The rice blast fungus MoRgs1 functioning in cAMP signaling and pathogenicity is regulated by casein kinase MoCk2 phosphorylation and modulated by membrane protein MoEmc2

GTP-binding protein (G-protein) and regulator of G-protein signaling (RGS) mediated signal transduction are critical in the growth and virulence of the rice blast pathogen Magnaporthe oryzae. We have previously reported that there are eight RGS and RGS-like proteins named MoRgs1 to MoRgs8 playing distinct and shared regulatory functions in M. oryzae and that MoRgs1 has a more prominent role compared to others in the fungus. To further explore the unique regulatory mechanism of MoRgs1, we screened a M. oryzae cDNA library for genes encoding MoRgs1-interacting proteins and identified MoCkb2, one of the two regulatory subunits of the casein kinase (CK) 2 MoCk2. We found that MoCkb2 and the sole catalytic subunit MoCka1 are required for the phosphorylation of MoRgs1 at the plasma membrane (PM) and late endosome (LE). We further found that an endoplasmic reticulum (ER) membrane protein complex (EMC) subunit, MoEmc2, modulates the phosphorylation of MoRgs1 by MoCk2. Interestingly, this phosphorylation is also essential for the GTPase-activating protein (GAP) function of MoRgs1. The balance among MoRgs1, MoCk2, and MoEmc2 ensures normal operation of the G-protein MoMagA-cAMP signaling required for appressorium formation and pathogenicity of the fungus. This has been the first report that an EMC subunit is directly linked to G-protein signaling through modulation of an RGS-casein kinase interaction.

Introduction Heterotrimeric G-protein signaling plays a fundamental role in regulating various cellular developmental processes in eukaryotic organisms. The largest family of membrane receptors, G-protein-coupled receptors (GPCRs), sense extracellular stimuli by activating G-proteins, consisting of three canonical subunits Gα, Gβ, and Gγ, and downstream effector molecules that modify cellular behaviors [1][2][3]. In a conserved fashion, activated GPCRs induce conformational changes in the GDP-bound Gαβγ heterotrimer leading to the dissociation of GTPbound Gα from the Gβγ heterodimer [4]. GTP-bound Gα can then convey signals to adenylyl cyclases (ACs) that modulate the levels of cyclic adenosine monophosphate (cAMP) and cAMP-dependent protein kinase A (PKA) pathways [5][6][7]. The Gβ subunit was also shown to be involved in both cAMP/PKA and Pmk1 MAPK pathways important in appressorium formation and virulence of the fungus [8][9][10][11]. The regulator of G-protein signaling proteins (RGS) is known to serve as GTPase-activating proteins (GAP) for various Gα subunits of heterotrimeric G proteins [12][13][14]. In addition, some RGS proteins contain additional domains such as the GoLoco motif that inhibits GDP dissociation [15].
Appressoria are the unique infectious structures produced by the rice blast fungus Magnaporthe oryzae during infection. With its robust turgor pressure, mature appressoria effectively penetrate the host cell allowing rapid invasion and colocalization [16][17][18][19]. G-protein-mediated cAMP signaling has been demonstrated to be crucial for various aspects of signal transduction and substrate transport during appressorial biogenesis in M. oryzae [20][21][22][23]. Previous studies have demonstrated that Pth11 (a non-canonical GPCR), MoMagA (Gα), and MoRgs1 (RGS) mediate the perception of hydrophobic and hard surface cues leading to the activation of the cAMP-PKA pathway and normal appressorial formation [20,[24][25][26].
MoRgs1 is one of the eight RGS and RGS-like proteins whose function is linked to multiple Gαs in M. oryzae [21,22,[26][27][28][29]. MoRgs1 have the N-terminal DEP domains involved in Gprotein signaling and a dynamic tubulo-vesicular localization [21,30,31]. Despite its important function in M. oryzae, regulation of MoRgs1 remains not clear. Previous studies in the budding yeast Saccharomyces cerevisiae suggested that the RGS protein Sst2 is regulated by a phosphorylation feedback loop involving the MAP kinase Fus3 in response to pheromone stimulation [32,33]. A phosphoproteome study indicated that MoRgs1 could be subject to phosphorylation regulation [34]; however, further studies were not done until today.
The CK2 protein kinase is an evolutionally conserved serine/threonine kinase in eukaryotes functioning in many cellular processes [35,36]. A typical Ck2 holoenzyme is tetrameric, with two catalytic and two regulatory subunits [37]. CK2 was found to function as a member of diverse kinases that regulate GPCRs [38]. In the chestnut blight fungus Cryphonectria parasitica, CK2 phosphorylates the phosducin-like protein Bdm-1 that is required for the Gβ subunit protein stability and virulence [39]. A previous study identified one catalytic subunit Ckα and two regulatory subunits Ckβ1 and Ckβ2 from M. oryzae, which were named MoCka1, MoCkb1, and MoCkb2, respectively [40]. MoCka1 encodes an essential function, whereas MoCkb1 and MoCkb2 have a role in regulating the pathogenicity of M. oryzae [37,40].
The ER membrane protein complex (EMC) is conserved proteins identified in mammals and yeasts [41]. Yeast has eight EMC proteins that aggregate around the ER compartment [41][42][43]. Emc2, Emc8, and Emc9 do not contain ER targeting signals, thus constituting a part of the EMC cytoplasmic interface [44]. Interestingly, Emc2 contains the conserved tetratricopeptide repeat (TPR) domains for protein-protein interactions [45][46][47][48]. EMC functions typically as an insertase and/or membrane protein chaperone required for the biogenesis of dominant proteins in lipid homeostasis and signal transduction [44,49,50]. To further explore the regulatory mechanism of MoRgs1, we here demonstrated that MoRgs1 is subject to MoCk2 regulation through protein phosphorylation, and this regulation is required for MoRgs1 GAP function. We also identified that MoCk2-dependent MoRgs1 phosphorylation is modulated by MoEmc2. These concordant cellular processes ensure a critical role of G-protein/cAMP-PKA signaling in regulating appressorium formation and pathogenicity of M. oryzae.

