The Formaldehyde Dehydrogenase SsFdh1 Is Regulated by and Functionally Cooperates with the GATA Transcription Factor SsNsd1 in Sclerotinia sclerotiorum

S. sclerotiorum is a pathogenic fungus with sclerotium and infection cushion development, making S. sclerotiorum one of the most challenging agricultural pathogens with no effective control method. We identified important sclerotium and compound appressorium formation determinants, SsNsd1 and SsFdh1, and investigated their regulatory mechanism at the molecular level. SsNsd1 and SsFdh1 are zinc finger motif-containing proteins and associate with each other in the nucleus. On other hand, SsNsd1, as a GATA transcription factor, directly binds to GATA-box DNA in the promoter region of Ssfdh1. The SsNsd1-SsFdh1 interaction and nuclear translocation were found to prevent efficient binding of SsNsd1 to GATA-box DNA. Our results provide insights into the role of the GATA transcription factor and its regulation of formaldehyde dehydrogenase in stress resistance, fungal sclerotium and compound appressorium development, and pathogenicity.

The GATA-type transcription factors (TF) commonly occur in fungi, plants, and metazoans, with a DNA-binding motif, usually constituted by a four-cysteine zinc finger (ZnF), specifically binding to a six-base-pair (A/T)GATA(A/G) DNA sequence (7,8). As transcriptional activators or repressors, the regulatory roles of GATA TFs are diverse in controlling the expression of downstream genes and governing cell differentiation and development. In fungi, GATA TFs play key roles in nitrogen metabolism, light perception, siderophore biosynthesis, and mating-type switching (9)(10)(11). In S. sclerotiorum, there are nine GATA-type TFs (12). We studied one of these GATA-type TFs, S. sclerotiorum Nsd1 (SsNsd1), orthologous to Aspergillus nidulans nsdD (for "never in sexual development D"), and found that SsNsd1 is functionally involved in asexual-sexual development, sclerotium development, and compound appressorium formation (6). However, there is little information on the SsNsd1-mediated signal pathways, its interacting partners, and its role in nitrogen metabolism in S. sclerotiorum.
To study the SsNsd1-mediated regulatory pathway, we used a yeast two-hybrid (Y2H) method to identify its interacting proteins and identified a formaldehyde dehydrogenase (SS1G_10135), SsFdh1, in S. sclerotiorum. The functions of SsFdh1 were characterized by a gene knockout strategy with a focus on nitrogen metabolism, sclerotium development, and virulence on host plants. We further characterized the importance of key Cys residues in the ZnF motifs for SsNsd1-SsFdh1 interaction, SsNsd1 GATA-box DNA binding, and protein functions. This information reveals the biological functions of SsFdh1 and SsNsd1-associated regulatory pathways for nitrogen metabolism and development in S. sclerotiorum.

RESULTS
SsFdh1 functions in formaldehyde detoxification and sodium nitroprusside (SNP) tolerance. We previously reported that Ssnsd1 was involved in S. sclerotiorum asexual-to-sexual development (6). To get further insight into the SsNsd1-mediated regulatory mechanism involved in the development of S. sclerotiorum, we used SsNsd1 as bait to search against the S. sclerotiorum cDNA library via a Y2H approach and identified 40 SsNsd1-interacting protein candidates (see Data Set S1 in the supplemental material). Among them, one formaldehyde dehydrogenase (SS1G_10135), SsFdh1, was further confirmed to interact with SsNsd1 in yeast cells (Fig. S1A).
