Genome-Wide Gene Expression Analyses of the AtfA/AtfB-Mediated Menadione Stress Response in Aspergillus nidulans

The bZIP transcription factors (TFs) govern regulation of development, secondary metabolism, and various stress responses in filamentous fungi. In this work, we carried out genome-wide expression studies employing Illumina RNAseq to understand the roles of the two bZIP transcription factors AtfA and AtfB in Aspergillus nidulans. Comparative analyses of transcriptomes of control, ΔatfA, ΔatfB, and ΔatfAΔatfB mutant strains were performed. Dependence of a gene on AtfA (AtfB) was decided by its differential downregulation both between the reference and ΔatfA (ΔatfB) strains and between the ΔatfB (ΔatfA) and the ΔatfAΔatfB strains in vegetatively grown cells (mycelia) and asexual spores (conidia) of menadione sodium bisulfite (MSB)-treated or untreated cultures. As AtfA is the primary bZIP TF governing stress-response in A. nidulans, the number of differentially expressed genes for ΔatfA was significantly higher than for ΔatfB in both mycelial and conidial samples, and most of the AtfB-dependent genes showed AtfA dependence, too. Moreover, the low number of genes depending on AtfB but not on AtfA can be a consequence of ΔatfA leading to downregulation of atfB expression. Conidial samples showed much higher abundance of atfA and atfB mRNAs and more AtfA- and AtfB-affected genes than mycelial samples. In the presence of MSB, the number of AtfB- (but not of AtfA-) affected genes decreased markedly, which was accompanied with decreased mRNA levels of atfB in MSB-treated mycelial (reference strain) and conidial (ΔatfA mutant) samples. In mycelia, the overlap between the AtfA-dependent genes in MSB-treated and in untreated samples was low, demonstrating that distinct genes can be under AtfA control under different conditions. Carbohydrate metabolism genes were enriched in the set of AtfA-dependent genes. Among them, AtfA-dependence of glycolytic genes in conidial samples was the most notable. Levels of transcripts of certain secondary metabolitic gene clusters, such as the Emericellamide cluster, also showed AtfA-dependent regulation. Genes encoding catalase and histidine-containing phosphotransfer proteins showed AtfA-dependence under all experimental conditions. There were 23 AtfB-dependent genes that did not depend on AtfA under any of our experimental conditions. These included a putative α-glucosidase (agdB), a putative α-amylase, calA, which is involved in early conidial germination, and an alternative oxidase. In summary, in A. nidulans there is a complex interaction between the two bZIP transcription factors, where AtfA plays the primary regulatory role.


Introduction
Basic domain leucine zipper-type (bZIP) transcription factors are members of a complex regulatory network, playing a crucial role in the maintenance and differentiation of cells as well as the coordination of stress responses in eukaryotes [1].
In Aspergillus oryzae, conidia of the ∆atfA mutant were more sensitive to oxidative stress than those of the ∆atfB [20,21]. Some genes involved in the oxidative stress defense, e.g., putative catalase, thioredoxin and glutathione metabolic genes, were repressed in the ∆atfA mutant according to the microarray analysis, which confirms the observed stresssensitive phenotype of the ∆atfA mutant [21]. In A. oryzae, atfA is involved in conidial storage stability [21]. Therefore, conidia of the ∆atfA showed lower germination rate compared to the control [21]. Most likely, atfA controls glutamate biosynthesis, which is necessary for germination of conidiospores [21]. atfB expression was significant in the late phase of culture growth and coincided with the initiation of conidiation in A. oryzae [21]. Furthermore, AtfB-regulated genes, such as catA (encoding a catalase) or a putative trehalose-6-phosphate synthase gene are most likely associated with conidial development and conidial stress tolerance [21]. bZIP transcription factors can form homodimers with themselves and heterodimers with other bZIPs and interact physically with stress-signaling proteins as well [22]. For example, in Schizosaccharomyces pombe, Atf1 and Pcr1 bZIPs form heterodimers and activate the majority of stress genes [23][24][25]. Evaluation of microarray data confirmed, however, that some stress genes are regulated by Atf1 independently of Pcr1 under osmotic stress (elicited with 0.4 M KCl) [24]. Moreover, Atf1 physically interacts with Cid2 poly(A) polymerase to regulate further genes [26].
