicum, and A. thaliana only exhibited one gene which is classified into OsCERK1 clade (Figure 1B). This result suggested that the OsCERK2 clade is specific for Oryza specie.

OsCERK2 can complement oscerk1-KO mutation in AM colonization

To evaluate the possible function of OsCERK2 in AM symbiosis, we over-expressed OsCERK2 under the control of OsACT1 promoter in the pSTARA vector into the oscerk1 knockout mutant plants (OsCERK2ox;oscerk1). Expression levels of OsCERK2 in the roots of the F₂ generation plants, OsCERK2ox;oscerk1 #6, #9, and #12, were 4, 8 and 11 times higher than nontransformants (WT), respectively. The roots of these overexpression lines, oscerk1 and WT, were inoculated with R. irregularis and evaluated for the mycorrhizal phenotypes described previously (Miyata et al. 2014). Interestingly, arbuscular development was observed in all the lines of OsCERK2ox;oscerk1 at 2 WAI, although oscerk1 did not establish any AM colonization at this early stage (Figure 2A–C) as reported previously (Miyata et al. 2014). The number of infection units in all the lines of OsCERK2ox;oscerk1 was significantly higher than in those of oscerk1 (Figure 2E). Additionally, we examined the expression of marker genes of AM colonization in these OsCERK2ox;oscerk1 lines. Expression levels of AM1 and AM3 were slightly upregulated in OsCERK2ox;oscerk1 #6 and #12 at 2 WAI (Figure 2G, H). These results indicated that OsCERK2 could trigger AM symbiosis, though the colonization level was not the same as WT.

Figure 2. OsCERK2 had potential function in AM symbiosis. (A–D) Trypan blue staining of OsCERK2ox;oscerk1 #9 (A), oscerk1 (B), WT (C), and OsCERK2-DVox/oscerk1 #4 (D) at 2 WAI with R. irregularis. Scale bars=100 µm. ab, arbuscles; ih, internal hypha; eh, external hypha; hp, hyphopodia. (E, F) Number of infection units/plants in WT, oscerk1 and (E) OsCERK2ox;oscerk1 #6, #9, and #12 and (F) OsCERK2-DVox/oscerk1 #4 and #10. (G, H) The expression level of (G) AM1 and (H) AM3 in WT, oscerk1, OsCERK2ox;oscerk1 #6, #9, and #12 at 2 WAI. Error bars indicate the SD from 3 replicates for WT and oscerk1 with more than 5 replicates for both OsCERK2ox;oscerk1 and OsCERK2-DVox/oscerk1.

We also demonstrated the same experiment using the kinase-dead OsCERK2 (OsCERK2-DVox;oscerk1) which contained a single amino acid substitution, D to V, at position 443 (Supplementary Figure S1A). Roots of OsCERK2-DVox;oscerk1 #4 and #10 showed 4 and 3 times higher expression of OsCERK2 than the those of WT, respectively, and did not restore the defects of arbuscular development of the oscerk1 (Figure 2D, F). These results confirmed that the kinase activity of OsCERK2 is indispensable for AM colonization in OsCERK2ox;oscerk1.

OsCERK2 also has potential to mediate chitin-triggered immunity

It is suggested by our previous study that OsCERK1 is a major receptor-like-kinase for the generation of chitin-induced defense response, because its mutant of knock-out OsCERK1 (oscerk1) almost abolished the activity for chitin signaling in rice suspension-cultured cells (Kouzai et al. 2014). To evaluate the possible function of OsCERK2 for chitin signaling in rice, we analyzed the levels of chitin-induced ROS in the suspension-cultured cells derived from OsCERK2ox;oscerk1 (#7 and #8) and OsCERK2-DVox;oscerk1 (#1 and #2). Expression levels of OsCERK2 in OsCERK2ox;oscerk1 #7 and #8 were 15 and 14 times higher than WT and those of OsCERK2-DVox;oscerk1 #1 and #2 were 10 and 13 times higher, respectively. Both cultured cell lines of OsCERK2ox;oscerk1 #7 and #8 accumulated the production of ROS under treatment with N-acetylchitooctaose ((GlcNAc)₈) even though it is lower extent than that of WT (Figure 3A, B). Conversely, the cell lines of oscerk1 and kinase-inactive OsCERK2-DVox;oscerk1 markedly decreased the chitin-induced ROS generation (Figure 3C, D). These results indicated that OsCERK2 has the potential to initiate the chitin-triggered immunity.