MoRgs1 is phosphorylated during the developmental and invasive stages of M. oryzae
To determine whether MoRgs1 is subject to phosphorylation regulation, we performed an in vivo phosphorylation assay using Mn 2+ -Phos-tag gel electrophoresis analysis. We generated a MoRgs1-GFP fusion construct and introduced it into the ΔMorgs1 strain. Cell extracts were obtained from mycelial, conidial, and appressorial stages treated either with or without a phosphatase or a phosphatase inhibitor. The results showed that a specific mobility shift pattern of MoRgs1-GFP consistent with that MoRgs1 could be phosphorylated in M. oryzae (S1 Fig).

MoCk2 is a protein kinase that phosphorylates MoRgs1
To identify proteins that bind to and potentially phosphorylate MoRgs1, we performed a yeast two-hybrid (Y2H) screen to identify MoRgs1-interacting proteins. We constructed a MoRgs1-BD bait vector and screened a cDNA activator library of M. oryzae. Following gene sequencing analysis of several binding partners, we identified MoCkb2 (encoded by loci MGG_05651) (S1 Text).
As the MoCk2 holoenzyme is composed of MoCka1 and MoCkb1 and MoCkb2, we tested the interactions of all three proteins with MoRgs1 by in vivo co-IP and split YFP BiFC assays and by in vitro Y2H and GST-pulldown assays. The results showed that MoRgs1 interacts with all three in vivo (Figs 1A, 1B, 1C and S2B), but with MoCka1 and MoCkb2 in vitro (Figs 1D and S2A). The BiFC assay found that fluorescence appeared at the inner PM and dynamic tubulo-vesicular compartments that are similar to the localization of MoRgs1 [30,31] (S2B Fig).
Previous studies indicated that both yeast CK2 catalytic subunits are essential and that the regulatory subunit of CK2 is required for modulating substrate specificity and stabilizing against denaturing forces [37,40,51]. To examine whether MoCkb1 and MoCkb2 have a role in regulating MoRgs1 through their kinase function, we performed an in vivo phosphorylation assay that found MoCkb2, but not MoCkb1, is involved in MoRgs1 phosphorylation (Figs 1E and S3A). To confirm the phosphorylation between MoCk2 and MoRgs1, we performed an in vitro fluorescence detection in tube (FDIT) assay that showed MoRgs1 phosphorylation is dependent on both MoCka1 and MoCkb2 (Fig 1F)

MoCk2 phosphorylates S396, S399, S585, S696, and S700 residues of MoRgs1
To identify the phosphorylation sites of MoRgs1 by MoCk2, we compared LC-MS/MS analysis data between the wild-type (WT) and ΔMockb2 strains and identified five differentiated serine phosphorylation sites, including S396, S399, S585, S696, and S700 (Figs 1G and S4). To test if these five serine residues are MoCk2-dependent phosphorylation sites, we generated all five Serine (S) to Alanine (A) and Aspartic acid (D) site-directed mutagenesis constructs fused with either GFP or GST to mimic the sustainable unphosphorylated (MoRgs1 5A ) and phosphomimetic (MoRgs1 5D ) status. Phosphorylation assays with in vivo Mn 2+ -Phos-tag gel analysis and the in vitro FDIT method confirmed that all five serine residues of MoRgs1 are the phosphorylation sites by MoCk2 (Fig 1H and 1I).

MoRgs1 phosphorylation is required for cAMP signaling
To test the roles of MoRgs1 phosphorylation, we evaluated the in vitro GAP function by measuring levels of free phosphate release using an ATPase/GTPase activity assay kit. The mimic sustainable unphosphorylated MoRgs1 5A loses its GAP function in comparison to MoRgs1 and mimic phosphomimetic MoRgs1 5D , indicating MoRgs1 phosphorylation is required for its GAP function (Fig 2A). Considering the GAP function of MoRgs1 is critical for maintaining the intracellular cAMP levels in M. oryzae [21,22], we tested the levels in ΔMorgs1 and site-directed mutagenesis transformants by HPLC. The mimic sustainable unphosphorylated MoRgs1 5A transformants have higher cAMP levels than the wild-type strains (Fig 2B), indicating jeopardizing MoRgs1 phosphorylation equals to the loss of the GAP function.
In addition, we tested the interaction among GDP-bound MoMagA with MoRgs1, MoRgs1 5A , and MoRgs1 5D by co-IP assays. The results showed that MoRgs1 and MoRgs1 5A interact with MoMagA, but not MoRgs1 5D (S5 Fig), which indicated that phosphorylated MoRgs1 would be dismissed from the heterotrimeric G-protein Gα subunit.

