Magnaporthe oryzae Auxiliary Activity Protein MoAa91 Functions as Chitin-Binding Protein To Induce Appressorium Formation on Artificial Inductive Surfaces and Suppress Plant Immunity

The rice blast fungus Magnaporthe oryzae generates infection structure appressoria in response to surface cues largely due to functions of signaling molecules, including G-proteins, regulators of G-protein signaling (RGS), mitogen-activated protein (MAP) kinase pathways, cAMP signaling, and TOR signaling pathways. M. oryzae encodes eight RGS and RGS-like proteins (MoRgs1 to MoRgs8), and MoRgs1, MoRgs3, MoRgs4, and MoRgs7 were found to be particularly important in appressorium development. To explore the mechanisms by which these proteins regulate appressorium development, we have performed a comparative in planta transcriptomic study and identified an auxiliary activity family 9 protein (Aa9) homolog that we named MoAa91. We showed that MoAa91 was secreted from appressoria and that the recombinant MoAa91 could compete with a chitin elicitor-binding protein precursor (CEBiP) for chitin binding, thereby suppressing chitin-induced plant immunity. By identifying MoAa91 as a novel signaling molecule functioning in appressorium development and an effector in suppressing host immunity, our studies revealed a novel mechanism by which RGS and RGS-like proteins regulate pathogen-host interactions.


FIG 1
Expression dynamics of genes regulated by RGS proteins in M. oryzae. (A) Hierarchical cluster analysis was performed to compare gene expression levels in the Guy11, ΔMorgs1, ΔMorgs3, ΔMorgs4, and ΔMorgs7 strains at 0, 1, 4, and 24 hpi. In general, the heat map indicates the pairwise distance between samples. The color bar represents the distances between the samples, with yellow signifying a small distance between the samples, while lighter blue signifies a larger distance. Gene expression levels in the ΔMorgs4 mutant strain were different from those in the ΔMorgs1, ΔMorgs3, ΔMorgs7, and Guy11 strains at same time point postinfection. Gene expression levels at 24 hpi were different from those seen at the initial infected stages (0, 1, and 4 h). In general, the levels of gene expression of the Guy11, ΔMorgs1, ΔMorgs3, and ΔMorgs7 strains exhibited less variability (yellow). However, the level of gene expression of the ΔMorgs4 strain showed more variability (light blue). (B) Principal-component analysis (PCA) recapitulates the observation from the HCL analysis. The dotted circle at lower left indicates the level of gene expression of the ΔMorgs4 mutant, the dotted circle at lower right indicates the gene expression of the Guy11, ΔMorgs1, ΔMorgs3, and ΔMorgs7 strains at 24 hpi, and the upper part indicates expression of the Guy11, ΔMorgs1, ΔMorgs3, and ΔMorgs7 strains at the initial infected stages (0, 1, and 4 hpi).
Three cluster A genes (MGG_06494, MGG_07101, and MGG_09307) were randomly selected to verify the binding sequence with MoMsn2 using qRT-PCR and electrophoretic mobility shift assays (EMSAs) (Fig. S5). The combined results suggested that MoMsn2 indeed regulated the expression of the genes that possess the putative MoMsn2 binding motif within their promoter sequences. In addition, we verified that MoMsn2 showed binding to the promoter sequence of MoAA91 by the use of EMSAs (Fig. 3D to F) and also found that the expression of MoAA91 was upregulated in the ΔMomsn2 mutant (Fig. 3G). We further analyzed the expression of MoMSN2 and MoAA91 during infection in the wild-type Guy11 strain and the ΔMorgs mutants and found that the two genes had opposing transcription profiles (Fig. 3H), suggesting that MoAA91 is negatively regulated by MoMsn2.
MoAa91 is important for pathogenicity. To examine the role of MoAa91 in the virulence of M. oryzae, conidial suspensions were sprayed onto susceptible rice (CO39) seedlings and injected into rice leaf sheaths. Seven days after inoculation, the ΔMoaa91 mutant caused fewer and more-restricted lesions on the rice leaves ( Fig. 4A and B). The results showed that the activity of the DNA was retarded by the addition of MoMsn2 in the second lane, suggesting that MoMsn2 protein was able to bind with the DNA, while proteinase K was able to dissolve the MoMsn2 protein and release DNA, and the His protein was not able to bind with DNA. (E) Increasing amounts of MoMsn2 were incubated with 1,000-bp DNA of the promoter of MoAA91. The complexes were resolved by electrophoresis on a 1.5% agarose gel. More MoMsn2 (Continued on next page) According to the "lesion type" scoring assay (30), most of the lesions caused by the ΔMoaa91 mutant were of types 1 to 3, with very few being of type 4 and none of type 5 (Fig. 4C). The mean lesion density per unit area of the ΔMoaa91 mutant was significantly lower than that of Guy11 (Fig. 4D). The level of fungal growth in rice was also significantly lower in the ΔMoaa91 mutant (Fig. 4E). These results indicated that MoAa91 had a role in the full virulence of M. oryzae.
