The strain Guy11 (isolated from Guyana, which has relatively higher virulent) was used as a wild-type in this work. The Guy11, mutants and complementary strains were cultured at 28°C in CM, MM, and SDC media for 7 days, respectively. Fungal mycelia were harvested after growth in liquid CM media with shaking for 48 h at 28°C and were used for DNA and RNA extraction. Fungal protoplasts were prepared and transformed as described previously (Sweigard et al., 1992). Transformants were selected on TB3 media with hygromycin B or zeocin (Yang et al., 2019). For conidiation statistics, mycelial blocks were cultured on SDC media for 7 days in the dark and then moved to constant illumination under fluorescent light for 3 days (Zhang et al., 2009).

The approximately 1.0 kb up- and downstream- sequences of MoPDX1 (MGG_05980) were separately amplified from M. oryzae genomic DNA with the primers shown in Table S1 . They were ligated to the pCX62 vectors in two steps to generate the MoPDX1 deletion vectors, which were confirmed by sequencing. Then, the deletion vectors were transformed into the Guy11 protoplasts, which were selected on TB3 media supplemented with hygromycin B. Candidate transformants were verified with inner and outer primers (in Table S1 ) of MoPDX1 through PCR and further verified by Southern blot analysis ( Figure S1 ). For complementary strains, fragments including the full-length coding region without the termination code and approximately 1.5 kb native promoters were cloned into the pYF11 vectors, which express green fluorescence protein (GFP), and sequenced. Then, the complementary vectors were transformed into the MoPDX1 deletion mutants and selected by zeocin-resistance. Candidate strains were verified by GFP screening and Western blot assays.

To test the function of vitamins in the development of M. oryzae, hyphal blocks of Guy11, two ΔMopdx1 mutants, and complementary strains were cultured on MM media supplemented with 10^-4 mg/mL of VB1 (Vitamin B1, Cat#V8020, Solarbio), VB2 (Riboflavin, Cat#V8060, Solarbio), VB3 (Nicotinic acid, Cat#N8060, Solarbio), VB6 (Pyridoxine HCl, Cat#V8030, Solarbio), VH (D-biotin, Lot F801BA0013, Sangon Biotech, Shanghai, China), or PABA(4-aminobenzoic acid, Lot F906BA0033, Sangon Biotech, Shanghai, China) alone or lacking only VB1, VB3, and VB6, at 28°C for 7 days in the dark. The hyphal blocks of the indicated strains cultured on MM media were used as controls. For the invasive growth complementation assay, 10^-4 mg/mL of VB6 were added in the conidial suspensions of the MoPDX1 deletion mutants, then injected into the 21-d rice sheath cells, results were observed at 24 and 48 hours post inoculation (hpi), respectively. For the virulence complementation assay, 25 μL of conidial suspensions (5×10⁴ spores/mL) of mutants supplemented with 10^-4 mg/mL VB6 were inoculated on the surface of 7-day barely leaves. For the appressorium turgor complementation assay, 10^-4 mg/mL of VB6 were added in the conidial suspensions of indicated strains, 25 μL of them were cultured on the hydrophobic slides (Fisherbrand, 12540A, Germany), different concentrations of 1-4 M glycerol were used to treat the appressoria for 3 min to detect the their incipient cytorrhysis. At least 100 appressoria were counted at 14 and 24 hpi for each experiments, respectively.

The indicated strains were reactivated from PDA media and cultured on CM media at 28°C for 5 d in the dark. Then, the samples were shaken in liquid CM media for 48 h, and the mycelium pellets were collected and quickly ground into powders in liquid nitrogen. The material extraction method was described in the previous study (Liu et al., 2016a). Intracellular VB6 in different samples were detected by HPLC-MS (Agilent 6460, USA) analysis. The steps were as follows: at 0~5 min, the C18 column (2.1× 50 mm, 2.7 μm; Agilent, USA) was rinsed with 100% water (pH 3.0-4.0) containing 5 mM ammonium acetate; at 6~7 min, the C18 column was rinsed with 100% acetonitrile; and at 8~15 min, the C18 column was rinsed again with 100% water (pH 3.0-4.0) containing 5 mM ammonium acetate. The flow rate was 0.2 mL per minute, and the temperature of the C18 column was 35°C. The peak flow of the VB6 standard was at approximately1.25 min. The intracellular VB6 content in different samples was calculated according to a previously described method (Liu et al., 2016a).