MoRgs1 phosphorylation is required for appressorium formation and pathogenicity
Because MoRgs1 is involved in G-protein cAMP signaling and the ΔMorgs1 strains maintain normal appressorium on non-inductive surfaces [21,22,52], we tested the roles of MoRgs1 phosphorylation in appressorium formation at the hydrophobic and hydrophilic surfaces, respectively. The mass of two dome-shaped appressoria was generated when the phosphorylation of MoRgs1 was compromised (MoRgs1 5A ) at the hydrophobic surface ( Fig 2C). In addition, when MoRgs1 losses phosphorylation (MoRgs1 5A ), it promotes appressorium formation at the hydrophilic surface ( Fig 2C). Therefore, the phosphorylation of MoRgs1 activated by MoCk2 is important for invasive structure morphogenesis in M. oryzae.
Because the appressorium functions as an invasive structure, we hypothesized that MoRgs1 phosphorylation plays a significant role in the pathogenicity of M. oryzae. To determine its function, we observed the invasive hyphae (IH) growth into epidermal cells of rice leaf sheath by confocal microscopy (Fig 2D). We also evaluated virulence by leaf spraying and in vivo sheath injection assays (Fig 2E and 2F). The results indicated that the balance of MoRgs1 phosphorylation is critical for tbe pathogenicity of M. oryzae. The sustainable unphosphorylated MoRgs1 showed virulence defects that were similar to the ΔMorgs1 strain. Even though the sustainable phosphomimetic MoRgs1 partially suppresses the defect, it still showed a significant difference from that of the wild type and complemented strains.
In addition, phenotypes growth, biomass, conidiation, and penetration were assessed for MoRgs1 and the site-directed mutagenesis strains (S2 Text). Consistent with the pathogenicity Conidia from the WT, ΔMorgs1, MoRGS1 5A (ΔMorgs1/MoRGS1 5A ), MoRGS1 5D (ΔMorgs1/MoRGS1 5D ), and complemented MoRGS1 (ΔMorgs1/MoRGS1) strains were dropped onto corresponding surfaces and photographed by confocal microscopy at 24 h. Values are means of three replications and SD ( �� P < 0.01, n = 100). (D) Detailed observation with statistical analysis for infectious hyphal growth in rice sheath cells at 24 hpi. Four milliliters of conidial suspension (1 × 10 5 spores/ml) of each strain were used for the injection. Appressorium penetration sites were observed, and IH were rated from type 1 to type 4 (n = 100). Error bars represent SD from three independent replicates. (E) Rice spraying assays in vivo and lesion area statistics. Conidial suspensions (5 × 10 4 spores/ml) were sprayed onto 2-week-old rice seedlings (CO-39). Diseased rice leaves were photographed and the percentage of per 5 cm length leaf lesion area were analyzed by ImageJ after 7 days of inoculation ( �� P < 0.01, n = 10). Error bars represent SD from three independent replicates. (F) Rice sheath injecting assays in vivo and lesion area statistics. Conidial suspensions (2 × 10 5 spores/ml) were sprayed onto 4-week-old rice seedlings (CO-39). Diseased rice leaves were photographed, and the percentage of per 5 cm length leaf lesion area was analyzed by ImageJ after 5 days of inoculation ( �� P < 0.01, n = 10). Error bars represent SD from three independent replicates. White triangles point out the injection sites. results, both sustainable unphosphorylated and phosphomimetic MoRgs1 exhibit significant defects compared to WT and the complemented strains. All of the above data showed that the phosphorylation of MoRgs1 has a significant role in G-protein signaling required for appressorium formation and pathogenicity.

MoEmc2 interacts with MoRgs1 and all three subunits of MoCk2
To further characterize the modulation of MoRgs1 phosphorylation by MoCk2 in G-protein/ cAMP signaling, we identified a tetratricopeptide repeat (TPR) domain-containing protein, MoEmc2, encoded by the gene locus MGG_07480 (S1 Text) from a Y2H screening. MoEmc2 shares a high amino acid sequence homology to yeast Emc2, an evolutionarily conserved ER membrane protein complex subunit (S6A Fig). To further validate this conclusion, we conducted a yeast complementation assay based on that the yeast ΔScemc2 strain has a growth defect at the restrictive high temperature [49,53]. We introduced MoEMC2 into the yeast ΔScemc2 strain using the expression vector pYES2 and the results showed that MoEmc2 did restore the growth to the yeast ΔScemc2 strain (S6B Fig).
To validate the interaction between MoRgs1 and MoEmc2, we performed Y2H, co-IP, and BiFC assays. All data showed that MoRgs1 interacts with MoEmc2 in vivo and in vitro (Fig 3A,  3B and 3G). Interestingly, confocal fluorescence in the BiFC assay is similar to that for MoRgs1-MoCk2 interactions, nearby the PM and dynamic tubulo-vesicular compartments at the germ tube hooking stage ( Fig 3G). This suggests that MoEmc2 shares a similar localization as MoRgs1-MoCk2 interactions. We again tested the interactions among all three subunits of MoCk2 and MoEmc2 by Y2H, co-IP, and BiFC assays. Unlike MoRgs1, MoEmc2 directly interacts with all three subunits MoCka1, MoCkb1, and MoCkb2 in vivo and in vitro (Fig 3C-3G). In addition, confocal fluorescence in the BiFC assay confirmed that MoEmc2 interacts with MoCk2 holoenzyme subunits at the inner PM and tubule-vesicular compartments ( Fig 3G).

