Basic Leucine Zipper (bZIP) Domain Transcription Factor MBZ1 Regulates Cell Wall Integrity, Spore Adherence, and Virulence in Metarhizium robertsii *

Background: The bZIP-type transcription factors are widely distributed in eukaryotes to control an array of biological activities. Results: MBZ1 from Metarhizium robertsii regulates fungal growth, cell wall integrity, spore adherence, and virulence against insects. Conclusion: MBZ1 is required for development and virulence of insect pathogenic fungi. Significance: The orthologs of bZIP-type transcription factors are functionally divergent in different organisms. Transcription factors (TFs) containing the basic leucine zipper (bZIP) domain are widely distributed in eukaryotes and display an array of distinct functions. In this study, a bZIP-type TF gene (MBZ1) was deleted and functionally characterized in the insect pathogenic fungus Metarhizium robertsii. The deletion mutant (ΔMBZ1) showed defects in cell wall integrity, adhesion to hydrophobic surfaces, and topical infection of insects. Relative to the WT, ΔMBZ1 was also impaired in growth and conidiogenesis. Examination of putative target gene expression indicated that the genes involved in chitin biosynthesis were differentially transcribed in ΔMBZ1 compared with the WT, which led to the accumulation of a higher level of chitin in mutant cell walls. MBZ1 exhibited negative regulation of subtilisin proteases, but positive control of an adhesin gene, which is consistent with the observation of effects on cell autolysis and a reduction in spore adherence to hydrophobic surfaces in ΔMBZ1. Promoter binding assays indicated that MBZ1 can bind to different target genes and suggested the possibility of heterodimer formation to increase the diversity of the MBZ1 regulatory network. The results of this study advance our understanding of the divergence of bZIP-type TFs at both intra- and interspecific levels.


Transcription factors (TFs) containing the basic leucine zipper (bZIP) domain are widely distributed in eukaryotes and display an array of distinct functions. In this study, a bZIP-type TF gene (MBZ1) was deleted and functionally characterized in the insect pathogenic fungus Metarhizium robertsii. The deletion mutant (⌬MBZ1) showed defects in cell wall integrity, adhesion to hydrophobic surfaces, and topical infection of insects.
Relative to the WT, ⌬MBZ1 was also impaired in growth and conidiogenesis. Examination of putative target gene expression indicated that the genes involved in chitin biosynthesis were differentially transcribed in ⌬MBZ1 compared with the WT, which led to the accumulation of a higher level of chitin in mutant cell walls. MBZ1 exhibited negative regulation of subtilisin proteases, but positive control of an adhesin gene, which is consistent with the observation of effects on cell autolysis and a reduction in spore adherence to hydrophobic surfaces in ⌬MBZ1. Promoter binding assays indicated that MBZ1 can bind to different target genes and suggested the possibility of heterodimer formation to increase the diversity of the MBZ1 regulatory network. The results of this study advance our understanding of the divergence of bZIP-type TFs at both intra-and interspecific levels.
Much of our knowledge about the features and functions of transcription factors (TFs), 2 including a group of structurally and functionally related members containing the conserved basic leucine zipper (bZIP) domain, comes from the discovery and study of AP-1 (activating protein 1) (1). The bZIP TFs are one of the largest families of dimerizing TFs and are widely distributed in the genomes of all eukaryotes (2). The prototypical bZIP TF was first discovered over 30 years ago in humans (3), and members are now classified into 19 families regulating a plethora of biological functions, including the cell cycle, development, reproduction, metabolism, and programmed cell death (2). Besides animals, bZIP-type TFs have also been well characterized in plants by mediating pathogen defense, abiotic stress response, hormone signaling, energy metabolism, and senescence (4).
In fungi, a bZIP-type TF was first characterized as YAP1 (yeast activating protein 1) in Saccharomyces cerevisiae (5), and then members of the family (YAP1-YAP8) were found and verified to be functionally distinct in yeast (6). Different bZIP-type TFs have also been characterized in various filamentous fungal species, including Aspergillus nidulans (7,8), Neurospora crassa (9), and the plant pathogens Fusarium graminearum (10 -12) and Magnaporthe oryzae (13)(14)(15). Deletions and functional characterizations of these genes revealed that bZIP-type TFs are involved in mediating fungal development, sexuality, stress responses, secondary metabolisms, and especially the virulence of plant pathogens. For example, the bZIP-type TF FGZIF1 was identified to be involved in the control of mycotoxin deoxynivalenol production, sexuality, and virulence in F. graminearum (11). MOAP1 of M. oryzae was functionally verified as a positive regulator of different target genes involved in growth, development, and pathogenicity (14). Genome analyses of insect fungal pathogens such as Metarhizium robertsii (16), Cordyceps militaris (17), and Beauveria bassiana (18) also identified an array of bZIP-type TFs present in their genomes. However, none of these has been functionally studied.