Figure 3. OsCERK2 restored defection of chitin-triggered immunity in oscerk1. (A–D) ROS production in (A) OsCERK2ox;oscerk1 #7 (circle) and #8 (triangle), (B) WT (BL2), (C) oscerk1, and (D) OsCERK2-DVox;oscerk1 #1 (circle) and #2 (triangle). Suspension-cultured rice cells (40 mg) were incubated with 1 nM GN8 (filled circle or triangle) or sterile H₂O (empty circle or triangle with dotted line) for 0, 30, 60, 90, and 120 min.

To clarify whether the ED of OsCERK2 can bind to chitin oligosaccharides, we transiently expressed an OsCERK2-Halo, comprised of the OsCERK2 ED, TM, juxtamembrane (JM) region, and Halo-Tag at the C-terminus, in N. benthamiana. The OsCERK2-Halo solubilized from the microsomal membranes (MF) in the N. benthamiana leaves was used for affinity labeling with biotinylated (GlcNAc)₈ (GN8-Bio). The solubilized rice MF was found to contain CEBiP, which is known to bind to chitin. The GN8-Bio-tagged CEBiP band was detected using the anti-biotin antibody and by the anti-CEBiP antibody, and the biotinylated band was abolished by the addition of excess (GlcNAc)₈ before affinity labeling with GN8-Bio (Supplementary Figure S2A). Whereas OsCERK2-Halo and OsCERK1-Halo bands were found in solubilized N. benthamiana MF by using an anti-Halo antibody, no GN8-Bio tagged bands were detected by anti-biotin antibody in either Halo-tagged proteins (Supplementary Figure S2A).

We further analyzed the binding activity of OsCERK2-Halo by using colloidal chitin (Supplementary Figure S2B). We previously used similar ED–TM–JM–Halo tagged proteins of At/OsCERK1 (AtCERK1-Halo, OsCERK1-Halo) to evaluate their binding activity to colloidal chitin and demonstrated that the AtCERK1-Halo bound to chitin but OsCERK1-Halo did not (Shinya et al. 2012). The result of the binding assay with the solubilized MF protein indicated that OsCERK2-Halo did not bind to colloidal chitin, similar to OsCERK1 but different from AtCERK1 (Supplementary Figure S2B).

Subcellular localization of OsCERK2 was analyzed by detecting the transiently expressed GFP-tagged C-terminus of OsCERK2 (OsCERK2-GFP) in N. benthamiana leaves using confocal microscopy. The results indicated that OsCERK2-GFP was present in the plasma membrane (Supplementary Figure S2C).

Comparison of mycorrhizal phenotype in osccamk and oscerk1 at 5 WAI

As described in previously, AM colonization was strongly repressed in the oscerk1 mutants. However arbusculers were observed in oscerk1 mutant plants at 5 WAI, even though the frequency is lower than that of WT. On the other hands, any colonization in the cortex was not detected in the osccamk mutant at in 6 WAI (Banba et al. 2008; Gutjahr et al. 2008; Nakagawa and Imaizumi-Anraku 2015). Although the reported phenotypes of osccamk and oscerk1 appeared to be different, it is worth comparing their phenotypes under the same experimental conditions.

Thus, we performed an AM inoculation assay using oscerk1 and osccamk mutants and compared the AM phenotypes. Because the genetic background of osccamk and oscerk1;oscerk2 was different (the osccamk mutant plants were WT Nipponbare (+/+) background whereas oscerk1 and oscerk2 mutant plants were produced based on WT Nipponbare (Pii), we used two different WT control plants for this experiment. Though more than 86% of WT (Pii) and WT (+/+) plants initiate the AM symbiosis and produce multiple arbuscules at 2 WAI, the number of plants that produce arbuscules was extremely low or zero in oscerk1 and osccamk (Supplementary Table S2). However, at 5 WAI, all the tested oscerk1 mutant plants produced arbuscules, while the osccamk plant did not (Figure 4A, B, Supplementary Table S2). This result suggested that the phenotypes of oscerk1 and osccamk were remarkably different.

Figure 4. Mycorrhizal phenotype in oscerk2, oscerk2;oscerk1, oscerk1, and osccamk. (A–B) Trypan blue staining of (A) WT (Pii), oscerk1, oscerk2, and oscerk2;oscerk1 and (B) WT (+/+) and osccamk inoculated with R. irregularis and incubated for 2 weeks and 5 weeks. Scale bars=100 mm. ab, arbuscles; ih, internal hypha; eh, external hypha; hp, hyphopodia; v, vesicles. (C) The number of infection units/plants in WT, oscerk1, oscerk2, oscerk2;oscerk1, WT (+/+), and osccamk at 2 WAI. (D) Root length colonization ratio in WT, oscerk1, oscerk2, oscerk2;oscerk1, WT(+/+), and osccamk at 5 WAI.