MoEmc2 interacts with MoRgs1 at the N terminus
To explore the MoEmc2-MoRgs1 interaction, we used SMART (http://smart.embl-heidelberg. de/) to predict protein interactive domains and motifs. MoEmc2 contains at least one centrally located TPR domain (160-193aa), whereas MoRgs1 contains two DEP domains at the N terminal and one RGS domain at the C terminal. The DEP domain at the N terminal is suggested to primarily function by targeting the C terminal RGS core domain to the punctate vesicular structures [22,31]. We divided the sequence of MoEmc2 amino acids into three regions, including the N terminal domain (NTD), the TPR regions (TPR), and the C terminal domain

MoEmc2 is required for MoCk2-dependent MoRgs1 phosphorylation
To further test the roles of MoEmc2 in MoCk2-dependent MoRgs1 phosphorylation, we performed targeted gene replacement and complement assays [54]. Putative hygromycin-resistant transformants were screened and mutants confirmed by Southern blot analysis (S8 Fig). Two independent ΔMoemc2 transformants showed the same defective phenotypes. In addition, the complemented strains were also phenotyped to be similar to the WT strains. Only one mutant and one complemented strain were used for further analysis.
Because MoEmc2 contains the TPR domain for protein-protein interaction, and it interacts with all three subunits of the MoCk2 holoenzyme and MoRgs1, we hypothesized that it might act as a scaffold in MoRgs1 phosphorylation. The interactions among MoCk2 holoenzyme subunits with MoRgs1 were tested in the wild type and ΔMoemc2 strains by co-IP assays (Figs 4A and S9C). As expected, the loss of MoEmc2 hinders the interactions among MoCk2 holoenzyme subunits with MoRgs1. Based on the defect in interaction, we further tested the effect of MoEmc2 in modulating MoCk2-dependent MoRgs1 phosphorylation. In vivo phosphorylation assay was performed. Similar to MoCk2-mediated MoRgs1 phosphorylation, MoEmc2 is required for MoRgs1 phosphorylation ( Fig 4B). We also tested the phosphorylation sites in the ΔMoemc2 mutant by LC-MS/MS and found that all five serine residues are unphosphorylated (S4G Fig). The results indicated that MoEmc2 is also required for MoRgs1 phosphorylation.
Considering MoEmc2 is a membrane protein, we hypothesized it might modulate the subcellular localization of MoCk2 holoenzyme subunits or/and MoRgs1. We observed the subcellular localization of MoCk2 and MoRgs1 by laser scanning microscopy (LSM) in the wild type and ΔMoemc2 strains at the germ tube hooking stage. The subcellular localization of all three subunits of MoCk2 at tubule-vesicular compartments and inner PM is modulated by MoEmc2, while it does not affect MoRgs1 (Figs 4C, 4D, S9A and S9B). Together, these results indicated that MoEmc2 modulates the subcellular localization of MoCk2 to govern MoRgs1 phosphorylation.

MoEmc2 is required for the development, differentiation, and full virulence of M. oryzae
As MoRgs1 phosphorylation is required for the development and pathogenicity of M. oryzae, we further explored the roles of MoEmc2 in development and differentiation. We tested the growth, conidiation, biomass, penetration, and the collapsed appressorium rate. Significant defects were observed between the ΔMoemc2 and the wild type and complemented strains (S3 Text). All of the results indicated that MoEmc2 is required for the development and reproduction of M. oryzae.
During the pathogen-host interaction, M. oryzae perceives the hard, hydrophobic surface of rice leaf and generates an infectious structure appressorium. G-protein/cAMP signaling usually promotes the germination of conidia, and the Pmk1 MAPK pathway is also required for appressorium maturation and IH growth [8,16,26,29,55]. Because MoEmc2 is involved in G-protein signaling, we determined the roles of MoEmc2 in the pathogen-host infection by rice spraying, in vivo sheath penetration, and in vitro assays (S10A-S10F Fig). Consistent with a role in modulating the development, MoEmc2 is also critical for the virulence and invasive growth of M. oryzae.
As appressorium is important for invasion, we tested the appressorium formation rate at continuous time points and calculated it at the hydrophobic and hydrophilic surfaces at 24 h. The results indicated that MoEmc2 not only regulates conidia germination at the early stage but also dominates appressorium maturation (S10G and S10H Fig). In addition, similar to MoRgs1, MoEmc2 is required for appressorium formation at the hydrophilic surface, even though with a lower rate (S10I Fig). As for morphology, MoEmc2 is required for normal appressorium development (S10K Fig). However, when we tested the cellular cAMP levels in the transformants, MoEmc2 and MoRgs1 showed differential effects (S10J Fig). Unlike the negative regulation in G-protein/cAMP signaling of MoRgs1, MoEmc2 exhibits a positive effect similar to the adenylyl cyclase MoMac1. The reasons for this discrepancy remain to be explored.

Activating MoRgs1 phosphorylation partially inhibits deficiency of the ΔMoemc2 mutant
Because sustainable activated MoRgs1 phosphorylation could partly inhibit the phenotype defect in M. oryzae, we tested the effects of phosphomimetic MoRgs1 5D in the ΔMoemc2 mutant. First, appressorium formation was observed and calculated at the hydrophobic and hydrophilic surfaces. The results showed that activating phosphorylation of MoRgs1 strengthens the defect of appressorium formation rate at the hydrophobic surface but partly suppresses the appressorium formation at the hydrophilic surface, suggesting the disturbance of MoRgs1 phosphorylation is, to some extent, responsible for the defect in ΔMoemc2 (Fig 5A).
To determine whether the phosphorylated MoRgs1 can rescue the invasive growth defect of the ΔMoemc2 mutant, we performed injection assays on rice sheath and performed a microscopic observation. The results showed that the phosphorylated MoRgs1 partly rescues the IH growth of the ΔMoemc2 mutant ( Fig 5B). Moreover, we tested the virulence of M. oryzae by rice leaves spraying and rice sheath injection assays. Consistent with IH growth, MoEmc2 is required for the virulence of M. oryzae (Fig 5C and 5D). Growth, conidiation, biomass, penetration, and collapsed appressorium rate were all evaluated by activating the phosphorylation of MoRgs1 in the ΔMoemc2 mutant, and the MoRgs1 5D partly suppresses the defect of the ΔMoemc2 mutant (S3 Text). All of these results indicated that MoEmc2 is required for G-protein cAMP signaling to regulate appressorium formation and pathogenicity of M. oryzae.

