MoErs1 is a virulent cytoplasmic effector secreted by M. oryzae
In previous studies, we found that the Qc-SNARE protein MoSyn8 mediates intracellular trafficking in M. oryzae22. To examine how MoSyn8 affects effector secretion, total secreted proteins were extracted from 4-day cultures of the ∆Mosyn8 mutant and wild-type Guy11 strains grown in nitrogen starvation minimal medium (MMN) liquid culture, which mimics the early infection process of M. oryzae23, and separated using 2-DE. We performed comparative secretome analysis to have successfully identified MoErs1 (Effector Regulated by MoSyn8) (Extended Data Fig. 1 and Extended Data Table 1). MoErs1 was predicted to encode a protein of 214 amino acid residues with an N-terminal signal peptide (SP) (GenBank accession no. OK562582) (Extended Data Fig. 2). To confirm the secretion of MoErs1, the native promotor-driven MoERS1-FLAG and SLP1-FLAG, and the constructive promotor-driven GFP-FLAG fusion genes were generated and transformed into Guy11. Secreted proteins were collected from a cultured liquid complete medium (CM), MMN, and the M. oryzae-infected rice leaves at 48 h and analyzed by SDS-PAGE. Finally, we found that MoErs1 and Slp1 can be detected in cultures of MMN and infected rice leaves but not CM (Extended Data Fig. 3a), suggesting that MoErs1 might be secreted during infection.
The expression pattern of MoERS1 during various growth stages was examined using real-time quantitative PCR (qPCR). MoERS1 transcripts were significantly higher at mycelia, 24- and 36- hours post-inoculation (hpi) that had a different transcript pattern with well-known effector Bas4 and AvrPi9 (Extended Data Fig. 3b)24, 25. To test the role of MoErs1 in virulence, we created the ∆Moers1 knockout mutant using a recombinant exchange technique (Extended Data Fig. 3c). The ∆Moers1 mutant was moderately reduced in vegetative growth, but conidia formation, tube germ growth, and appressorium formation were the same as Guy11 and the complemented ∆Moers1/MoERS1 strain (Extended Data Table 2). Intriguingly, in conidial suspension spray assays, we observed a drastic reduction in disease severity as lesion area and fungal DNA were decreased by nearly 60% on leaves infected with the ∆Moers1 mutant compared with controls (Fig. 1a-c). Specifically, the typical disease lesions caused by Guy11 and the complemented strain produced abundant conidia (> 80%, n = 100), whereas the necrotic lesions caused by the ∆Moers1 mutant failed to produce any conidia (Fig. 1d, e). These results indicated that MoErs1 plays an important role in virulence despite some reduction in vegetative growth.
To examine the secretion of MoErs1 into the rice cell, a native MoERS1 promotor-driven MoERS1-RFP fusion gene was generated and transformed into the ∆Moers1 and ∆Mosyn8 mutants, respectively. Observation of ~ 100 infectious sites revealed that more than 70% of sites showed the accumulation of red fluorescence in biotrophic interface complexes (BICs in the ∆Moers1 mutant but not in the ∆Mosyn8 mutant (Fig. 1f). In addition, the native promotor-driven MoERS1-RFP with a nuclear localization signal (MoErs1-RFP-NLS) and Slp1-GFP fusion genes were generated and co-transformed into the ∆Moers1 mutant. Red fluorescence was found in BIC, and the nuclei of host cells invaded by the complemented strain, and green fluorescence was readily detected in the extra-invasive hyphal membrane (EIHM) which has no detectable red fluorescence (Fig. 1g). These observations indicated that MoErs1 is a secreted cytoplasmic effector protein required for virulence.
Moers1 Has A Role In Invasive Growth And Suppression Of Host Innate Immunity
To explore how MoErs1 contributes to virulence, we performed infection assays on rice sheaths. The results showed that invasive hyphae (IH) growth was significantly restricted at 24 hpi in the ∆Moers1 mutant (> 85%, n = 100), and the IH failed to expand to adjoining cells even at 48 hpi (> 80%, n = 100) (Extended Data Fig. 4a). This data suggested that MoErs1 is required for invasive growth and lesion formation.