Ssfdh1 knockouts (KOs; KO.1 and KO.2) (Fig. S1B) and genetic complementation strains (C-22), by introduction of an Ssfdh1 genomic region, were generated in S. sclerotiorum (Fig. 1). We obtained two KOs (ΔSsfdh1 KOs; KO.1 and KO.2) and one complementation strain (C22) after molecular confirmation performed with PCR ( Fig. S1C and D) and Southern blotting (Fig. 1A). We further observed the hyphal morphology under light microscopy and found that all of the mutants and the control had similar mycelial branching patterns and mycelial convergent states at the early stages (Fig. S1E), suggesting that we could use these strains for further analysis under different conditions. We assayed enzyme activity with different strains using different substrates. ΔSsfdh1 KO strains exhibited significantly impaired FDH activity against formaldehyde com-pared to the wild type (WT), while introduction of Ssfdh1 to KO.1 complemented this compromised enzyme activity (Fig. 1B). In this assay, we did not observe any differences among the tested strains under conditions of supplementation of the mycelial crude extract with either methyl alcohol or methanoic acid (Fig. 1B). The FDH activity of SsFdh1 was further characterized by measuring radial hyphal growth in the presence of Southern blot assay was performed to validate the presence or absence of selected DNA fragments, such as the Ssfdh1 genomic DNA region (Probe 1), a partial hygromycin resistance gene (Hyg) fragment (Probe 2), and a partial G418 resistance gene fragment (Probe 3). (B) SsFdh1 detoxifies formaldehyde. Relative enzyme activity levels of formaldehyde dehydrogenase were tested with 1.5 ml formaldehyde or methyl alcohol or methanoic acid as a substrate. (C) Deletion of the Ssfdh1 gene inhibits vegetative growth under conditions of formaldehyde or sodium nitroprusside (SNP) stresses. Mycelium-colonized agar plugs of WT, KO.1, KO.2, and C-22 strains were cultured on PDA medium containing 0.3 mM formaldehyde or 0.25 or 0.5 mM SNP at 28°C for 6 days. (D) Mycelium diameters of colonies of the tested strains were measured, and growth inhibition was evaluated by 6 days after inoculation (DAI). (E) Deletion of Ssfdh1 affects sclerotium development. Mature sclerotia were collected from a 10-day culture on autoclaved smashed potato agar (SPA) medium at 25°C. (F) Comparison of sclerotium dry weights and numbers among WT, KO, and C-22 strains. Sclerotia were collected from 15-cm-diameter SPA plates for statistics analysis. (G) Expression profiles of sclerotium-associated genes in Ssfdh1 deletion mutants. Relative expression levels of sclerotium development-associated genes were investigated in hyphal tissue in Ssfdh1 deletion mutant and WT strains. The constitutively expressed histone H3 gene (XM_001589836.1) was used as the reference gene to standardize data. The experiments were performed in triplicate (*, P Ͻ 0.05; ***, P Ͻ 0.01 [Student's t test]). Values are means Ϯ standard errors (SE).
formaldehyde. In this assay, the growth of KOs was completely inhibited on potato dextrose agar (PDA) medium containing 0.3 mM formaldehyde (Fig. 1C). Again, the introduction of WT Ssfdh1 into the KO line rescued this phenotype, since the C-22 strain could grow as well as the WT strain (Fig. 1C). The KOs were also evaluated for their ability to resist exogenous NO (nitric oxide) stress. The level of mycelial growth of all of the tested strains was reduced after 6 days on PDA medium containing sodium nitroprusside (SNP), a NO donor. However, the KOs displayed more-severe growth inhibition ( Fig. 1C and D). Under these conditions, the growth of strain C-22 was similar to that of the WT strain.
Deletion of Ssfdh1 affects sclerotium development. The sizes and dry weights of mature sclerotia from KOs were significantly lower than those of the WT (Fig. 1E and F), while the numbers of mature sclerotia from KOs were similar to those seen with the WT (Fig. S1F and G). Introduction of Ssfdh1 complemented the sclerotium development phenotypes (Fig. 1E and F).
Ssfdh1 is associated with compound appressorium formation and pathogenicity. The virulence of KOs was strongly reduced on healthy bean leaves ( Fig. 2A and C) at 2 days after inoculation (DAI). Both KO strains exhibited a decreased ability to infect healthy host leaves compared to the WT and C-22 strains and developed smaller lesion areas ( Fig. 2A and C). The development of compound appressoria plays an important role in pathogenicity (5,6). We found that, at 1 DAI, the KOs had developed fewer pigmented compound appressoria than did the WT and C-22 strains on glass slides or parafilm ( Fig. 2E to G). What is more, KO.1 and KO.2 did not produce normal compound appressoria of the same size as those generated by WT and C-22 (Fig. 2E).
We tried to evaluate the transcript levels of infection cushion generation-related genes, such as Ssemp1, Ssmst12, Sspls1, Sschm1, Ssmas2 (26), and SscpkA (27), via qRT-PCR in the WT and ΔSsfdh1 strains. The mRNA levels of these genes were diminished in the ΔSsfdh1 mutant during compound appressorium development (Fig. 2H). We asked whether the impaired pathogenicity on the healthy leaf tissues was caused by impaired infection cushion formation, and we inoculated the wounded common bean leaves with different strains. The inoculation results showed that the KOs successfully infected the wounded leaves in a manner similar to that seen with the WT and C-22 strains (Fig. 2B). However, the lesions produced by the KOs were less extensive (smaller in size) than those produced by the WT and C-22 strains ( Fig. 2B and D), indicating that SsFdh1 is also involved in postpenetration pathogenicity.
Ssfdh1 is important in osmotic and oxidative stress resistance. To study how Ssfdh1 affects penetration-independent pathogenesis-associated stress adaptation, we cultured the WT, KO, and C-22 strains on PDA containing different stress agents. Under osmotic stress conditions, such as 1 M KCl, 1 M NaCl, or 1 M sorbitol, KOs showed inhibited hyphal growth compared to the control strains as determined by analysis of the visible hyphal growth on the plates ( Fig. S2A and B). The KOs also showed smaller colony diameters under conditions of SDS treatment, which can destroy cells by dissolving proteins and lipids in the cell membrane ( Fig. S2A and B). We also subjected different strains to oxidative stress by adding H 2 O 2 , and we found that when we enhanced the oxidative stress by increasing the concentration of H 2 O 2 from 5 to 20 mM, the hyphal growth of the WT, the KOs, and C-22 was increasingly inhibited ( Fig. S2C and D). However, the KOs were more sensitive to H 2 O 2 than the control strains under the same treatment conditions ( Fig. S2C and D).
