Transcriptomic, Protein-DNA Interaction, and Metabolomic Studies of VosA, VelB, and WetA in Aspergillus nidulans Asexual Spores

Filamentous fungi produce a vast number of asexual spores that act as efficient propagules. Due to their infectious and/or allergenic nature, fungal spores affect our daily life. Aspergillus species produce asexual spores called conidia; their formation involves morphological development and metabolic changes, and the associated regulatory systems are coordinated by multiple transcription factors (TFs).

In this study, we aimed to determine the cross-regulatory mechanisms of VosA/ VelB/WetA in fungal conidiation using comparative transcriptomic and metabolomic analyses of WT and null mutants of wetA, velB, and vosA in A. nidulans conidia. In addition, the direct targets of these regulators were identified by combining the results from the VosA-and VelB-chromatin interactions using ChIP sequencing (ChIP-seq) analysis with WetA direct targets identified in a previous study (25). The results clarify the detailed molecular mechanisms by which VosA/VelB and WetA control defined common and distinct regulons and increase the overall understanding of the regulatory networks that govern fungal cell differentiation and metabolism.

RESULTS
VosA-, VelB-, and WetA-mediated gene regulation in A. nidulans conidia. To understand the conserved and divergent regulatory roles of VosA, VelB, and WetA in A. nidulans conidia, a comparative analysis of gene expression differences between the WT and null mutant conidia was carried out (Fig. 1). Totals of 40.98% (4,503/10,988), 45.61% (5,012/10,988), and 51.96% (5,729/10,988) of genes of the A. nidulans genome are differentially regulated in the DvosA, DvelB, and DwetA mutant conidia, respectively, suggesting that the three regulators have a broad regulatory effect on conidia (see Fig. S1 in the supplemental material). A total of 2,143 differentially expressed genes (DEGs) between the WT and the DvosA, DvelB, and DwetA mutant conidia were identified ( Fig. 1A) (fold change of .2.0 for upregulation or downregulation and q value [false discovery rate {FDR}] of ,0.05). The mRNA expression levels of 890 genes were downregulated in all three mutant conidia compared with the WT conidia. However, in all three mutant conidia, the mRNA levels of 1,253 genes were upregulated. Among them, the mRNA expression levels of a variety of genes associated with asexual development and signal transduction were affected by these three TFs (Tables S1 and S2). Certain developmental regulator genes such as abaA, brlA, flbC, nosA, nsdC, velC, vapA, and esdC were upregulated in all three null mutants (Table S1). Genes associated with heterotrimeric G-protein signal transduction (GprC, GprC, GprK, GprM, FlbA, and RgsB) and the mitogenactivated protein (MAP) kinase pathway (MpkB and ImeB) were also upregulated by these TFs in conidia (Table S2). However, several genes related to sporulation, spore wall formation, and structural integrity, including rodA, conJ, tpsA, wA, vadA, and atfB, were downregulated in all three null mutants (Table S1). Importantly, 748 and 769 DEGs were down-or upregulated by both DvosA and DvelB mutant conidia, respectively, but not DwetA mutant conidia, while the mRNA levels of 2,792 genes were affected solely in the wetA-null mutant conidia. Put together, these results suggest that VosA and VelB share more DEGs, while the WetA regulon has many more uniquely regulated genes.
To gain further insight into the regulatory roles of these TFs, functional category analyses using Gene Ontology (GO) terms were carried out (Fig. 1B). The results of the GO analysis demonstrated that several genes involved in the monocarboxylic acid metabolic process, the oxidation-reduction process, the trehalose metabolic process, and the cellular carbohydrate metabolic process were downregulated in all three mutant conidia, whereas a large number of genes associated with the secondary metabolic biosynthetic process, the chitin biosynthetic process, asexual sporulation resulting in formation, and the (1-3)-b-D-glucan metabolic process were upregulated in these mutant conidia. The VosA-and VelB-specific downregulated genes were enriched in functional categories that included the cellular catabolic process, protein localization, and the acetate catabolic process. The functional GO categories associated with the VosAand VelB-specific upregulated genes were the secondary metabolic biosynthetic process, the steroid metabolic process, and transport (Fig. S2A). Interestingly, a large number of genes involved in the RNA metabolic process were downregulated in DwetA mutant conidia but not in DvosA or DvelB mutant conidia (Fig. S2B).