To test whether MoErs1 inhibits the host immune response associated with the rapid production of reactive oxygen species (ROS), we used 3,3’-diaminobenzidine (DAB) staining to estimate ROS production in rice sheathes. ROS was rarely found in infection by Guy11 and the complemented strain but readily detected in cells infected with the ∆Moers1 mutant (Extended Data Fig. 4b, c). We also pretreated the rice sheathes with diphenyleneiodonium (DPI), an inhibitor of NADPH oxidases involved in ROS production26, and observed IH growth of ~ 100 appressorial penetration sites by rating the hyphal growth from level I to IV (Extended Data Fig. 4d). The results showed that the IH of the ∆Moers1 mutant could usually expand, similar to Guy11, at 24 and 48 hpi (Extended Data Fig. 4e). We also generated transgenic rice lines overexpressing signal peptide-less MoErs1 (MoERS1∆SP-OX) in the TP309 background. The MoERS1∆SP-OX rice lines were more susceptible to blast, and the ∆Moers1 mutant was as virulent as Guy11 with typical lesions (Extended Data Fig. 5). These data indicated that MoErs1 has a critical role in promoting IH growth and suppressing host immunity.
Plant pathogens often secrete effectors to interfere with host immunity during infection3. To test whether the deletion of MoERS1 results in a defect in effector secretion, the AVR-Pia and AVR-Piz-t genes fused with a C-terminal GFP were expressed in the Guy11 and ∆Moers1 strains. We found that both Guy11 and the ∆Moers1 mutant expressing Avr-Pia or Avr-Piz-t fail to produce any lesions on LTH-Pia (Pia R gene monogenetic rice line) or LTH-Piz-t (Piz-t R gene monogenetic rice line) rice lines (Extended Data Fig. 6a, b), indicating that MoErs1 does not interfere with the recognition of Pia/Avr-Pia and Piz-t/Avr-Piz-t.
We further observed the localization of Avr-Pia and Avr-Piz-t in Guy11 and the ∆Moers1 mutant during infection. GFP was normally detected in the BICs of Guy11 and the ∆Moers1 mutant (> 80% of 100 imaged infection sites) (Extended Data Fig. 6c, d). We then observed the localization of the apoplastic effector Slp127. GFP was readily detected in the EIHM of Guy11 and the ∆Moers1 mutant (> 85% imaged of 100 infection sites) (Extended Data Fig. 6e). These data suggested that host immunity suppressed by MoErs1 is independent of the secretion of other effector proteins.
Moers1 Molecular Structure
To understand the MoErs1 functional mechanism, we determined the crystal structure of MoErs1 at 2.5-Å resolution. MoErs1 was first overexpressed in Escherichia. coli BL21 (DE3) and purified. The structure of MoErs1 was resolved using the single-wavelength anomalous diffraction method28. The model was then refined with Rwork and Rfree values of 21.6% and 25.0% (Extended Data Table S3). The final model of MoErs1 contains one MoErs1 in the asymmetric unit (PDB: 7VS2). MoErs1 adopts a typical β-trefoil fold where strands 2, 3, 4, 11, 12, and 13 form a β -barrel covered by three two-stranded antiparallel β-sheets (Extended Data Fig. 7a). The disulfide bond between C42 on the short α-helix and C105 on the β-strand 5 plays a critical role in the overall rigidity of the MoErs1 structure (Extended Data Fig. 7a, b), which helps the short α-helix folds back and allows the short helix and β-strand 1 to lean on the surface of β-barrel. The β-trefoil fold is often found in Künitz-type protease inhibitors29. However, structural similarity search using Dali30 yielded the best match with the Künitz-type protease inhibitor and water-soluble chlorophyll protein (WSCP) at a root-mean-square Ca deviation (r.m.s.d.) of 3.4 Å (Extended Data Fig. 7c). These results indicated that MoErs1 might function as an inhibitor of plant-origin proteases.
Moers1 Interacts With Osrd21 On The Plasma Membrane
WSCP from Brassicaceae (PDB: 5HPZ) inhibits the activity of A. thaliana PLCP AtRD2131, 32. To test whether MoErs1 inhibits proteinases similar to Brassicaceae WSCP, we searched for AtRD21 homologs in Oryza sativa L. database and identified a gene locus (Loc_Os04g57440.1) that we named OsRD21.