To understand the mechanism underlying the reduced virulence, the appressorium turgor was measured. We found that the collapse rates in the presence of 1, 2, 3, and 4 M glycerol were higher in the ΔMoaa91 mutants (Fig. 4F), indicating that MoAa91 had influenced appressorium turgor generation. We also performed penetration assays using detached rice sheaths to assess the role of MoAa91 during the expansion of invasive hyphae (IH) in plant cells, using previously reported methods and criteria (31,32). At 30 hpi, more than 40% of the IH of Guy11 and the complemented strains were type 4. In contrast, less than 12% of IH was type 4 in the ΔMoaa91 mutant, while there were more types 1 to 3 seen with the ΔMoaa91 mutant than with Guy11 and the complemented strains (Fig. 4G). These findings suggested that MoAa91 was important for penetration and invasive hyphal growth.
MoAa91 is involved in regulating the normal appressorium formation on artificial inductive surfaces. By observing appressorium formation on artificial inductive surfaces, we found that the ΔMoaa91 mutant had a delay in appressorium differentiation compared with the wild-type Guy11 strain and the complemented strain. The percentage of appressorium formation was significantly lower in the ΔMoaa91 mutant at 4 and 6 hpi, but it became less distinguishable after 8 hpi ( Fig. 5A and B). However, the germ tubes of the ΔMoaa91 mutant were longer, and the appressoria were also smaller and not fully developed until 24 hpi ( Fig. 5A and C). These results suggested that MoAa91 contributed to appressorium development on the artificial inductive surface.
Surprisingly, we found that the ΔMoaa91 mutant formed normal germ tubes and appressoria on onion epidermis, which was different from the results seen with an artificial inductive surface (Fig. 5D). Given that the artificial inductive surface lacked the signaling molecules from the rice leaf surface, such as the cutin monomer (1,16hexadecanediol) or wax (1-octacosanol), we speculated that MoAa91 played a role in the recognition of surface signals. To test this hypothesis, we examined appressorium formation of the ΔMoaa91 mutant on an artificial inductive surface coated with cutin and wax or in the presence of rice and onion epidermis extracts and found that the ΔMoaa91 mutant produced normal germ tubes and appressoria at 24 hpi under all of these conditions (Fig. 5E). These results suggested that the deletion of MoAA91 did not impair the ability of the fungus to sense plant surface signals during appressorium formation.
MoAa91 is secreted during appressorium development and is synergistic with cellulase. To explore the mechanism underlying the action of MoAa91 in appressorium formation and pathogenicity, we first analyzed its structure and found that MoAa91 contains a signal peptide, a Glyco_hydro_61 domain, and a C-terminal chitin-binding domain (Fig. 6A). The signal peptide was validated (Fig. S4B) based on a method previously described by Oh et al. (33). We then investigated whether MoAa91 was expressed and secreted during appressorium formation. MoAa91 was highly expressed spores/ml) of wild-type strain Guy11 and of the mutant and complemented strains were inoculated on rice (cv. CO39), and the rice was incubated for 7 days. Three independent experiments were performed. (B) Leaves from 3-week-old rice seedlings were injected with a conidial suspension (5 ϫ 10 4 spores/ml). Diseased leaves were photographed 7 days after inoculation. The arrows indicate the injection points. A total of 60 healthy rice seedlings were used for each strain in experiments performed three times. (C) Quantification of lesion types. Scoring was as follows: 0 ϭ no lesion, 1 ϭ pinhead-sized brown specks, 2 ϭ 1.5-mm-diameter brown spots, 3 ϭ 2-to-3-mm-diameter gray spots with brown margins, 4 ϭ many elliptical gray spots longer than 3 mm, 5 ϭ coalesced lesions infecting 50% or more of the leaf area. Lesions were photographed and measured or scored at 7 days postinoculation (dpi), and experiments were repeated three times with similar results. Numbers within an area of 1.8 cm 2 of 20 rice leaves infected by each strain were counted. Asterisks indicate significant differences (Duncan's new multiplerange tests, P Ͻ 0.01). (D) Bar chart of mean lesion density of seedlings infected with wild-type strain Guy11 and the ΔMoaa91 mutant per unit area. Mean lesion density was significantly reduced in the ΔMoaa91 mutant infections. Error bars represent the standard deviations, and asterisks represent significant differences (Duncan's new multiple-range tests, P Ͻ 0.01). (E) The severity of blast disease was evaluated by quantifying M. oryzae genomic 28S ribosomal DNA (rDNA) relative to rice genomic Rubq1 DNA (7 dpi). The mean values of three determinations with standard deviations are shown. Asterisks represent significant differences (Duncan's new multiple-range tests, P Ͻ 0.01). (F) Appressorium turgor was measured by an incipient cytolysis (cell collapse) assay in the presence of 1, 2, 3, and 4 M glycerol. The percentage of collapsed appressoria was recorded by