Putative direct targets of VosA, VelB, and/or WetA in conidia. Our previous studies reported that VosA contains the velvet DNA-binding domain, which recognizes the VosA-binding motif in certain promoter regions (29). To identify the VelB direct target genes and compare the putative direct target genes of VosA and VelB, ChIP experiments followed by high-throughput sequencing of the enriched DNA fragments were carried out. ChIPs from strains containing FLAG epitope-tagged versions of VosA and VelB were compared to ChIPs from WT conidia that did not contain the FLAG epitope. Totals of 1,734 and 655 genes that were VosA and VelB peak associated, respectively, were identified using the same analysis pipeline as the one described previously (25) (Fig. 2). To identify the VosA/VelB response elements, DNA sequences in the 100 bp surrounding each peak were subjected to Multiple Em for Motif Elicitation (MEME) analysis, which led to the predicted VosA response element (VoRE) and the predicted VelB response element (VbRE) ( Fig. 2A). Interestingly, the predicted VbRE (59-CCXTGG-39) was quite similar to the predicted VoRE (59-CCXXGG-39). The VoRE was found in 278/1,404 peak sequences, had an E value of 1.6e256, and was the only motif identified by MEME with an E value of ,1. The VbRE was found in 188/511 peak sequences, had an E value of 2.4e285, and was one of only two motifs identified by MEME with an E value of ,1 (the other motif had an E value of 4.0e25 and was found in only 72 peak sequences).
We then compared the results of the ChIP-seq and RNA sequencing (RNA-seq) analyses to identify potential direct target genes of the three TFs (Table S3). There were 66 genes associated with the peaks of all three TFs (Table S4). Among them, 22 genes, including flbA, xgeA, atfB, tpsA, vadA, cetA, nopA, and ppsA, were DEGs in all three null mutants (Fig. 2B). Importantly, 532 genes were considered to be potential direct target genes for both VosA and VelB but not WetA. A total of 166 genes were upregulated in both DvosA and DvelB mutant conidia. These genes, including brlA, fadA, rosA, steA, steC, and veA, were found to be involved primarily in asexual or sexual developmental processes. Taking these results together with the previously reported results (27,35), we suggest that VosA works with VelB and that the VosA-VelB complex coordinates the processes involved in conidial production and maturation in A. nidulans.
Roles of VosA, VelB, and/or WetA in conidial wall integrity. Previous studies have shown that the deletion of vosA, velB, or wetA leads to decreased amounts of trehalose and increased b-glucan levels in conidia (25,30). The results of transmission electron microscopy analyses revealed that three TFs are needed for the proper formation of the conidial wall (25,30,37), suggesting that these genes play a conserved role in regulating the expression of genes associated with conidial structural integrity. High-performance liquid chromatography (HPLC) analysis demonstrated that the trehalose contents of the three null mutant conidia were dramatically decreased (Fig. 3A). The mRNA expression levels of most genes involved in trehalose biosynthesis were downregulated ( Fig. 3B and Table S5). In addition, tpsA, a putative trehalose synthase gene, is the direct target of three TFs (Fig. 2B). These results suggest that three TFs directly or indirectly control the mRNA levels of genes associated with trehalose biosynthesis, thereby regulating the trehalose contents in conidia. Most genes associated with chitin and b-(1,3)-glucan biosynthesis were upregulated in the DvosA, DvelB, and DwetA mutant conidia ( Fig. 3C and D). These results suggest that VosA, VelB, and WetA govern the mRNA expression of genes associated with conidial wall integrity in A. nidulans. Alterations to primary metabolites in DvosA, DvelB, and DwetA conidia. As mentioned above, the deletion of vosA, velB, or wetA led to alterations in the mRNA expression of genes involved in metabolic processes (glycerol metabolic process, ketone metabolic process, and amino sugar metabolic process) and amino acid metabolism (Table S6), implying that the amounts of primary metabolites may be affected by the absence of vosA, velB, or wetA in conidia. To test this hypothesis, the abundances of several primary metabolites involved in the tricarboxylic acid (TCA) cycle and amino acid biosynthesis were examined in WT and mutant conidia (Fig. 4). The abundances of pyruvate, a-ketoglutarate, and malate were increased in the conidia of the three null mutants. The abundances of acetyl-CoA and succinate were decreased in both DvosA and DvelB, but not DwetA, mutant conidia. The amounts of lactate in both DvosA and DvelB mutant conidia were significantly large compared with those in WT conidia.