For appressorial formation, spores of the indicated strains were collected with a double filter cloth, and 20 μL of a 5×10⁴ spores/mL suspensions were dripped onto hydrophobic slides. At least 100 appressoria were counted at 24 hpi. For turgor pressure, 1-4 M glycerol was used to treat the appressoria for 3 min at 14 and 24 hpi, and at least 100 appressoria were counted to observe their collapse rate. For rice spray assays, 5×10⁴ spores/mL conidia suspensions of the indicated strains were sprayed onto the surface of 11-d rice CO-39 (a susceptible variety to rice blast) seedlings. After incubation in a moist chamber at 28°C for 24 h in the dark, the samples were transferred into another moist chamber at 28°C for 6 days under continuous12 h/12 h light and dark conditions. For rice leaf sheath injection assays, 15×10⁴ spores/mL conidial suspensions were injected into 21-d rice leaf sheaths, and the conditions of disease occurrence were consistent with those of the rice spray assay. The experiments were repeated three times with the same results.

For expression analysis assays of MoPDX1 at different stages of M. oryzae, hyphal blocks of Guy11 strains were cultured in liquid CM media for 48 h at 28°C, conidia of Guy11 were collected in SDC media, 5×10⁴ spores/mL conidial suspensions of Guy11 were sprayed onto 11-d rice seedlings. Samples were collected at 8, 24, 48 and 72 hpi, respectively. RNA was extracted from all samples using an RNA extraction kit (Invitrogen, USA). The concentration of total RNA was measured and then reverse transcribed (R223-01, Vazyme). Quantitative RT–PCR (qRT–PCR) was performed on a Bio-Rad CFX96 instrument (Bio-Rad, USA) according to the manufacturer’s instructions. The expression of MoPDX1 at the mycelium stage (MY) was set at 1.0. To evaluate the expression level of PR genes, 5×10⁴ spores/mL conidial suspensions of Guy11 and mutant strains were singly sprayed onto 11-d rice seedlings, and samples were collected at 48 hpi. Sterile water sprays were used as negative controls. The stable-expression rice gene EF1-α (primers in Table S1 ) was set as an internal control. All experiments were performed for three independent replicates with the same results.

Hyphal blocks of complementary strains were cultured in liquid CM media at 28°C for 48 h, filtered, and ground into powders in liquid nitrogen as soon as possible. Powders were lysed with 1 mL of protein lysis buffer [10 mM Tris/Cl(pH7.5); 150 mM NaCl; 0.5 mM EDTA; 0.5% NP-40] and 100 mM protease inhibitor PMSF (Cat No. ST506, Beyotime), shaken every 10 min, and then centrifuged at 1,2000 rpm at 4°C for 20 min, the supernatant was used as the total protein. Total protein was co-incubated with commercial GFP beads (Lot 91009001A-04, Ychromotek) at 4°C overnight. Then, the beads were centrifuged at 500 rpm at 4°C for 2 min and washed five times with washing buffer [10 mM Tris/Cl(pH7.5); 150 mM NaCl; 0.5 mM EDTA]. The proteins were eluted within the elution buffer (200 mM glycine, pH2.5), which were neutralized with 1M Tris-base (pH10.4). The candidate interacting proteins of MoPdx1 were further analyzed through mass spectrography (the instrument was LTQ Orbitrap Velos) at Shenzhen BGI Co., LTD, China. The single GFP interaction proteins were used as controls.

Identified peptides were blasted in the protein database of M. oryzae (https://fungidb.org/fungidb/showQuestion.do?questionFullName=UniversalQuestions.UnifiedBlast). The proteins were analyzed in the GO (Gene Ontology), KOG (Eukaryotic orthologous groups), and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway databases, respectively.

One-way ANOVA analysis with IPM SPSS Statistics Software was used to conduct variance analysis. LSD test was used to make multiple comparisons, the mean difference is significant at the 0.05 level in the whole text. Student-Newman-Keuls test was used to display the means for groups in homogeneous subsets. Three replicates were set for each sample, and all experiments were performed for three independent replicates with the same results.

We identified a gene MoPDX1 encoded a pyridoxine biosynthesis protein in M. oryzae. The transcript level of the gene was analyzed at different stages, such as the mycelium stage (MY), conidial stage (CO), and infectious stage at 8, 24, 48, and 72 hpi. Results showed that the gene MoPDX1 was significantly upregulated at the early infection stage, approximately 2.4 folds, 3.8 folds, and 1.7 folds at 8, 24, and 48 hpi compared with that of at the mycelium stage, respectively. However, there were no significant differences at the conidial stage and the infectious stage at 72 hpi ( Figure 1 ). Suggesting that MoPdx1 might play important roles in the early stage of infection of M. oryzae.