observing at least 100 appressoria, and the experiment was repeated three times. Error bars represent standard deviations, and asterisks represent significant differences (Duncan's new multiple-range tests, P Ͻ 0.01). (G) Penetration assays in rice sheath. IH growth on rice cells was observed at 30 hpi, and 4 types of IH were quantified and statistically analyzed (type 1, no penetration; type 2, a single primary hypha; type 3, extended but limited to one cell; type 4, spread to adjacent cells). Error bars represent standard deviations. A total of 50 invasive cells were statistically analyzed, and the experiment was repeated three times. at 4 hpi, and the expression weakened gradually to become nearly nondetectable at 12 hpi on the artificial inductive surface (Fig. 6B). Similar results were observed on barley leaves and rice sheaths (Fig. S4C). However, the level of MoAA91 transcription during appressorium formation reached a higher value at 8 hpi, and the expression of MoAA91 increased from 4 to 8 hpi, while the green fluorescent protein (GFP) fluorescence weakened gradually (Fig. S4D). We postulated that MoAa91 was secreted during this period. Therefore, we examined the expression of MoAa91 in the extracellular fluid (EF) collected from the complemented strain during appressorium formation on an artificial inductive surface by Western blotting. We found that the level of MoAa91 increased in EF and decreased in appressoria from 4 to 12 hpi (Fig. 6C), consistent with the idea that MoAa91 is secreted from the appressoria.
We also assessed the MoAa91 enzyme activity and found that MoAa91 alone showed no hydrolase activity but showed synergistic activity when incubated with the cellulase (Fig. 6D). On the basis of a previous study (34), we predicted that residues 20, 102 (histidines), and 186 (tyrosine) formed the active sites of MoAa91 (Fig. S6) and we therefore constructed a point mutation of MoAa91, namely, MoAa91 DTB (H20A/H102A/ The mean appressorium diameter was recorded by measuring at least 100 appressoria for each strain, and the experiment was repeated three times. Error bars represent standard deviations, and the asterisk represents a significant difference (Duncan's new multiple-range tests, P Ͻ 0.01). (D) Conidia of the wild-type Guy11 and ΔMoaa91 mutant strains were inoculated onto onion epidermis in 20-l droplets followed by incubation for 4 or 6 h. (E) Appressorium development was observed in the wild-type Guy11 and ΔMoaa91 mutant strains in the presence of cutin (1,16hexadecanediol) and wax (1-octacosanol) followed by extraction of rice leaves and onion epidermis for 24 h, respectively. Bar ϭ 10 m.
T186A), and analyzed its enzyme activity. The result revealed that these residues were required for the synergistic activity of MoAa91 with the cellulase (Fig. 6D). But we determined that this synergistic activity is not necessary for the function of appressorium formation and virulence on the basis of the experiments described below.
MoAa91 acts as an inducer to regulate appressorium differentiation. MoAa91 is not involved in the recognition of physical or chemical stimuli during appressorium differentiation ( Fig. 5D and E). To explore how the deletion of MoAA91 could cause defects in appressorium differentiation on artificial inductive surfaces, appressorium formation was examined by inoculating ΔMoaa91 conidia mixed with EF on an artificial inductive surface. After 4 hpi, the conidia of the ΔMoaa91 mutant formed normal germ tubes and appressoria ( Fig. 7A and B), indicating that some components of the Guy11-EF rescued appressorium formation by the ΔMoaa91 mutant, whereas boiled Guy11-EF failed to suppress this defect ( Fig. 7A and B). Interestingly, adding heterologously expressed MoAa91 to the conidial suspension on an artificial inductive surface suppressed the appressorium formation defects of the ΔMoaa91 mutant. Moreover, increasing the concentration of Guy11-EF or MoAa91 protein enhanced the ratio of two appressoria formed from one conidium (Fig. 7C). Surprisingly, adding the nonactive MoAa91 protein (MoAa91 DTB ) also suppressed the defects of the mutant (Fig. 7A), indicating that the enzyme activity of MoAa91 is not required for appressorium differentiation. In addition, to further assess the role of the motifs, we divided MoAa91 into two parts, the N-terminal domain (i.e., the Glyco_hydro_61 domain; indicated as "N" in Fig. 6A) and the C-terminal domain (indicated as "C" in Fig. 6A), and obtained the ΔMoaa91/MoAA91 N -GFP and ΔMoaa91/MoAA91 C -GFP transformants. We found that the defect in the appressorium differentiation of these two types of transformants was similar to that of the ΔMoaa91 mutant, as well as to that seen after adding purified MoAa91 N or MoAa91 C proteins expressed in Escherichia coli to the conidial suspension ( Fig. 7A and B). On the basis of these results, we speculated that MoAa91 was secreted into extracellular space and that the two motifs of MoAa91 were essential for the functions of the appressorium inducer that regulated the appressorium formation on artificial inductive surfaces.
