Magnaporthe oryzae Transcription Factor MoBZIP3 Regulates Appressorium Turgor Pressure Formation during Pathogenesis

The devastating fungus Magnaporthe oryzae (M. oryzae) forms a specialized infection structure known as appressorium, which generates enormous turgor, to penetrate the plant cells. However, how M. oryzae regulates the appressorium turgor formation, is not well understood. In this study, we identified MoBZIP3, a bZIP transcription factor that functioned in pathogenesis in M. oryzae. We found that the pathogenicity of the MoBZIP3 knockout strain (Δmobzip3) was significantly reduced, and the defect was restored after re-expression of MoBZIP3, indicating that MoBZIP3 is required for M. oryzae virulence. Further analysis showed that MoBZIP3 functions in utilization of glycogen and lipid droplets for generation of glycerol in appressorium. MoBZIP3 localized in the nucleus and could bind directly to the promoters of the glycerol synthesis-related genes, MoPTH2, MoTGL1 and MoPEX6, and regulate their expression which is critical for glycerol synthesis in the appressorium turgor pressure generation. Furthermore, the critical turgor sensor gene MoSln1 was also down regulated and its subcellular localization was aberrant in Δmobzip3, which leads to a disordered actin assembly in the Δmobzip3 appressorium. Taken together, these results revealed new regulatory functions of the bZIP transcription factor MoBZIP3, in regulating M. oryzae appressorium turgor formation and infection.


Introduction
Rice blast disease caused by the devastating fungus Magnaporthe oryzae (M. oryzae) is the greatest threat to the global rice harvest and leads to serious economic losses worldwide every year [1]. Rice blast are spread by wind or splash dispersal of the conidias to the host surface [2]. After the conidias landing and adhering to the rice leaf cuticle, M. oryzae will undergo extensive developmental changes while building elaborate infection structures, invading plant cells, and finally proliferating inside the host cells [3,4]. The conidias quickly germinate and form a dome-shaped infection structure at the tip called appressorium with great internal turgor, which is required for cuticle penetration [5]. Thus, developing a mature appressorium is an essential prerequisite for M. oryzae to successfully invade the plant cells.
The appressorium initiates and develops at the germ tube tips. It accumulates multiple and high concentrations of metabolites which generates enormous turgor pressure (>eight MPa) [6,7]. Relying on the turgor pressure, a narrow penetration hypha develops from the base of the appressorium and mechanically punches the plant cuticle [8]. Since the differentiation of infection structures is without exogenous nutrients, it underlines the fact that appressorium development is nutrified by compounds transferred from the conidia. 2

of 14
The most abundant solute in appressorium is glycerol and the other polyols, and the germinating conidia utilize intrinsic glycogen and lipid droplets, to provide sufficient energy and materials for appressorial morphogenesis and turgor generation [5,7]. The turgor-sensing histidine-aspartate kinase, Sln1, could enable the appressorium to sense the increasing turgor and then intrigue a set of downstream signals in the appressorium for facilitating penetration of the host cells [9]. Thus, complex enzymatic catalysis, cellular events and signaling are involved in appressorium turgor formation and regulation. Currently, we have not totally understood the regulation mechanism in these processes.
Transcription factors generally include the BZIP proteins, MYB-like proteins, MADSbox proteins, helix-loop-helix proteins, zinc-finger proteins, and homeobox proteins [10]. They could bind to specific promoter sequences to regulate expression of the target genes and function as key regulatory elements in the control of gene expression in development and plant infection in fungi [11]. The BZIP proteins belong to one of the largest transcription factor families and are widely distributed in eukaryotes. Members of this family play a diverse variety of regulatory roles in fungi. For example, the BZIP protein FlbB of Aspergillus nidulans (A. nidulans) and Aspergillus fumigatus (A. fumigatus) functions in asexual development [12,13]. And MetR protein from A. nidulans was reported to be critical in fungal development through regulation of sulfur metabolism [14]. In M. oryzae, over twenty members of BZIP proteins are encoded in the genome, and the members such as MoAP1, MoBZIP13, MoBZIP22, MoMetR, MoBZIP5 and MoMeaB were shown to play critical roles in M. oryzae sporation, appressiorium formation or host infection [15][16][17]. These findings provided new insights for understanding the crucial roles of bZIP proteins in both development and pathogenesis in M. oryzae.
In this study, we transcriptionally analyzed the M. oryzae BZIP genes in strain Y34 which was isolated in a field of the Yunnan province of China [18,19]. A total of 22 BZIP genes were found through BLAST search and of which, the transcripts of MoBZIP3, MoBZIP16, MoBZIP17, MoBZIP20, MoBZIP23, MoBZIP24 were increased in abundance in Y34 at appressorium or host infection stages. Deletion of each of these six BZIP genes showed that only ∆mobzip3 exhibited an obviously weaker virulence than the control, indicating that this protein plays a role in Y34 pathogenesis. Infection analysis showed that ∆mobzip3 was compromised in rice sheath cell penetration. Furthermore, utilization of glycogen and lipid droplets in appressorium was disordered and glycerol accumulation in ∆mobzip3 appressorium was decreased, suggesting that the turgor pressure was compromised. MoBZIP3 could directly bind the promoters of the glycerol synthesis-related genes, such as MoPTH2, MoTGL1 and MoPEX6 and regulated their expression. In addition, MoBZIP3 also functions in the appressorium turgor sensor protein, MoSln1-mediated turgor recognition. The expression of MoSln1 was reduced and the protein subcellular localization was changed in ∆mobzip3. These resulted in an abnormal assembly of actin ring in the specialized cell which is critical for the host penetration of M. oryzae. In conclusion, we demonstrated the biological functions of the bZIP protein in regulation of both glycerol synthesis and MoSln1 mediated-turgor sensing in appressorium.