In this study, we performed RNAseq-based transcriptome analysis in ∆atfA, ∆atfB, ∆atfA∆atfB and control strains from MSB-treated and untreated surface cultures in vegetative and conidial development stages in order to understand more deeply the regulatory functions of AtfA and AtfB. We focused on the possible interactions between AtfA and AtfB during the evaluation process.
For RNA sequencing, freshly grown conidiospores (10 5 suspended in 5 µL aliquots of PBS-0.1% Tween 20) were spread on NMM plates with or without 0.04 MSB (menadione sodium bisufite), and surface cultures were incubated at 37 • C. Mycelia were collected before the formation of conidiophores, while conidia were washed from the surface culture with PBS-0.1% Tween 20 and separated from the vegetative tissue with centrifugation and filtering through Miracloth.

RNA Sequencing
Total RNA were isolated from the menadione sodium bisulfite (MSB, a superoxide generating agent)-treated and untreated cultures of the THS30.3 (control), ∆atfA, ∆atfB, ∆atfA∆atfB strains. Samples were taken from 20-33 hour-old (mycelial samples) and from 3-day-old (conidial samples) surface cultures of the strains. Total RNA was isolated according to Chomczynski, 1993 [28]. RNA sequencing, from library preparation to generation of fastq.gz files, was carried out at the Genomic Medicine and Bioinformatic Core Facility, Department of Biochemistry and Molecular Biology, Faculty of Medicine, University of Debrecen, Debrecen, Hungary. Libraries were prepared with the TruSeq RNA Sample preparation kit (Illumina) according to the manufacturer's protocol. Conidial and mycelial samples were sequenced (single-read 75 bp sequencing on an Illumina HiScan SQ instrument; Illumina, San Diego, CA, USA) separately, but each library pool belonging to the same cell type was sequenced in one lane of a sequencing flow cell. Depending on the sample type, 14-39 million reads per sample (mycelial samples) and 10-28 million reads per sample (conidial samples) were obtained. The FastQC package (http://www.bioinformatics.babraham.ac.uk/projects/fastqc, accessed on 25 January 2023) was used for quality control. Reads were aligned to the genome of A. nidulans FGSC A4 with hisat2 (version 2.1.0) [29]. The successfully aligned reads varied between 77-96% (mycelial samples) and 78-94% (conidial samples). DESeq2 (version 1.24.0) [30] was used to determine differentially expressed genes. Since conidial samples originated in two separate experiments, the batch effect was taken into consideration during the identification of differentially expressed genes in this case. RPKM values (reads per kilo base per million mapped reads) were also calculated with the edgeR package ("rpkm" function) [31] and used to visualize transcription activities of selected genes.

Evaluation of the Transcriptome Data
Transcriptomes were characterized with three types of features: Strain (control, ∆atfA, ∆atfB, ∆atfA∆atfB), treatment (untreated, MSB-treated) and cell type (mycelium, conidium). Since mycelial and conidial cultures were studied separately (e.g., the age of the studied cultures and the efficiency of RNA isolation were different), mycelial and conidial transcriptomes were not compared. We compared only transcriptomes (A vs. B) differing only in either strain or treatment. log 2 FC was calculated using the DESeq2 software, using B as reference. Differentially expressed genes were considered "upregulated" if log 2 FC > 1, "downregulated" if log 2 FC < −1 and "regulated" if |log 2 FC| > 1. "MSB stress-responsive" genes refer to genes "regulated" in an A MSB treated vs. B untreated comparison for any strain. The "atfA gene-deletion-responsive" genes are genes "regulated" in either the A ∆atfA vs. B control or A ∆atfA∆atfB vs. B ∆atfB comparisons. The "atfB gene-deletion-responsive" genes were defined similarly. The atfA atfB gene-deletion-responsive genes were defined as genes "regulated" in the A ∆atfA∆atfB vs. B control comparison.
We applied the following assumptions and simplifications during the evaluation of the data: There are genes that are AtfA-dependent but not AtfB-dependent (AA genes) and genes that are AtfB-dependent but not AtfA-dependent (BB genes).