MoAa91 is involved in suppressing host defense. Reactive oxygen species (ROS) function as secondary signals that mediate plant defense (35)(36)(37)(38). As the ΔMoaa91 mutant showed restricted growth of infectious hyphae, we hypothesized that MoAa91 was involved in suppressing the host defense. To test this, host-derived ROS was detected by staining with 3,3=-diaminobenzidine (DAB) at 30 hpi of the rice cells infected with the ΔMoaa91 mutant; we found that 82% of the infected cells stained brown, compared with 12% and 15% of those infected with the wild-type and complemented strains, respectively. This finding indicated that the ΔMoaa91 mutant failed to suppress ROS surrounding the site of infection (Fig. 8A).
Further, we employed diphenyleneiodonium (DPI), an inhibitor of the activity of NADPH oxidases, which is necessary for ROS generation in plants (39)(40)(41)(42). Upon in Escherichia coli were added into the conidial suspension of the ΔMoaa91 mutant and suppressed the appressorium defects of mutants on inductive surfaces. However, the appressoria were abnormal after treatment with boiled Guy11-EF. The ΔMoaa91/MoAA91 N -GFP and ΔMoaa91/MoAA91 C -GFP transformants showed defects in appressorium differentiation similar to that shown by the ΔMoaa91 mutant, as well as to that seen after addition of heterologously expressed MoAa91 N or MoAa91 C proteins to the conidial suspension. Apart from the Guy11 control (first column), columns 5 to 11 represent proteins added into the conidial suspension of the ΔMoaa91 mutant. (B) Normal appressorium formation rates at different time points were calculated and statistically analyzed. (The appressorium formed by the ΔMoaa91 mutant had long germ tubes, while the appressorium formed by the wild-type strain had short germ tubes.) The percentage present at a given time was recorded by measuring at least 100 conidia for each treatment, and the experiment was repeated three times. Error bars represent standard deviations, and asterisks represent significant differences (Duncan's new multiple-range tests, P Ͻ 0.01). (C) Two appressoria were formed at the tip of one germ tube treated with 2ϫ or 3ϫ Guy11-EF and purified MoAa91. Formation rates of two appressoria were calculated and statistically analyzed. "2 x" or "3 x" refers to 2ϫ or 3ϫ Guy11-EF and purified MoAa91, respectively, amounts which can restore normal appressorium formation. The percentage at a given concentration was recorded by observing at least 100 conidia for each treatment, and the experiment was repeated three times. Error bars represent standard deviations, and asterisks represent significant differences (Duncan's new multiple-range test, P Ͻ 0.01). treatment with 0.5 M DPI, the infectious hyphae of the ΔMoaa91 mutant spread to the neighboring cells (Fig. 8A). Treatment with catalase (CAG; Sigma), which degrades H 2 O 2 surrounding the infection site, suppressed the infection defect of the ΔMoaa91 mutants ( Fig. 8A and B). These results demonstrated that inhibiting ROS generation or scavenging ROS suppressed the infection defects of the ΔMoaa91 mutants.
We then examined the transcript levels of pathogenesis-related (PR) genes, including PBZ1, LOX1, PR1A, and PAD4, which are involved in the jasmonic acid (JA) pathway or the salicylic acid (SA) pathway (26,42). The transcript levels of these PR genes were higher in rice leaves infected with the ΔMoaa91 mutant than in those infected with the wild-type Guy11 strain (Fig. 8C). On the basis of these results, we concluded that restriction of the ΔMoaa91 mutant in rice cells might be due to host-derived ROS accumulation at the infection site and the subsequent activation of a strong host defense response.
MoAa91 is a competitive inhibitor of rice receptor CEBiP that suppresses chitin-triggered plant immune responses. Given that MoAa91 is a secreted protein and that deletion of the MoAA91 gene leads to the accumulation of ROS in infected cells, we postulated that MoAa91 acted as an effector once secreted into the host cell or the apoplastic space to suppress plant immunity. Therefore, we added the respective recombinant MoAa91 and MoAa91 DTB proteins to the conidial suspension of the ΔMoaa91 mutant and sprayed the suspension onto rice seedlings. The disease symptoms were severe, with more type 4 and 5 lesions produced than were seen with the ΔMoaa91 mutant alone (Fig. 9A and B). We then examined the ROS surrounding the infected site and found that the percentages of cells stained by DAB in the ΔMoaa91 mutant expressing MoAa91 and MoAa91 DTB were similar to those seen with the wild-type strain (Fig. 9C and D). These results suggested that heterologously expressed MoAa91 or MoAa91 DTB was able to suppress the defense response of rice and facilitate the growth of the invasive hyphae.
