System-Wide Characterization of MoArf GTPase Family Proteins and Adaptor Protein MoGga1 Involved in the Development and Pathogenicity of Magnaporthe oryzae

Magnaporthe oryzae is the causal agent of rice blast, representing the most devastating diseases of rice worldwide, which results in losses of amounts of rice that could feed more than 60 million people each year. Arf (ADP ribosylation factor) small GTPase family proteins are involved in vesicle trafficking and organelle maintenance in eukaryotic cells. To investigate the function of Arf family proteins in M. oryzae, we systematically characterized all seven Arf proteins and found that they have shared and specific functions in governing the growth, development, and pathogenicity of the blast fungus. We have also identified the pathogenicity-related protein MoGga1 as the common adaptor of MoArf1 and MoArl1. Our findings are important because they provide the first comprehensive characterization of the Arf GTPase family proteins and their adaptor protein MoGga1 functioning in a plant-pathogenic fungus, which could help to reveal new fungicide targets to control this devastating disease.

We also introduced MoArf1, MoArl1, MoArl3, and MoCin4 into the S. cerevisiae ΔScarf1, ΔScarl1, ΔScarl3, and ΔSccin4 mutants, respectively, and found that they restored the sensitivity to various concentrations of hygromycin B (HygB) (Fig. S2). Although functional complementation of MoArf6, MoArl8, and MoSar1 was not possible, we nevertheless named them on the basis of the high sequence conservation and on data from previous studies (35,38). Additionally, we identified MoGga1 (MGG_00852) as the sole Gga protein homolog of M. oryzae.
Expression patterns and targeted deletions of MoARF genes. To characterize the functions of MoArf proteins, we performed qRT-PCR analysis to examine their transcriptional patterns in various growth stages. The transcript level of MoARF1 was 3.8-fold higher at the conidial stage than that at the mycelium stage. The MoARL1 transcription was upregulated during the early infection stage, with its transcript level increased by 20-, 55-, and 32-fold at 8, 24, and 48 h postinoculation (hpi), respectively, relative to that of the mycelium stage. The level of MoSAR1 transcription was at least 18-fold higher than that seen at the mycelium stage in all observed infection stages. The transcriptional profiles of MoARF6, MoARL3, MoARL8, and MoCIN4 remained relatively constant (Fig. S3A). These results indicated the relative importance of MoArf1 in conidiation and of MoArl1 and MoSar1 in infection.
We have next obtained ΔMoarf6, ΔMoarl1, ΔMoarl3, ΔMoarl8, ΔMocin4, and ΔMogga1 mutant strains by targeted gene disruption and verified their genotypes by Southern blotting (Fig. S3E). We also complemented the mutant strains with the respective wild-type (WT) genes that rescued all of the mutant defects. For reasons unknown, the ΔMoarf1 and ΔMosar1 mutants could not be obtained despite the screening of ϳ5,000 transformants each. Disruption of ScARF1/ScARF2 and ScSAR1 in S. cerevisiae, as well as AnARFA and AnSAR1, the respective paralogue of ARF and SAR in A. nidulans, was lethal (31,(39)(40)(41). To circumvent this, we introduced the nitrate reductase MoNIA1 promoter, a conditional promoter described previously in studies of Z. tritici and Aspergillus fumigatus (42,43), into MoARF1 and MoSAR1. We obtained one ΔMoarf1/CPR mutant and one ΔMosar1/CPR mutant, confirmed by Southern blotting (Fig. S3F), and found that the respective WT genes rescued the mutant defect.
The nitrate reductase promoter was induced by nitrate or other secondary nitrogen sources and repressed by ammonium or other primary nitrogen sources (43). We found that the levels of expression of MoARF1 in the ΔMoarf1/CPR mutant and MoSAR1 in the ΔMosar1/CPR mutant were 4.9-fold and 15.1-fold compared to that in strain Guy11, respectively, under NaNO 3 induction conditions (Fig. S3B). We next used NaGlu (C 5 H 8 NNaO 4 ) as the sole nitrogen source and found that the level of expression of MoARF1 in the ΔMoarf1/CPR mutant was about 0.5-fold lower than that in strain Guy11 at the concentrations tested (115, 230, and 460 mM). However, the expression levels of MoSAR1 in the ΔMosar1/CPR mutant were comparable to those in strain Guy11 at 115 mM and 230 mM NaGlu but not at 460 mM, where the expression level in the ΔMosar1/CPR mutant was found to be 0.53-fold lower than that in Guy11 (Fig. S3C). For this reason, we selected 460 mM NaGlu for further studies.