-There are genes regulated by both AtfA and AtfB. Some of these genes need both atfA and atfB to reach their normal physiological activity detected in the control (deletion of either gene reduces their activities at least to the level of the double mutant; AB genes). Some of the genes, regulated by both AtfA and AtfB, need either atfA or atfB for their normal activity (deletion of either gene does not reduce their transcriptional activity; A/B genes). Some of them need both atfA and atfB for normal activity and deletion of at least one of the two genes reduces the activity to a level between the control and the double mutant (A-B genes). For clarity, see Figure S1.
The following gene sets were constructed for mycelial and conidial transcriptomes: Set0: MSB stress-responsive downregulated genes of the control strain Set0 + : MSB stress-responsive upregulated genes of the control strain The following gene sets, containing downregulated genes, were also constructed for untreated and MSB-treated, mycelial and conidial transcriptomes.
From the ∆atfA vs. control, ∆atfB vs. control and ∆atfA∆atfB vs. control comparions ( Figure S1A AtfA-dependent and AtfB-dependent genes were regarded as the union of the appropriate gene sets defined above: Set26 = Set21∪Set23∪Set24∪Set25 (AtfA-dependent genes) Set27 = Set22∪Set23∪Set24∪Set25 (AtfB-dependent genes) The gene sets containing the corresponding upregulated genes were marked from Set1 + to Set27 + . Note, AtfA-dependent genes (similarly to the AtfB-dependent genes) were determined from two comparisons of the four strains (∆atfA vs. control and ∆atfA-∆atfB vs. ∆atfB as well as ∆atfB vs. control and ∆atfA-∆atfB vs. ∆atfA) to reduce the number of misidentified genes. These comparisons were carried out under two culturing conditions (untreated and MSB-treated cultures) in two types of cells (mycelium and conidium). This way we obtained four AtfA-dependent and four AtfB-dependent gene sets. Since AtfA-or AtfBdependence can depend on the culturing conditions and the cell types, these gene sets were studied separately.
AtfA-and AtfB-dependent gene sets were characterized by gene set enrichment analyses. For it, "Functional Catalogue" (FunCat), "Gene Ontology" (GO) and "Kyoto Encyclopedia of Genes and Genomes pathway" (KEGG pathway) terms were used with the FungiFun2 package (https://elbe.hki-jena.de/fungifun/fungifun.php, accessed on 25 January 2023) applying default settings. Only hits with a corrected p-value < 0.05 were regarded as significantly enriched in the studied gene set.
The enrichment of the following gene groups in the AtfA-and AtfB-dependent gene sets were also tested by the Fisher's exact test with the "fisher.test" function of R project (www.R-project.org/, accessed on 25 January 2023): "Lactose utilization" genes. This gene group contains the Leloir and oxido-reductive pathways of galactose utilization [32] as well as known and putative β-galactosidase and lactose permease genes according to Fekete et al. 2012Fekete et al. , 2016 and Gila et al. 2022 [35].
Phosphorelay response regulator activity, iron-sulfur cluster assembly, "Respiration", and "Transcription factor" genes. These groups were constructed based on the related GO terms and their child terms [35,36]. "Secondary metabolism cluster" genes. Manually or experimentally determined secondary metabolite cluster genes collected by Inglis et al. 2013 [38] and gene set enrichment analysis was carried out with the clusters separately.

Deletion of atfA Downregulates atfB
Mycelial and conidial transcriptomes from four strains (control, ∆atfA, ∆atfB, ∆atfA ∆atfB) at two different culturing conditions (untreated, MSB-treated) were determined. Changes in either feature (strain, treatment) had substantial effects on the transcriptomes ( Figure S2). Genes responsive for gene deletions (in untreated and MSB-treated cultures), for MSB treatment (in the control strain) or that showed AtfA-and/or AtfB-dependence were identified in both mycelial and conidial samples (Tables 1, 2 and S1).