Furthermore, we dissected the function of MoAa91 by separating it into the N-terminal glycoside hydrolase 61 domain and the C-terminal chitin-binding domain. We found that the ΔMoaa91/MoAA91 C -GFP transformants suppressed the defect in pathogenicity of the ΔMoaa91 mutant (Fig. 9A). This result was in agreement with the finding that adding purified MoAa91 C proteins to the conidial suspension rescued the defects of ΔMoaa91 mutant (Fig. 9C and D). On the basis of these results, it is suggested that the C-terminal domain was important in the pathogenicity of MoAa91.
As MoAa91 C contains the chitin-binding domain, we hypothesized that MoAa91 was capable of binding specific polysaccharides, including chitin. After incubation of the purified MoAa91 protein with insoluble chitin (chitin beads and crab shell chitin), xylan polysaccharides, and cellulose (Avicel), we observed that MoAa91 coprecipitated with insoluble chitin at higher levels than with polysaccharides (Fig. 10A). As previous studies had demonstrated that rice PRR CEBiP directly bound chitin oligosaccharides, leading to subsequent strong activation of the plant defense response (43), we purified CEBiP:His with insoluble chitins (Fig. 10A). Therefore, we hypothesized that MoAa91 might be involved in disrupting chitin-induced sensing mediated by CEBiP between the fungus and the rice host. To test whether MoAa91 competes with CEBiP for chitin binding, a competition assay was performed by incubation of the CEBiP:His protein (10 g) with chitin beads (100 l) followed by the addition of MoAa91 (Fig. 10B). The results showed that the level of CEBiP-bonded chitin became significantly reduced as the concentration of MoAa91 increased, suggesting that MoAa91 indeed competes with CEBiP in chitin binding.

DISCUSSION
RNA-Seq analysis revealed that MoRgs1, MoRgs3, MoRgs4, and MoRgs7 regulated the expression of various genes during the early infection stages and that the regulation levels appeared to be distinct and to vary with the different RGS proteins. Of note, there were significant differences in the genes regulated by MoRgs4. On analyzing the transcription pattern of 215 DESGs, 19 different expression patterns were found, including patterns corresponding to cluster A, cluster B, and cluster C. Cluster A contains five genes coregulated by the aforementioned four MoRgs, including MoAa91, which was shown to be a protein that is secreted during appressorium formation. While the receptor(s) specific to MoAa91 is not known, we did find that MoAa91 functioned as an inducer of appressorium formation on artificial inductive surfaces. Despite the finding that disruption of MoAA91 did not impair the ability of M. oryzae to sense the plant surface cue, evidence was accumulated to demonstrate that MoAa91 is involved in suppressing the host-derived defense response to promote infection.
We found that most of the DESGs showed one of the three expression patterns, i.e., the cluster A, cluster B, or cluster C expression pattern. Cluster B showed the expression pattern of pathogenicity-related genes, and cluster C showed the expression pattern of effector genes, including BAS1, BAS3, BAS4, and AvrPiz-t (26). From these two clusters, several novel pathogenicity-related genes and effector genes were identified, such as IUG18, IUG6, and IUG9 (26). Only cluster A, which showed an "M"-like expression pattern, has not been reported previously. Here, we identified five genes that were coregulated by four MoRgs proteins with the cluster A pattern. Other genes with a similar expression pattern, such as those encoding MoCbp1 and MoCsr1, are also involved in appressorium development (27,28). Thus, the cluster A expression pattern appears to be important for identifying novel genes and effector genes related to appressorium formation. Further study showed that various genes in the cluster A were regulated by transcription factor MoMsn2, suggesting that MoMsn2 may be the key regulator of MoRgs-mediated signaling during appressorium formation in M. oryzae.
In M. oryzae, conidia germinate to form germ tubes that further differentiate into appressoria through sensing surface hardness, hydrophobicity, and additional physical or chemical cues (44). Multiple putative sensors that respond to physical and chemical cues on rice leaves have been reported previously (45). Pth11 is involved in recognizing surface cues on rice leaves and is required for appressorium formation (8). The surface mucin protein MoMsb2 and extracellular chitin-binding protein MoCbp1 are important for appressorium formation on artificial inductive surfaces (27). MoMsb2 is also critical for sensing hydrophobicity and cutin monomers on the rice surface (3). Studies also identified the conserved fungal-specific extracellular membranespanning (CFEM) GPCR WISH that was required for appressorium differentiation on an inductive surface or rice leaf. WISH has a critical role in the recognition of hydrophobicity and appressorium morphogenesis (46). In the current study, MoAa91 did not seem to have a role in sensing plant surface signals as the ability of the ΔMoaa91 mutant to form appressoria on leaf surfaces was not affected, even though it was delayed on artificial surfaces. It was reported previously that the chitin-deacetylase activity of MoCbp1 plays a vital role in the induction phase of appressorium formation (47). MoAa91 showed synergistic activity with cellulase, but such synergy remained minor, similarly the levels reported in a previous study of other Aa9 protein homologs (34). In addition, both MoAa91 and MoAa91 DTB were able to overcome the defect in appressorium formation on artificial surfaces. We therefore speculated that M. oryzae perceived the structure of MoAa91 protein, rather than the enzymatic activity, in regulating appressorium development. On plant surfaces, such as rice leaves, despite MoAa91 still being secreted, the fungus may preferentially perceive stimuli from cutin and waxes.