The abundances of 13 amino acids (alanine, isoleucine, methionine, leucine, phenylalanine, tryptophan, valine, threonine, serine, asparagine, glutamine, aspartate, and glutamate) were affected in at least one null mutant. Moreover, the levels of nine amino acids were high in all three mutant conidia. The effects of deleting vosA-velB or wetA on the abundances of glutamate, glutamine, aspartate, and asparagine differed. The deletion of wetA caused decreased levels of glutamate, glutamine, and asparagine in conidia, whereas the levels of these amino acids were increased or not affected by the absence of vosA or velB. The genes involved in the biosynthesis of these amino acids and primary metabolites were differentially regulated in the three null mutants. Overall, these results show that WetA, VosA, and VelB regulate the expression of genes involved in both the TCA cycle and amino acid biosynthesis; however, the three lists of primary metabolites that they affect contain both shared and unique molecules.
Abundances of secondary metabolites in DvosA, DvelB, and DwetA conidia. Previous studies found that these three TFs are important for the production of several Regulatory Roles of VosA, VelB, and WetA in Conidia ® secondary metabolites in Aspergillus species (34,36,40). In addition, according to the GO analysis results, the deletion of vosA, velB, or wetA results in an alteration of the mRNA expression of biosynthetic gene clusters involved in the production of multiple secondary metabolites, including monodictyphenone, sterigmatocystin, and asperfuranone ( Fig. 1, Fig. S2, and Table S7). To elucidate the conserved and divergent regulatory effects of secondary metabolism in the three conidial mutants, the secondary metabolites were extracted and subjected to liquid chromatography-mass spectrometry (LC-MS) analysis. A principal-component analysis showed differences between the four different conidial samples (Fig. S3). The secondary metabolite content of the WT conidia was relatively similar to that of the DwetA conidia, indicating similar abundances and types of secondary metabolites. Conidia from the DvosA and DvelB mutants clustered far apart, which suggested that a unique set of secondary metabolites or different levels of metabolites were expressed and extracted. This is interesting considering that the two TFs can interact and that their binding motifs and regulated gene lists were so similar to one another ( Fig. 1A and Fig. 2A).
Next, we applied analysis of variance to identify the most different molecular entities detected as mass/charge (m/z) values and retention time (RT) pairs in the LC-MS analysis-derived metabolomics data. As shown in Fig. 5, the abundances of several secondary metabolites were different in the positive and negative ionization modes. For example, the abundance of arugosin A was high in the DwetA conidia, compared with the WT conidia, but not in the DvosA and DvelB mutant conidia.