Characterization of growth and conidiation. We examined vegetative growth on complete medium (CM), minimal medium (MM), oatmeal medium (OM), and straw decoction and corn (SDC) medium. A significant growth reduction was observed in the ΔMoarf6, ΔMoarl1, ΔMoarl3, and ΔMocin4 mutant strains but not the ΔMoarl8 and ΔMogga1 mutants (see Table S1 in the supplemental material). We then examined the vegetative growth of the ΔMoarf1/CPR or ΔMosar1/CPR mutant on MM containing 460 mM NaGlu as the sole nitrogen source and found that the mutants showed significantly reduced colony diameters compared to Guy11 and complemented strain ΔMoarf1/CPR-C or ΔMosar1/CPR-C (Table S1). We also found the ΔMoarf6, ΔMocin4, and ΔMogga1 mutants showed 0.57-, 0.02-, and 0.55-fold reductions, respectively, in conidiation compared with the WT Guy11 strain (Table S1). MoGga1 are required for full virulence. Spraying with the  equal conidial suspensions, pin-sized specks were observed on leaves incubated with  the ΔMoarl1 mutant, in contrast to the WT blast lesions caused by the ΔMoarf6, ΔMoarl3,  and ΔMoarl8 mutants (Fig. 2A; see also Table S1). A similar result was observed in a detached barley leaf assay (Fig. 2B). As conidiation was severely impaired in the ΔMocin4 strain, hyphae were used for inoculation of detached rice and barley leaves, resulting in a nearly complete absence of disease symptoms ( Fig. 2A and B; see also Table S1). Conidial suspensions of ΔMoarf1/CPR and ΔMosar1/CPR mutants with 460 mM NaGlu were also tested for pathogenicity. The mutants caused typical disease symptoms similar to those caused by Guy11 and complemented strains in the injected-rice assays (Table S1). However, there were no differences between the ΔMoarf1/CPR, ΔMosar1/CPR, and Guy11 strains in the expression levels of MoARF1 or MoSAR1 in planta (Fig. S3D). Nevertheless, our results suggested that MoArl1 and MoCin4 are important in pathogenicity.

MoArl1, MoCin4, and
To further examine MoArl1 and MoCin4 functions in infection, we carried out the penetration and colonization assay. In rice sheath infection, we observed 100 penetration sites and rated them from type 1 to type 4 (type 1, no penetration; type 2, penetrating peg formed; type 3, spreading but limited to one cell; type 4, spreading to neighboring cells) at 48 hpi. In the Guy11 and ΔMoarl1/MoARL1 strains, more than 90% appressoria penetrated the rice cells, with more than 75% consisting of type 3 and type 4. In contrast, more than 50% of the penetration sites were type 1 and type 2 in the ΔMoarl1 mutant (Fig. 2C). To determine the possible reasons for the reduced pathogenicity of the ΔMocin4 mutant, mycelia of the Guy11, ΔMocin4, and ΔMocin4/MoCIN4 strains were inoculated onto detached barley leaves. We observed 100 penetration sites and also rated these from type 1 to type 4 (type 1, no penetration; type 2, penetrating peg formed; type 3, two or three invasive hyphae; type 4, more than three invasive hyphae) at 36 hpi. More than 45% of the penetration sites displayed type 3 and type 4 in the Guy11 and ΔMocin4/MoCIN4 strains, whereas more than 90% of the appressorium-like structures could not penetrate the barley cells in the ΔMocin4 mutant (Fig. 2C). These results demonstrated that MoArl1 and MoCin4 are involved in both penetration and colonization of the host cells.
As appressorium-mediated host penetration requires strong turgor pressure, we examined the turgor of appressorium or appressorium-like structures using a cytorrhysis assay (44). The results showed that the ΔMoarl1 and ΔMocin4 mutants displayed a higher collapse rate than strain Guy11 and the complemented strains, indicating that MoArl1 is required for the normal turgor of appressorium and MoCin4 for that of appressorium-like structures (Fig. 2D). The septin (Sep) ring is required for turgor generation in M. oryzae (45), we therefore expressed Sep5-GFP (Sep5-green fluorescent protein) in ΔMoarl1 and ΔMocin4 mutants. The appressorium showed a Sep5-GFP ring in Guy11 but a disorganized mass in the ΔMoarl1 mutant; however, the appressoriumlike structures did not exhibit a Sep5-GFP ring in either Guy11 or the ΔMocin4 mutant (Fig. 2E). Additionally, we also found that the ΔMogga1 mutant showed reduced pathogenicity compared with Guy11 and the complemented strain (Fig. 2F).