Pathogens secrete various effectors that are hypothesized to facilitate effective host infection (48). Some effectors are secreted by appressoria even before the penetration of the host (49). A Colletotrichum higginsianum effector candidate (ChEC) was expressed in appressoria only before penetration and might have been secreted through a pore in melanized appressoria (50)(51)(52). The unique M. oryzae AVR gene ACE1 encodes avirulence-conferring enzyme 1, which is localized specifically to the cytoplasm of appressoria. An uncharacterized secondary metabolite synthesized by the AceI enzyme is secreted from the appressorium, triggering Pi33-mediated resistance in rice (53)(54)(55). Our results revealed that the level of expression of MoAA91 was increased in the appressorium stage and that MoAa91 was secreted during appressorium development. Interestingly, heterologously expressed MoAa91 or MoAa91 DTB was able to suppress the ROS and partially restore the pathogenicity defect. Therefore, it is reasonable to postulate that MoAa91 is an effector involved in disrupting host-derived immunity. It is not clear how MoAa91 is secreted into the host cells or into the apoplastic spaces of host cells. Previous studies showed that there are two independent secretion pathways involved in effector transport into or around the host cells: the biotrophic interfacial complex (BIC) pathway and the extrainvasive hyphal membrane (EIHM) pathway (49). However, no fluorescence of GFP-MoAa91 was observed in rice cells during infection. Therefore, we speculated that MoAa91 was likely secreted through an appressorium pore similar to that seen with ChEC of C. higginsianum.
To understand the mechanism by which MoAa91 subverts the host-derived immunity, we truncated MoAa91 into a C-terminal chitin-binding domain and an N-terminal glycoside hydrolase 61 domain. Analysis revealed that the chitin-binding domain was able to complement the defects in virulence of the ΔMoaa91 mutant sufficiently. Therefore, we propose that MoAa91 suppression of the host defense depends on the chitin-binding activity. Chitin oligosaccharide released from the cell wall of M. oryzae was sensed by CEBiP in rice and triggered chitin-mediated immunity. We speculated that MoAa91 bound with chitin or fungal polysaccharides to block the immune recognition by CEBiP. The secreted protein LysM protein 1 (Slp1) functions in the apoplast space by binding of chitin, thereby preventing the initiation of patterntriggered immunity by CEBiP (18). Also, Ustilago maydis secretes Rsp3 protein, which binds to fungal hyphae for protection against maize alpha-fetoprotein (AFP) proteins and interacts with membrane-bound receptor kinases containing the DUF26 domain to block signaling and prevent maize immune responses (56). In our study, we found that MoAa91 showed a high affinity in chitin binding in competition with CEBiP, indicating that M. oryzae might secret MoAa91 to compete with CEBiP. As MoAa91 and MoSlp1 showed similar functions in chitin binding, we performed a qRT-PCR assay to reveal the relationship between MoAa91 and MoSlp1 during infection. The results showed that both MoAa91 and MoSlp1 were expressed in the early stage of infection (8 to 24 hpi) and that the expression level of MoAa91 reached a high value at 8 hpi in comparison to that seen at 24 hpi with MoSlp1 (see Fig. S7F and G in the supplemental material). On the basis of these results, we speculated that MoAa91 and MoSlp1 might collaboratively function over the duration of the early infection to subvert host-derived immunity through chitin binding. We have proposed a working model of the regulatory mechanism underlying the function of MoAa91 during appressorium formation and plant infection in M. oryzae (Fig. 11).

MATERIALS AND METHODS
Fungal strains and RNA samples. M. oryzae strain Guy11 was used as the wild type and ΔMorgs1, ΔMorgs3, ΔMorgs4, ΔMorgs7, and ΔMomsn2 mutants as the mutant strains in this study (21,29). Rice plants (cv. CO-39; 2 weeks old) were inoculated (by standard spray inoculation) with conidial suspensions (3 ϫ 10 6 spores/ml, in a 0.2% [wt/vol] gelatin solution) of the Guy11, ΔMorgs1, ΔMorgs3, ΔMorgs4, or ΔMorgs7 strain. The inoculated plants were placed in a sealed plastic box in the dark for 24 h at 28°C, and leaf tissues were collected at 0, 1, 4, and 24 h postinoculation (hpi). Recovered samples were immediately frozen in liquid nitrogen, lyophilized, and stored at -80°C until needed.