Identification of BZIP Transcription Factor Genes in Y34
Since the genome of M. oryzae Y34 was sequenced and annotated [19], We searched and blast the BZIP domain in the NCBI Genome Database (www.ncbi.nlm.nih.gov/genome) (accessed on 10 August 2021). A total of 22 BZIP genes were found in Y34 genome. Compared with the annotated BZIP genes in isolates of Guy11 and KJ201, no MoBZIP11 and MoBZIP14 were present in the Y34 genome, but additional two BZIP genes (MoBZIP23 and MoBZIP24) were identified [15,16]. We then analyzed the expression profiles of these BZIP genes in Y34 through quantitative real-time PCR analysis. As shown in Figure 1, MoBZIP1, MoBZIP4, MoBZIP8, MoBZIP9, MoBZIP18 (MoHAC1) and MoBZIP19, showed the highest transcription levels in conidium stage than those of other stages. And in the infection stage, MoBZIP20, MoBZIP23 and MoBZIP24 showed relative higher expressions than in the mycelial or conidium stage. These results suggested that BZIP genes in Y34 were differently expressed and may play potentially various roles at developmental or infection stages.
showed the highest transcription levels in conidium stage than those of other stages. And in the infection stage, MoBZIP3, MoBZIP12 (MoMEAB), MoBZIP13 (MoHAPX), MoBZIP16, MoBZIP17 (MoATFA), MoBZIP20, MoBZIP23 and MoBZIP24 showed relative higher expressions than in the mycelial or conidium stage. These results suggested that BZIP genes in Y34 were differently expressed and may play potentially various roles at developmental or infection stages.

Figure 1.
Expression profiles of BZIP genes in Y34. Y34 conidia (1 × 10 5 conidia/mL) was sprayed on rice leaves. mRNAs were extracted at different infection stages (3,6,12,24,36 and 48 h after spray, respectively). mRNAs from conidia and mycelia were also included in this experiment. cDNAs were then generated using these mRNAs. Quantitative real-time PCR analysis of the expression of BZIP genes was conducted. The experiments were repeated three times with similar results. Error bars represent ±SE of three biological replicates. Primers used in this experiment were listed in Supplemental Table S1.

MoBZIP3 Plays a Role in M. oryzae Infection
To characterize the function of BZIP genes in M. oryzae, we next carried out gene replacement experiments to knockout genes which expressed at the infection stages in M. oryzae. A total of eight genes were revealed to express at a high level at the infection stages ( Figure 1). We successfully knocked out six genes including MoBZIP3, MoBZIP16, MoB-ZIP17, MoBZIP20, MoBZIP23 and MoBZIP24 except for MoBZIP12 and MoBZIP13, though we repeated for several times. All deletion mutants were confirmed by polymerase chain reaction (PCR) using sets of PCR primers (Supplemental Figure S1).
Next, we cultured these deletion mutants in the complete medium (CM) and straw rice bran (SRB) medium. We found that none of the knockout mutants exhibited an obvi-Figure 1. Expression profiles of BZIP genes in Y34. Y34 conidia (1 × 10 5 conidia/mL) was sprayed on rice leaves. mRNAs were extracted at different infection stages (3,6,12,24,36 and 48 h after spray, respectively). mRNAs from conidia and mycelia were also included in this experiment. cDNAs were then generated using these mRNAs. Quantitative real-time PCR analysis of the expression of BZIP genes was conducted. The experiments were repeated three times with similar results. Error bars represent ±SE of three biological replicates. Primers used in this experiment were listed in Supplemental Table S1.