To further dissect the roles of VosA, VelB, and WetA in secondary metabolism, we focused on some known secondary metabolites, including sterigmatocystin, emericellamide, and austinol (Fig. 6). Sterigmatocystin is a precursor of aflatoxins, and its biosynthetic gene cluster and intermediates have previously been studied (41,42). The amount of sterigmatocystin in the DvelB conidia was significantly decreased compared with that in the WT conidia, but the DvosA and DwetA conidia contained similar amounts of sterigmatocystin (Fig. 6A). However, the amounts of sterigmatocystin intermediates were different in DvosA and DwetA conidia. Levels of norsolorinic acid and nidurufin were low in the DvelB and DwetA conidia, while the level of versiconol was  high only in the DvelB conidia. The RNA-seq results indicated that the mRNA levels of almost all of the genes in the sterigmatocystin gene cluster were increased in both the DvosA and DwetA conidia, whereas the mRNA expression of these genes in the DvelB conidia was less consistent. In particular, the mRNA levels of stcL, stcN, stcQ, stcS, stcT, stcU, stcV, and stcW were decreased in the DvelB conidia compared with the WT conidia. These results suggest that VosA and VelB play diverse roles in the regulation of sterigmatocystin biosynthesis.
Emericellamide compounds are cyclopeptides that are produced by several Aspergillus species (43,44). The abundances of these compounds, relative to WT production, were high in DvosA and DvelB conidia, and the mRNA levels of easA, easB, easC, and easD were also high in both mutant conidia, implying that VosA and VelB repress emericellamide biosynthesis in WT conidia (Fig. 6B). In the DwetA conidia, however, the mRNA expression of the emericellamide gene cluster was increased, but the quantity of emericellamide compounds did not increase, suggesting that the regulatory mechanism of emericellamide biosynthesis in the DwetA conidia is more complex than the influence of DvosA and DvelB on emericellamide production in conidia. In the three types of null mutant conidia, the abundances of two fungal meroterpenoids, austinol and dehydroaustinol (45), were decreased, compared with the WT conidia (Fig. 6C). Furthermore, the expression levels of several austinol cluster genes were decreased in the DvelB and DwetA conidia. Taken together, these results demonstrate that the ways in which VosA, VelB, and WetA govern the expression of secondary metabolite gene clusters, and the production of their associated metabolites, in A. nidulans conidia are divergent from one another.

DISCUSSION
Asexual developmental processes in filamentous fungi are regulated by a variety of TFs (6). These TFs orchestrate the spatial and temporal transcriptional expression of development-specific genes, leading to physiological and metabolic changes. During the processes of conidium formation from phialides and conidial maturation, conidiumspecific TFs, including VosA, VelB, and WetA, regulate spore-specific gene expression patterns and metabolic changes (25,30). In this study, we investigated the transcript and metabolite changes that are regulated by VosA, VelB, and WetA in A. nidulans conidia.
Transcriptomic analyses indicated that about 20% of the A. nidulans genome (2,143 genes) is differentially expressed in DvosA, DvelB, and DwetA mutant conidia. ChIP-seq results identified 66 direct target genes that are shared between VosA, VelB, and WetA in conidia. These results offered some explanation of how these TFs control phenotypic changes in conidia. First, the deletion of vosA, velB, or wetA caused increased mRNA expression of certain development-specific genes, including abaA (23), brlA (19), flbA (46), flbC (47), nsdC (48), nosA (49), and mpkB (50), which are involved in the formation of asexual and sexual structures during the early and middle stages of conidium formation, but decreased transcript accumulation of spore-specific genes such as vadA (51), catA (52), wA (53), conF (54), conJ (54), cetA (55), cetJ (56), and cetL (56), which are important for conidial germination, morphogenesis, and dormancy (see Table S1 in the supplemental material). Alteration of the mRNA expression levels of development-specific genes or spore-specific genes affect spore maturation, dormancy, and germination. For example, misscheduled expression of key asexual developmental regulators, especially BrlA and AbaA, can affect proper sporulation (9,57). In the case of the sporespecific genes, the deletion of vadA or catA affects conidial germination and the conidial stress response (51,52,58). Based on these results, we propose that alteration of the mRNA expression levels of development-specific genes or spore-specific genes caused by the deletion of vosA, velB, or wetA affect conidial maturation, dormancy, morphology, and germination. However, the detailed molecular mechanism of how three TFs act as activators or repressors for the expression of development-specific genes and spore-specific genes will be elucidated in further studies.