SsNsd1 directly regulates Ssfdh1 transcripts. We profiled the expression levels of Ssnsd1 and Ssfdh1 during different developmental stages and found that the expression levels of Ssnsd1 and Ssfdh1 were spatially and temporally synchronized (Fig. 3A). Both Ssnsd1 and Ssfdh1 showed higher transcription accumulation levels in the hyphal stage than in the sclerotium (S1 to S5) and apothecium (A1 to A6) stages (Fig. 3A). To explore the genetic relationship between Ssnsd1 and Ssfdh1, we determined the levels of expression of both genes in different genetic backgrounds by qRT-PCR. The qRT-PCR results showed that the expression levels of Ssnsd1 were similar in the WT and ⌬Ssfdh1 strains (Fig. 3B), whereas the transcription level of Ssfdh1 in the ⌬Ssnsd1 strain was drastically reduced relative to the WT (Fig. 3C), suggesting that SsNsd1 regulates Ssfdh1 transcripts. SsNsd1 contains a zinc finger (ZnF) GATA binding domain at the region spanning amino acids (aa) 320 to 385 of its C terminus, which may recognize and bind a core (A/T)GATA(A/G) consensus sequence (11). We then scanned a 2-kb Ssfdh1 promoter region and identified two core GATA sequences located at nucleotides Ϫ446 and Ϫ439 ( Fig. 3D; see also Data Set S2).
Next, we asked whether SsNsd1, as a GATA TF, could directly bind to the GATA sequences in the Ssfdh1 promoter region. We performed an in vitro electrophoretic mobility shift assay (EMSA) with a probe containing two predicted GATA consensus sequences as indicated in Fig. 3D. SsNsd1 increasingly bound to the DNA probe along with increasing recruitment of SsNsd1 protein (Fig. 3E). In the full EMSA, SsNsd1 efficiently bound to the GATA-box DNA probe (Fig. 3F, lane 2). The specificity of the SsNsd1-DNA complex was confirmed by adding an excess of unlabeled DNA (Fig. 3F, . A quantity of SsNsd1 protein (500 ng, 600 ng, 700 ng, 800 ng, 900 ng, or 1,000 ng) was incubated with a 30-bp Ssfdh1 promoter probe. With the increase in the SsNsd1 amount, more DNA shift was observed with less free probe accumulation at the bottom. (F) Further confirmation of SsNsd1 GATA-box binding. The labeled DNA probe was preincubated with 500 ng SsNsd1, and then a 100-fold excess of unlabeled special competitor (unlabeled DNA probe), a 100-fold excess of unlabeled mutant competitor (unlabeled mutant DNA probe), or 1 g anti-GATA TF commercial antibody was added. The presence or absence of the reaction content is indicated with a minus sign or a plus sign, respectively. (G) Ssnsd1 and Ssfdh1 expression levels of the WT strain in the mycelial stage of growth was tested by qRT-PCR. The WT strain was cultured on the PDA medium with formaldehyde, methyl alcohol, or methanoic acid. The histone H3 gene was used as the reference gene to standardize the data. Statistically significant differences are indicated by asterisks (**, lane 3). As expected, a shift was detected when the unlabeled DNA was mutated (Fig. 3F, lane 4). We used commercial anti-GATA TF antibody to visualize the super shift of the protein-DNA complex (Fig. 3F, lane 5), and the results clearly showed a supershifted band. These data suggest that SsNsd1 can directly bind to the GATA sequence in the Ssfdh1 promoter. qRT-PCR evaluation of Ssfdh1 and Ssnsd1 transcripts in WT hypha treated with formaldehyde, methyl alcohol, and methanoic acid showed that both Ssnsd1 and Ssfdh1 had significantly higher mRNA levels after formaldehyde treatment (Fig. 3G). Together, these data suggest that SsNsd1 has the potential to directly regulate the Ssfdh1 gene by recognizing its GATA-box.
Overexpression of Ssfdh1 in the ⌬Ssnsd1 strain partially restored the sclerotia and compound appressorium deficiency. We sought additional evidence for SsNsd1 regulation of Ssfdh1 by comparing development-associated phenotypes of WT, ΔSsfdh1, and ΔSsnsd1 strains. The Ssnsd1 and Ssfdh1 deletion mutants shared similar vegetative developmental phenotypes (Fig. 4A to C). Phenotypically, both the ΔSsnsd1 and ⌬Ssfdh1 strains exhibited defective development with respect to the size and dry The WT had a high (85%) carpogenic germination rate, but only about half of the sclerotia from the ⌬Ssfdh1 mutant formed apothecia, while the ⌬Ssnsd1 mutant lacked the ability to produce apothecia (Fig. 4B) (6). With regard to pathogenicity, the WT strain had fully colonized detached healthy leaves by 3 DAI, while colonization was delayed by 3 to 4 days in the ⌬Ssfdh1 mutant and the ⌬Ssnsd1 mutant was nonpathogenic on unwounded host tissue (Fig. 4C).