MoArl1 and MoCin4 are required for normal vesicle trafficking. To investigate the roles of MoArl1 and MoCin4 in vesicle trafficking, we examined the uptake of endocytic tracer FM4-64 dye. FM4-64 was internalized by Guy11 and the complemented strains after 1 min of incubation, but no definitive dye staining pattern was observed in the ΔMoarl1 and ΔMocin4 mutants. After 5 min for the ΔMoarl1 mutant and 15 min for the ΔMocin4 mutant, uptake of the stains was seen, and until 10 min for the ΔMoarl1 mutant and 30 min for the ΔMocin4 mutant, the level of staining was comparable to that seen with Guy11 ( Fig. 3A and B). This result suggested that MoArl1 and MoCin4 are required for endocytic uptake of FM4-64.
A recent study showed that vesicle trafficking is required for maintaining the plant-fungus interface and effector secretion (46); therefore, we investigated the roles of MoArl1 and MoCin4 in those processes. Guy11, ΔMoarl1, and ΔMocin4 strains were transformed with GFP-labeled Avr-Pia and AvrPiz-t, which preferentially accumulated in Bar, 10 m. DIC, differential inference contrast. (F) Pathogenicity and penetration assays for ΔMogga1 mutant. The criteria of the classification were the same as those described for ΔMoarl1. Error bars represent standard deviations of results from three replicates. a biotrophic interfacial complex (BIC) and translocated to the plant cell cytoplasm (47). We found that both Avr-Pia and AvrPiz-t accumulated in BIC in the rice cells infected by the Guy11 and ΔMoarl1 strains (Fig. 3C). In barley cells, we also observed that Avr-Pia and AvrPiz-t accumulated in BIC in Guy11 but that more than 80% Ϯ 5.0% of ΔMocin4infected cells showed no observable BIC and Avr-Pia and AvrPiz-t were not detected (Fig. 3D). These results indicated that MoCin4 is required for BIC formation and for normal effector deployment.
MoCin4 is involved in the scavenging of reactive oxygen species (ROS). Plants protect themselves against pathogens by evolving ROS, while pathogens evolve effector and antioxidation systems to neutralize ROS (48)(49)(50)(51). Since MoArl1 and MoCin4 function in vesicle trafficking and are required for pathogenicity, we measured the level of host ROS production using 3,3=-diaminobenzidine (DAB). We found that 12% of the rice cells infected by the ΔMoarl1 mutant stained brown, similarly to Guy11 and the complemented strain ( Fig. 4A and B). However, 58% of barley cells infected by the ΔMocin4 mutant stained brown compared to 18% and 17% of those infected by the Guy11 strain and the complemented strain, respectively ( Fig. 4C and D).
MoArl1 is localized to the cytoplasm and the Golgi, whereas MoCin4 is restricted to the cytoplasm. To detect the localizations of MoArl1 and MoCin4, the native promoter of MoARL1 and the entire MoARL1 gene were fused with the green fluorescent protein (GFP). For MoCin4, since we could not detect any GFP (data not shown) using its native promoter, we opted to use the strong constitutively activated ribosomal protein 27 (RP27) promoter (52). MoArl1 was distributed throughout the cytoplasm but it also appeared as green fluorescence punctate. Since S. cerevisiae Arl1 To do this, we introduced the Golgi marker protein MoSft2-RFP (MoSft2-red fluorescent protein) (36,55) and found that the green fluorescence punctate indeed colocalized with MoSft2-RFP in conidia, germ tubes, appressoria, and the vegetative and invasive hyphae (Fig. 5). To quantify the efficiency of colocalization, the images were subjected to Pearson's colocalization analysis, which yielded the values of 0.39 Ϯ 0.02 and 0.38 Ϯ 0.01, respectively, in conidia and hyphae. MoCin4 appeared to be distributed throughout the cytoplasm (Fig. S4). The expression levels of MoArl1 and MoCin4 were verified by Western blotting (Fig. S5A and B).