Another important phenotype of the DvosA, DvelB, and DwetA mutant conidia was the differences in conidial wall integrity and the components of the conidial wall (25,30). As shown in Fig. 3, most of the genes involved in chitin and b-glucan biosynthesis were upregulated in all three mutant conidia. The dynamic expression of these genes is required mainly for the remodeling of the cell wall during isotropic growth and mobilization of energy for differentiation (59) but is not required in dormant conidia. However, by altering the mRNA expression of these genes in the mutant conidia, the dormancy of conidia could be broken, affecting long-term viability as well as conidial germination.
Another feature of fungal spores is their ability to resist various environmental stresses (1). However, DvosA, DvelB, and DwetA mutant conidia are more sensitive to several environmental stresses (25,35). It is speculated that this is regulated by alterations in the expression of genes involved in environmental stress tolerance. The data that we show here support this hypothesis. First, these regulators govern the mRNA expression of genes involved in the trehalose biosynthetic pathway, thereby affecting the amount of conidial trehalose, a key component in stress protection and fungal virulence (60). Second, VosA, VelB, and WetA directly or indirectly regulate genes previously associated with stress responses. CatA is a spore-specific catalase, and compared with WT spores, catA deletion mutant spores are sensitive to oxidative stress (52). AtfB is a bZIP TF (61), and the AtfB homolog is crucial for the stress response in Aspergillus oryzae conidia (62). These two genes are putative direct target genes of the three regulators reported in this study, and the mRNAs of catA and atfB can be positively regulated by VosA, VelB, and WetA in conidia ( Fig. 2 and Table S3). Along with these genes, the mRNA level of hogA, a key component of osmotic stress signaling (63), was downregulated in all mutant conidia. These results contribute to our understanding of the ways in which these three regulators influence the environmental stress response in conidia.
VosA, VelB, and WetA are key functional regulators in the formation of conidia and control spore-specific gene expression. However, our data have shown that their gene regulation networks are slightly different. RNA-seq results showed that VosA and VelB coregulate the expression of spore-specific genes. Importantly, the predicted VbRE is quite similar to the predicted VoRE ( Fig. 2A). In addition, biochemical results from previous studies (27,35) suggested that VosA and VelB form a heterocomplex in asexual spores. However, WetA is not directly related to VosA and VelB. WetA's putative binding site is different from the VosA/VelB binding site. Moreover, the WetA peak-associated genes and the VosA/VelB peak-associated genes did not overlap much. These results imply that WetA-mediated gene regulation may be different from the VosA-or VelB-mediated gene regulatory network.
The velvet domain is a fungus-specific DNA-binding domain that recognizes specific DNA sequences. Previously, Ahmed et al. proposed that the VosA velvet domain recognizes a DNA sequence (59-TGGCCGCGG-39) based on ChIP-chip analysis and electrophoretic mobility shift assays (EMSAs) (29). Further EMSAs demonstrated that both TGG and CCGCGG sequences are necessary for DNA binding of the VosA velvet domain. In the present study, we conducted ChIP-seq analyses in conidia and proposed the predicted VbRE (59-CCXTGG-39) and VoRE (59-CCXXGG-39) (Fig. 2). In our experimental results, the TGG sequence does not appear for the VbRE or VoRE, but the 59-CCXXGG-39 sequence is conserved in the VbRE and VoRE. The reason why these DNA sequences are not the same is likely because the experimental methods and analyses are different from those used to obtain the previous results. Ahmed et al. used 15 DNA sequences based on chromatin immunoprecipitation with microarray technology (ChIP-chip) analysis and EMSAs, whereas the motif in Fig. 2A was built from running MEME with every peak sequence that we identified. Nevertheless, the 59-CCXXGG-39 sequence appears common in previous and current results. Based on these data, we propose that the 59-CCXXGG-39 sequence may be crucial for DNA binding of the velvet domain, and further studies will be needed to fine-tune the precise velvet proteinbinding sequence.