We then expressed the SsFdh1 protein under the control of a strong OliC promoter in the ΔSsnsd1 background and found that overexpression of Ssfdh1 did not provide full rescue but did restore some of the deficiencies in production of sclerotia and compound appressoria (Fig. 4D and E). The ΔSsnsd1 mutant was unable to generate compound appressoria (6), while overexpression of Ssfdh1 partially restored the loss of compound appressoria (Fig. 4F). Consequently, overexpressing Ssfdh1 in the ΔSsnsd1 background allowed host penetration but symptom development was delayed (Fig. 4G).
SsNsd1 interacted with SsFdh1 in a disulfide bond-dependent manner. To further confirm the interaction between SsFdh1 and SsNsd1, we transiently coexpressed SsFdh1-3xFLAG and SsNsd1-green fluorescent protein (SsNsd1-GFP) in Nicotiana benthamiana leaves and immunoprecipitated the SsNsd1-GFP with GFP-IP and co-IP results showed that SsFdh1 associates with SsNsd1 (Fig. 5A).
To study how SsNsd1 interacts with SsFdh1 in S. sclerotiorum, the three-dimensional (3D) structures of SsFdh1 and SsNsd1 were modeled. SsFdh1 (aa 8 to 374) was predicted to belong to the Zn-dependent dehydrogenase family due to BLAST homology results in which a zinc ion was found in the predicted structure (Fig. S3A) (28). The active sites of ADH were predicted with two divalent zinc ion-cysteine sites (Cys 44 and Cys 173) that aid catalytic activities and contribute to substrate binding (29) and that are strictly conserved in the SsFdh1 protein (Fig. S3A). The modeled SsNsd1 showed a ZnF module possessing four cysteine residues (Cys 331, Cys 334, Cys 353, and Cys 356) coordinated to a single zinc ion (Fig. S3B) which can stabilize protein structure and function in DNA binding or in protein-protein interactions (30). The ZnF domain is involved in interactions with other ZnF proteins (31). Thus, we assumed that the cysteine residues in this region were crucial for their functions and interaction. Since cysteine-rich protein can form disulfide bonds, we hypothesized SsNsd1 might interact with SsFdh1 through disulfide bond formation. To test this hypothesis, we mutated the cysteine to an alanine residue (C to A) in SsNsd1 and SsFdh1 (Fig. S4A). As the SsNsd C353/C356 mutations still maintained the interaction with SsFdh1 (Fig. S4B), Cys residues C353 and C356 of SsNsd1 did not contribute to this interaction. We examined the possibility that C331 and C334 were involved in the interaction ( Fig. 5B and C). In yeast, SsFdh1 C44 interacted with SsNsd1 C331 , but not with SsNsd1 C334 , and SsFdh1 C173 interacted with SsNsd1 C334 , but not with SsNsd1 C331 (Fig. 5B). The mutant with the SsFdh1 C44/C173 double mutation did not maintain the interaction with SsNsd1, and the SsNsd1 C331/C334 double mutant was not able to interact with SsFdh1 (Fig. 5B). These results were further confirmed in Arabidopsis thaliana protoplast with a bimolecular fluorescence complementation (BiFC) assay (Fig. 5C). These results confirmed the SsFdh1-SsNsd1 interaction and indicated that the correct interaction was mediated by disulfide bonds (SsFdh1 C44-S-S-C331 SsNsd1 and SsFdh1 C173-S-S-C331 SsNsd1) as shown in Fig. 5D.
SsNsd1 enriched SsFdh1 in the nucleus. Our BiFC results showed that SsFdh1 interacted with SsNsd1 in the nucleus (Fig. 5C). However, results obtained using a protein subcellular localization prediction tool (PSORT) indicated that SsNsd1 is a nucleus-localized protein and SsFdh1 was predicted to be cytoplasmic. We hypothesized that the interaction between SsNsd1 and SsFdh1 might change the native subcellular localization of SsFdh1. To test this hypothesis, a localization assay was performed in S. sclerotiorum. First, SsNsd1-GFP was expressed in hyphae of the WT strain and SsNsd1 was clearly observed to localize in the nucleus (Fig. 6A). We found that SsFdh1 localized in the nucleus and cytoplasm when SsFdh1-mCherry was ex- pressed in WT hyphae but that SsFdh1 was mostly localized in the cytoplasm when expressed in the ΔSsnsd1 mutant (Fig. 6B). This result illustrated that SsNsd1 recruits SsFdh1 to the nucleus. SsNsd1-GFP and SsFdh1-mCherry were then coexpressed in the ΔSsnsd1 mutant (Fig. 6C). SsNsd1 was localized in the nucleus, and SsFdh1 exhibited both nuclear and cytoplasmic localization (Fig. 6C). When SsNsd1 C331/C334 -GFP and SsFdh1-mCherry were coexpressed in the ΔSsnsd1 mutant, Nsd1 C331/C334 was still localized in the nucleus, but SsFdh1 localized to the cytoplasm. This appears to have been due to its inability to interact with SsFdh1 (Fig. 6C).