Insect pathogenic fungi such as M. robertsii, Metarhizium acridum, and B. bassiana diverged after mammalian and plant pathogens (19) and have been developed as promising and environmentally friendly mycoinsecticides (20,21). Molecular pathogenesis studies have revealed the functions of genes involved in insect cuticle adherence, degradation, penetration, immune invasion, and production of insecticidal mycotoxins (22,23). However, the TFs involved in regulation of virulencerelated genes are still unknown.
In this study, the bZIP-type TF gene MAA_01736, designated MBZ1, was deleted and functionally characterized in the insect pathogen M. robertsii. This gene was previously found to be highly transcribed by the fungus during early infection (16). Our functional studies showed that MBZ1 mediates both positive and negative regulation of different target genes involved in growth, sporulation, cell wall integrity, spore adherence, and virulence against insect hosts in M. robertsii.
Bioinformatics Analysis-Genome-wide surveys for bZIP domain-containing TFs were performed by BLASTp searches of the genomes of M. robertsii and other fungi. The protein sequences of functionally verified bZIP-type TFs from yeast and other species were found and aligned with those from M. robertsii with ClustalX 2.0 (30). A maximum likelihood tree was generated using a Dayhoff model with 1000 replicates for bootstrapping using the program MEGA6 (31). For analysis of the putative MBZ1-binding site, the promoter regions (ϳ1.5 kb upstream of each start codon) of selected genes were retrieved and analyzed using the weight matrix-based program Match (version 1.0). Subcellular Localization and Transcriptional Activation Assays-For the subcellular localization assay of MBZ1, the HSP70 gene (MAA_02081) promoter, GFP, and full-length MBZ1 genes were amplified individually by PCR, and the cassette was generated by fusion PCR with different primer pairs ( Table 1). The product was cloned into binary vector pDHT-BAR to produce plasmid pBAR-HSP70p-GFP-linker-MBZ1 with a Clon-Express kit (Vazyme). The resulting vector was used for transformation of the M. robertsii strain. For the transcriptional activation test, full-length cDNA of MBZ1 was amplified from a cDNA library with primers 1736CF and 1736CR and cloned into the NdeI and BamHI sites of the yeast vector pBKT7 (Invitrogen) under the control of the GAL4 promoter to generate plasmid pBKT7-MBZ1 for transformation into the AH109 yeast strain. Putative prototrophic transformants were selected using a Trp-free synthetic dropout medium and further transferred into Trp/His/Ade-free synthetic dropout plates for transcriptional activation assays. The negative control was made by transforming yeast with the blank vector, and the positive control yeast was transformed with the pBKT7 vector containing the GAL4 activation domain sequence (29).
Gene Deletion and Complementation-Targeted gene deletion of the MBZ1 gene was performed by homologous recombination via Agrobacterium-mediated fungal transformation as described previously (32,33). In brief, the 5Ј-and 3Ј-flanking sequences were amplified using the genomic DNA as a template with primer pairs TF1736UF/TF1736UR and TF1736DF/ TF1736DR, respectively ( Table 1). The respective products were digested with the restriction enzymes EcoRI, XhoI, and XbaI and then inserted into the corresponding sites of the binary vector pDHT-BAR (conferring ammonium glufosinate resistance) to produce plasmid pBAR-MBZ1-KO for fungal transformation. For null mutant complementation, the ORF of the MBZ1 gene was amplified with its promoter and terminator regions using primer pairs 1736SF and 1736SR. The product was digested with XbaI and then inserted into vector pDHT-BEN to generate pBEN-MBZ1 (conferring resistance to benomyl) (34). The mutants were verified by PCR and RT-PCR analyses with primers TF1736F and TF1736R ( Table 1). The complemented mutant was designated Comp.