Localization of MoArl1 is nucleotide dependent. To further study the conserved GTP/GDP binding motifs for localization of MoArl1, we observed the localization pattern of three point-mutated alleles, namely, MoArl1 T31N -GFP, MoArl1 Q71L -GFP, and MoArl1 N126I -GFP, with results that showed dominant-negative, constitutively active, and allele-altering nucleotide exchange rates of MoArl1, respectively (53,56). We found  that MoArl1 Q71L -GFP colocalized with MoSft2-RFP with weak cytosolic distributions in hyphae and conidia and that the Pearson's colocalization values were 0.71 Ϯ 0.06 and 0.69 Ϯ 0.04, compared with 0.38 Ϯ 0.02 and 0.40 Ϯ 0.03 for MoArl1 and MoSft2, respectively ( Fig. 6A and B). In contrast, we found that MoArl1 T31N -GFP and MoArl1 N126I -GFP lost the punctate signal distribution pattern and that the Pearson's values were 0.03 Ϯ 0.01/0.05 Ϯ 0.02 and 0.08 Ϯ 0.03/0.04 Ϯ 0.02, respectively ( Fig. 6A and B). The results reported above implied a recycled model for MoArl1 in which the GTP-bound form is associated with the Golgi and disassociates from the Golgi upon hydrolyzation into the GDP-bound form. The latter reassociates with the Golgi while being activated to the GTP-bound form (Fig. 6C).
The N-myristoylation motif is essential for function and Golgi localization of MoArl1. In addition to the GTP/GDP binding motif, MoArl1 also has an N-myristoylation motif. We mutated the conserved glycine to alanine (MoArl1 G2A ) at the myristoyl acceptor site. The MoArl1 G2A mutant exhibited a moderate level of growth between those seen with the WT and ΔMoarl1 strains (  respectively (Fig. 8C). The level of expression of the MoArl1 G2A allele was verified by Western blotting (Fig. S5A and S6D). These results suggested an essential role of the N-myristoylation motif in the Golgi localization and function of MoArl1.
MoGga1 interacts with both MoArl1 and MoArf1 in the Golgi. The ScArl1 yeast regulates three pathways in the Golgi, including the transport of ScGas1 to the plasma membrane, the targeting of ScImh1 with the Golgi, and the recruitment of ScGga to the Golgi (57,58). Among them, Gga cooperates with ScArl1 Q71L to favor its interactions with downstream proteins during vesicle transport (26). We sought to investigate whether such functional relationships exist in M. oryzae by performing the yeast two-hybrid (Y2H) assay. Since previous studies showed that full-length ARF sequences contain membrane binding domains that may interfere with Y2H (53, 59, 60), a truncated form, lacking the first 17 N-terminal hydrophobic amino acids, was used instead. Y2H revealed an interaction between MoArl1 Q71LΔ17N and MoGga1. In addition, we tested the interactions of the other six truncated GTP-bound MoArf proteins with MoGga1 and showed that MoArf1 Q71LΔ17N , but not truncated GDP-bound MoArl1 T31NΔ17N and MoArf1 T31NΔ17N , also interacted with MoGga1 ( Fig. 9A and B).
To examine whether the interactions also exist in vivo, we employed the bimolecular fluorescence complementation (BiFC) assay. The transformants coexpressing MoGga1-YFP N (N-terminal MoGga1-yellow fluorescent protein) and MoArl1-YFP C exhibited punctate yellow signals in conidia, germ tubes, appressoria, and vegetative and invasive hyphae (Fig. 9C). The similar punctate signals were also observed in the transformants coexpressing MoGga1-YFP N and MoArf1-YFP C (Fig. 9D). The interactions between MoGga1 and MoArl1 and between MoGga1 and MoArf1 were further validated by the in vivo coimmunoprecipitation (Co-IP) assay, and the results indicated that both MoArl1 and MoArf1 interact with MoGga1 and that MoArl1 also interacts with MoArf1 in vivo (Fig. 9E). We also tested the interactions among all 7 Arf proteins; however, we did not find interactions within the Arf family members (Fig. S7), with the exception of the interaction between MoArl1 and MoArf1 described above.