To further explore the function of MoPdx1 in M. oryzae, the MoPDX1 gene was replaced with the hygromycin (HPH) with the one-step method. Southern blotting confirmed that two transforments were successfully knocked out and that only one copy of HPH was inserted into the genome of M. oryzae ( Figure S1 ). Furthermore, complementary strains were acquired. Vegetative growth of wild-type Guy11, two MoPDX1 deletion mutants, and complementary strains on CM, MM, and SDC media were measured. The results showed that the vegetative growth of the MoPDX1 mutants on the three kinds of media all significantly slowed compared to that of Guy11 and complementary strains, especially on MM media, on which the mutants nearly stopped growing ( Figures 2A, B ). Given that MoPdx1 encoding a protein involved in the biosynthesis of pyridoxine, one of VB6 compounds, and there are only basic trace substances in MM media, we want to determine whether vegetative growth defects of the MoPDX1 mutants could be mitigated by the addition of various vitamin B elements, including VB1, VB2, VB3, VB6, VH, and PABA, which are components of CM media. The results showed that only the exogenous addition of VB6 could entirely rescue the growth defects of two MoPDX1 deletion mutants, while addition of VB1, VB2, VB3, VH, and PABA individually or together could not ( Figure 3 and Table 1 ). Moreover, the VB6 content in different strains in CM media was measured with HPLC-MS and it was found that there was no significant difference between the WT and MoPDX1 deletion mutants ( Figure S2 and Table S2 ). Indicating that MoPdx1 regulates the vegetative growth via de novo VB6 biosynthesis not in uptake process in M. oryzae.

We further evaluated conidiation, conidial germination, appressorium formation, and turgor pressure in Guy11, two MoPDX1 deletion mutants, and their complementary strains. The results showed that there was no significant difference in conidiation, conidial germination, or appressorium formation between Guy11 and the deletion mutants ( Table 2 ). However, appressorium collapse rate in the two MoPDX1 deletion mutants rose dramatically at 2-4 M glycerol compared with that of in Guy11 and complementary strains, approximately 64.8% under the treatment of 2 M glycerol, 84.2% under the treatment of 3 M glycerol and 91.9% under the treatment of 4 M glycerol; while it was significantly reduced in the deletion mutants at 1 M glycerol ( Table 2 ). The results indicate that the deletion of MoPDX1 affects the appressorium turgor pressure of M. oryzae.

The appressorium is an important structure of M. oryzae which produces penetrating pegs that successfully penetrate into the host cell. Deletion of MoPDX1 affected the appressorium turgor of M. oryzae; therefore, virulence assays were performed. The 5×10⁴ spores/mL of conidial suspensions of Guy11, two MoPDX1 deletion mutants and complementary strains were sprayed on 11-d rice CO-39 seedlings, and disease symptoms were observed at 7 days post inoculation (dpi). The two MoPDX1 deletion mutants presented the most restricted lessons and bits of typical lesions of rice blast compared to that caused by Guy11 and complementary strains ( Figures 4A, C ). In addition, lesion biomass assays of all the diseased leaves suggested that the content of M. oryzae was markedly reduced treated with the two ΔMopdx1 mutants ( Figure 4D ). Moreover, conidial suspensions (15×10⁴ spores/mL) of the different strains was injected into 21-d rice sheath cells and observed at 5 dpi. Lesion incidence was in keeping with that of rice spraying, with lesions reduced by two-thirds in the two ΔMopdx1 mutants ( Figure 4B ). Overall, the results suggested that MoPdx1 is involved in the pathogenicity of M. oryzae.

To further explore the causes of the reduced pathogenicity of the two ΔMopdx1 mutants, the growth of infectious hyphae in 21-d rice sheath cells injected with conidial suspensions of Guy11, two MoPDX1 deletion mutants, or the complementary strains were observed at 24 and 48 hpi, respectively. Four types of invasive hyphae are shown in Figure 5A . In Guy11 and complementary strains, type 1 invasive hyphae accounted for 8%, 30% of cells were type 2, 60% of cells were type 3 and 2% of cells were type 4 at 24 hpi. In the two ΔMopdx1 mutants, 18% of cells were type 1, 56% of cells were type 2 and 24% were type 3 ( Figure 5B ). Most of the cells (approximately 65%) were type 4 and 25% of cells were type 3 in Guy11 and complementary strains at 48 hpi. However, in the two mutant strains, 30% were still type 2, 34% were type 3, and 26% were type 4 ( Figure 5B ). In addition, conidial suspensions of MoPDX1 deletion mutants were added 10^-4 mg/mL of VB6 solutions, and then injected into the rice shealth cells, observed the invasive growth at 24 and 48 hpi. Results found that addition of VB6 could rescue the defects of invasive growth of the mutants whether inoculated 24 or 48 h ( Figure 5B ). The results suggest that de novo VB6 plays an important role in invasive growth of M. oryzae.