Extraction and Quantification of Cell Wall Chitin-To quantify the chitin component of fungal cell walls, mycelia were harvested from SDB cultures; washed twice with sterile-distilled water; ground in liquid nitrogen; extracted with buffer containing 50 mM Tris-HCl (pH 7.8), 2% SDS, 0.3 M ␤-mercaptoethanol, and 1 mM EDTA; and incubated at 100°C for 15 min. After centrifugation at 8000 ϫ g, the pellet was washed three times with water and completely dried. The samples were dissolved in water to make a solution of 25 mg/ml. In total, 5 mg each of cell wall samples were acidified in 6 M HCl at 100°C for 4 h. Quantification of chitin was conducted by measuring the acid-released glucosamine from chitin using the substrate p-dimethylaminobenzaldehyde as a chromogen (35). The absorbance was measured at 520 nm using a BioPhotometer (Eppendorf), and the quantity of glucosamine was calculated in reference to the standard curve established using pure glucosamine (Sigma) (36).
Conidial Adhesion and Hydrophobicity Assays-Quantitative adhesion assays were conducted by labeling the fungal spores with FITC and incubating the microtiter plates in the dark for 1 h (25). After removing the unbound cells, the fluorescence of each sample was measured with a Varioskan Flash microplate reader (Thermo Scientific). Variation in spore surface hydrophobicity between the WT and mutants was examined using hexadecane as described previously (26).
Insect Bioassays-Virulence tests were conducted for the WT, ⌬MBZ1, and Comp against newly emerged last-instar larvae of silkworms (Bombyx mori) and wax moths (Galleria mellonella). Topical infections were performed by immersing insect larvae for 30 s in a spore suspension containing 1 ϫ 10 7 conidia/ml. Each treatment had three replicates with 15 insects each, and the experiments were repeated three times. Mortality was recorded every 12 h. For injection assays to bypass insect cuticle barriers, each insect was injected in the second proleg with 10 l of a spore suspension containing 1 ϫ 10 6 conidia/ml. The median lethal time (LT 50 ) was calculated as described previously (34). Dead insects were removed and placed on moist filter paper in Petri dishes. To determine and compare the differences in fungal penetration processes between the WT and mutants, insect cadavers were fixed in Carnoy's solution (60% CGGGCCCCCCCTCGAGGCAGTAGAGGTGTAGTTG For generating plasmid pBAR-HSP70p-GFP-linker-MBZ1 Hsp70R TCATAGACACGGAACAAT For RT-PCR and qRT-PCR analyses AGTTATTACCCTCGAGCGTCCAATGCTGCAAGACTA For generating p178-Xpromoter plasmids for yeast one-hybrid tests ethanol, 30% chloroform, and 10% glacial acetic acid), dehydrated in alcohol, and embedded in paraffin wax for cutting (10 m thick) with a rotary microtome (Leica) (32). Appressorium Induction and Mycotoxin Quantification-To examine MBZ1 regulation of fungal differentiation, conidia of the WT and mutants were used for infection structure induction. The hind wings from cicada (Cryptotympana atrata) were collected and surface-sterilized in 37% H 2 O 2 for 5 min, washed twice in sterile water, and immersed in a conidial suspension (2 ϫ 10 7 spores/ml) for 20 s. The inoculated wings were incubated on 0.8% water agar at 25°C for 24 h. Appressoria were induced on a hydrophobic surface in 1% glycerol-buffered minimum medium (34). To examine whether MBZ1 is involved in regulation of secondary metabolism, the induction, extraction, and quantification of cyclopeptide destruxins were performed as described previously (37).
Protease Activity Assay and Western Blotting-For protease activity assays, conidia of the WT, ⌬MBZ1, and Comp were inoculated in SDB at a final concentration of 10 6 conidia/ml and incubated at 25°C for 48 h at 180 rpm. The mycelia were then washed twice with sterile-distilled water, transferred to a basal salt medium supplemented with 1% (w/v) homogenized silkworm pupae, and incubated at 25°C for 48 h at 180 rpm (27). The alkaline protease activities were assayed using the substrate N-succinyl-Ala 2 -Pro-Phe p-nitroanilide (Sigma) as described previously (38). The protein concentration of each sample was determined using the Bradford method (34). All experiments were repeated twice, and each sample had three replicates. To confirm protease expression, Western blot analysis was conducted using an antibody against Metarhizium protease PR1A as described previously (39).
Field Emission Scanning Electron Microscopy Analysis-To compare the differences in cell surface structure between the WT and mutants, fungal conidia were incubated in SDB at 25°C for 7 days. The harvested mycelia were washed and fixed overnight in 0.1 M phosphate buffer (pH 7.2) containing a final concentration of 2.5% glutaraldehyde. The samples were dehydrated with a series of ethanol solutions (50 -100%, v/v) and coated with platinum. Observations were conducted using a field emission scanning electron microscope (Zeiss) operating at 3 kV (40).