RD21 is a multicellular organelle localized protein reported to be accumulated in the vacuole and ER bodies33 and also PM and apoplastic spaces11. To determine the localization of OsRD21, the GFP-tagged full-length OsRD21 was transiently expressed in N. benthamiana and the rice protoplast. The green fluorescence was focused on the cell periphery and endoplasmic reticulum (ER) naturally in N. benthamiana (Extended Data Fig. 8a, b). When plant cells are treated with a high concentration of salt leads to plasmolysis and fluorescence remains localized at the plasma membrane (PM). In addition, OsRD21-GFP colocalized with RFP-tagged Remorin (StREM1.3), a PM-localized protein34, 35, and RFP with a signal peptide and ER retention signal HDEL, when they were coexpressed in N. benthamiana, respectively (Extended Data Fig. 8a, b). However, in the rice protoplast, OsRD21-GFP colocalized with Remorin-RFP on PM without showing any ER localization (Extended Data Fig. 8c). We also showed that GFP-tagged MoErs1 without the signal peptide localizes to the cytoplasm of the rice protoplast (Extended Data Fig. 8d). All these tagged proteins can be normally expressed in planta (Extended Data Fig. 8e).
To test whether OsRD21 is secreted into the apoplast, the Flag-tagged GFP (a negative control), the apoplastic effector Slp1 (a positive control), and proOsRD21 were transiently expressed in N. benthamiana. Apoplastic and intracellular leaf extracts were separated, and immunoblots with the anti-Flag antibody. We found that OsRD21 accumulates in intracellular compartments but not the apoplast. We also confirmed this result in transgenic rice lines overexpression proOsRD21:FLAG (OsRD21-OX) driven by actin1 promoter, MoERS1∆SP-OX and OsAO4-OX (Fig. 2a)36. This data indicated that OsRD21 mainly localizes on the PM rather than the apoplast.
We then examined and verified whether MoErs1 interacts with OsRD21 by carrying out yeast two-hybrid (Y2H) and co-immunoprecipitation (co-IP) assays of proteins transiently co-expressed in N. benthamiana (Extended Data Fig. 9a, b). To investigate the location of MoErs1-OsRD21 interaction, we performed the bimolecular fluorescence complementation (BiFC) assay. MoErs1 tagged with N-terminal YFP (MoErs1-nYFP) and OsRD21 tagged with C-terminal YFP (OsRD21-cYFP) were co-expressed with Remorin-RFP in N. benthamiana and the rice protoplast. The result indicated that the interaction mainly occurs at the PM (Fig. 2b-e).
Moers1 Inhibits The Activity Of Rice Plcp Osrd21 To Promote Virulence
To identify the amino acids of MoErs1 required for binding to OsRD21, molecular modeling, and docking analysis were carried out. The OsRD21 3D model closely resembles to that of barley EP-B2 (Extended Data Fig. 10)37. ClusPro was used to predict that interactions occur between MoErs1 loop2 (L2, red), loop4 (L4, chocolate), loop8 (L8, magenta), and β-strand 11 (β11, blue) with OsRD21 (Fig. 3a)38. Specifically, L2 (Ser64, Glu67, Phe71, and Pro72) of MoErs1 is predicted to intrude into the active site region of OsRD21 containing Cys165 and His302, thereby blocking its proteolytic activity31. Moreover, Ser64 and Glu67 in L2, are predicted to form hydrogen bonds with Gln201 and Ser299 of OsRD21, respectively (Fig. 3a). We also predicted that Arg178 and Asp180 in β11, Arg95 in L4, and Gln160 in L8 form hydrogen bonds (yellow dotted line) or noncovalent binding forces (orange dotted line) with Gln201, Asn202, Asp235, and Arg295, of OsRD21, respectively (Fig. 3a). Together, these hydrogen bonds and noncovalent binding forces may stabilize the observed MoErs1-OsRD21 interaction.
To verify this modeling prediction, alanine substitution in each of the four regions was carried out, and Y2H and co-IP revealed mutations in L2 (positions 64, 67, 71, and 72) and β11 (positions 178 and 180) nearly abolished the MoErs1-OsRD21 interaction. In addition, L4 (position 95) and L8 (position 160) are also required for full binding activities (Extended Data Fig. 9a, b). Finally, a microscale thermophoresis (MST) assay and measurement of dissociation constants (Kd) revealed that MoErs1 binds more tightly to OsRD21 than any of its mutated variants (Extended Data Fig. 9c).