MoEmc2 is localized at the ER, late endosome, and PM
To examine the subcellular localization of MoEmc2 at the conidia and germ tube hooking stages during infection, we employed the Dye ER-Tracker to image ER, FM4-64 to label the PM and trafficking vesicles [56][57][58], and the late endosome (LE) marker protein GFP-MoRab7. Conidia and hooking germ tubes were observed and photographed by LSM. The co-localization was evaluated by the fluorescence intensity curve. All of the results suggested that, except for ER localization, MoEmc2 still locates at the late endosome and the inner plasma membrane at the conidia and germ tube hooking stages (S11 Fig).
To further determine whether MoEmc2 targets or migrates through the intracellular membrane, we treated MoEmc2-GFP transformants with ER-Golgi trafficking inhibitors brefeldin A (BFA) and 4-bromobenzaldehyde N-2, 6-dimethylphenyl (EGA), which is known as an inhibitor of early endosome (EE) to LE transport [28,[59][60][61]. After observing hundreds of conidia, the fluorescence of MoEmc2 was observed at the inner PM, indicating no MoEmc2 trafficking between the ER and the PM (S12A and S12B Fig). In addition, EGA treatment did not co-localize MoEmc2 with the EE marker MoRab5, following Pearson correlation coefficient analysis with ImageJ Coloc2 (S12C and S12D Fig). The results indicated both inhibitors have no effect on the subcellular localization of MoEmc2, suggesting that MoEmc2 is a membrane component of ER, LE, and inner PM organelles.

Discussion
G-protein/cAMP signaling is required for appressorium formation and pathogenicity in the rice blast fungus M. oryzae [28,62,63]. The regulators of G-protein signaling, in particular MoRgs1, also plays an important role in these function [21]. As RGS proteins have a conserved function by modulating Gα activities through their intrinsic GAP functions, we here explored MoRgs1 regulatory mechanisms. We have demonstrated that the MoRgs1 GAP function requires MoCk2-dependent phosphorylation. The loss of phosphorylation resulted in an attenuation in the GAP function and dysregulated intracellular cAMP levels that are similar to those of the ΔMorgs1 mutant. Sustainable phosphomimetic MoRgs1 could suppress the defect and restore intracellular cAMP levels to the extent of the wild type and complemented strains, even though it does not interact with the GDP-bound Gαs. This is consistent with that, following accelerated GTPase hydrolysis of Gαs, MoRgs1 dissociates from the GDP-bound Gαs. We have also found that MoEmc2, as an EMC subunit, modulates the subcellular localization of MoCk2 through a direct interaction, which also regulates MoRgs1 phosphorylation. In addition, we have provided evidence that the balance of MoRgs1 phosphorylation is equally important for appressorium development and pathogenicity of M. oryzae.
Phosphorylation is an important cellular regulatory mechanism that is widely involved in most cellular processes, including cell growth, development, division, signal transduction, and protein synthesis [64]. A previous study identified four phosphorylation sites S159, S399, S401, and S510 in MoRgs1 and suggested that hyper-phosphorylated MoRgs1 may be essential for pathogenicity [34]. We here identified additional phosphorylation sites by MoCk2 and that MoCka1 and MoCkb2 subunits are directly involved in this phosphorylation of MoRgs1. The newly identified five MoCk2-dependent serine phosphorylation sites S396, S399, S585, S696, and S700 are not in consensus with the minimum canonical sequence of Ser-Xaa-Xaa-acidic (S represent for serine, and X can be any amino acid) for the CK2 target site [65,66]. It should be noted that this study examined only the direct subunits of MoCk2 in MoRgs1 phosphorylation.
As one of the core subunits of yeast EMC, Emc2 is usually recognized to promote the posttranslational insertion of tail-anchored (TA) membrane protein with its first transmembrane domain (TMD) into the ER and biogenesis [44,50,53,67]. Except for ER localization, MoEmc2 is also located at the late endosome and the inner PM at the conidia and germ tube hooking stages. This multicellular organelle localization indicates that MoEmc2 may have multiple roles, including that in directly regulating signal transduction. To further explore whether MoEmc2 directly anchors the organelle membrane components or shuffles protein trafficking between multicellular organelles, we employed the ER-Golgi trafficking inhibitor BFA and early to late endosome inhibitor EGA that disrupt protein trafficking. Both BFA and EGA treatment had no effects, suggesting MoEmc2 directly integrates with the internal membrane components without undergoes dynamic trafficking.
Previous studies also indicated that EMC promotes phospholipid transfer during lipid metabolism [49], which may satisfy the requirement during appressorium maturation and appressorial host penetration [17,18]. Consistent with this, we found that MoEmc2 is critical for proper turgor pressure [17,18,68]. Intriguingly, we found a discrepancy in that intracellular cAMP levels in the ΔMoemc2 strain were lower despite that MoEmc2 has a positive role in promoting MoRgs1 phosphorylation. We propose that the roles of MoEmc2 in regulating Gproteins signaling may be complexed, and additional studies are needed to dissect these roles.
In summary, the rice blast fungus M. oryzae senses the physical cues and activates G-proteins signaling during its interaction with the host. MoRgs1 plays an important function in modulating MoMagA-mediated cAMP signaling for appressorium formation and pathogenicity. In addition, MoRgs1 is subject to MoCk2-dependent protein phosphorylation, while this phosphorylation is also modulated by MoEmc2 (Fig 6). This multiple levels and complex regulatory circuitry likely ensure a proper balance between signal transduction and cellular homeostasis required for being a pathogen microorganism.