RNA Isolation and Quantitative Real-time PCR (qRT-PCR) Analysis-The mycelia from the WT, ⌬MBZ1, and Comp that were harvested from 2-day old SDB cultures and from protease induction medium were used for RNA extraction. The samples were homogenized in liquid nitrogen, and total RNA was isolated using TRIzol reagent (Invitrogen) and treated with DNase I (Takara) in a column. cDNA of each sample was generated using an AffinityScript multiple-temperature cDNA synthesis kit (Toyobo). Using the primers for different target genes (Table 1), qRT-PCR analyses were performed using a SYBR Premix Ex Taq kit (Takara) with an ABI Prism 7000 sequence detection system (Applied Biosystems). The relative expression of each gene was normalized with reference to the expression of the ␤-tubulin gene (32).
Yeast One-hybrid Tests-For yeast one-hybrid tests, the cDNA sequence of MBZ1 was amplified from a cDNA library with primers 1736HF and 1736HR (Table 1) and cloned into the SpeI and SacI sites of vector pPC86 under the control of the GAL4 promoter to generate plasmid pPC86-MBZ1. To test the ability of MBZ1 to bind to the promoter regions of selected genes, promoter fragments were individually amplified by PCR with different primer pairs (Table 1) and inserted into the XhoI site of the p178 vector to control the lacZ reporter gene (29). The resultant plasmids were named p178-Xpromoter (where X represents each selected gene). pPC86-MBZ1 and each individual p178-Xpromoter were transformed into yeast strain EGY48 using a small-scale yeast transformation protocol (24). To examine the effect of autoactivation, the yeast cells were transformed only with each p178-Xpromoter vector. Positive colonies were selected on appropriate selection plates (Ura/Trpfree synthetic dropout medium with X-gal). A negative control was generated by transforming the yeast cells with vector pPC86-MBZ1 and a blank vector (p178). A positive control was generated by transforming the yeast cells with vector pPC86-OSBZIP58 and p178-HA2, as used in a previous study (29). The

RESULTS
Characteristics of MBZ1-Genome-wide analysis indicated that there are 24 bZIP domain-containing TFs encoded in the genome of M. robertsii, whereas 14 -26 bZIP-type TFs are present in yeast and other fungi (Fig. 1A). Overall, the number of bZIP TFs present in fungi are Ͼ2-fold fewer than those found in humans (56 bZIP genes) and plants (e.g. 75 in Arabidopsis, 89 in rice, and 125 in maize) (4). A phylogenetic analysis based on the bZIP domain sequences indicated that M. robertsii bZIP-type TFs could be classified into at least 10 subfamilies (Fig. 1B). Subfamilies 4 -6 have no homologs in yeast. The MBZ1 protein belongs to subfamily 5 and shares higher homology with its counterparts FGZIF1 (FGSG_01555, 57% identity) of F. graminearum and MOZIF1 (MGG_03288, 45% identity) of M. oryzae (10), but is only moderately homologous to MeaB (AN4900, 11% identity) of A. nidulans (7). All these TFs belong to subfamily 5 (Fig. 1B). In particular, subfamily 1 contains the yeast proteins CST6, ACA1, and SKO1 (Fig. 1B), which belong to the cAMP response element-binding protein (CREB) family of bZIP-type TFs that bind the consensus palindromic site 5Ј-TGACGTCA-3Ј (41).
To verify the nuclear localization of MBZ1, a GFP gene was fused with the 5Ј terminus of MBZ1 for transformation of M. robertsii. Fluorescence microscopy examinations verified that the GFP signal could be detected only in the nuclei of fungal spores, germlings, and appressoria of M. robertsii ( Fig. 2A). Consistent with our previous RNA-seq analysis (16), a qRT-PCR analysis confirmed that MBZ1 was most highly transcribed by the fungus during infection structure appressorium formation compared with the stages of fungal conidiation, saprophytic growth, and differentiation of hyphal bodies (blastospores formed in the insect body cavity) (Fig. 2B). A transcriptional activation test indicated that, relative to a negative control, yeast cells transformed with a vector containing MBZ1 cDNA could survive on an auxotrophic medium, as could the positive control yeast cells engineered with a GAL4 activation domain (Fig. 2C). The results thereby confirmed the typical TF characteristics of MBZ1: nuclear localization and transcriptional activation.