RNA isolation, library construction, and sequencing. Total RNA was extracted from samples using an Invitrogen kit as described previously (42). The RNA integrity numbers (RIN) of all samples were above Isolation of poly(A) mRNA from total RNA and construction of cDNA libraries were performed according to methods described previously by Dong et al. (26). Finally, the cDNA libraries were loaded onto the flow cell channels of an Illumina HiSeq 4000 platform for paired-end 151-bp-by-2 sequencing at the Beijing Genomics Institute (BGI; Shenzhen, China) (57).
After discarding low-quality raw reads (retaining the connectors and discarding the reads whose content had more than 5% unknown bases), the clean reads from each library were assembled for M. oryzae. Gene expression levels were measured in the RNA-Seq analysis and expressed as the number of exon model reads per kilobase per million (RPKM) (58). Expression levels of differentially expressed genes and the corresponding P values were determined using the DEG fold change (log 2 ratio) values estimated according to the normalized gene expression level in each sample. We used the absolute value of a log 2 ratio of Ͼ1, P values of Ͻ0.001, and false-discovery-rate (q) values of Ͻ0.001 as the thresholds to judge differentially expressed genes.
Quantitative RT-PCR assay. Quantitative real-time RT-PCR (qRT-PCR) assays were performed as previously described (26,42). RNA-Seq expression profiles were validated by qRT-PCR. Primer pairs used as described in this section are listed in Table S1G in the supplemental material.
Gene disruption and complementation. To generate MoAA91 gene replacement vector pCX62, approximately 1-kb upstream and 1-kb downstream fragments were amplified with primer pairs (Table S1G). The resulting PCR products were ligated to the hygromycin resistance cassette released from pCX62, as previously described (26). Putative mutants were screened by PCR and confirmed by Southern blotting. The complement assay was performed according to our previously established methods (26,29).
EMSA. The full-length DNA of the MoAA91 promoter was amplified from Guy11 genomic DNA using primers MoAA91-Pro-F/MoAA91-Pro-R (Table S1G). The DNA fragment from the MoAA91 promoter was end labeled with Alex 660 by PCR amplification using the 5=-Alex 660-labeled primer. The MoMsn2 protein was expressed and purified from Escherichia coli strain BL21 by the use of the pGEX4T-2 construct containing an N-terminal ␤-glucuronidase (GST) tag coding sequence. The purified protein was mixed with Alex 660-labeled DNA, incubated for 20 min at 25°C in binding buffer, and separated by agarose gel electrophoresis. Gels were visualized directly using a Li-COR (Lincoln, NE, USA) Odyssey scanner with excitation at 700 nm (41).
Pathogenicity, plant infection, and rice-sheath penetration assays; appressorium formation assay; and appressorium turgor determination. Conidia were harvested from 10-day-old straw decoction and corn (SDC) agar cultures, were filtered through three layers of lens paper, and were resuspended to a concentration of 5 ϫ 10 4 spores/ml in a 0.2% (wt/vol) gelatin solution. For the leaf assay, more than 20 leaves from 2-week-old seedlings of rice (Oryza sativa cv. CO39) were used for spray inoculation. Spray inoculation and pathogenicity assays were performed according to methods previously described by Zhang et al. (21). At the same time, more than 20 leaves from 3-week-old seedlings of rice (Oryza sativa cv. CO39) were used for injection inoculation. A 1-ml syringe was used with a concentration of 5 ϫ 10 4 spores/ml in a 0.2% (wt/vol) gelatin solution for injection of the spore suspension into the cavity of each of the rice sheaths, as the standard method of preparation of infected leaf-sheath material involves filling the hollow space in the center of the sheath with a spore suspension.
For infection assays performed with rice tissues, 3-week-old rice cultivar CO39 was inoculated with 100 l conidial suspension (1 ϫ 10 5 spores/ml in a 0.2% [wt/vol] gelatin solution) on the inner leaf-sheath cuticle cells followed by incubation under humid conditions at 28°C. The leaf sheaths were observed under a Zeiss Axio Observer A1 inverted microscope at 30 hpi. A total of 5 leaf sheaths were used for each strain, and this experiment was performed three times.
For the appressorium formation assay, droplets (30 l) of the conidial suspension were placed on plastic coverslips (Fisher Scientific, St. Louis, MO, USA) under humid conditions at 28°C (59). The appressorium turgor was determined by cell collapse assay using a 1-to-4 molar concentration of glycerol solution. The percentages of conidium-germinating and conidium-forming appressoria were determined by microscopic examination of at least 100 conidia.