Strains and culture conditions
The M. oryzae Guy11 was used as the wild-type strain for transformation, and all strains were cultured on complete medium (CM) plates at 28˚C for 7 d [22]. For vegetative growth, small mycelium blocks were collected from 4-day-old colonies and transferred into fresh media, followed by incubation at dark 28˚C. The radial growth was measured following incubation for 7 days. Mycelia were harvested from liquid CM medium and used for DNA, RNA tests, and protein extractions. For conidiation, mycelium blocks were maintained on SDC (100 g of straw, 40 g of corn powder, 15 g of agar in 1 L of distilled water) agar media at 28˚C for 7 days in the dark followed by 3 days continuous illumination under fluorescent light.

Phylogenetic tree analysis and yeast complementation assays
The proteins with high query coverage to MoEmc2 were acquired from the NCBI database website (https://www.ncbi.nlm.nih.gov/). Amino acid sequence alignment was made with CLUSTAL W analysis and the phylogenetic tree was built by MEGA 7.0. To test the homology of MoEMC2 with ScEMC2, the full-length cDNA of MoEMC2 was constructed with vector pYES2 (Invitrogen). Following PCR sequencing and screening on SD medium without uracil, the MoEMC2-pYES2 constructs were transformed into the yeast ΔScemc2 mutant (BY4741 loss of YJR088C). Yeast colonies were cultured on YPD medium (10 g Yeast Extract; 20 g Peptone 20 g D-Glucose) and adjusted to OD600 value 0.1. Followed by 10-fold serial dilutions were grown on SD-Met-Leu-His-Ura (galactose) plates at 30˚C for 4 d and photographed.

Targeted MoEMC2 deletion and ΔMoemc2 complementation
The MoEMC2 gene deletion mutant was conducted by using the standard one-step gene replacement strategy. First, two approximate 1.0 kb of sequences flanking of MoEMC2 were amplified with two primer pairs, the products of MoEMC2 were digested with restriction endonucleases (XhoI and EcoRI, BamHI and SpeI) and ligated with hygromycin-resistance cassette (HPH) released from pCX62. The protoplasts of WT for targeted gene deletion were transformed with vectors with the hygromycin resistance marker gene cassette in which the marker gene was inserted into the middle of two flanking sequences of the MoEMC2 gene. For screening hygromycin-resistant transformants, CM plates were supplemented with 250 μg/ml hygromycin B (Roche, USA). We generated the complementation construct pYF11-MoEMC2, the gene sequence containing the full-length of the MoEMC2 gene gDNA sequence and 1.0 kb long native promoter with GFPEMC2F/GFPEMC2R. A yeast strain XK1-25 was co-transformed with the recombinant sequence and XhoI-digested pYF11 empty vectors. After 3-5 days of transformation and incubation, we screened the positive yeast transformants with polymerase chain reaction (PCR). Then we transformed the positive reconstructive plasmid into E. coli and with PCR test for acquiring positive clone of pYF11-MoEMC2. To generate the complemented strains, the pYF11-MoEMC2 construct containing the bleomycin-resistant (BLM) gene was introduced into the ΔMoemc2 mutant for M. oryzae transformants screening.

Appressorium formation and virulence assays
We harvested conidia and filtered them through two layers of Miracloth (EMD Millipore Corporation, 475855-1R) with distilled water. For appressorium formation assays, droplets (30 μl) of conidial suspension were placed on plastic coverslips (Fisher Scientific, St Louis, MO, USA) or GelBond PAG film (GE Healthcare, 80112936) under humid conditions at 28˚C [26]. All the samples were observed under Zeiss Axio Observer A1 inverted microscope (40 ×).
For virulence assays, 0.2% (w:v) gelatin was added to conidia suspension (5 × 10 4 spores/ml) to promote adhesion. Two weeks old seedlings of rice (Oryza sativa cv. CO39) were used for rice spraying assays. We inoculated 5 ml of conidial suspension each treatment onto rice seedlings with a sprayer. We kept inoculated seedlings in a growth chamber at 28˚C with 90% humidity, and the first 24 h was in the dark, followed by a 12 h -12 h light-dark cycle. The degree of disease was judged at 7 days post-inoculation. The degree of disease lesions was analyzed by ImageJ.
For the sheath injection assays in vivo, conidia suspensions (2 × 10 5 spores/ml) were injected into 3-week-old rice sheath with syringes and incubated in the same conditions as rice spraying assays. After 5 days post-inoculation, tissues that emerge from the sheath were collected and photographed. For the penetration assays in vitro, the conidia suspension (2 × 10 5 spores/ml) was injected into 3-week-old rice sheath with syringes and harvested after 36 h post-incubation at dark 28˚C. Epidermal cells in the leaf sheath were removed for microscopic observation. Each experiment was repeated with three biological repetitions and three replicates under the same experimental conditions (e.g., temperature, humidity, illumination, and phase of host plants).