Deletion of MBZ1 Impairs Cell Wall integrity-We performed experiments to test the sensitivities of the WT, ⌬MBZ1, and Comp in response to different cell wall stressors or biosynthesis inhibitors. The observations indicated that the MBZ1 null mutant (but not the WT and Comp) lost the ability to grow on PDA supplemented with Congo red (Fig. 5A). However, no significant differences were found between the WT and mutants regarding growth in the presence of tunicamycin, Calcofluor White, amphotericin B, CuSO 4 , sorbitol, DTT, or SDS (data not shown). Congo red also inhibited mutant spore germination in SDB. Relative to the WT (68.89 Ϯ 7.83%) and Comp (58.40 Ϯ 8.90%), the germination rates of ⌬MBZ1 spores (11.95 Ϯ 2.80%) were significantly reduced (p Ͻ 0.001) after incubation for 24 h (Fig. 5B).
As indicated above, ⌬MBZ1 became sensitive to Congo red, but not Calcofluor White. We speculated that the cell wall components, especially the chitin content, would be altered in the MBZ1 null mutant. To test this, we performed a chitin quantification assay, which revealed that the level of chitin in the ⌬MBZ1 cell wall (8.62 Ϯ 0.82%) was significantly increased (p Ͻ 0.05) compared with the WT (4.19 Ϯ 0.95%) and Comp (3.74 Ϯ 0.11%) (Fig. 5C). In addition, we performed cell wall binding assays with concanavalin A, G. simplicifolia lectin II, and G. nivalis lectin (24,25). The results demonstrated a stronger fluorescent signal for ⌬MBZ1 cells when stained with G. simplicifolia lectin II, suggesting again a higher level of chitin accumulation compared with the WT and Comp. No apparent  The results indicate no significant differences between the WT and mutants, suggesting that MBZ1 is not involved in controlling destruxin production. MARCH 27, 2015 • VOLUME 290 • NUMBER 13 JOURNAL OF BIOLOGICAL CHEMISTRY 8223 differences were found between the WT and mutants for the treatments with concanavalin A and G. nivalis lectin (Fig. 5D).

MBZ1 Regulates Fungal Development and Virulence
Impairment of Conidial Hydrophobicity and Adherence to Hydrophobic Surface-Consistent with the impairment of cell wall integrity, the hydrophobicity defect in the ⌬MBZ1 spores was also found during the preparation of spore suspensions. Thus, unlike the WT and Comp, ⌬MBZ1 conidia were more readily dispersible in water without the detergent Tween 20 (Fig. 6A). Ultrastructure examinations of spore surface features did not reveal any morphological difference between the WT and mutants (Fig. 6B). However, the measurements of cell surface hydrophobicity indicated that the hydrophobicity index of ⌬MBZ1 spores (0.39 Ϯ 0.09) was significantly reduced (p Ͻ 0.01) compared with the WT (0.88 Ϯ 0.02) and Comp (0.81 Ϯ 0.05) (Fig. 6C). In addition, a quantitative assay of cell adhesion ability demonstrated that ⌬MBZ1 had a significantly impaired capacity (p Ͻ 0.001) to bind to a hydrophobic surface compared with the WT and Comp (Fig. 6D).
Deletion of MBZ1 Impairs Fungal Virulence-To examine the effect of MBZ1 deletion on fungal virulence, we performed both topical infection and injection bioassays with last-instar silkworm and wax moth larvae. The LT 50 values were estimated and compared among the WT, ⌬MBZ1, and Comp. After top-ical infection of silkworms, the differences were significant between the WT (LT 50 ϭ 2.8 Ϯ 0.09 days) and ⌬MBZ1 (LT 50 ϭ 3.9 Ϯ 0.13 days) ( 2 ϭ 25.12, p Ͻ 0.0001) and between ⌬MBZ1 and Comp (LT 50 ϭ 2.8 Ϯ 0.08 days) ( 2 ϭ 29.33, p Ͻ 0.0001) (Fig. 7A). However, there was no significant difference in injection assays among the WT (LT 50 ϭ 2.8 Ϯ 0.10 days), ⌬MBZ1 (LT 50 ϭ 2.7 Ϯ 0.10 days), and Comp (LT 50 ϭ 2.7 Ϯ 0.10 days) (p Ͼ 0.5) (Fig. 7B). Similar patterns were observed during the topical infection and injection assays using the wax moth larvae (Fig. 7, C and D). Thus, MBZ1 is required for the virulence of M. robertsii by playing a role in fungal adhesion to and penetration of the insect cuticle. The results are consistent with the qRT-PCR data that the expression of MBZ1 was highly activated by the fungus during early infection, but not during growth in insect blood (Fig. 2B). In support of this, the melanization spots characteristic of fungal penetration were not evident on the insects topically treated with the ⌬MBZ1 spores, but were abundant on the insects treated with the WT and Comp conidia 48 h post-inoculation (Fig. 7E).