During the asexual development of A. nidulans, the abundance of amino acids other than phenylalanine changes, and the expression of genes related to amino acid biosynthesis is altered (64). Overall, our analyses confirmed that the amounts of most amino acids, and the expression of related genes, increased in all mutant spores. In addition, the abundances of metabolites involved in the TCA cycle increased in all mutant conidia. However, the abundances of some primary metabolites such as glutamate, glutamic acid, lactate, and acetyl-CoA were decreased in the DwetA conidia (Fig. 4). It is not yet clear how these metabolic changes affect spore production and maturation, and further studies will be needed to understand this.
Our multi-omics analyses found that VosA, VelB, and WetA regulate the expression of several secondary metabolite gene clusters (Table S7) and the production of secondary metabolites, especially sterigmatocystin, in conidia. The process of sterigmatocystin production and its regulation involves 25 genes, and this metabolite is produced via steps involving several intermediate products. In DvosA conidia, the mRNA expression of sterigmatocystin gene clusters was induced, and the amounts of sterigmatocystin produced were similar to those in the WT conidia. These results were similarly observed in sexual spores (34). While the DvosA conidia contained sterigmatocystin, the metabolite was not detected in DvelB conidia. We reported that the VosA-VelB complex is a functional unit in conidia, but this particular result indicates that VosA and VelB play different roles in sterigmatocystin production. It is possible that VelB forms another complex, such as the VelB-VeA-LaeA complex (40), to participate in sterigmatocystin production in conidia. For the DvelB conidia, we speculated that the mRNA expression levels of genes such as stcB, stcC, stcF, and stcI, which are associated with the early stages of sterigmatocystin biosynthesis, were increased, and that the amount of versiconol, a putative sterigmatocystin/aflatoxin intermediate, was also increased in comparison with the WT. However, the mRNA levels of genes associated with the late phase of sterigmatocystin biosynthesis, such as stcL, stcN, stcQ, and stcT, were decreased in DvelB conidia. It might be possible that VelB (or VelB/VeA/LaeA) can regulate some expression of sterigmatocystin gene clusters by epigenetic means rather than through the canonical method of aflR expression or activity. Although changes in the expression of secondary metabolite gene clusters and secondary metabolites affected by three TFs were studies, detailed molecular mechanisms have not been studied yet. Therefore, it is necessary to study how these three TFs work together or separately through further research. In the DvosA and DwetA conidia, the mRNA levels of most of the genes in the sterigmatocystin gene cluster were increased compared to those in WT conidia, but the amounts of sterigmatocystin were similar to those in WT conidia. There are some speculations about this phenomenon. The expression of genes may not directly affect the biosynthesis of secondary metabolites. Alternatively, the translation of mRNA molecules to proteins and the posttranslational modification of those metaboliteproducing proteins are two factors that can create discrepancies between RNA and metabolite abundances. To further explain this, further experiments should be conducted to determine how the three TFs regulate the biosynthesis of secondary metabolites.
In conclusion, this study provides a systematic dissection of the gene regulatory network and molecular mechanisms of VosA, VelB, and WetA (Fig. 7). In conidia, VosA, VelB, and WetA directly or indirectly control the expression of spore-specific or development-specific genes, thereby altering conidial wall integrity and conidial viability. In addition, these TFs regulate multiple secondary metabolite gene clusters, thus inducing secondary metabolic changes. These results provide an advance in the knowledge of conidial formation and will provide the basis for future insights into spore formation in other filamentous fungi.

MATERIALS AND METHODS
Strains, media, and culture conditions. The fungal strains used in this study are listed in Table 1. Fungal strains were grown on solid or liquid minimal medium with 1% glucose (MMG) and appropriate supplements for general purposes as previously described (65). For conidium samples, WT and mutant conidia were inoculated onto solid MMG plates and incubated for 48 h. Next, conidia were collected from plates using Miracloth (Calbiochem, San Diego, CA, USA) and stored at 280°C.