We further observed the localization of MoGga1 and found that it also appears as green punctate, in a manner similar to that seen with the GTP-bound MoArl1, and that most of the GFP signals colocalized with MoSft2-RFP. The Pearson's values were 0.48 Ϯ 0.06 and 0.45 Ϯ 0.03 in conidia and hyphae, respectively (Fig. S8A). Additionally, we observed that MoArf1 partially localized to the Golgi in conidia and hyphae and that the Pearson's values were 0.35 Ϯ 0.02 and 0.36 Ϯ 0.03, respectively (Fig. S8B). The results reported above led us to investigate whether the yellow punctate fluorescence seen with the interaction (Fig. 9) represents Golgi structures. We observed their colocalization with MoSft2-RFP and found that they colocalized with each other, with Pearson's values of 0.42 Ϯ 0.05 and 0.41 Ϯ 0.03 for the strains coexpressing MoGga1-YFP N , MoArl1-YFP C , and MoSft2-RFP and 0.43 Ϯ 0.04 and 0.44 Ϯ 0.02 for the strains coexpressing MoGga1-YFP N , MoArf1-YFP C , and MoSft2-RFP strains in conidia and hyphae, respectively ( Fig. S8C and D). These results suggested that MoGga1 interacts with MoArl1 and MoArf1 in the Golgi. The localization and function of MoGga1 are dependent on its interaction with MoArf1 and MoArl1. To examine how MoGga1 is recruited to the Golgi, we first observed the localization of MoGga1 in the ΔMoarl1 mutant and found no significant differences in punctate signals between the mutant and Guy11 (Fig. 10A). Since a ΔMoarf1 mutant was Arf GTPase Family in M. oryzae not available, we treated the WT Guy11 strain containing MoGga1-GFP with brefeldin A (BFA), which inhibits the exchange of GDP to GTP for Arf proteins (61). The punctate GFP signals were significantly reduced following treatment (Fig. 10B), suggesting that localization of MoGga1 is dependent on MoArf proteins.
We found that MoGga1 contains conserved VHS, GAT, and GAE domains (Fig. 10C). The GAT domain was previously identified as an Arf-binding domain, and point mutations in the GAT domain were shown to affect its interaction with Arf proteins (62)(63)(64). We identified isoleucine in MoGga1 at position 208 (Fig. 10C), which corresponds to the critical residue present in other Gga proteins (64). We then changed the isoleucine 208 residue to asparagine and found that MoGga1 I208N failed to interact with MoArf1 Q71L and MoArl1 Q71L (Fig. 10D and E). Meanwhile, we also found that MoGga1 GAT (GAT domain of MoGga1 alone) still interacted with MoArf1 Q71L and MoArl1 Q71L (Fig. 10D and E).
To test the functions of these interactions, we fused MoGga1 I208N and MoGga1 GAT with a GFP tag before the transformation of the ΔMogga1 mutant. We obtained strains MoGga1 I208N and MoGga1 GAT (verified by Western blotting; Fig. S5C) and found that the MoGga1 I208N strain showed dramatically decreased levels of punctate GFP signals and that the MoGga1 GAT strain exhibited homogeneous GFP signals throughout the cytoplasm (Fig. 10F). The effect of point and domain mutations was further determined by infection tests. The MoGga1 I208N strain resulted in some lesions, which were less extensive than those seen with Guy11 but more extensive than those seen with the ΔMogga1 mutant. The MoGga1 GAT strain caused lesions that were similar in extent to those seen with the ΔMogga1 mutant ( Fig. 10G and H). These results indicated that interactions between MoGga1 and MoArf1/MoArl1 are required but not sufficient for the localization and the function of MoGga1.

DISCUSSION
The Arf GTPase family proteins are involved in many cellular processes, including vesicle trafficking, cytoskeletal organization, signaling transduction, and organelle maintenance in diverse organisms ranging from yeast to animals (11,28). However, the functions of Arf GTPases in filamentous plant pathogens remain poorly understood. We found that Arf proteins of M. oryzae have specific as well as shared functions governing the growth, development, and pathogenicity of the fungus. Specifically, MoArf6 and MoCin4 are involved in growth and conidiation. MoArl1 and MoArl3 are positive regulators of vegetative growth, whereas MoArl8 is dispensable for most of the functions tested. Additionally, experiments that used a strategy employing a conditional promoter led to the conclusion that MoArf1 and MoSar1 are also important for growth.