Negative Regulation of Protease Activity-The impact of MBZ1 deletion on spore adhesion and hydrophobicity contributes to virulence defects in ⌬MBZ1. To examine whether any other virulence factors were affected, we performed infection structure induction. The results indicated that the spores of ⌬MBZ1 could similarly form appressoria like the WT and Comp on both hydrophobic surfaces and insect cuticles (Fig. 8).
In addition, we found that the biosynthesis of insecticidal cyclopeptide destruxins was not affected in ⌬MBZ1 compared with the WT and Comp (Fig. 4B). We also tested the expression of cuticular degradation proteases. After growth in SDB for 48 h, unexpectedly, the subtilisin-like protease activity of ⌬MBZ1 was significantly increased (p Ͻ 0.001) compared with the WT and Comp. However, the differences were not significant among the isolates when they were grown in the inductive medium containing insect homogenates (Fig. 9A). These observations were confirmed by Western blot analysis using an antibody against PR1A, an extracellular protease (Fig. 9B). In addi-tion, consistent with a previous analysis (42), the expression of proteases was switched off in both the WT and ⌬MBZ1 when the fungi were grown in insect blood (Fig. 9B). These findings suggest that MBZ1 is involved only in the negative control of protease expression during fungal saprophytic growth in artificial and nutrient-rich media, but not in insect component-containing substrates. This was further evidenced when the cell surface mucilage-like structures found in ⌬MBZ1 (but not in the WT and Comp) were self-digested when grown in SDB (Fig.  9C). This phenotype is similar to fungal cell autolysis (43).
Verification of Transcriptional Control of Target Genes-As indicated above, the cell wall integrity of ⌬MBZ1 was impaired, and expression of proteases was activated in artificial medium. To verify the control of downstream genes by MBZ1, the puta-FIGURE 6. Analyses of spore hydrophobicity and adherence ability. A, variation in spore dispersal ability between the WT and mutants in water and 0.1% Tween 20. Scale bar ϭ 10 m. B, spore surface structures examined by field emission scanning electron microscopy analysis. Scale bar ϭ 1 m. C, variation in spore surface hydrophobicity between the WT and mutants. **, p Ͻ 0.01, WT (or Comp) versus ⌬MBZ1. D, variation in spore ability to adhere to hydrophobic surfaces between the WT and mutants. ***, p Ͻ 0.001, WT (or Comp) versus ⌬MBZ1.
tively targeted genes were selected for qRT-PCR analysis and yeast one-hybrid tests for promoter binding. These included the cell wall biosynthesis-related genes, protease genes, and the adhesin gene MAD1 (Table 2). After growth in SDB for 48 h, the putative ␤-1,3-glucanosyltransglycosylases (MAA_01794, similar to yeast GAS2, 39% identity; and MAA_02558, similar to yeast GAS3, 39% identity) and cell wall glucanosyltransferases (MAA_02062 and MAA_04636, both similar to the yeast chitin transglycosylase CRH1 with 44 and 31% identities, respectively) were significantly down-regulated in ⌬MBZ1 compared with the WT (Table 3). In contrast, the gene MAA_07202, which is highly similar to the yeast glutamine:fructose-6-phosphate amidotransferase GFA1 (60% identity) involved in chitin biosynthesis (44,45), was up-regulated by Ͼ6-fold in ⌬MBZ1 ( Table 3). The adhesin MAD1 (MAA_03775) mediates conidial adherence to hydrophobic surface in Metarhizium (46). Relative to the WT, the MAD1 gene was down-regulated by 4-fold in ⌬MBZ1. This was therefore consistent with the defect of spore adherence ability in ⌬MBZ1 (Fig. 6C). Consistent with the increase in protease activity in SDB (Fig. 9, A and B), the subtilisin-like protease genes, e.g. the PR1A gene (MAA_05675), were significantly up-regulated in ⌬MBZ1 compared with the WT (Table 3).