MoBZIP3 Plays a Role in M. oryzae Infection
To characterize the function of BZIP genes in M. oryzae, we next carried out gene replacement experiments to knockout genes which expressed at the infection stages in M. oryzae. A total of eight genes were revealed to express at a high level at the infection stages ( Figure 1). We successfully knocked out six genes including MoBZIP3, MoBZIP16, MoBZIP17, MoBZIP20, MoBZIP23 and MoBZIP24 except for MoBZIP12 and MoBZIP13, though we repeated for several times. All deletion mutants were confirmed by polymerase chain reaction (PCR) using sets of PCR primers (Supplemental Figure S1).
Next, we cultured these deletion mutants in the complete medium (CM) and straw rice bran (SRB) medium. We found that none of the knockout mutants exhibited an obvious change in growth rate when compared to the wild type (WT) (Figure 2A,B; Supplemental Figure S2A). We then analyzed the pathogenesis of these mutants using the same amounts of conidias to infect the rice leaves (Oryza sativa cv. Nipponbare). Rice seedlings were sprayed with conidias of these M. oryzae mutants and only mutant of MoBZIP3 (∆mobzip3) displayed decreased pathogenicity. It developed an obvious de-creased number of lesions at the rice leaves than the other types of M. oryzae. And the complemented strain (∆mobzip3-c) could recovered its pathogenecity ( Figure 2C,D; Supplemental Figure S2B,D). Punch inoculated assay also showed that infection ability of ∆mobzip3 was decreased ( Figure 2E,F; Supplemental Figure S2C,E). In all, these results indicate that MoBZIP3 is required for full pathogenicity of M. oryzae and its deletion reduces virulence.
Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW ous change in growth rate when compared to the wild type (WT) (Figure 2A,B; Su mental Figure S2A). We then analyzed the pathogenesis of these mutants using the amounts of conidias to infect the rice leaves (Oryza sativa cv. Nipponbare). Rice seed were sprayed with conidias of these M. oryzae mutants and only mutant of Mo (Δmobzip3) displayed decreased pathogenicity. It developed an obvious decreased ber of lesions at the rice leaves than the other types of M. oryzae. And the complem strain (Δmobzip3-c) could recovered its pathogenecity ( Figure 2C,D; Supplemental F S2B,D). Punch inoculated assay also showed that infection ability of Δmobzip3 w creased ( Figure 2E,F; Supplemental Figure S2C,E). In all, these results indicate that ZIP3 is required for full pathogenicity of M. oryzae and its deletion reduces virulenc

MoBZIP3 May Involve in Appressorium Turgor Formation
Penetration assays using rice sheath tissues were carried out to investigate how ZIP3 functions in M. oryzae pathogenesis. It has been defined that four types of inv hyphal (IH) could be classified during host infection: (type one, appressorium form and no hyphal penetration; type two, IH with less than two branches; type three, IH more than two branches; type four, IH that fully occupies a plant cell and penetrate neighboring cells) [20]. To further observe this process, we labeled the WT, the Δm and the complemented strains with GFP ( Figure 3A). Penetration assays were then ducted by observing 100 appressorium for each GFP labeled M. oryzae strain and cla ing their IH types at 36 h after infection. We found that in WT the four types of IH 3.3%, 17.7%, 26.5% and 52.5% IH respectively. In contrast, those in the Δmobzip3 42.6%, 35.4%, 14.5%, and 7.5% ( Figure 3B). These results indicated that type 1 IH in zip3 occupies the most part and penetration of Δmobzip3 were delayed at the early tion stages.

MoBZIP3 May Involve in Appressorium Turgor Formation
Penetration assays using rice sheath tissues were carried out to investigate how MoBZIP3 functions in M. oryzae pathogenesis. It has been defined that four types of invasive hyphal (IH) could be classified during host infection: (type one, appressorium formation and no hyphal penetration; type two, IH with less than two branches; type three, IH with more than two branches; type four, IH that fully occupies a plant cell and penetrates into neighboring cells) [20]. To further observe this process, we labeled the WT, the ∆mobzip3 and the complemented strains with GFP ( Figure 3A). Penetration assays were then conducted by observing 100 appressorium for each GFP labeled M. oryzae strain and classifying their IH types at 36 h after infection. We found that in WT the four types of IH were 3.3%, 17.7%, 26.5% and 52.5% IH respectively. In contrast, those in the ∆mobzip3 were 42.6%, 35.4%, 14.5%, and 7.5% ( Figure 3B). These results indicated that type 1 IH in ∆mobzip3 occupies the most part and penetration of ∆mobzip3 were delayed at the early infection stages. The penetration assay above indicated that the appressorium penetration ability was decreased in Δmobzip3. This prompted us to evaluate the turgor pressure inΔmobzip3 appressorium. The incipient collapse assay [21] showed that the collapse rate of appressorium of Δmobzip3 upon glycerol treatment was significantly higher than the WT and the complemented strain ( Figure 3C,D), suggesting that turgor pressure of Δmobzip3 appressorium was decreased compared with that of the WT strain.

MoBZIP3 Functions in Utilization of Glycogen and Lipid Droplets in Appressorium
Glycerol accumulation generated the turgor pressure in the appressorium and production of glycerol was mainly from transfer and utilization of glycogen and lipids in the conidia [22,23]. Thus, to investigate the glycogen distribution and lipid turnover of Δmob-zip3, we stained glycogen and lipids using potassium iodide and Nile red in appressorium, respectively. We found that at eight h after conidia developed on a hydrophobic surface, glycogen was transferred from conidia to appressorium in WT. And then the amount of glycogen in appressorium gradually decreased. In contrast, in Δmobzip3 appressorium, the amount of glycogen did not decrease significantly as compared with WT. Even at twenty four h appressorium, about 50% glycogen was not transferred (Figure 4 A-D). Similarly, the distribution of lipids droplets exhibited the same pattern as glycogen in Δmobzip3, as shown in Figure 4E-H. In all, these results indicated that MoBZIP3 is required for lipid droplets and glycogen utilization in appressorium development. The penetration assay above indicated that the appressorium penetration ability was decreased in ∆mobzip3. This prompted us to evaluate the turgor pressure in∆mobzip3 appressorium. The incipient collapse assay [21] showed that the collapse rate of appressorium of ∆mobzip3 upon glycerol treatment was significantly higher than the WT and the complemented strain ( Figure 3C,D), suggesting that turgor pressure of ∆mobzip3 appressorium was decreased compared with that of the WT strain.