Double knock-out oscerk2;oscerk1 mutant established AM symbiosis in later stage

Comparing the mycorrhizal phenotype in oscerk1 and osccamk suggested that the redundant factor works for AM symbiosis in the oscerk1 mutant. Because these experiments revealed that OsCERK2 could trigger AM symbiosis and chitin-triggered immunity, OsCERK2 could be the candidate receptor with a redundant role. Thus, we generated and obtained one line of oscerk2 mutant by homologous recombination (Supplementary Figure S3A, B). The large insertion of hygromycin phosphotransferase gene (HPT) broke the OsCERK2 gene, and the expression of OsCERK2 was completely diminished in the oscerk2 mutant, whereas the expression level of OsCERK1 was not affected (Supplementary Figure S3C). To examine the redundant role of OsCERK1 and OsCERK2 in AM symbiosis, we produced the double knockout mutant, oscerk2;oscerk1, by crossing the oscerk2 with previously developed oscerk1-117 mutant plants reported in Miyata et al. 2014.

AM colonization assay was performed using the WT (Pii), oscerk1, oscerk2, oscerk2;oscerk1, WT (+/+), and osccamk plants and evaluated the phenotypes at both the early (2 WAI) and later (5 WAI) stages. At 2 WAI, the numbers of infection units decreased largely in oscerk1 and oscerk1;oscerk2, compared to that of WT, while the AM colonization completely diminished in osccamk mutants (Figure 4A–C). The numbers of the infection units in oscerk2 plants and WT (Pii) were not statistically different (Figure 4C). This result indicated that OsCERK1 played a much more important role in AM symbiosis than OsCERK2.

Because of the difficulty of counting the number of infection units in fully colonized roots, we analyzed the root length colonization ratio in these mutants to evaluate the phenotype of AM colonization at 5 WAI as described by Banba et al. (2008) and Gutjahr et al. (2008). The root length colonization was less than 10% in oscerk1, and no vesicles developed under our experimental condition, whereas both WT (Pii) and WT (+/+) produced several vesicles obviously and its root colonization rate were approximately 20%. There were no arbuscules in osccamk at 5 WAI (Figure 4D). However, contrary to our expectation, the oscerk1;oscerk2 double knock-out mutant produced arbuscules at 5 WAI, and there was no significant difference of root length colonization between oscerk1 and oscerk2;oscerk1 (Figure 4C, D). These results indicated that the reason for the characteristic mycorrhizal phenotype in oscerk1 was not the redundancy of OsCERK2. On the other hand, a slight but significant reduction of AM colonization in oscerk2 was observed at 5 WAI. This result, along with phenotype of OsCERK2ox;oscerk1 suggested that OsCERK2 had a partial role in AM symbiosis, though its contribution was lower than that of OsCERK1 (Figure 4D).

Discussion

Similar to OsCERK1, OsCERK2/OsRLK10 exhibits two opposing biological activities, AM symbiosis and chitin-triggered defense signaling. The phylogenic analysis revealed that OsCERK2 and OsCERK1 had the highest sequence/structure similarity among ten LysM-RLKs in rice. These results suggested that these two molecules may be generated by gene duplication. Interestingly, although OsCERK1 grouping genes could be found in various plants, the OsCERK2 group was only conserved in the genus Oryza.

Investigation of the evolutional trajectory of these wild rice species (Wang et al. 2018) revealed O. brachyantha as the most ancient wild rice among the nine Oryza species we studied (Stein et al. 2018), and that it possesses genes in both the OsCERK2 and OsCERK1 clades. In contrast, L. perrieri, located in the nearest outgroup of the genus Oryza, possesses only one OsCERK1 type protein, and no evidence of OsCERK2 orthologs. These observations suggest that L. perrieri and O. brachyantha may have separated ∼15 MYA (Guo and Ge 2005; Wang et al. 2018), therefore the duplication event of OsCERK2 and OsCERK1 appears to have happened in the same era. Stein and colleagues have described that L. perrieri lineage have been experienced genome conversion event. This may be related to the duplication of OsCERK1 and OsCERK2 (Stein et al. 2018). They also demonstrated that many defense-related genes were appeared in the gene set that emerged within the Oryzeae including L. perrieri and Oryza species. This analysis implies that sophisticated immune system was strongly demanded in Oryza species during the course of evolution and this necessity lead the acquisition of further defense genes, including OsCERK2.

In this report, we showed that OsCERK1 and OsCERK2 have common functional features in the chitin-induced ROS burst and AM symbiosis. In addition to the functional similarity indicated from the results of OsCERK2ox;oscerk1, binding assay using GN8-Bio and colloidal chitin revealed that OsCERK2 did not bind chitin oligosaccharides directly as same as OsCERK1 which was shown not to bind chitin in the previous report (Shinya et al. 2012). Such a functional and biochemical similarities between OsCERK1 and OsCERK2 further supported the redundant function of these molecules in immune and symbiotic responses in rice.