Construction of the MoMagA-RFP plasmid vector
Tagging at the C terminal or the N terminal of MoMagA is known to jeopardize the Gα function, but a previous report indicated that introducing a tag peptides coding sequence in-frame into the αB-αC loop of MoMagA would effectively address this issue [30]. We thereby introduced an RFP fusion into MoMagA between amino acids 113-120.

Yeast two-hybrid assays
Bait constructs were conducted by cloning full-length cDNAs of target genes into pGADT7 and pGBKT7 (LMAI Biotechnology, LM1010). Prey constructs and bait constructs were confirmed by DNA sequencing and co-transformed into the yeast strain AH109 (LMAI Biotechnology, LM1010) following the recommended protocol (BD Biosciences Clontech). Transformants screened by synthetic dextrose medium minus leucine, tryptophan, adenine, and histidine (SD-Leu-Trp-Ade-His) were selected. Yeast strains for positive and negative controls were provided by the BD library construction and screening kit.

Co-immunoprecipitation (co-IP) assays
To confirm the interactions between MoRgs1 with all three MoCk2 holoenzyme subunits MoCka1, MoCkb1, and MoCkb2 in vivo, full-length gDNA with 1500 bp native promoter region of MoCka1, MoCKB1, and MoCKB2 were cloned into pXY203, a vector with a S tag [69]. Full-length gDNA of MoRGS1 was constructed on pYF11 labelled by the GFP tag [69]. All constructs were introduced into the protoplasts of WT strains Guy11 by co-transformation. Total proteins were extracted from transformants co-expressing MoRgs1-GFP with MoCka1-S, MoCkb1-S, and MoCkb2-S, then incubated with anti-GFP affinity beads (Smart lifesciences, SA070001). Proteins bound to the beads were eluted after a series of washing steps

Bimolecular fluorescence complementation (BiFC) assays
The protein MoRgs1, MoCka1, MoCkb1, MoCkb2, and MoEmc2 fusion constructs were generated by cloning full-length gDNA with 1500 bp native promoter region into pHZ65 and pHZ68 (BiFC vectors were graciously provided by Dr. J. R. Xu of Purdue University, USA), respectively [70]. Constructs pairs of MoRgs1-YFP N and all three MoCk2 holoenzyme subunits MoCka1-YFP C , MoCkb1-YFP C and MoCkb2-YFP C were co-introduced into the protoplasts of WT strain, respectively. Transformants screened by two antibiotic hygromycin (Solarbio Life Sciences, H8080) and zeocin (Thermo Fisher Scientific, R25001) were isolated and confirmed by PCR analyses. Similar methods were used to detect interactions of MoEm-c2-YFP N with MoRgs1-YFP C and all three MoCk2 holoenzyme subunits MoCka1-YFP C , MoCkb1-YFP C , and MoCkb2-YFP C , respectively.

Assays under confocal laser scanning microscope observation
All strains transformed with fluorescent labeling were observed under the confocal laser scanning microscope (Zeiss LSM710, 63 × oil). The filtered channels set: GFP (excitation spectra: 488 nm, emission spectra: 510 nm), RFP (excitation spectra: 555 nm, emission spectra: 584 nm), YFP (excitation spectra: 514 nm, emission spectra: 528 nm). Exposure time: 800 ms. Insets highlight areas analyzed by line-scan. Bar = 10 μm. ImageJ software was applied to calculate the Pearson correlation coefficient for analyzing co-localization of GFP fusion protein with RFP fusion protein if needed. One area of interest was photographed with GFP and RFP channels, respectively. Photographs were opened using ImageJ. Image type was set to 8 bits, respectively. The 'Colocalization Coloc2' in the 'Analysis' section was applied to all images.

Intracellular cAMP level measurement by HPLC
M. oryzae strains or transformants were harvested in liquid medium CM for 48 h. Samples were lyophilized for 24 h and grounded into fine powders in liquid nitrogen. Intracellular cAMP was extracted following previously established procedures [21]. The intracellular cAMP levels were quantified by HPLC (High-Performance Liquid Chromatography). The relevant experimental procedures were assessed, as described previously [71].

In vivo phosphorylation assays using Phos-tag gel electrophoresis
The MoRgs1-GFP fusion construct was introduced into WT, ΔMockb1, ΔMockb2, and ΔMoemc2 strains. The total protein extracted from mycelia was resolved on 8% SDS-PAGE prepared with 50 μM acrylamide-dependent Phos-tag ligand and 100 μM MnCl 2 as described [72,73]. Gel electrophoresis was performed with a constant voltage of 80 V for 6 h. Before transferring, gels were equilibrated in transfer buffer with 5 mM EDTA for 20 min two times and followed by transfer buffer without EDTA for another 20 min. Protein transferred from the Mn 2+ -phos-tag acrylamide (NARD institute Limited company, 18D-01) gel to the PVDF membrane (EMD Millipore Corp., ISEQ00010) was performed for 40 h (depend on different proteins) at 80 V at 4˚C, then it was analyzed by immunoblotting using the anti-GFP antibody.