MoBZIP3 Functions in Utilization of Glycogen and Lipid Droplets in Appressorium
Glycerol accumulation generated the turgor pressure in the appressorium and production of glycerol was mainly from transfer and utilization of glycogen and lipids in the conidia [22,23]. Thus, to investigate the glycogen distribution and lipid turnover of ∆mobzip3, we stained glycogen and lipids using potassium iodide and Nile red in appressorium, respectively. We found that at eight h after conidia developed on a hydrophobic surface, glycogen was transferred from conidia to appressorium in WT. And then the amount of glycogen in appressorium gradually decreased. In contrast, in ∆mobzip3 appressorium, the amount of glycogen did not decrease significantly as compared with WT. Even at twenty four h appressorium, about 50% glycogen was not transferred ( Figure 4A-D). Similarly, the distribution of lipids droplets exhibited the same pattern as glycogen in ∆mobzip3, as shown in Figure 4E-H. In all, these results indicated that MoBZIP3 is required for lipid droplets and glycogen utilization in appressorium development. Significant differences of Δmobzip3 compared with the WT strain were estimated by students t test, n ≥ 50 cells, ** p < 0.01. Images showing the distribution of lipid droplets in WT (E) and Δmobzip3 (F) during appressorium development. Samples were stained with Nile red and observed with fluorescence microscopy. The lipid droplets show a red signal fluorescence. Bar = 10 μm. The percentage of lipid droplets in conidia (G) and appressoruin (H) during appressorium development. Error bars represent the standard deviation. Significant differences of Δmobzip3 compared with the WT strain were estimated by Student's t-test, n ≥ 50 cells, ** p < 0.01.

MoBZIP3 May Function as a Working Transcription Factor in Regulating Glycerol Synthesis Genes Expression
We next investigated the localization of MoBZIP3. MoBZIP3-GFP driven by MoBZIP3 native promoter was expressed in the nuclear localization signal (NLS)-mCherry labeled Δmobzip3 mutant. Localization of MoBZIP3 was then observed in the M. oryzae conidia, germ tube, appressorium and invasive hyphal. The results showed that MoBZIP3-GFP could colocalize with NLS-mCherry in the nucleus in these cells ( Figure 5A). We also found that the MoBZIP3 in M. oryzae all contains a conserved basic region leucin zipper domain at the C-terminal. However, it also appears non-conserved at the N-terminal of this protein through alignment of amino acid of MoBZIP3 from Y34, P131 and Guy11 strain (Supplemental Figure S3A). And we further found that the differences in the amino acid at the N-terminal may result in different biochemical activity in the DNA binding. Y34 MoBZIP3 exhibited higher binding activity to A-box DNA fragment than that from Guy11 (Supplemental Figure S3B,C). The site mutation experiment further showed that MoBZIP3 specifically binds to A-box DNA fragment (TACGTA) (Supplemental Figure  S3D). Error bars represent the standard deviation. Significant differences of ∆mobzip3 compared with the WT strain were estimated by Student's t-test, n ≥ 50 cells, ** p < 0.01.

MoBZIP3 May Function as a Working Transcription Factor in Regulating Glycerol Synthesis Genes Expression
We next investigated the localization of MoBZIP3. MoBZIP3-GFP driven by MoBZIP3 native promoter was expressed in the nuclear localization signal (NLS)-mCherry labeled ∆mobzip3 mutant. Localization of MoBZIP3 was then observed in the M. oryzae conidia, germ tube, appressorium and invasive hyphal. The results showed that MoBZIP3-GFP could colocalize with NLS-mCherry in the nucleus in these cells ( Figure 5A). We also found that the MoBZIP3 in M. oryzae all contains a conserved basic region leucin zipper domain at the C-terminal. However, it also appears non-conserved at the N-terminal of this protein through alignment of amino acid of MoBZIP3 from Y34, P131 and Guy11 strain (Supplemental Figure S3A). And we further found that the differences in the amino acid at the N-terminal may result in different biochemical activity in the DNA binding. Y34 MoBZIP3 exhibited higher binding activity to A-box DNA fragment than that from Guy11 (Supplemental Figure S3B,C). The site mutation experiment further showed that MoBZIP3 specifically binds to A-box DNA fragment (TACGTA) (Supplemental Figure S3D). Given that lipid catabolism is a major source of glycerol, and glycogen metab supplies the energy for turgor pressure during appressorium maturation [24,25 MoBZIP3 may function in these processes, we then analyzed expression of several which played critical roles in lipid and glycogen metabolism. These included trehal phosphate (T6P) synthase (TPS1), carnitine acetyl transferase family gene PTH2, iso lyase gene (ICL1), peroxin gene (PEX6) and multifunctional β-oxidation protein (MFP1) [23,[26][27][28]. We found that expression of these genes was significantly down lated in Δmobzip3 appressorium ( Figure 5B).
The results above indicated that MoBZIP3 may function as a transcription fac gene expression regulation. And expression of MoPTH2, MoICL1 and MoPEX6 in zip3 was decreased to 30% at least when compared with WT, indicating that MoBZIP function with other complex proteins or directly to regulate their expressions. Then trophoretic mobility shift assays (EMSA) were performed using the biotin-labeled moters of MoPTH2, MoICL1, MoPEX6 and in vitro expressed MoBZIP3. The r showed that MoBZIP3 could directly bind the promoters of these genes ( Figure 5 addition, the EMSA results further showed that MoBZIP3 specifically binds to the promoters which contains an A-box (Supplemental Figure S3E).