We previously found that ArfGAP protein MoGlo3, mediating endocytosis and vesicle trafficking, regulates growth, conidiation, and pathogenesis in the blast fungus (36). Intriguingly, interactions between MoGlo3 and all MoArf proteins cannot be established by the Y2H assay (data not shown). Because M. oryzae is a fungal pathogen, we focused our research efforts on pathogenicity and found that MoArl1 and MoCin4 are required for the full virulence of the fungus. The ΔMoarl1 and ΔMocin4 mutants showed reduced virulence due to the defect in appressorial penetration and invasive hyphal growth. But the mechanisms responsible for the reduced pathogenicity of the two mutants also showed some differences; ΔMoarl1 mutants exhibited lower appressorial turgor levels and disorganized septin rings, while ΔMocin4 mutants exhibited lower appressorial turgor levels and defects in scavenging of ROS. A recent study in C. albicans revealed a novel role of CaArl1 in virulence (17), and our characterization of MoArl1 is in accordance with that finding. In addition, since MoCin4 is homologous to yeast ScCin4 and human HsArl2, which regulate normal microtubule stability (65) and mitochondrial and microtubule morphology (66,67), respectively, our identification of MoCin4 from M. oryzae may represent the first study result indicating that modulation of such a protein can also impact virulence.
MoArl1 is localized to the Golgi and the cytoplasm, consistent with previously performed studies of other systems (53,62). Our results also revealed that MoArl1 Q71L -GFP localizes to the Golgi more efficiently than MoArl1-GFP does and that MoArl1 T31N -GFP and MoArl1 N126I -GFP are cytosolic. The MoArl1 Q71L form is likely to be locked in the constitutively activated GTP-bound state, whereas the MoArl1 T31N and MoArl1 N126I forms likely maintain the GDP-bound state (53,56). We demonstrated that the localization of MoArl1 in the Golgi and the cytoplasm is nucleotide dependent and could be regulated by GEFs and GAPs (12,13). Moreover, we also showed that MoArl1 G2A -GFP is distributed throughout the cytoplasm. Together with the incomplete recovery of these point-mutated isoforms for suppressing the defect in growth and pathogenicity of the ΔMoarl1 mutant, we propose that the cycled localization of MoArl1 between the Golgi and the cytoplasm is essential for its normal function.
The Golgi functions as the central hub in the conventional secretory pathway of fungi that sorts protein cargos to the plasma membrane, extracellular cells, or the recycled system (30,68,69). To further elucidate the functions of MoArl1 with respect to the Golgi, we identified the sole Gga protein, MoGga1, as a MoArl1 Q71L -interacting protein in M. oryzae. There are some contradictions between our results and those obtained with this interaction in other systems; one study showed that GTP-bound ScArl1 could not interact with ScGga1 or ScGga2 (63), whereas other studies revealed that ScArl1 and ScGga2 may interact with each other, either directly or indirectly, and that ScGga2 functions as a monomeric adaptor protein of ScArl1 in clathrin coat formation (26,70). Our results demonstrated that MoArl1 Q71L , but not MoArl1 T31N , directly interacts with MoGga1. On the basis of the cytoplasm localization of MoArl1 T31N , the Golgi localization of MoArl1 Q71L , and interaction of MoArl1 Q71L with MoGga1 on the Golgi, we concluded that MoArl1 is localized to the Golgi, where it interacts with MoGga1 for function.
The localization of ScGga2 was previously shown to exhibit a slight change in the ΔScarl1 mutant in S. cerevisiae (26). Our results showed that MoArl1 does not affect the localization of MoGga1. Again, previous studies have indicated a central role of human HsArf1 in the recruitment of HsGga1, HsGga2, and HsGga3 (25,71). On the basis of the finding that both MoArl1 Q71L and MoArf1 Q71L interacted with MoGga1 and that the levels of MoGga1-GFP punctate signals were markedly reduced following treatment with BFA, we hypothesized that MoArl1 and MoArf1 cooperate to interact and recruit MoGga1 to the Golgi.
Human HsGga proteins must interact with HsArf proteins for proper localization and function, whereas these interactions play a minor role in the Golgi localization and in the function of ScGga proteins in S. cerevisiae (27,64). This raises the issue of how this is different in M. oryzae. Point mutation of the common binding area for MoArf1 and MoArl1 in MoGga1 abrogated its interaction with both proteins and dramatically affected its localization, resulting in an incomplete recovery of the pathogenicity defect of the ΔMogga1 mutant. These results indicated the importance of the interactions between MoGga1 with MoArf1 and MoArl1 for the proper localization and function of MoGga1. Taking the data together, we have demonstrated that MoGga1, whose localization and function depend on its interaction with MoArf1 and MoArl1, acts as the common adaptor of MoArf1 and MoArl1. The growing identification of specific and common proteins interacting with Arf proteins indicated that Arf GTPase family members do not only work alone but also can cross talk with each other (53,72,73).