In vitro phosphorylation analysis
GST-MoCka1, GST-MoCkb1, GST-MoCkb2, His-MoRgs1, and His-MoRgs1 5A were expressed in E. coli DE3 cells and purified. A rapid and cost-effective fluorescence detection in tube (FDIT) method was used to analyze in vitro protein phosphorylation [74]. The Pro-Q Diamond Phosphorylation Gel Stain (Thermo Fisher Scientific, P33301) is used for phosphorprotein gel-staining. For protein kinase reactions, 2 μg MoRgs1 (MoRgs1 5A ) was mixed with MoCka1, MoCkb1, and MoCkb2 in a kinase reaction buffer (100 mM PBS [Beyotime Biotechnology, ST476], pH 7.5, 10 mM MgCl 2 , 1 mM ascorbic acid [Sigma-Aldrich, A5960]), with the appearance of 50 μM ATP (Sigma-Aldrich, FLAAS) at room temperature (RT) for 60 min, 10 folds of cold acetone was added to stop the reaction. For protein in tube staining, samples were homogenized and suspended in Mili-Q water at the concentration of 0.2 μg/μl. The pellet was rinsed with 0.5 ml cold acetone and centrifuge to remove the supernatant twice. The pellet was air-dried and dissolved in 200 μl of Mili-Q water and moved to a black 96 well plate (Corning, 3925). Fluorescence signal at 590 nm (excited at 530 nm) was measured in a Cytation3 microplate reader (Biotek, Winooski, VT, USA) [73].

LC-MS-MS analysis
To identify phosphorylation sites of the substrate, proteins were extracted and separated on 10% SDS-PAGE. In-gel digestion (or elution digestion), the stained protein bands were in-geldigested after overnight incubation in a destaining solution, followed by drying in an acetonitrile solution and equilibration in 20 mM ammonium bicarbonate. The liquid was removed, and pellets were washed by acetonitrile and dehydrated in a speed vac until protein bands are completely opaque. Proteins were reduced with 10 mM DTT at 56˚C for 1 h and alkylated immediately using 55 mM iodoacetamide (IAM) in the dark at room temperature for 1 h. Liquid removal, acetonitrile wash, and dehydration were again performed. Trypsin solution (1 μg/μl in 20 mM ammonium carbonate, pH 8.9) was added, and then a buffer was added until the protein bands were restored in size (rehydrated). The digestions were carried out overnight at 37˚C. The resulting peptides were recovered by two extractions of 20 min each, with 100 μL of a solution of 60% acetonitrile acid. When the highly hydrophobic peptides were expected, the third extraction with 60% acetonitrile plus 0.1% formic acid was performed. The extracts were combined and concentrated to~20 μL in a speed vac.
Peptide identification and quantification were carried out on the Mascot software Revision 2.3.01 using the TAIR database search algorithm and the integrated false discovery rate (FDR) analysis function. For mass spectrometry analyzing the phosphorylation sites, we thank the Beijing Protein Innovation Co., Ltd., for technological assistance.

GTPase activity assays
The fusion constructs His-MoRgs1, His-MoRgs1 5A , His-MoRgs1 5D , and His-MoMagA were expressed in E. coli DE3 cells and purified. Then we conducted GTPase activity assays with the ATPase/GTPase Activity Assay kit (MAK113; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) instructions. First, the phosphate standards were set as indicated in the kit instructions to plot a standard curve. Second, a series of dilutions of purified proteins were performed in assays buffer. The sample reactions and the control well were set up according to the scheme. The reaction was incubated for the desired period of time (10 s, 30 s, 1 min, 2 min, 5 min, and 10 min) at room temperature. Reagent (200 ml) was added to each well, and incubation was carried out for an additional 30 min at room temperature to terminate the enzyme reaction and generate the colorimetric product separately. Absorbance at 600-660 nm [maximum absorbance at 620 nm (A620)] was read. We calculated the change in absorbance values (DA620) for the samples by subtracting the A620 of the control well (A620) control from the A620 of the sample well (A620) sample. The concentration (mM) of free phosphate [Pi] was computed in the sample from the standard curve. The formula was: Enzyme activity (units/l) = [Pi] (mM) × 40 ml � [10 μl × reaction time (min)]. One unit is the amount of enzyme that catalyzes the production of 1 mmol of free phosphate per minute under the assay conditions. The relevant experimental procedures were assessed, as described previously [75].

Statistical analysis
Each experiment was performed with three replicates and represented as the mean ± standard deviation (SD). The significant differences between treatments were statistically determined by one-way analysis of variance (ANOVA) comparison and followed by Duncan's new multiple range tests if the ANOVA analysis is significant at P < 0.01 as described previously [76].

Accession numbers
Genes reported in this article can be found in the GenBank database under the following acces- (A and B) Rice spraying assays and lesion area statistics. Conidial suspensions (5 × 10 4 spores/ ml) were sprayed onto 2 week-old rice seedlings (CO-39). Diseased rice leaves were photographed and percentages of per 5 cm length leaf lesion area were analyzed by ImageJ after 7 days of inoculation. Values are means of three replications and SD ( �� P < 0.01, n = 10). (C and D) Rice sheath injecting assays in vivo and lesion area statistics. Conidial suspensions (2 × 10 5 spores/ml) were sprayed onto 4 week-old rice seedlings (CO-39). Diseased rice leaves were photographed and percentages per 5 cm length leaf lesion area were analyzed by ImageJ after 5 days of inoculation. Values are means of three replications and SD ( �� P < 0.01, n = 10). White triangles point out the injection sites. (E and F) Rice sheath injecting assays in vitro and classification statistics. Invasive hyphae (IH, n = 100) in rice cells were observed at 36 hpi and 4 types of were quantified and statistically analyzed. Error bars represent SD from three independent replicates. (G and H) Appressorium formation assays and statistics analysis. Conidia of the WT, ΔMoemc2 and complemented ΔMoemc2 (ΔMoemc2/MoEMC2) strains were dropped on hydrophobic surfaces and the dynamics of appressorium formation were photographed at various times ( �� P < 0.01, n = 100). Bar = 10 μm.