In silico analysis identified the presence of the conserved motif 5Ј-TGACG-3Ј in the promoter regions of all selected genes (Fig. 10A), which is the functional half-site of the CREBtargeting palindrome (5Ј-TGACGTCA-3Ј) (47), suggesting that MBZ1 is also the CREB site or half-site activator. Heterologous expression of MBZ1 in Escherichia coli failed to obtain soluble protein for the electrophoretic mobility shift assay. To verify the binding of MBZ1 to the promoter regions of these FIGURE 7. Insect bioassays. A, survival of silkworm larvae following topical infection with the spore suspensions (1 ϫ 10 7 spores/ml) of the WT and mutants. B, survival of silkworm larvae following injection with the spore suspensions (10 l, 1 ϫ 10 6 spores/ml) of the WT and mutants. Control insects were treated with sterile water. C, survival of wax moth larvae following topical infection with the spore suspensions (1 ϫ 10 7 spores/ml). D, survival of wax moth larvae following injection with the spore suspensions (10 l, 1 ϫ 10 6 spores/ml). Control insects were treated with sterile water. Error bars represent S.D. E, impairment of topical infection in ⌬MBZ1. The wax moth larvae were topically infected for 48 h and embedded for cutting and microscopic observations. Relative to the WT and Comp, many fewer melanization spots were observed on the insects treated with ⌬MBZ1 spores. Scale bar ϭ 1 mm. target genes, we performed yeast one-hybrid assays. The yeast cells transformed with only the individual promoter region could not express ␤-galactosidase, i.e. the preclusion of selfactivation (Fig. 10B). Except for the promoters of MAD1 and the GFA1-like gene (MAA_07202), the ␤-galactosidase activity was activated by MBZ1 when under the control of the selected gene promoters (Fig. 10C). However, consistent with a previous report (28), the reporter enzyme activity varied among the controls with different gene promoters. The highest binding and activation activities were observed for MBZ1 toward the GAS2 (MAA_01794) and GAS3-like (MAA_02558) genes (Fig. 10C). Interestingly, the full CREB site is present in the promoter of MAA_01794, but it is the 5Ј-TGACGTTG-3Ј for MAA_02558 (Fig. 10A).

DISCUSSION
In this study, we characterized the functions of the bZIP-type TF MBZ1 in the insect pathogenic fungus M. robertsii. Gene deletion, biochemical analyses, and insect bioassays indicated that MBZ1 contributes to the regulation of fungal growth,  conidiation, cell wall integrity, and virulence. Enzyme activity, Western blotting, and qRT-PCR analyses demonstrated that MBZ1 is involved in both the positive and negative transcriptional control of various target genes, which is similar to the regulation pattern of the C2H2-type pH-responsive regulator of M. robertsii (32). Similar to virulence defects in gene deletion mutants of FGZIF1 in F. graminearum and MOZIF1 in M. oryzae (10), ⌬MBZ1 was impaired in killing insects. Interestingly, ⌬FGZIF1 does not show defects in maintenance of cell wall integrity (i.e. not sensitive to Congo red), conidiation, and growth (10). This is in contrast to our observations for ⌬MBZ1, which suggests that, in addition to the functional divergence of paralogous bZIP-type TFs (15), the orthologous TFs could play divergent roles in different organisms. Deletion of the bZIP-type TF gene FGAP1 does not affect the pathogenicity of F. graminearum to wheat (11), but loss of the orthologous gene MOAP1 in M. oryzae completely abolishes its pathogenicity to rice (14). Likewise, other studies have found that the null mutants of MOATF1 of M. oryzae and FGATF1 of F. graminearum have reduced virulence, but the former are more sensitive to oxidative stress (12,13). We also observed that deletion of MBZ1 affected fungal pigmentation (Figs. 3A and 4A), but not production of insecticidal mycotoxin destruxins (Fig. 4B). In F. graminearum, however, deletion of FGZIF1 significantly impairs fungal production of the mycotoxin deoxynivalenol (a type B trichothecene) (10). Thus, the functional divergence of these orthologs in the different fungal species would question the assumption of ortholog conjecture, which suggests that orthologous genes carry out equivalent biological functions in different organisms (48). However, it remains to be determined whether the orthologs from plant pathogens could fully complement the phenotypes of ⌬MBZ1 in M. robertsii.