To test whether stress conditions can change the interaction between SsFdh1 and SsNsd1, coexpression strains were exposed to formaldehyde or H 2 O 2 (Fig. S5) . SsFdh1 was still found in the nucleus of compound appressoria when coexpressed with SsNsd1 under conditions of formaldehyde or H 2 O 2 stress (Fig. S5A). However, this nuclear localization was lost when SsNsd1 C331/C334 -GFP and SsFdh1-mCherry were coexpressed in the ΔSsnsd1 mutant under the same stress conditions. The same results were seen in the hyphae of the ΔSsnsd1 mutant (Fig. S5B).
To study the interaction of SsFdh1 and SsNsd1 during the infection process, the coexpression strains were inoculated on onion epidermal cells. SsFdh1 could be observed in the nucleus of compound appressoria induced on onion epidermal cells. This nucleus localization was lost when SsNsd1 C331/C334 -GFP and SsFdh1-mCherry were coexpressed in the ΔSsnsd1 mutant (Fig. S5C). A similar result was observed in the hyphae of the ΔSsnsd1 mutant (Fig. S5D).
SsFdh1 interacts with SsNsd1 and prevents SsNsd1 binding to DNA. We performed in vitro EMSA with a GATA-box DNA probe such as we used in the experiments represented by Fig. 3, with SsNsd1 WT protein, and with SsNsd1 protein with point mutations. SsNsd1 C331/C334 lost the capacity to bind DNA (Fig. 7A). ZnFs are generally regarded as DNA-binding motifs; however, some reports have implicated ZnFs in the mediation of protein-protein interactions (11) and ZnF domains interacting with a protein might lose the ability to bind DNA (31). To test whether the SsFdh1-SsNsd1 association interferes with SsNsd1-DNA binding, GATA probe, SsNsd1 protein, SsFdh1, and SsFdh1 C44/C173 were tested with the EMSA. Panel B of Fig. 7 shows that SsNsd1 efficiently bound to the GATA probe as described for Fig. 3. In contrast, surprisingly, the coexistence of SsFdh1 and SsNsd1 completely eliminated the DNA affinity of SsNsd1 and no shift band was observed (Fig. 7B, lane 3). Disruption of the SsFdh1-SsNsd1 interaction by point mutations (SsFdh1 C44/C173 ) allowed DNA binding of SsNsd1 (Fig. 7B, lane  4). We also observed clearly supershifted bands when we used anti-GATA TF commercial antibody (Fig. 7B, lane 5). These data suggest that the SsNsd1-SsFdh1 association interferes with SsNsd1-DNA binding.
We then predicted a 3D structure using the SsNsd1-DNA binding model and a simulated pattern diagram of SsNsd1 interacting with SsFdh1 by hypothetically combining their individual protein models (Fig. 7C and D). The results showed that SsFdh1 binding to SsNsd1 protein blocks efficient recognition of or binding to GATA-box DNA by GATA factor SsNsd1 (Fig. 7E).
The interaction core Cys residues are essential for function of SsFdh1 and SsNsd1. The four cysteine residues play crucial roles in SsNsd1-SsFdh1 interactions. To study the biological relevance of these cysteine residues, knockout mutants and corresponding strains complemented with double point mutations were phenotypically analyzed during compound appressorium development and stress resistance (Fig. 8). In the context of compound appressorium development and pathogenicity, the KO strains complemented with Cys mutations could not rescue the development and pathogenicity deficiency phenotypes, since those complementations exhibited effects on compound appressorium development and pathogenicity similar to those seen with the KOs (Fig. 8A to D). In addition, point mutations of SsFdh1 or SsNsd1 had similar levels of sensitivity to H 2 O 2 and formaldehyde stress (Fig. 8E). These results demonstrate that core Cys residues play a role in biological functions that is as important as their role in maintaining the interactions of SsFdh1 and SsNsd1.