Glycerol Synthesis Is Impaired in Δmobzip3
Glycerol production in appressorium is dependent on complex enzymatic ca and cellular events. β-Oxidation is a critical reaction for metabolization of fatty ac glycerol production. MoPEX6 was previously reported to function in β-Oxidation o acids [23,29] and in our experiments MoPEX6 was down regulated in Δmobzip3. To f explore that MoBZIP3 may function in glycerol metabolism, we validated the expr of MoPEX6 in Δmobzip3. MoPEX6-GFP driven by MoPEX6 native promoter was expr in WT and theΔmobzip3 strain. Fluorescence observation showed that MoPEX6-GF Given that lipid catabolism is a major source of glycerol, and glycogen metabolism supplies the energy for turgor pressure during appressorium maturation [24,25], and MoBZIP3 may function in these processes, we then analyzed expression of several genes which played critical roles in lipid and glycogen metabolism. These included trehalose-6-phosphate (T6P) synthase (TPS1), carnitine acetyl transferase family gene PTH2, isocitrate lyase gene (ICL1), peroxin gene (PEX6) and multifunctional β-oxidation protein gene (MFP1) [23,[26][27][28]. We found that expression of these genes was significantly down regulated in ∆mobzip3 appressorium ( Figure 5B).
The results above indicated that MoBZIP3 may function as a transcription factor in gene expression regulation. And expression of MoPTH2, MoICL1 and MoPEX6 in ∆mobzip3 was decreased to 30% at least when compared with WT, indicating that MoBZIP3 may function with other complex proteins or directly to regulate their expressions. Then electrophoretic mobility shift assays (EMSA) were performed using the biotin-labeled promoters of MoPTH2, MoICL1, MoPEX6 and in vitro expressed MoBZIP3. The results showed that MoBZIP3 could directly bind the promoters of these genes ( Figure 5C). In addition, the EMSA results further showed that MoBZIP3 specifically binds to the gene promoters which contains an A-box (Supplemental Figure S3E).

Glycerol Synthesis Is Impaired in ∆mobzip3
Glycerol production in appressorium is dependent on complex enzymatic catalysis and cellular events. β-Oxidation is a critical reaction for metabolization of fatty acids in glycerol production. MoPEX6 was previously reported to function in β-Oxidation of fatty acids [23,29] and in our experiments MoPEX6 was down regulated in ∆mobzip3. To further explore that MoBZIP3 may function in glycerol metabolism, we validated the expression of MoPEX6 in ∆mobzip3. MoPEX6-GFP driven by MoPEX6 native promoter was expressed in WT and the ∆mobzip3 strain. Fluorescence observation showed that MoPEX6-GFP exhibited as bright spots in WT conidia, germ tube and appressorium. In contrast, the fluorescence intensity in the ∆mobzip3 strain was much lower ( Figure 6A). Quantitative analysis of the fluorescence intensity further supported the conclusion ( Figure 6B). hibited as bright spots in WT conidia, germ tube and appressorium. In contrast, the fluorescence intensity in the Δmobzip3 strain was much lower ( Figure 6A). Quantitative analysis of the fluorescence intensity further supported the conclusion ( Figure 6B). Furthermore, we measured the amount of glycerol in WT and Δmobzip3. The results showed that glycerol is generated rapidly in germ tube and appressorium in WT. However, in Δmobzip3, the amount of glycerol was much less produced when compared with those in the WT ( Figure 6C). Together, these results indicated that MoBZIP3 functions in appressorium glycerol production.

MoSln1 Mediated Appressorium Turgor Formation Is Affected in Δmobzip3
It was revealed that the turgor-sensing histidine-aspartate kinase, MoSln1, could enable the appressorium to sense turgor threshold and intrigue the appressorium to penetrate the host cells [9]. And our results indicated that MoBZIP3 may function in glycerol production, which could generate turgor pressure in appressorium. Thus, we explored whether MoBZIP3 functions in MoSln1 initiated turgor-driven plant infection. As MoB-ZIP3 is a transcription factor, we first examined the expression level of MoSln1 in WT and Δmobzip3. The expression of MoSln1 decreased about 28% in Δmobzip3 when compared with WT in the appressorium cell ( Figure 7A) indicating that MoSln1 is also a potential target for MoBZIP3. Furthermore, we measured the amount of glycerol in WT and ∆mobzip3. The results showed that glycerol is generated rapidly in germ tube and appressorium in WT. However, in ∆mobzip3, the amount of glycerol was much less produced when compared with those in the WT ( Figure 6C). Together, these results indicated that MoBZIP3 functions in appressorium glycerol production.