MSB treatment and deletion of the atfB gene did not upregulate or downregulate the atfA gene in atfA + strains ( Figure 1). In contrast, the presence of MSB (in the case of mycelial samples) and the deletion of atfA (in both mycelial and conidial samples) downregulated the atfB gene ( Figure 1). This means that AtfA can affect the transcription of AtfB-dependent genes via atfB transcription; therefore, some of the genes putatively regulated by both AtfA and AtfB may be genes that were regulated directly only by AtfB. These genes can potentially occur in any gene sets but especially in those where the effect of atfB deletion was stronger than or equal with that of the atfA gene deletion: AB (Set23), A-B1 and A-B5 (Set25) gene sets (Tables 1 and 2). Importantly, no genes belonging to the A-B5 (Set25) gene set were identified (Tables 1 and 2).  114 genes a -The lists of genes belonging to the sets are available in Table S1. b -Significant enrichment (Fisher's exact test, p < 0.05). MSB stress treatment upregulated 485 genes and downregulated 1070 genes (Set0 + and Set0, respectively). c -Figures in parentheses shows the overlap between conidial and mycelial samples as well as the percentage of overlap relative to the size of the mycelial set. In the case of the mycelial and conidial Set0 + and Set0 gene sets, the overlaps were 45 upregulated and 225 downregulated genes, which represent 6% and 25% of the mycelial gene sets, respectively.

Most of the AtfB-Dependent Genes Show AtfA-Dependence in Mycelia of Untreated Cultures
Many more atfA gene-deletion-responsive genes were found in untreated mycelial samples than atfB gene-deletion-responsive genes ( Table 1). The difference between the ∆atfA and the ∆atfB strains was more obvious in the case of the downregulated genes than with the upregulated ones (Table 1). Altogether, 329 AtfA-and 96 AtfB-dependent genes were identified in these cultures, and most of the AtfB-dependent genes showed AtfA-dependence as well ( Table 1). The high number of AtfA-dependent genes relative to the number of AtfB-dependent genes concurs well with atfA gene deletion having stronger transcriptomical ( Figure S2, Table 1) and physiological [19] consequences than atfB gene deletion.
According to the type of possible interactions between the regulatory effects of AtfA and AtfB, genes that showed both AtfA-and AtfB-dependence were grouped into three sets: AB, A/B, and A-B ( Figure 2, Table 1). The most interesting group was the AB set. The transcriptional pattern of the related genes ( Figure 2, Table S2) suggests that both AtfA and AtfB were needed for their normal ("wild type") activity. Besides the genes regulated by AtfA via regulation of the atfB gene, it is possible that some of these genes were regulated by an AtfA-AtfB heterodimer. The majority of the genes under both AtfA and AtfB regulations belonged to the A/B or A-B gene sets (Tables Table 1 and S2), suggesting that the missing transcription factor was completely (A/B) or at least partially (A-B) substituted with the other transcription factor. In the case of the most A-B genes, atfA gene deletion had a stronger consequence than that of atfB (A-B2 and A-B4 genes) (Figure 2, Table S2).

AtfB Regulates Only Few Genes in Mycelia of MSB-Treated Cultures
The transcription of fewer genes was affected by atfA and/or atfB gene deletions on MSB than in untreated cultures (Table 1). Only a few AtfB-dependent genes (10 genes) were found in this case, and most of these genes were AtfA-dependent too (Table 1). This concurs well with the observation that atfB was downregulated by MSB stress in the control strain ( Figure 1). Not surprisingly, the ∆atfB mutant was as sensitive to MSB stress as the control strain [19].
The overlaps between the MSB-treated and untreated cultures in the cases of the gene-deletion-responsive gene sets and the AtfA-dependent gene sets were relatively small (Table 1), supporting the view that AtfA regulates (directly or indirectly) different genes in mycelia under different culturing conditions [16][17][18]. Surprisingly, the AtfAdependent gene set of MSB-treated cultures, similarly to the AtfA and AtfB-dependent gene sets of untreated cultures, was enriched in MSB stress-responsive downregulated genes (Table 1), i.e., many genes that were downregulated by the presence of MSB showed further downregulation in the absence of AtfA. This behavior suggests that one of the main functions of AtfA in cultures that have adapted to the presence of MSB is not to keep high the transcriptional activity of genes upregulated by the stress treatment but to prevent the excessive downregulation of genes downregulated under this stress. Of course, this does not exclude that several genes are upregulated by AtfA during the early stress response of MSB stress.