DISCUSSION
We previously characterized the functions of SsNsd1 in S. sclerotiorum (6) in the context of sclerotium development and pathogenicity. In this study, we identified SsFdh1 as an SsNsd1-interacting protein using the Y2H method and this interaction was further confirmed by different methods (Fig. 5; see also Fig. S1A). SsFdh1 is classified as a glutathione-dependent FDH with common activity involving detoxification of endogenous and exogenous formaldehyde (32). As expected, the ΔSsfdh1 mutants showed lethal phenotypes in the presence of formaldehyde stress (Fig. 1). FDHs also have S-nitrosoglutathione (GSNO) reductase activity (17), controlling GSNO and Snitrosothiol (SNO) levels in addition to detoxifying formaldehyde in cells (33). In yeast, fdh mutant cells show increased susceptibility to nitrosative challenges, indicating that GD-FDHs provide protection against nitrosative stress (15). In line with this, we also observed that Ssfdh1 mutants displayed greater sensitivity to exogenous SNP stress than the controls, as did their counterpart in filamentous fungus M. oryzae (Fig. 1) (14). Endogenous NO in fungi can also be involved in conidiation germination (34) and in formation of infection structures (35). Thus, loss of Ssfdh1 might affect growth and development by affecting Ssfdh1-mediated NO metabolism during these biological processes.
Sclerotium development involves several distinct stages (36) and is tightly regulated by many intrinsic genetic factors. Losing the capacity to produce normal sclerotia in S. sclerotiorum could disrupt the disease cycle and the ability to cause disease (6,23,25). We found that the ΔSsfdh1 mutant lost its capability to produce normal sclerotia and compound appressoria and showed severely impaired pathogenicity compared to the control strains (Fig. 2). Similarly, after deletion of its homologous protein, MoSFA1, M. oryzae could still form appressoria, whereas the turgor pressure and virulence of appressoria were severely reduced (14). The ΔSsfdh1 mutant produced lesions that were less extensive than those produced by the WT and C-22 strains on wounded host, indicating a penetration-independent virulence strategy (Fig. 2B). To cope with pathogen infection, production of reactive oxygen species (ROS) and of reactive nitrogen species (RNS) is one of a variety of mechanisms that plants have evolved to combat pathogen attack (37). Thus, plant pathogens encounter stresses imposed by hosts upon penetration into host cells, and the resistance or adaptation of pathogens to these stresses has been widely regarded as contributing to pathogenesis (38). We observed attenuated virulence in host infection with Ssfdh1 KO mutants, and those KOs showed significantly decreased resistance to stress agents, such as salts, oxidants, and SDS (Fig. S2). This indicated that SsFdh1 might play an important role in infection-associated stresses beyond formation of compound appressoria.
Nitric oxide (NO) is a short-lived, endogenously produced radical that acts as a signaling molecule (39), and GD-FDHs with GSNO reductase activity, critical for nitrosative stress, are evolutionarily present in yeast, humans, and plants (17). For example, elaborate enzymatic defenses against the nitrosative stress mounted by the host are used to promote fungal virulence of Cryptococcus neoformans (18). MoSFA1-mediated NO metabolism functions in redox homeostasis and protein S-nitrosylation in response to development and host infection of M. oryzae (14). In this study, the ΔSsfdh1 mutant showed sensitivity to SNP stress, confirming its GSNO reductase activity. Given the positive roles of FDHs in protecting pathogens against infection-associated stresses, SsFdh1 could help overcome host defense and increase S. sclerotiorum pathogenicity. This result supports the attenuated virulence of the mutants observed on the wounded host (Fig. 2B).
The GATA-type ZnF factor NsdD regulates the early stages of the sexual reproduc-tion pathway in A. nidulans (40,41). An important and complex NsdD signaling pathway involving brlA, rosA, nosA, nsdB, and NsdD in A. nidulans has been shown to regulate sexual and asexual development, pathogenicity, and physiology (41,42). In S. sclerotiorum, deletion of Ssnsd1 reduced the expression of Ssfdh1 (Fig. 3), indicating that SsFdh1 might function as a downstream target regulated by SsNsd1. In support of this, SsNsd1 could directly bind to the GATA-box DNA in EMSA (Fig. 3). However, binding sites of other TFs, such as TFIID, CTF, and AP-1, were also found in the Ssfdh1 promoter region, indicating that other TFs or other GATA factors might be involved in the regulation of Ssfdh1 (Fig. 3). This accounted for the detectable expression of Ssfdh1 in the ΔSsnsd1 mutant. Similarly, NsdD repressed conidiation by binding to the GATA-box upstream of brlA, but loss of flbC or flbD in the absence of nsdD resulted in delayed activation of brlA, suggesting positive roles of other