MoSln1 Mediated Appressorium Turgor Formation Is Affected in ∆mobzip3
It was revealed that the turgor-sensing histidine-aspartate kinase, MoSln1, could enable the appressorium to sense turgor threshold and intrigue the appressorium to penetrate the host cells [9]. And our results indicated that MoBZIP3 may function in glycerol production, which could generate turgor pressure in appressorium. Thus, we explored whether MoBZIP3 functions in MoSln1 initiated turgor-driven plant infection. As MoBZIP3 is a transcription factor, we first examined the expression level of MoSln1 in WT and ∆mobzip3. The expression of MoSln1 decreased about 28% in ∆mobzip3 when compared with WT in the appressorium cell ( Figure 7A) indicating that MoSln1 is also a potential target for MoBZIP3.
MoSln1 could sense the turgor threshold and its subcellular localization is sensitive to changes in turgor. During appressorium maturation, MoSln1 gradually accumulates to the appressorium pore along with the increasing turgor pressure [9]. We then investigated the distribution of MoSln1 in ∆mobzip3. MoSln1-GFP driven by its native promoter was expressed in WT and the ∆mobzip3 strain. It could be observed that MoSln1-GFP accumulated at the appressorium pore at 24 h after conidia germed on glass coverslips. But in the ∆mobzip3 strain, MoSln1-GFP did not accumulate at the appressorium pore. Instead, the MoSln1-GFP fluorescence signals were distributed at the periphery of the cell membrane ( Figure 7B,C).
Appressorium pore accumulation of MoSln1 could recruit the actin cytoskeleton assembly as filament actin (F-actin) ring in the infection cell, which is critical for M. oryzae penetration into the host cells [30]. We then investigated the actin assembly in ∆mobzip3. We expressed the actin scaffold protein, Septin5-GFP and the actin labeling peptide, lifeact-mCherry to label the actin cytoskeleton [30,31]. In the WT appressorium, both the Septin5- MoSln1 could sense the turgor threshold and its subcellular localization is sensitive to changes in turgor. During appressorium maturation, MoSln1 gradually accumulates to the appressorium pore along with the increasing turgor pressure [9]. We then investigated the distribution of MoSln1 in Δmobzip3. MoSln1-GFP driven by its native promoter was expressed in WT and theΔmobzip3 strain. It could be observed that MoSln1-GFP accumulated at the appressorium pore at 24 h after conidia germed on glass coverslips. But in the Δmobzip3 strain, MoSln1-GFP did not accumulate at the appressorium pore. Instead, the MoSln1-GFP fluorescence signals were distributed at the periphery of the cell membrane ( Figure 7B,C).
Appressorium pore accumulation of MoSln1 could recruit the actin cytoskeleton assembly as filament actin (F-actin) ring in the infection cell, which is critical for M. oryzae penetration into the host cells [30]. We then investigated the actin assembly in Δmobzip3. We expressed the actin scaffold protein, Septin5-GFP and the actin labeling peptide, lifeact-mCherry to label the actin cytoskeleton [30,31]. In the WT appressorium, both the Sep-tin5-GFP and the lifeact-mCherry assemble as ring structures. But in Δmobzip3, abnormal actin structures formed ( Figure 7D-G).
In all, these results indicated MoBZIP3 functions in arppressorium turgor formation and affected MoSln1-mediated actin cytoskeleton assembly during M. oryzae infection. In all, these results indicated MoBZIP3 functions in arppressorium turgor formation and affected MoSln1-mediated actin cytoskeleton assembly during M. oryzae infection.