AtfB, Similarly to AtfA, Increases Importance in Conidia
The transcriptional activity of both atfA and atfB was much higher in conidia than in mycelial samples (Figure 1). Not surprisingly, huge numbers of atfA and/or atfB genedeletion-responsive and AtfA-and/or AtfB-dependent (Table 2) genes were recorded in conidial samples, demonstrating that AtfA and AtfB are more important regulators of the physiology of (germinating) conidia than of the vegetative mycelia. MSB treatment did not reduce the abundance of atfB transcript in conidia ( Figure 1); however, the number of AtfBdependent genes decreased more radically in the presence of MSB than did AtfA-dependent genes ( Table 2).
Again, most of the AtfB-dependent genes showed AtfA-dependence as well (Tables 2 and S2). Among the genes showing both AtfA-and AtfB-dependence, the AB, A/B, A-B2 and A-B4 genes were the most abundant (Tables 2 and S2). Accordingly, AtfB regulates only few genes independently of AtfA (BB genes, Tables 2 and S1). Some genes are regulated together with AtfA (AB genes, Table 2), and in the case of the most AtfBdependent genes AtfA can replace the missing AtfB (A/B, A-B2, A-B4 genes, Tables 2 and S2).
MSB treatment substantially modified the transcriptome of both conidia and mycelia: 485 and 786 upregulated and 1070 and 912 downregulated MSB stress-responsive genes were found in conidial and mycelial samples, respectively. The effect of MSB treatment on the conidial transcriptome suggests that stresses affecting the physiology of mycelia also affect the transcriptome of conidia produced by the stress-treated mycelia. Interestingly, these changes modified only slightly the regulatory role of AtfA and AtfB in conidia: The overlaps between the AtfA-(AtfB-) dependent genes of conidia from untreated and MSBtreated cultures were large, in contrast to those in mycelial samples (Tables 1 and 2). It can be understandable if we assume that the transcriptional changes of mycelia reflect how vegetative cells adapted to the presence of MSB, while the transcriptional changes in conidia show how the "experiences" of mycelia are implemented into the germination strategies of conidia. In other words, conidia do not have to adapt to all consequences of long-term MSB treatment; they have to prepare only for the increased possibility of MSB stress during their germination.
Gene sets identified with conidial samples showed surprisingly low overlap with the appropriate mycelial gene sets (Table 2). Even in the case of the AtfA-dependent genes, the overlap was only around 50% (Table 2). This huge difference between mycelial and conidial samples suggests that AtfA and AtfB had different functions and that atfA and atfB gene deletions had different consequences in mycelia and in conidia.

AtfA Affects Carbohydrate Metabolism and Light Dependent Processes
Gene set enrichment analyses were carried out with four AtfA-and four AtfB-dependent gene sets (identified in untreated mycelial and conidial samples as well as in MSB-treated mycelial and conidial samples) (Tables 3, 4, S3 and S4). The AtfA-dependent gene sets were enriched with carbohydrate metabolism genes. Among them, AtfA-dependence of glycolytic genes in the case of the conidial samples is the most notable (Table S3). Phosphorelay response regulator genes were enriched in all AtfA-dependent gene sets but the untreated mycelial samples (Tables 4 and S4), while enrichment of iron-sulfur cluster assembly genes was characteristic for the AtfA-dependent genes of conidial samples (Tables 4 and S4). Interestingly, antioxidant enzyme genes were enriched only in the AtfA-dependent gene set of conidial samples from untreated cultures (Tables 4 and S4). Enrichment of TCA cycle and respiration genes was characteristic for the AtfA-dependent gene set of conidial samples from MSB-treated cultures. Most of the abovementioned genes were regulated only by AtfA (AA genes) (Table S4). Certain secondary metabolite cluster genes also showed AtfA-dependent regulation (Tables 4 and S4). Among them, the Emericellamide cluster is notable since, depending on the treatment, four or five genes out of the five cluster genes were AtfA-dependent, including the easB gene (AN2547) encoding the polyketide synthase ( Figure S3, Table S3). Interestingly, in untreated cultures, these genes showed both AtfA-and AtfB-dependence (A/B genes), while in MSBtreated cultures, where atfB was downregulated (Figure 1), they were only AtfA-dependent (AA genes).