activators in conidiation development of A. nidulans (42). Deletion of Ssfdh1 produced only partial developmental phenotypes compared to the Ssnsd1 mutants, while overexpression of Ssfdh1 did not restore the full function of the Ssnsd1 mutant, indicating that Ssfdh1 was only one of the downstream genes that SsNsd1 could regulate during sclerotium development, compound appressorium formation, and pathogenicity. In A. flavus, NsdD is essential for transcriptional control of a series of genes that are vital for the morphological development of sclerotia and the production of conidia (43). In this regard, SsNsd1could also regulate several downstream targets during compound appressorium formation (12). ZnFs can act as protein-protein recognition platforms in addition to their familiar role as DNA recognition motifs (30). In vertebrates, the GATA-1 factors contain two ZnF domains, termed the N-finger and the C-finger domains. Generally, the C-finger domain is involved in sequence-specific DNA binding, and the N-finger is involved in the mediation of protein-protein interactions with other ZnF proteins (44). In fungi, the GATA factors contain only one DNA-binding domain (11). We confirmed that SsNsd1 could bind GATA-box DNA in vitro and that SsNsd1 physically associates with SsFdh1 in a disulfide bond-dependent manner. Mutagenesis analysis reveals the importance of Cys residues in the ZnF domain in both proteins. When mutated (C331/C334), SsNsd1 could neither bind to the GATA-box sequence from Ssfdh1 promoter region nor associate with SsFdh1, clarifying the mechanism by which ZnF proteins might interact with other ZnF proteins. It is likely that fungal GATA proteins utilize the same ZnF domain for both DNA and other ZnF protein binding. This dual role appears to be important for regulation as the SsNsd1-SsFdh1 interaction occurs in the nucleus and prevents SsNsd1 GATA-box binding. SsNsd1 recruits SsFdh1 to the nucleus by physical interactions that might prevent the binding of SsNsd1 to its target DNA regions, while SsNsd1 regulates Ssfdh1 transcriptionally, suggesting the presence of a regulatory loop by which the SsNsd1-SsFdh1 interaction and nuclear localization can interfere with the SsNsd1 factor to activate or repress downstream genes. However, we observed consistent phenotypes in Ssnsd1 and Ssfdh1 deletion mutants, indicating that SsNsd1 functions partially through SsFdh1 and that the expression of Ssfdh1 is regulated by other unidentified factors. Similarly, two ZnFs of EVI1 (ecotropic viral integration site 1) protein have been demonstrated to interact with the ZnF of GATA1 and compete with GATA1 for DNA-binding sites, resulting in repression of gene activation by GATA1 (31). ZnF motifs of GATA factor are highly conserved in different organisms, and we also found a conserved GATA-box in the promoter region of FDH homologs from different species (Fig. S6). Consistent with the observations that GATA factors and FDHs play key roles in nitrogen metabolism among many different fungal species (9-11, 14, 15, 17), our results reveal a possible mechanism by which SsNsd1 regulates Ssfdh1 and functionally joins with SsFdh1 to repress further transcription of Ssfdh1 (Fig. S7), ensuring proper accumulation of FDH in fungi nucleus, where it could help maintain NO homeostasis around the chromatin region.
Transformants were cultured on PDA medium with 100 g ml Ϫ1 hygromycin B (Roche, Indianapolis, IN, USA) or 100 g ml Ϫ1 G418 (Sigma-Aldrich, Shanghai, China). Dry stocks of hyphae were maintained at -20°C. Escherichia coli strain DH5␣ (TaKaRa, Dalian, China) was used to propagate plasmids, and the Agrobacterium tumefaciens EHA105 strain was used for the fungal transformation (45). Sclerotia generated on PDA plates were gathered, air-dried, and stored at -20°C.
Arabidopsis thaliana Col-0 and Nicotiana benthamiana were grown under greenhouse conditions at 21 to 23°C under conditions of a 16-h light/8-h dark cycle with 60% relative humidity (46). Common beans (Phaseolus vulgaris) and tomato (Solanum lycopersicum) were grown under fluorescent lighting in a temperature range of 22 to 25°C in the laboratory. The seeds were planted in a potting soil mix (vermiculite; humus ϭ 1:2).
Yeast two-hybrid (Y2H) assay and sequence analysis of SsFdh1. A Matchmaker Gold Y2H system (Clontech, CA, USA) was used to screen for interactors as well as for identification of protein-protein interactions. The bait protein SsNsd1 was used to search against a full-length cDNA library of S. sclerotiorum which was constructed with cDNA from fresh WT mycelial tissues according to the instructions provided for the Matchmaker GAL4-based system. SsFdh1 (SS1G_10135) was identified as one of SsNsd1-interacting proteins. For protein-protein interaction assay, the coding sequences of Ssnsd1 and Ssfdh1 were inserted into yeast vectors pGBKT7 and pGADT7, respectively. The Y2H assay was performed according to the standard operational procedures (47).