In summary, we have characterized seven Arf GTPase family members that regulate growth, development, and pathogenicity in M. oryzae and we have also identified MoGga1 as a sole common adaptor protein for MoArf1 and MoArl1. Further investigations of MoArf proteins and their additional interacting partners are warranted to elucidate the dynamic and multiple networks of this important group of small GTPase proteins in M. oryzae.

MATERIALS AND METHODS
Strains and culture conditions. M. oryzae Guy11 was used as the wild-type strain in this study. All strains were cultured on CM agar plates in the dark at 28°C, unless indicated otherwise. The strains were incubated in liquid CM for 2 days in darkness for extraction of DNA, RNA, and protein.
Arf GTPase Family in M. oryzae ® Phylogenetic tree construction and sequence alignment. All of the Arf proteins of M. oryzae, F. graminearum, Z. tritici, A. nidulans, N. crassa, C. albicans, and S. cerevisiae were obtained from the NCBI database (https://www.ncbi.nlm.nih.gov/) or from FungiDB (http://fungidb.org/fungidb/). The phylogenetic tree was constructed using MEGA 5.05 programs with 1,000 bootstrap replicates and the neighborjoining method. The alignment of MoArf proteins was performed by the use of CLUSTAL_W programs.
Growth, conidiation, and turgor assays. For vegetative growth, small blocks of strains were cultured on the plates of CM, OM, MM, and SDC for 7 days and then measured and analyzed (74). For the growth of the CPR mutant, the NaNO 3 of MM was replaced by 460 mM NaGlu. For conidiation assay, the strains were cultured on SDC in the dark for 7 days, followed by 3 days of constant illumination under fluorescent light, and then the conidia were collected and analyzed (75,76). The turgor assays for appressorium or appressorium-like structures were as described previously (77,78).
S. cerevisiae ⌬Scarf mutant complementation. The full-length cDNAs of MoArf proteins, which were amplified using primers (see Table S2 in the supplemental material), were cloned into pYES2 vector using the GAL1 promoter, induced by galactose treatment, and repressed by glucose treatment. After sequencing, the fused constructs were transformed into the corresponding S. cerevisiae ΔScarf mutants (BY4741 mutants ΔYDL192W, ΔYOR094W, ΔYBL164C, ΔYPL051W, and ΔYMR138W). Putative transformants were selected on Sabouraud dextrose (SD) medium lacking uracil. For complementation assays, yeast strains were cultured in liquid yeast extract-peptone-dextrose (YPD) overnight and were diluted to an optical density at 600 nm (OD 600 ) of 0.1, and then 5-l volumes of 10-fold serial dilutions were grown on SD-Met-Leu-His-Ura (galactose) plates with or without HygB. The S. cerevisiae ΔScarf mutants and BY4741-expressed pYES2 strains were controls.
Reverse transcription-PCR (RT-PCR), quantitative RT-PCR, and gene expression analysis. For RT-PCR, RNA was reverse transcribed into first-strand cDNA with a reverse transcription kit (Vazyme). The correction of the putative MoARF gene model was performed with the primers (Table S2). The qRT-PCR was performed with an Applied Biosystems 7500 real-time PCR system as described previously (79). The relative quantification transcriptional levels of all MoARF genes were normalized to that of ACTIN (MGG_03982).
Gene deletion, complementation, and amino acid substitution. The gene deletion mutants were generated by the standard one-step gene replacement strategy as previously described (80). The CPR deletion mutants were generated as described previously in Z. tritici and A. fumigatus (42,43). Two approximately 1.0-kb DNA fragments flanking the promoter of MoARF1 or MoSAR1 and the promoter of MoNIA were amplified using the primer pairs (Table S2). The downstream flanking sequence and the promoter of MoNIA were further amplified by overlap PCR. The upstream flanking sequence and overlapped PCR products were digested by restriction endonucleases and ligated with the same enzymes used to cleave pCX62 vector, respectively. The verified plasmids were transformed into Guy11, and putative mutants were screened by PCR and further confirmed by Southern blotting.