Altogether 87 genes showed AtfA-dependence in all the four AtfA-dependent gene sets. Most of them encode proteins with unknown functions ( Table 5). The genes with known or predicted function includes the catA catalase, six genes involved in carbohydrate metabolism as well as the hk-8-1 and hk2 putative histidine-containing phosphotransfer protein genes and 10 genes involved in light sensing and light response (Tables S1 and S4). Only 23 AtfB-dependent genes were found that never showed AtfA-dependence (Table S1). Out of them, the following four genes are notable: AN8953 (agdB), putative α-glucosidase and AN3402 (amyB), putative α-amylase genes; AN7619 (calA), involved in early conidial germination; and AN2099, putatively encoding alternative oxidase.

Discussion
bZIP-type transcription factors are important regulators of developmental processes, stress responses and secondary metabolite production in filamentous fungi [19,[39][40][41][42]. They can act as homodimers, and they can also regulate processes forming heterodimers with other bZIP-type transcription factors or physically interact with other signaling proteins [22]. In A. fumigatus, AtfA physically interacts with other three bZIP transcription factors, namely AtfB, AtfC and AtfD, as well as with the MAPK SakA to coordinate stress responses [11]. According to the stress sensitivity assays, the ∆atfA∆atfB double-gene deletion mutant was as sensitive to the oxidative stress generating menadione sodium bisulfite or to the cell wall stress-generating agents calcofluor white (CFW) and CongoRed as the corresponding single mutants in A. fumigatus [11]. In A. parasiticus, AtfB and AP-1 bZIPs form functionally active heterodimers and regulate aflatoxin production and oxidative stress responses [39]. In the case of A. nidulans, Lara-Rojas et al. 2011 [2] suggested possible physical interaction between AtfA and AtfB. Here we studied genome-wide transcriptional changes in mycelia and conidia of ∆atfA, ∆atfB, ∆atfA∆atfB gene-deletion mutants and the control strain in the presence and absence of MSB to collect data on the possible interactions between AtfA and AtfB.
The high transcriptional activity of atfA ( Figure 1) and the genome-wide transcriptional as well as phenotypic consequences of atfA gene deletion (Tables 1 and 2; [15,19]) suggest the importance of AtfA-dependent regulations in both mycelia and conidia. The data also support the view that AtfA was a more important regulator in conidia than in vegetative mycelia (Figure 1, Tables 1 and 2). This concurs well with the observations of Hagiwara et al. 2008 [12] and Balázs et al. 2010 [14] that AtfA protects conidia under different temperatures as well as against oxidative stress. AtfA plays a paramount role in the regulation of conidium-specific genes in other Aspergillus species as well. In a comprehensive study, more than 50% of the conidia-associated genes (CAGs) proved to be atfA-dependent in A. fumigatus, A. oryzae and A. niger [43].
Many genes regulated (directly or indirectly) by AtfA have been identified so far in Aspergillus species. These genes-among others-encode antioxidant proteins, heat shock proteins, phosphorelay response regulators, iron-sulfur cluster assembly proteins, enzymes involved in trehalose and glycogen formation, utilization of different carbohydrates or secondary metabolite synthesis as well as light response of conidia [11,[16][17][18]21,[43][44][45][46][47]. Our data support the role of AtfA in the transcription of antioxidant enzyme and phosphorelay response regulator genes, trehalose and glycogen metabolism genes, glucose utilization genes as well as secondary metabolite cluster genes (Tables 3, 4 and S4). The cytoplasmic phytochrome FphA (acting as red-light sensor) activates AtfA via the high-osmolarity glycerol (HOG) MAPK pathway in A. nidulans [46,47]. Not surprisingly, several lightdependent genes were identified as AtfA-dependent in our study (Tables 5 and S1). Among them, cryA encoding a putative UV-A/blue light sensor (cryptochrome) [48] is particularly interesting. Yu et al. 2016 [46] found that the blue-light-dependent activation of the HOG pathway depends on FphA only, but not on the blue-light sensor LreA-LreB complex. One explanation of this observation is that another blue-light receptor (e.g., CryA) is involved in this process and its activity is somehow regulated by FphA [46]. The AtfA-dependent transcription of cryA (Tables 5 and S1) also supports the view that there is interaction between the blue-and red-light dependent signaling pathways.