SmartBLAST was used to search for homologs and to localize highly conserved domains (48). Multiple alignments of protein sequences were performed using BioEdit version 7.0.5 (Tom Hall, CA, USA). The 3D structural modeling of SsFdh1 and SsNsd1 protein was performed on SWISSMODEL (http://www.expasy .ch). The structural model of SsFdh1 and SsNsd1 was visualized with the PyMOL program (http://www .pymol.org/pymol). Manual model building and experiments involving interactions of protein were performed manually using RasMol 2.7.5 (http://www.openrasmol.org/). Transcription factor binding sites in a 2-kb Ssfdh1 promoter region were determined with LASAGNA-Search 2.0 (http://biogrid-lasagna .engr.uconn.edu/lasagna_search/). Gene replacement and complementation. All the primers used in this study are listed in Table S1 in the supplemental material. The Ssfdh1 gene was deleted by double homologous recombination. The 5= region (ϳ0.9 kb) and the 3= region (ϳ1.0 kb) of the Ssfdh1 gene were amplified from the genomic DNA (gDNA) of WT S. sclerotiorum and were cloned into the pXEH vector (46) to generate a pXEH-R-L construct containing a hygromycin phosphotransferase (hph) cassette driven by a trpC promoter. Hygromycinresistant transformants obtained by protoplast transformation (25), namely, knockout (KO) strains, were verified by PCR. All of the transformants were purified by three rounds of hyphal-tip transfer.
For genetic complementation of Ssfdh1 KOs, a genomic region containing the full-length fragment of Ssfdh1 (ϳ3.9 kb), including 944-bp 5= upstream and 1,027-bp 3= downstream of the coding sequence, was amplified from the WT S. sclerotiorum gDNA. This fragment was then cloned into pD-NEO1 carrying the strong promoter trpC, which was constructed from vectors pSilent-Dual1 and pD-NAT1 (46). KO strain Ssfdh1-KO.1 was used for genetic complementation. The complementation transformants were screened on PDA medium with 100 g ml Ϫ1 G418 and then verified by PCR.
To identify mutants, gDNA was extracted from the WT, the KOs, and a complementation strain (C-22) for Southern blot analysis. Extracted gDNA was digested with XbaI and BamHI (TaKaRa, Dalian, China), separated on 0.8% agarose gel by electrophoresis, and transferred to a nylon membrane (Hybond-N ϩ ; GE Healthcare). Probe 1 was the PCR-amplified product of an 878-bp fragment from the Ssfdh1 gene. Probe 2 was a 768-bp fragment amplified from the hph gene. Probe 3 was a 672-bp fragment amplified from a G418 resistance gene.
Morphological characterization. Agar-mycelium plugs from the WT strain, KOs, and strain C-22 were placed on PDA medium at 25°C. The diameters of colony on PDA medium were measured at different time points to evaluate the radial mycelial growth. For quantification of compound appressoria, fresh mycelial plugs were inoculated on paraffin film and then incubated under suitable humidity conditions for 3 days. For morphological observation of compound appressoria, a fresh PDA-colonized agar plug (5 mm in diameter) was placed on a glass slide (Sail brand) and was placed in a box at room temperature for 24 h. Compound appressoria were observed with an optical microscope (BA310Met-T; Ted Pella, Xiamen, China). Mature sclerotia of each strain were collected from smashed potato agar (SPA) medium and photographed at 7 days after inoculation (DAI) using a Sony SELP1650 digital camera (Sony, Tokyo, Japan). Sclerotium development was determined by measuring dry weight and size. For induction of apothecia, mature sclerotia were cultured in a 15°C incubator at 70% humidity and with constant illumination. Germination rates (corresponding to the proportion of sclerotia producing apothecia) were calculated after 60 days as described previously (6).
Enzyme activity assay. UV spectroscopy was used to analyze enzyme activities (38). Different strains were cultured in liquid potato dextrose medium (24 h shake/24 h static, normal illumination) at room temperature for 7 days. The supernatant was removed by centrifuging 25-ml cultures at 7,000 rpm and 4°C. To obtain intracellular crude enzymes from hypha, mycelia were added with 50 ml 0.05 M potassium phosphate buffer solution (pH 7.5) and washed twice; the same buffer solution was then added to the mycelium pellet (1.5:1 ratio [vol/vol]) to suspend the pellet. The mycelium cells were then crushed by the use of an ultrasonic cell crusher (2 horn; 22% power; ultrasound 2 s, stop 3 s) on ice for 15 min and were centrifuged at 15,000 rpm for 30 min at 4°C to remove cell debris. Enzyme activity reactions were performed by combining 1 ml of crude extract from different genotypes with 1.5 ml various reaction buffers, such as 0.1 mM formaldehyde, methyl alcohol, or methanoic acid in 0.05 M potassium phosphate buffer solution (pH 7.5) at 37°C for 1 h. Radiation absorbance at 540 nm was determined by the use of a UV spectrometer. Enzyme activity was defined using the following equation: (units per milliliter) ＝ ml Ϫ1 bleomycin (61). For the colocalization assay, SsNsd1 and SsNsd1 C331/C334 were cloned into pCB-GFP vector, the resulting constructs were used to transform protoplasts of ⌬Ssnsd1 containing Ssfdh1-mCherry, and the transformants were screened on PDA with 50 g ml Ϫ1 chlorimuron-ethyl (62). Fluorescence localization was observed with a fluorescence inverted microscope (emission: GFP, 488 nm; mCherry, 587 nm) with nuclei stained with DAPI.