Cutin monomer, wax, and plant extraction treatments. Cutin monomer (1,16-hexadecanediol; Sigma-Aldrich, United Kingdom) was eluted at 200 M. Wax (1-octacosanol; Sigma-Aldrich, United Kingdom) was dissolved to 4 mg/ml in chloroform. And aliquots of 10 l were dropped onto microscope glass surface (Gold Seal, Portsmouth, NH), as previously described (3,60). The onion epidermis (0.7 g/5 ml double-distilled water [ddH 2 O]) and rice leaves of 2-week-old seedlings (0.43 g/2 ml ddH 2 O) were ground. After centrifugation, the supernatant was subjected to 100ϫ dilution. Drops of 20-l conidial suspensions were placed on cutin-coated or wax-coated areas, and then the drops were mixed with 1-l plant extractions and were placed onto plastic coverslips and then assayed for appressorium formation.
DAB staining and penetration assay performed with DPI or CAG treatment. For DAB staining assay, rice tissues infected by M. oryzae strains at 30 hpi were stained with 1 mg/ml DAB (Sigma-Aldrich) solution (pH 3.8) for 8 h and destained with an ethanol/acetic acid solution (ethanol/acetic acid ϭ 98:2 [vol/vol]) for 1 h. For evaluation of the growth of IH in ROS-suppressed rice sheath, 0.5 M DPI or 0.2 U catalase of Aspergillus niger (CAG; Sigma) dissolved in 10 mM (NH4) 2 SO 4 was mixed with the conidial suspensions (1 ϫ 10 5 spores/ml) to suppress the stress imposed by ROS. The details of the methods used were as previously described (39)(40)(41)(42). All the samples were observed under a Zeiss Axio Observer A1 inverted microscope (40ϫ).
In vitro protein purification assays. To construct the plasmids corresponding to GST-MoAA91 and GST-MoAA91 DTB , His-MoMSN2, and His-CEBiP, full-length cDNA of MoAA91 lacking the signal peptide was amplified and inserted into the vector pGEX4T-2, full-length cDNA of MoMSN2 was amplified and inserted into the vector pET-32A, and full-length cDNA of CEBiP lacking the signal peptide and the transmembrane domain was amplified and inserted into the vector pET-32A, respectively. These plasmids were then expressed in E. coli strain BL21, and bacterial cells were collected and treated with lysis buffer (10 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.5 mM EDTA, 0.5% Triton X-100). To confirm the expression of the GST or His fusion proteins, bacterial lysates were separated by the use of SDS-PAGE gel followed by Coomassie blue staining (see Fig. S7A to D in the supplemental material). In the assay used for purification of protein (His-MoMsn2/CEBiP and GST-MoAa91/MoAa91 DTB ), bacterial lysate containing His-MoMsn2/CEBiP protein was incubated with 30 l nickel-nitrilotriacetic acid (Ni-NTA) agarose beads (Invitrogen, Shanghai, China) and bacterial lysate containing GST-MoAa91/MoAa91 DTB protein was incubated with GST beads for 1 h at 4°C. The beads were then washed five times, and the elution was reduced using glutathione (Abmart, Shanghai, China).
Synergistic cellulase hydrolysis with MoAa91. For the hydrolysis of cellulose, 1% (wt/vol) Avicel PH-101 (Sigma-Aldrich, United Kingdom) was incubated with 1 mg of MoAa91 or MoAa91 DTB /g Avicel in the presence or absence of Celluclast (Sigma-Aldrich, United Kingdom) in 50 mM sodium acetate buffer (pH 5.0) at 50°C for 48 h. Sodium azide (NaN 3 ) was added at 0.02% (wt/vol) to all the synergistic reaction mixtures as an antibiotic or reducing agent. To terminate the enzyme reaction, the reaction mixture was boiled at 95°C for 5 min and was then subjected to centrifugation at 12,000 ϫ g for 5 min. Finally, the amount of reducing sugar present in the supernatant was quantified by 3,5-dinitrosalicylic acid (DNS) assay at 540 nm using D-glucose as the standard (61) (Fig. S7E). The extent of synergism was equal to the yield of reducing sugar from the hydrolysis of the MoAa91 or ⌬Moaa91 mutant with the cellulase.
Affinity precipitation of MoAa91 with polysaccharides. The affinity of MoAa91 for various polysaccharides was investigated by incubating 50 mg/ml of MoAa91 with 5 mg of chitin beads (New England Biolabs), crab shell chitin, xylan, or cellulose (Avicel; Sigma-Aldrich) as described previously (18,62). Protein and the polysaccharide were incubated at 4°C on a rocking platform in a final volume of 1 ml of water. After 16 h, the insoluble pellet fraction was centrifuged (5 min, 13,000 ϫ g), and the supernatant was collected. The insoluble fraction was pelleted and rinsed a further three times in distilled sterile water to remove unbound protein. Both the supernatant and the pelleted fractions were then boiled in 200 ml of 1% SDS solution before being examined by SDS-PAGE and Western blotting.
Data availability. The gene identity (ID) in this paper can be found in the FungiDB (https://fungidb .org/fungidb/). And the accession number for the transcriptome used in this study is GSE128219.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.