RNA sequencing analysis. To isolate total RNA for RNA sequencing (RNA-seq) analysis, total RNA from WT and mutant conidia was extracted using TRIzol reagent (Invitrogen, USA), according to the manufacturer's instructions, with modifications. To remove DNA contamination from the RNA samples, DNase I (Promega, USA) was added, and RNA was then purified using an RNeasy minikit (Qiagen, USA). Three technical replicates of each sample were analyzed. RNA sequencing was performed as previously described (34). RNA samples were submitted to the University of Wisconsin Gene Expression Center (Madison, WI, USA) for library preparation and sequencing. A strand-specific library was prepared using an Illumina TruSeq strand-specific RNA sample preparation system. The libraries of all the replicates were sequenced using an Illumina HiSeq 2500 system. Data analysis of the DvosA and DvelB RNA-seq experiments was performed using the same analysis pipeline as the one previously described for the DwetA RNA-seq analysis (25). Reads were mapped to the A. nidulans FGSC4 transcriptome using Tophat2 version 2.1.1 (66) and the parameter "-max-intronlength 4000." On average, 19.9 million reads per sample mapped to the genome, and the number of reads aligning to each gene was counted with HTseq-Count version 0.9.1 (67). DESeq version 1.14.1 (68)    was used to determine significantly differentially expressed genes, and genes were considered regulated if they exhibited an adjusted P value of ,0.05 and a log 2 fold change either greater than 1 or less than 21.
Chromatin immunoprecipitation sequencing analysis. Samples for chromatin immunoprecipitation sequencing (ChIP-seq) analysis were prepared according to methods described previously (29,30). DNA samples from each strain were extracted using a MAGnify chromatin immunoprecipitation system (Invitrogen, USA) according to the manufacturer's protocol, with modification. Two-day-old conidia from the WT strain or strains containing VosA-FLAG or VelB-FLAG were cross-linked, washed, homogenized with a Mini-Beadbeater 16 instrument (Biospec, USA), sonicated, and separated by centrifugation. The chromatin extracts were incubated with an anti-FLAG antibody-Dynabead complex. Next, samples were eluted from the beads at 55°C using proteinase K. Enriched DNA was purified using DNA purification magnetic beads. DNA samples from each strain were submitted to the University of Wisconsin Gene Expression Center (Madison, WI). Libraries were prepared using a TruSeq ChIP library preparation kit (Illumina, CA). The libraries of all the replicates were sequenced using an Illumina HiSeq 2500 system.
Raw reads were trimmed using Trimmomatic version 0.36 (69) and the parameters "ILLUMINACLIP:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36." Trimmed reads were mapped to the A. nidulans A4 genome using version 0.7.15 of BWA-MEM (70), and shorter split hits were marked as secondary alignments. Mapped reads with mapping quality (MAPQ) values of ,1 as well as unmapped, secondarily aligned, supplementary, and duplicated reads were discarded with SAMtools version 1.6 (71). On average, 2.3 million and 7.2 million reads per sample were used for peak calling in the VosA and VelB experiments, respectively. Mapped reads that survived our filter were pooled, and extension sizes were estimated with version 1.15.2 of SPP (72,73). Peaks were called with MACS2 (74) version 2.1.2 using the extension sizes estimated by SPP, a genome size of 2.93e7, and the "-nomodel" parameter. Peaks with a fold change of .2.0 and a q value of ,0.001 were further analyzed. Peak lists were combined from both of the VosA biological replicates, as .99% of the peaks from the first replicate were found in the second replicate. Motifs were identified in the 100 bp of sequences surrounding each peak summit using MEME-ChIP (75). Motifs that occurred zero times or once in the sequences around the peaks and that were 4 to 21 nucleotides (nt) long were further analyzed.