Stress tolerance of conidia highly depends on culturing conditions occurring during conidiogenesis [49][50][51][52]. Not surprisingly, MSB stress treatment (of mycelia) affected both the mycelial and conidial transcriptomes (Tables 1 and 2). The regulatory role of AtfA also depended on culturing conditions: Different genes showed AtfA-dependence in MSB stress-adapted and unstressed cultures (Tables 1 and 2). Importantly, the difference between the AtfA-dependent gene sets was more obvious in mycelial than in conidial samples (Tables 1 and 2). It is reasonable to assume that conidia, de facto, do not have to adapt to the presence of MSB. Conidiogen cells alter the mRNA content of conidia only to prepare them for the stresses that (according to their "experiences") most likely will occur during germination.
Transcriptional activity of atfB was low in mycelial samples; however, atfB mRNA was abundant in conidia (Figure 1), suggesting that this transcription factor may have a minor regulatory role during vegetative growth. The small transcriptomic (Tables 1 and 2) and phenotypic [19] consequences of atfB gene deletion relative to that of atfA also support this view.
The majority of the AtfB-dependent genes were AtfA-dependent as well ( Figure 2, Tables 1, 2 and S2), which concurs well with results of Sakamoto et al. 2009 [21], who also found that most of the stress-responsive genes regulated by AtfB were also AtfA-dependent in A. oryzae. Some of the genes that showed dual AtfA-and AtfB-dependent regulation needed both AtfA and AtfB for their "normal" expression (AB genes on Figure 2 and in Tables 1, 2 and S2). It is possible that some of them are regulated by an AtfA-AtfB heterodimer; however, without experimental justification, other possibilities cannot be ruled out. The majority of the AtfA-, AtfB-dependent genes were genes where one of the two transcription factors could completely or partially substitute the missing other transcription factor (A/B and A-B genes on Figure 2 and in Tables 1, 2 and Table S2). Some of these genes may be regulated by both transcription factors directly, which also allows physical interaction between the two transcription factors on the promoters. Moreover, atfB itself also showed AtfA-dependence; deletion of atfA downregulated atfB in both mycelial and conidial samples irrespectively of the presence of MSB (Figure 1). Therefore, some of the genes showing both AtfA-and AtfB-dependence can be AtfB-dependent genes regulated by AtfA only indirectly via atfB transcription. The interaction between AtfA and AtfB has also been suggested by the overexpression of atfB being able to compensate for the increased MSB sensitivity of the ∆atfA mutant [19]. Moreover, sterigmatocystin production was completely inhibited by atfA gene deletion; however, it was restored in the ∆atfA∆atfB mutant [19]. Importantly, we found a few AtfB-dependent genes that did not show AtfA-dependence (Tables 1, 2, S1 and S4). Among them, calA is notable, since it contributes to the germination of conidia [53,54] which may explain why conidia of the ∆atfB strain were sensitive to high temperature [19].
Our results support the view that (1) AtfA and AtfB have some regulatory functions in mycelia; however, they are more important regulators in conidia. (2) Besides regulating antioxidant enzyme genes, phosphorelay response regulator genes, secondary metabolite cluster genes, and light-dependent genes, AtfA also control genes of carbohydrate metabolism (e.g., trehalose and glycogen metabolism genes as well as glucose utilization genes) in A. nidulans, as it was also found in A. fumigatus [11]. (3) There should be a complex genetic and possibly physical interaction between the two transcription factors where AtfA is the dominant player, and the main function of AtfB is supporting the regulatory role of AtfA. Understanding the nature of the interaction between the two transcription factors needs further investigations: e.g., determining the AtfA-and AtfB-binding sites on the promoters at genome level by combining chromatin immunoprecipitation assays with sequencing (ChIP-Seq), and justifying the AtfA-AtfB heterodimer formation using a bimolecular fluorescence complementation (BiFC) technique. Both are in progress in our laboratory.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/cells12030463/s1, Figure S1: Identification of potentially AtfA and AtfB dependent genes; Figure S2: Principal component (PC) analysis of the transcriptomes; Figure S3: Transcriptional profile of easB encoding the polyketide synthase of the Emericellamide secondary metabolite cluster in mycelial samples; Table S1: Lists and characteristics of AtfA-and AtfB-dependent genes; Table S2: List, characteristics and RPKM values of genes that showed both AtfA-and AtfB-dependence in mycelial and conidial samples; Table S3: Results of the gene set enrichment analyses for AtfA-and AtfB-dependent genes; Table S4: Characterization of AtfA-and AtfB-dependent genes by their known or putative functions.