Discussion
Magnaporthe oryzae is the causal agent of rice blast, the most destructive disease of cultivated rice worldwide. The devastating fungi depend largely on their adaptability to environments. It has developed its own complex regulatory networks to control developmental processes, pathogenesis and responses to stresses. The BZIP transcription factors are involved in many critical processes in a diverse range of species. In fungi, there are multiple BZIP transcription factors and members of them in Aspergillus nidulans, Aspergillus fumigatus, Fusarium graminearum, Fusarium oxysporum which have been shown to be closely linked to various developmental and physiological processes including the fungal metabolism, development, stress responses and virulence [12,[32][33][34]. In M. oryzae, 22 BZIP genes were identified in strain Guy11 and KJ201. And of which, several members were characterized to play crucial functions in development and pathogenesis. For example, MoBZIP5, MoBZIP10, MoMeaB (MoBZIP12), MoHac1 (MoBZIP18) and MoMetR (MoBZIP22) were reported to function associated with many aspects of M. oryzae development and host infection [15][16][17]. Considering the diverse critical roles for BZIP proteins in M. oryzae, we attempted to identify new functional BZIP genes in the Y34 strain, which is isolated from the field and its pathogenicity is different from Guy11 [19]. In this study, we found that in Y34 strain, no BZIP11 and BZIP14 were present in the genome but additional BZIP22 and BZIP23 emerged. we then analyzed expression profiles of the 22 BZIP genes in Y34 and found that their expressions were quite different from those in the Guy11 and KJ201, indicating that the BZIP genes in Y34 may form a different regulatory network. We further knocked out six BZIP genes including MoBZIP3, MoBZIP16, MoBZIP17, MoBZIP20, MoBZIP23 and MoBZIP24 which showed high expression levels in M. oryzae infection stages ( Figure 1). And of which, ∆mobzip3 exhibited decreased infection capability (Figure 2) indicating that MoBZIP3 is a positive regulator for M. oryzae virulence in BZIP family. Virulence of the other five bzip mutants obtained in our study did not significantly change though they expressed at a high level in the infection stages. We speculated that some BZIP proteins in M. oryzae maybe function in an overlapping manner.
Previous work has shown that knockout of MoBZIP3 in strain Guy11 and KJ201 did not result in significant changes in the M. oryzae development and virulence. In line with their results, in Y34, this gene mutant also exhibits a normal development phenotype (Figure 2A), indicating that MoBZIP3 may not much function in the fungal development. However, we found that the pathogenesis of ∆mobzip3 was significantly decreased ( Figure 2B-D). We found that the N-terminal of Y34 BZIP3 protein was different from those of Guy11 and KJ201. It lacks 24 amino acids at the N-terminal when compared with the Guy11 (Supplemental Figure S3A,B). MoBZIP3 from Guy11 possesses a weaker DNA binding activity than the Y34 MoBZIP3 (Supplemental Figure S3C). Thus, we speculate that no significant changes in Guy11 MoBZIP3 mutant may associate with its weak DNA binding property. And the amino acid discrepancy in the Y34 strain leads to a stronger DNA binding activity, which may effectively promote the transcription of the targeted genes. Our results suggested that MoBZIP3 is a new member of functional BZIP genes in M. oryzae virulence.
The rice blast fungus employs a specialized infection structure called appressorium to penetrate into host plant leaf surface using huge invasive force [35]. The appressorium cell wall which is rich in melanin could act as a rigid structural barrier to help the generation of turgor [6]. The major solute that accumulates in appressorium is glycerol which gradually produces the turgor pressure in the cell [5]. Thus, the generation of glycerol is critical for the appressorium to achieve enough turgor pressure. Prior work showed that lipid degradation is a major route to generate glycerol in appressorium and in this process, the glycogen servers as the energy source [22]. And high levels of triacylcglycerol lipase activity for lipid conversion to both fatty acids and glycerol during turgor generation are proposed in the appressorium. During glycerol generation, fatty acid β-oxidation is required for activation of the glyoxylate cycle and gluconeogenesis, which is significant during plant infection by M. oryzae [22,36]. In our study, we found that degradation of lipids and glycogen was significantly decreased in the ∆mobzip3 appressorium ( Figure 4). The MoTPS1 gene which is involved in glycogen etabolism [26], MoICL1 gene which functions in glyoxylate cycle catalyzing [28] and MoPTH2, MoICL1, MoPEX6 which are critical for fatty acid β-oxidation, were all down regulated in expression in ∆mobzip3 ( Figures 5B and 6A,B). Furthermore, the glycerol accumulation was compromised in the ∆mobzip3 appressorium ( Figure 6C). Considering that MoBZIP3 localized in the M. oryzae nucleus ( Figure 5A) and it could directly bind to the promoters of MoPTH2, MoICL1, MoPEX6 ( Figure 5C), we could conclude that MoBZIP3 serves as a working transcription factor in M. oryzae pathogenesis and its functions were associated with regulation of expression of the glycerol generation related genes.
The appressorium produces turgor of up to 8.0 MPa through accumulating high concentrations of glycerol and other polyols [5]. Accompany with the generation of appressorium turgor, a narrow penetration hypha emerges from the base of the appressorium and breach the cuticle of the rice leaf. In this process, a sensor kinase MoSln1 could recognize the threshold of turgor and initiate a complex signaling pathway to generate invasive force to cause blast disease [9]. MoSln1 could change its subcellular localization from the cytoplasm to the appressorium pore in response to turgor increasing, which is critical for its function. In this study, we found that expression of MoSln1 was down regulated in ∆mobzip3 ( Figure 7A), and MoSln1 did not change its subcellular localization to the appressorium pore ( Figure 7B,C). These indicate that MoSln1 is a potential target for MoBZIP3. Prior work showed that MoSln1 could recruit septin proteins to the appressorium pore to orchestrate a toroidal network of F-actin, which governs the penetration of the appressorium [30]. But in our study, we found that both the Septin5-GFP and the F-actin ring were all deformed in ∆mobzip3 ( Figure 7D-G). Thus, we concluded that MoBZIP3 could directly regulate the expression of glycerol synthesis-related genes, and meanwhile, it may also have a function in MoSln1-mediated turgor sensing and penetration.