In spite of the functional and biochemical similarities of OsCERK1 and OsCERK2, the phenotypes of oscerk2 and oscerk1 were different. The mutation of oscerk1 was sufficient to diminish chitin-triggered immunity and caused a severe delay in AM colonization, whereas the oscerk2 single mutant demonstrated a slight reduction in AM colonization at 5 WAI. These results indicated that OsCERK1 plays more important roles than OsCERK2 for the initiation of AM symbiosis. The reason for this phenotypic difference between oscerk1 and oscerk2 could be the functional difference of these receptors. Despite the high similarity of these proteins, the 3 amino acid motif YAR, which is reported as a required motif for symbiosis, was not perfectly conserved in OsCERK2 and converted to “AR” in OsCERK2 instead of “YAR” (Supplementary Figure S1A; Nakagawa et al. 2011). This difference between OsCERK2 and OsCERK1 could also explained by the expression level of these genes. Analysis of the microarray data (TENOR; Transcriptome ENcyclopedia Of Rice) suggested that the expression level of OsCERK1 (https://tenor.dna.affrc.go.jp/EPV/Os08t0538300-01/) was higher than that of the partial sequence of OsCERK2 (https://tenor.dna.affrc.go.jp/EPV/Os09t0511000-01/, https://tenor.dna.affrc.go.jp/EPV/Os09t0511000-02/) in the leaf or root at 10 or 14 days after sawing (Mizuno et al. 2010; Oono et al. 2011, 2014). Further studies on the functional differences between OsCERK1 and OsCERK2 could lead to a better understanding on the precise mechanism of signal transduction for chitin-triggered immunity and AM symbiosis in rice.

In addition to the difference between OsCERK2 and OsCERK1, we focused on the characteristic mycorrhizal phenotype in oscerk1 as compared to osccamk in this research. oscerk1 plants demonstrated a severe delay in AM colonization, but AM fungi colonized in the oscerk1 mutant at 5 WAI, and the colonization ratio was lower than in WT plants. However, there was no colonization in osccamk under the same experimental conditions. Based on this difference, we hypothesized that the redundant role of OsCERK2 and OsCERK1 could be the reason for the oscerk1 phenotype. However, unexpectedly, oscerk2;oscerk1 established AM symbiosis at 5 WAI, and its phenotype did not significantly differ compared to oscerk1. Thus, it is difficult to explain the phenotypic difference between oscerk1 and osccamk from the redundant function of OsCERK1 and OsCERK2. There are ten RLKs and 12 LysM proteins without a kinase domain in rice, and most of them are still under investigation. Thus, the possibility that another gene could work redundantly cannot be ignored.

Concerning to the mycorrhizal phenotype of oscerk1, it is also worth to mention the bifunctionality of OsCERK1. Because OsCERK1 controls both immunity and symbiosis, there is a possibility that oscerk1 mutants allow AM fungi to invade in a later stage mainly due to its weak immunity. During the initiation process of RN symbiosis, nod factors transiently activate defense-related genes which are later suppressed after the initiation of symbiotic process in lotus (Nakagawa et al. 2011). Similar to this, the defense response is probably suppressed when AM symbiosis is initiated. In case of oscerk1, the lack of defense response against various PAMPs including chitin might leads the AM colonization, even though the efficiency of AM colonization is lower than WT. However, oscerk1 lacks the initiation pathway of AM symbiosis as well and how this mutant starts the AM symbiosis still remains elusive.

In conclusion, OsCERK2 is an Oryza specific gene that has a potential function in chitin-triggered immunity and AM symbiosis initiation, similar to OsCERK1. Along with OsCERK1, OsCERK2 could serve as a novel bifunctional receptor for both immunity and symbiosis. However, the functional efficiency of OsCERK2, especially for AM symbiosis, appears to be lower than OsCERK1. Considering the fact that CERK2 genes in Oryza species have been conserved since 150 MYA, it cannot be excluded that OsCERK2 has unknown functions more than those discussed here. Further clarification of the function of OsCERK2 and the comparison with OsCERK1 could lead to a better understanding of the initiation process of chitin-triggered immunity and AM symbiosis.

Abbreviations

AM

arbuscular mycorrhizal

CSSP

common symbiosis signaling pathway

(GlcNAc)₈

N-acetylchitooctaose

GN8-Bio

biotinylated (GlcNAc)₈

LysM

lysin motif

RLK

receptor-like kinase

RN

root nodule