For complementation and amino acid substitution experiments, the complement fragments, which contained the related genes and their 1.5-kb native promoters, were amplified with primer pairs (Table S2) and cotransformed with XhoI-digested pYF11 vector (bleomycin resistance) into the yeast XK1-25 strain (52) and then transformed into the Escherichia coli DH5␣ strain for further amplification. After sequencing, the fused-pYF11 plasmids were transformed into the related mutant for the corresponding complementation (47).
Pathogenicity assays. Equal volumes of conidial suspensions (5 ϫ 10 4 conidia/ml) with 0.2% (wt/vol) gelatin were inoculated on rice seedlings (Oryza sativa cv. CO39) or detached barleys. Inoculated plants were kept in the dark under conditions of 90% humidity for the first 24 h and then subjected to light/dark cycles for 5 to 7 days (81). For analysis of the pathogenicity of mutants with fewer conidia, mycelia cultured in liquid CM for 2 to 4 days were washed and inoculated onto the detached barleys or rice leaves as described previously (78,82). For analysis of the pathogenicity of CPR mutants, the conidial suspensions were supplemented with 460 mM NaGlu for the rice injection sheath assay (47,83). For the microscopic observation of penetration and invasive hyphae in plant tissues, conidia or mycelia were infected with rice sheaths or barley leaves and plant cells were microscopically observed after 36 h (barley leaves) or 48 h (rice sheaths) of inoculation.
Endocytosis and secretion assays. For endocytosis, the strains were cultured in liquid CM for 24 h and then stained with FM4-64 for several minutes. For secretion, the conidia or mycelia were used to infect rice sheath or barleys and then BICs and the localization of Avr-Pia and AvrPiz-t in the infected cells were observed. DAB staining. The infected rice sheaths or barley were incubated with 1 mg/ml DAB for 8 h, and the stained cells were subjected to washing in an ethanol/acetic acid solution (47:1 [vol/vol]) for 4 h and then observed.
Subcellular localization observation. To observe the subcellular localization, all of the proteins were fused on a GFP tag with their native promoters, with the exception of MoCin4, which showed no GFP signal; thus, the RP27 promoter was instead. All of the images were observed using confocal fluorescence microscopy (Zeiss LSM710 laser scanning microscope; 63ϫ oil).
Yeast two-hybrid assays. The truncated point-mutated MoArf proteins were cloned into pGADT7 as the prey constructs and MoGga1 was cloned into pGBKT7 as the bait construct using primers (Table S2). After sequencing, the prey and bait constructs were transformed into yeast strain AH109 in pairs. The Trp-positive (Trp ϩ ) and Leu ϩ transformants were isolated and assayed for growth on SD-Trp-Leu-His-Ade medium and for expression of the LacZ reporter gene (84).
Bimolecular fluorescence complementation (BiFC) assays. The MoGGA1-YFP N plasmid was constructed by cloning the MoGGA1 gene with its native promoter into pHZ65 vector. Similarly, the MoARL1-YFP C and MoARF1-YFP C plasmids were constructed by cloning the corresponding gene into pHZ68 vector. Construct pairs (the MoGGA1-YFP N and MoARL1-YFP C pair and the MoGGA1-YFP N and MoARF1-YFP C pair) were transformed into the protoplasts of Guy11. The transformants were selected by the use of both hygromycin and bleomycin and were then observed by confocal fluorescence microscopy (Zeiss LSM710 laser scanning microscope; 63ϫ oil).
Coimmunoprecipitation (Co-IP) assay. The MoARF and MoGGA1 genes with their corresponding native promoters were cloned into both pXY203 (S tag) vector and pYF11 (GFP tag) vector with primers (Table S2). After sequencing, construct pairs were transformed into the protoplasts of Guy11. The total proteins were extracted from transformants coexpressing the above two fusion constructs and incubated with anti-GFP beads (Abmart). After three washes, the elution of the proteins bound to anti-GFP beads was analyzed by Western blotting with anti-GFP (Abmart) (1:5,000) and anti-S (Abcam) (1:5,000) antibodies (85).
Brefeldin A (BFA) treatment. The harvested conidia were treated with 5 g/ml BFA (Sigma) dissolved in dimethyl sulfoxide (DMSO) for 2 min, and the DMSO treatment acted as a control. The treated conidia were then observed using confocal fluorescence microscopy (Zeiss LSM710 laser scanning microscope; 63ϫ oil).