To investigate whether MoErs1 inhibits the activity of OsRD21 through binding, we transiently co-expressed MoErs1-GFP and its mutants with the ProRD21-Flag in N. benthamiana, respectively. PLCP activity assessments using the previously established method39 showed that MoErs1 strongly inhibits the activity of OsRD21 than MoErs1L2, MoErs1β11, and MoErs1All, while MoErs1L4 and MoErs1L8 have no inhibitory activities (Fig. 3b). Importantly, we found that L2 and β11 regions have a more prominent role in the virulence of M. oryzae, in contrast to L4 and L8, which is consistent with their respective inhibitory activities (Fig. 3c-e).
To determine whether the inhibition of OsRD21 occurs during M. oryzae infection in a MoErs1-dependent manner, we further generated OsRD21 gene knockout (OsRD21-KO) transgenic rice lines (Extended Data Fig. 11a). The OsRD21-OX plant was inoculated with Guy11, ∆Moers1, ∆Moers1 expressing MoERS1 mutants with the native promoter, and the complemented strain with the native promoter (∆Moers1/MoERS1) and the constitutive rp27 promoter (∆Moers1/MoERS1rp27). After 48 hpi, leaves were harvested for total protein extraction and purification. OsRD21 activity assessment showed that ∆Moers1, MoErs1L2, and MoErs1β11 mutants could hardly inhibit the activity of OsRD21, while Guy11, MoErs1L4, MoErs1L8, and ∆Moers1/MoERS1 all have moderate inhibitory activities. The ∆Moers1/MoERS1rp27 strain with high MoErs1 expression levels significantly inhibited the activity of OsRD21 (Extended Data Fig. 11b). These results are consistent in that inhibition of OsRD21 is MoErs1-dependent.
To assess the contribution of OsRD21-mediated host resistance against M. oryzae, the susceptible rice plant TP309, OsRD21-KO, and OsRD21-OX were inoculated with Guy11, ∆Moers1, ∆Moers1 expressing MoERS1 mutants, ∆Moers1/MoERS1, and ∆Moers1/MoERS1rp27, respectively. The results showed that OsRD21-OX lines confer enhanced resistance against these strains with fungal growth decreased by > 60%. Growth was moderately compromised for ∆Moers1/MoERS1rp27 strains (Extended Data Fig. 11c, d). Notably, the virulence of ∆Moers1, MoErs1L2, and MoErs1β11 was significantly rescued on OsRD21-KO when compared to TP309 (Extended Data Fig. 11c, d). Importantly, OsRD21 transgenic rice plants remain susceptible to Bipolaris oryzae and Xanthomonas oryzae infection (Extended Data Fig. 11e, f). These results suggested that MoErs1 can function as a PLCP inhibitor to suppress OsRD21-mediated host immunity. Interestingly, MoErs1 could target multiple PLCPs (Extended Data Fig. 12), suggesting its strong ability of immune suppression.
Structure-based Design Of Diphenyl Ether Ester Compounds Inhibiting Moers1 Function
To examine MoERS1 conservation, we performed single-nucleotide polymorphism (SNP) analysis on sequenced rice blast isolates available from the NCBI database (Extended Data Table S4). Additionally, a BLASTp search failed to reveal any homologs of MoErs1 from other fungi (E-value < e-30). This prompted us to explore whether MoErs1 could be a specific target for small molecule compounds. According to the docking model, interaction sites between MoErs1 and OsRD21 are mainly within a long and narrow surface region, which suggests that flexible molecules may facilitate the binding. We thus designed diphenyl ether ester chemical compounds because of their excellent biological compatibilities and drug activities40. In addition, there are several amino acid residues for hydrogen bonding or hydrophobic interaction in the simulated active pocket. We further optimized the skeleton design and added a hydroxyl group to the diphenyl ether ester structure, named FY21001, which exhibits good molecular flexibility by forming hydrogen bonds with Phe72 of MoErs1 and maintaining hydrophobic interactions with adjacent amino acid residues (Fig. 4a and Extended Data Table S5). To determine the specificity of FY21001 to MoErs1, we performed an MST assay that showed a stronger binding ability of FY21001 to MoErs1 than OsRD21 (Fig. 4b), AtWSCP, or OsWSCP (Extended Data Fig. 13a). When the rice protoplast expressing MoErs1-nYFP and OsRD21-cYFP was treated with 500 µM FY21001, the in vivo interaction was completely abolished (Fig. 4c). We then performed a co-IP assay in the rice protoplast that showed FY21001 reduces the binding affinity between MoErs1 and OsRD21 in a dose-dependent manner, and, at 500 µM, FY21001 could completely abolish the interaction (Fig. 4d). These results suggested that FY21001 competes with OsRD21 in binding to MoErs1.