Primary metabolite analysis. WT and DwetA, DvosA, and DvelB mutant conidia were inoculated onto solid MMG plates and incubated for 48 h, and fresh conidia were then harvested using Miracloth with HPLC-grade water. For each sample, 2 Â 10 8 conidia were mixed with 500 ml HPLC-grade acetonitrile-methanol-water (40:40:20, vol/vol) and 300 ml beads, homogenized by using the Mini-Beadbeater, and centrifuged. The supernatant was filtered using a 0.45-mm polytetrafluoroethylene (PTFE) Mini-UniPrep filter vial (Agilent), collected, and immediately snap-frozen with liquid nitrogen. The samples were stored at 280°C until primary metabolite analysis.
The samples were then analyzed as described previously (79,80). Samples were analyzed using an HPLC-MS system consisting of a Dionex ultrahigh-performance liquid chromatography (UHPLC) instrument coupled by electrospray ionization (ESI) (negative mode) to a hybrid quadrupole-high-resolution mass spectrometer (Q Exactive orbitrap; Thermo Scientific) operated in full-scan mode. Metabolite peaks were identified by their exact mass and matching retention times to those of pure standards (Sigma-Aldrich).
Secondary metabolite analysis. The conidia of WT and DwetA, DvosA, and DvelB mutant strains were extracted by adding 1.5 ml of a methanol-acetonitrile (2:1) mixture followed by sonication for 60 min. The suspension was then left overnight before centrifugation at 14,000 rpm for 15 min. The supernatant (1 ml) was removed, filtered, and evaporated to dryness in vacuo. Extracts for the metabolomics analysis were normalized to 10 mg/ml in methanol for LC-MS analysis.
Analytical HPLC was performed using an Agilent 1100 HPLC system equipped with a photodiode array detector. The mobile phase consisted of ultrapure water (mobile phase A) and acetonitrile (mobile phase B) with 0.05% formic acid in each solvent. A gradient method from 10% mobile phase B to 100% mobile phase B in 35 min at a flow rate of 0.8 ml/min was used. The column (Phenomenex Kinetex C 18 , 5 mm by 150 mm by 4.6 mm) was reequilibrated before each injection, and the column compartment was maintained at 30°C throughout each run. All samples were filtered through a 0.45-mm nylon filter before LC-MS analysis.
Extracts from the WT and mutant conidia were analyzed in duplicate on an Agilent 1100 series LC-MS platform (81,82). The negative ionization mode was found to detect the most metabolites. The first 5 min of every run was removed due to a large amount of coeluting, low-molecular-weight, polar metabolites. Data sets were exported from Agilent's Chemstation software as .netCDF files and imported into MZmine 2.38 (83). Peak picking was performed with established protocols (84), resulting in 123 marker ions. Briefly, mass detection was centroid with a 5e2 minimum height. Chromatogram building was limited to peaks greater than 0.1 min with a tolerance of 0.05 m/z and a minimum height of 1e3. Data smoothing was performed at a filter width of 5. Chromatogram deconvolution was performed by utilizing a local minimum search with a chromatographic threshold of 95%, a minimum relative height of 10%, a minimum absolute height of 3e3, a minimum ratio of peak to edge of 1, and a peak duration range of 0.1 to 5.0 min. The spectra were de-isotoped with a 1-ppm m/z tolerance before all treatments were aligned, and duplicate peaks were combined with a tolerance of 0.1 m/z and a 3.0-min RT. Peak finder gap filling was performed with 50% intensity tolerance and 0.1 m/z tolerance. Peak lists were exported to Metaboanalyst (85), where missing values were replaced with half the minimum positive value, data were filtered by the interquartile range, and log transformation of the data was employed.
Data availability. All RNA-seq and ChIP-seq data files are available from the NCBI Gene Expression Omnibus database (wetA RNA-seq, accession number GSE114143; vosA and velB RNA-seq, accession number GSE154639; WetA ChIP-seq, accession number GSE114141; VosA and VelB ChIP-seq, accession number GSE154630).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.