Further, conidial suspensions of MoPDX1 deletion mutants with 10^-4 mg/mL VB6 solutions were then inoculated onto the detached 7-d barely leaves, and the lesion incidence was observed at 5 dpi. Results showed that the exogenous addition of commercial VB6 could also rescue the pathogenicity defect of the two ΔMopdx1 mutants on barely leaves ( Figure 6 ). Indicating that VB6 is involved in the virulence of M. oryzae.

In general, plants produce a broad-spectrum defense response against the attack of pathogens, including rapid generation of ROS, and changes in the expression levels of pathogenesis-related (PR) proteins (Bradley et al., 1992; Tanaka et al., 2003). We want to determine whether the reduced pathogenicity of the ΔMopdx1 mutants was due to defense responses activated by the host. 3,3’- diaminobenzidine (DAB) was used to stain the ROS in rice sheath cells after inoculation for 48 h. The results showed that the ROS level was not significantly different caused by WT and the two ΔMopdx1 mutants, and approximately 25% of rice sheath cells were stained with DAB ( Figure 7A ). In addition, the transcript levels of rice defense-related genes PBZ1 and CHT1 (Chen et al., 2014; Liu et al., 2016b) were measured in the rice leaves infected for 48 h by spores of Guy11 and ΔMopdx1 mutants, separately. Water-sprayed rice leaves were used as negative controls. The transcript levels of PBZ1 and CHT1 were also not significantly different between Guy11 and ΔMopdx1 mutants ( Figure 7B ). Results illustrated that deletion of MoPDX1 does not activate plant defense responses.

MoPdx1 co-expressed with a GFP protein to detect its localization at different development stages of M. oryzae, including the mecelium stage, infectious hyphal stages, conidial stage, and appressorium formation stage. The results showed that MoPdx1 was located in the cytoplasm at the mecelium stage, conidial stage, and infectious hyphal stages ( Figures 8A, B, E ). Furthermore, MoPdx1 was present in spore and germ tubes at 14 hpi and then transferred into the appressorium at 24 hpi ( Figures 8C, D ). Suggesting that MoPdx1 is located in the cytoplasm of M. oryzae.

To gain further insight into the function of MoPdx1 in M. oryzae, its interacting proteins were enriched by GFP beads using two strains of Guy11 expressing MoPdx1-GFP. A Guy11 strain with only GFP was used as a control. The results showed that 10 proteins had more than two peptides that potentially interacted with the MoPdx1 protein in both two MoPdx1-GFP expressing strains, except the proteins in only GFP expressing strain ( Table 3 ). Furthermore, two proteins previously studied by our laboratory encoded by MGG_11513 (Yang et al., 2018) and MGG_04719 (Yin et al., 2018) were selected to identify the interaction with MoPdx1 through yeast two-hybrid (Y2H), respectively. Both proteins could physically interact with MoPdx1 on -Trp/-Leu/-His medium under 5 mM 3-AT ( Figure S3 ), conforming the reliability of the Co-IP results. In this work, 146 proteins and 275 peptides were identified also in both two MoPdx1-GFP expressing strains, except the proteins in only GFP expressing strain, and further analyzed the function of these IP proteins, which might elucidate the function of MoPdx1 in M. oryzae. Eukaryotic Orthologous Groups (KOG) is a database for identifying orthologous and paralogous proteins. KOG database showed the IP proteins of MoPdx1 mainly focused on translation, ribosomal structure and biogenesis, general function prediction only, posttranslational modification, protein turnover, chaperones, signal transduction mechanisms, cytoskeleton, energy production and conversion, and RNA processing and modification ( Figure 9A ). The top 10 GO enrichment terms were binding, cellular process, catalytic activity, cell part, cell, metabolic process, organelle, macromolecular complex, organelle part, and structural molecular activity ( Figure 9B ). The top 10 pathways from KEGG pathway enrichment analysis were metabolic pathways, ribosome, biosynthesis of antibiotics, biosynthesis of secondary metabolites, biosynthesis of amino acids, carbon metabolism, protein processing in the endoplasmic reticulum, spliceosome, pyruvate metabolism, and endocytosis ( Table 4 ). Therefore, MoPdx1 takes part in multiple pathways of M. oryzae, especially in the ribosome and in biosynthesis of various substances.