FY21001 inhibits the PLCP inhibitor activity of MoErs1 in a dose-dependent manner (Fig. 4e). Moreover, alanine substitution mutations in L2 and β11 significantly reduced the interaction between FY21001 and MoErs1 (Extended Data Fig. 13b). We also obtained two derivatives of FY21001, FY21003, and FY21019 (Table S6), with strong binding to MoErs1 (Extended Data Fig. 14a). Further examination indicated that these derivatives significantly inhibit the function of MoErs1 by relieving the inhibition of OsRD21 protease activities (Extended Data Fig. 14b). These results demonstrated that diphenyl ether ester compounds are effective in inhibiting the function of MoErs1, and they possess the potential to disrupt the MoErs1-dependent virulence of M. oryzae.
Diphenyl Ether Ester Compounds Effectively Control Rice Blast
To examine whether these compounds can be explored as antifungal compounds to manage the rice blast, we applied FY21001 at a concentration of 500 µM in infection under laboratory condition. We found that FY21001 could significantly reduce the lesion area and hyphal growth on rice leaves (Fig. 5a). Further comparative infection assay using these three compounds indicated that FY21001 has the best preventive EC50 at a concentration of 231.07 µM, which is similar to tricyclazole (EC50 = 224.08µM) (Extended Data Table S7). There were some slight reductions in EC50 with FY21003 (258.9 µM) and FY21019 (246.69 µM) (Extended Data Table S7). To further test their applications, we carried out a preventive effect test using three different settings: (1) co-treatment with M. oryzae spores, (2) pre-treatment for 24 hours, and (3) post-treatment for 24 hours. The results showed that co- and pre-treatment has the best-controlling effect, followed by post-treatment (Fig. 5a-d). FY21001 also showed an effective preventive effect against the neck blast in natural rice blast nursery, similar to tricyclazole (Fig. 5e-g), a mainstream and high-efficiency fungicide for controlling rice blast; and the effect has a higher efficacy compared to cafenstrole and metazachlor (Extended Data Fig. 15)41. Moreover, FY21001 and tricyclazole-treated Nip (Nipponbare) rice led two more than twofold increase in grain yield compared with DMSO treated Nip control rice (Fig. 5h). There are six rice blast isolates (Jiangsu #1–6) separated from the nursery; all of them contain the MoERS1 gene and are without sequence polymorphism (Extended Data Table S4). Notably, FY21001 did not confer any resistance against B. oryzae and X. oryzae lacking the MoERS1 gene (Extended Data Fig. 16), and it does not appear to affect the development of M. oryzae (Extended Data Fig. 17). We further carried out infection assays on rice sheaths and found that FY21001 and its derivatives can significantly inhibit IH growth (Extended Data Fig. 18a). Finally, the DAB staining assay showed that the application of these compounds results in a reduced ability of M. oryzae to scavenge host ROS (Extended Data Fig. 18b). This effect is similar to that exhibited by the ∆Moers1 mutant.
To investigate whether FY21001 could induce rice immunity, we examined the transcriptional levels of different disease-resistance genes in the host, including PR1, PBZ1, AOS2, LOX1, and NADPH oxidases RBOHA and RBOHB. None of these genes showed any changes in expression in the presence of FY21001 (Extended Data Fig. 19a, b). To further determine whether FY21001 induces ROS burst in planta, the TP309, OsRD21-KO, and OsRD21-OX rice leaf disks were treated with FY21001, flg22, and DMSO, respectively. Interestingly, flg22 induced a higher ROS accumulation in the OsRD21-OX rice line than TP309 and OsRD21-KO. However, ROS levels were not affected in the presence of FY21001 or DMSO (Extended Data Fig. 19c), indicating that FY21001 cannot induce rice immune responses. These results suggested that FY21001 and its derivatives have a preventive protection role against M. oryzae infection.