The teleomorph of Metarhizium spp. is considered to be Metacordyceps spp.; however, their heterothallic sexual cycles are largely cryptic (49). Future studies are still required to investigate whether MBZ1 is involved in controlling sexuality in M. robertsii as FGZIF1 and FGATF1 control sex cycles in F. graminearum (10,12). The fungal cell wall plays essential roles in mediating growth, stress resistance, cell adhesion to hosts, and infection (50). We found that deletion of MBZ1 led to the down-regulation of GH72-like genes (MAA_01794 and MAA_02558) and yeast CRH1-like genes (MAA_02062 and MAA_04636), but led to the up-regulation of the GFA1-like gene MAA_07202. The GH72 family of glycosidases/transglycosidases plays important roles in cell wall maintenance (51). Deletion mutants of these genes have defects in cell adherence to hydrophobic surfaces and increases in chitin in the cell wall (52,53). The yeast chitin transglycosylase CRH1 acts to transfer chitin and cross-link it to ␤-1,3and ␤-1,6-glucans in the cell wall (54), and GFA1 contributes to the first step of chitin biosynthesis (44). Relative to the WT, differential regulation of these genes would be responsible for chitin accumulation in the ⌬MBZ1 cell wall and therefore the increased sensitivity to the chitin-specific dye Congo red (Fig. 5A). However, the Congo red sensitivity was not observed for ⌬FGZIF1 (10), but it was observed for ⌬FGATF1 of F. graminearum (12). This is again suggestive of the functional divergence of orthologous bZIP TFs among different species. In Candida albicans, deletion of the GH72-like PHR2 gene (homologous to yeast GAS2 and GAS3) additionally results in sporulation failure (52,53). Thus, the down-regulation of GH72-like genes could contribute, at least in part, to the defective conidiogenesis in ⌬MBZ1 (Fig. 3). However, it remains to be determined how MBZ1 is involved in mediating mycelial growth in M. robertsii; the effect has not been observed for its counterpart FGZIF1 in F. graminearum (10).
We tested virulence via both topical infection and injection assays against two different insect species and found that differences were observed only between the WT and ⌬MBZ1 with topical infections (Fig. 7), suggesting that the null mutant can evade host immunity within the insect hemocoel as well as the WT. Because ⌬MBZ1 could successfully produce the infection  structure appressorium (Fig. 8), failure to bind to the insect cuticle would be mainly responsible for the defective virulence in ⌬MBZ1 due to the reduction of its surface hydrophobicity and adherence ability. This resembles the effect of deletion of the adhesin MAD1 gene in M. robertsii (46). The subtilisin proteases are the key virulence factors of insect pathogenic fungi for degradation of insect cuticular proteins (42). Overexpression of the protease PR1A increases fungal virulence and causes extensive melanization of infected insects due to the activation of insect prophenoloxidases (39,55). Interestingly, the up-reg-ulation of proteases in ⌬MBZ1 was observed only when the fungus was grown in artificial medium, but not in insect component-related conditions. This was particularly clear when proteases were successfully switched off in insect hemolymph (Fig. 9B). The result suggests that there are additional TFs involved in fine-tuning protease expression in M. robertsii. In filamentous fungi, the Zn 2 Cys 6 domain-containing TF PrtT has been characterized as an activator of extracellular proteases (56). A homolog of Aspergillus PrtT is also present in the genome of M. robertsii (MAA_09737, 24% identity) (16). Thus, the mechanisms used by MBZ1 to regulate protease expression, as well as the possible interactions between MBZ1 and other TFs, remain to be clarified. The formation of either homodimers or heterodimers is the hallmark of bZIP-type TFs, which allows them to increase the complexity of regulatory networks (2). Our results indicate that deletion of MBZ1 resulted in up-regulation of a GFA1-like gene (MAA_07202) and down-regulation of the adhesin gene in M. robertsii (Table 3). Cell adhesion assays also revealed the significantly reduced ability of ⌬MBZ1 spores to bind to hydrophobic surfaces (Fig. 6D). The CREB half-site is present in the promoter regions (Fig. 10A); however, yeast one-hybrid assays showed that MBZ1 did not bind to the promoter regions of MAD1 and the GFA1-like gene (Fig. 10C). Precluding the possibility of a technical false-negative or false-positive issue, the findings suggest the possible formation of heterodimerized MBZ1 in the regulation of these genes, i.e. the indirect control of these target genes. The putative partner involved in dimerization with MBZ1 remains to be determined.
In conclusion, we characterized the bZIP-type TF MBZ1 in an insect pathogenic fungal model and established that the MBZ1 gene is required for virulence against insects by mediating cell wall integrity, cell surface hydrophobicity, and adherence to hydrophobic surfaces. Both negative and positive regulation of various target genes was evident, indicating that MBZ1 can function as either an activator or a suppressor. The results of this study also reveal the functional divergence of orthologous bZIP-type TFs in different fungal species, highlighting the need to elucidate orthologous gene function in different organisms.