Plant and Fungal Strain Growth Conditions
All M. oryzae strains used in this study were derived from the wild-type strain Y34 (kindly provided by Prof. LiHuang Zhu, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences). All strains were cultured on liquid or solid completed medium for growth and straw rice bran medium for conidia generation at 28 • C in the dark. Mycelia were harvested from liquid CM and used for genomic DNA and RNA extractions. Y34 susceptible rice (Oryza sativa cv. Nipponbare) were grown at 28 • C under 16 h light and 8 h dark condition and used for fungal infection analyses.

Targeted Gene Deletion and Plasmid Construction
The BZIP gene deletion mutants were generated using the standard one-step gene replacement strategy [16,31]. To construct plasmids expressing MoBZIP3-GFP, MoPEX6-GFP, MoSep5-GFP,~1.5-kb native promoter region from the M. oryzae genome were amplified and cloned into the pKNTG binary vector [37]. All constructs were cloned by homologous recombination (ClonExpress MultiS One Step Cloning Kit, Vazyme Biotech, Nanjing, China, C112); all primers with restriction enzyme sites are listed in the Supplemental Table S1.

RNA Extraction and qRT-PCR Analysis
Total RNA was extracted from the samples using a Total RNA Purification kit (Trans-Gen, Fuzhou, China, ET101-01) according to the manufacturer's protocol. For the gene expression analysis in leaf blades, total RNA was isolated from three 5-cm long leaf sections per plant spotted with conidia suspension. Data were normalized to the expression levels of Actin in M. oryzae [20]. qRT-PCR was performed using Perfect Start Green qPCR SuperMix (TransGen, Fuzhou, China, AQ601) with a Bio-Rad CFX96 real-time PCR detection system. Primer sequences are listed in Supplemental Table S1.

Pathogenicity Analysis Assay
The wild-type M. oryzae strains and each mutant were cultured on rice bran medium for 7 days. For spray inoculation of conidia, prepare conidia suspension 1 × 10 5 spores/mL in a 0.02% (w/v) tween-20 solution and spray it evenly on the 2-3 weeks rice leaves with sprayer. The inoculated plants were grown in a growth chamber at 28 • C, with high humidity in the dark for the first 24 h, followed by a 12-h/12-h light (20,000 Lux)/dark cycle. The development disease lesions were observed at 4-5 days after inoculation [38].
For plant sheath cells penetration assays, the conidia suspension of GFP labeled WT, ∆mobzip3 and the complemented strain was injected into the rice leaf sheath under high humidity, and the development and infection status of the rice blast fungus in the leaf sheath cells were observed under microscopy (Zeiss LSM880, Oberkochen, Germany, with a 40× objective) [20].

Protein Expression and Purification in Escherichia coli
ORF sequence of MoBZIP3 was cloned into the Pet-28a vector between the BamH1 and HindIII sites to generate. The fusion constructs were introduced into BL21 (DE3). Bacteria containing the plasmid was grown in Luria-Bertani (LB) medium containing 100 mg/mL Kan at 37 • C to OD600 = 0.6. Expression of the fusion protein His-MoBZIP3 was induced by the addition of 0.5 mM isopropyl b-D-1 thiogalactopyranoside (IPTG) and incubation at 16 • C for 14 h. Then His-tagged MoBZIP3 proteins were purified using nickel-nitrilotriacetic acid resin following procedures described by the manufacturer (TransGen, DP101).

Electrophoretic Mobility Shift Assay
Light Shift Chemiluminescent EMSA Kit (Beyotime Biotechnology, Shanghai, China, No. 3308) was used in this experiment. Biotin was labeled at 3 end and 5 end of primers for each gene. The biotin-labeled DNA from upstream of the ORF for each gene were obtained through PCR. The detailed procedure of EMSA follows the manufacturer's instructions. Photos obtained by ChemiDocTM Imaging Systems (Bio-Rad Laboratories, Hercules, CA, USA, ChemiDoc MP).

Lipid Droplets and Glycogen Staining
Lipid droplets and glycogen staining mainly followed the method described previously [39]. The suspension of conidia (20 µL) was cultured on a hydrophobic membrane to induce formation of appressorium at 25 • C in the dark for 0 h, 2 h, 8 h, 16 h, and 24 h, respectively. Nile red staining was carried out by using the staining solution (final concentration 1 mM) for 30 min. Then the distribution of lipid droplets was observed under fluorescence microscopy (Zeiss LSM880, with a 40×objective). Similarly, glycogen was stained with I2/KI solution (1 mM) for 1 min. And the images were obtained with Zeiss microscopy (Zeiss A1). To analyze the glycogen, image processing and measurements were performed using ZEN software (Zeiss 2.3 lite). Images were selected using the 'rectangle' tool and the mean value of the selected region was calculated using the "Measure" plugin. The experiment was repeated 3 times, and 100 conidia were counted for each treatment.

Accession Numbers
Sequence data for the genes described in this study can be found in the GenBank/EMBL data under the accession numbers: MoBZIP1 (ELQ37252.