Histone Acetyltransferases and Deacetylases Are Required for Virulence, Conidiation, DNA Damage Repair, and Multiple Stresses Resistance of Alternaria alternata

Histone acetylation, which is critical for transcriptional regulation and various biological processes in eukaryotes, is a reversible dynamic process regulated by HATs and HDACs. This study determined the function of 6 histone acetyltransferases (HATs) (Gcn5, RTT109, Elp3, Sas3, Sas2, Nat3) and 6 histone deacetylases (HDACs) (Hos2, Rpd3, Hda1, Hos3, Hst2, Sir2) in the phytopathogenic fungus Alternaria alternata by analyzing targeted gene deletion mutants. Our data provide evidence that HATs and HDACs are both required for mycelium growth, cell development and pathogenicity as many gene deletion mutants (ΔGcn5, ΔRTT109, ΔElp3, ΔSas3, ΔNat3, ΔHos2, and ΔRpd3) displayed reduced growth, conidiation or virulence at varying degrees. In addition, HATs and HDACs are involved in the resistance to multiple stresses such as oxidative stress (Sas3, Gcn5, Elp3, RTT109, Hos2), osmotic stress (Sas3, Gcn5, RTT109, Hos2), cell wall-targeting agents (Sas3, Gcn5, Hos2), and fungicide (Gcn5, Hos2). ΔGcn5, ΔSas3, and ΔHos2 displayed severe growth defects on sole carbon source medium suggesting a vital role of HATs and HDACs in carbon source utilization. More SNPs were generated in ΔGcn5 in comparison to wild-type when they were exposed to ultraviolet ray. Moreover, ΔRTT109, ΔGcn5, and ΔHos2 showed severe defects in resistance to DNA-damaging agents, indicating the critical role of HATs and HDACs in DNA damage repair. These phenotypes correlated well with the differentially expressed genes in ΔGcn5 and ΔHos2 that are essential for carbon sources metabolism, DNA damage repair, ROS detoxification, and asexual development. Furthermore, Gcn5 is required for the acetylation of H3K4. Overall, our study provides genetic evidence to define the central role of HATs and HDACs in the pathological and biological functions of A. alternata.


INTRODUCTION
The fundamental structural subunit of the chromatin is the nucleosome, which consists of approximately 200 bp of DNA wrapped around an octamer of core histone proteins H2A, H2B, H3, and H4 (Kornberg and Lorch, 1999). The reversible acetylation of specific lysine residues in N-terminal tails of core histones is maintained by histone acetyltransferases (HATs) and histone deacetylases (HDACs) and has been found to play a pivotal role in chromatin-regulated DNA events, such as DNA replication, damage repair, recombination, and gene expression in eukaryotes (Lee and Workman, 2007;Yang and Seto, 2008). HATs are classified into five families, including MYST (Ybf2/Sas3, Sas2, MOZ, Tip60), GNAT (Gcn5-related N-acetyltransferases), p300/CBP, nuclear receptor cofactors, and basal transcription factors (Jeon et al., 2014). HDACs are grouped into four groups based on phylogenetic analysis and sequence homology, including Class I and II (Rpd3, Hda1), Class III (Sir2 or Sir2like protein) and Class IV (HDAC11) (Gregoretti et al., 2004;Jeon et al., 2014).
In yeast, many HATs or HDACs have been found to be involved in transcriptional regulation and many cellular functions. Gcn5 belongs to the GNAT family and acts as a core subunit of several histone acetyltransferase (HAT) complexes, such as ADA, SAGA, or SALSA (Lee and Young, 2000). Gcn5 acetylates N-terminal lysine (K) residues of histones H2B (K11/16) or H3 (K14/18/23/27/36), and the loss of its function leads to defects in gene transcription, sexual differentiation and multiple stresses resistance (Helmlinger et al., 2008;Johnsson et al., 2009;Nugent et al., 2010). H3K9/27/56 can be acetylated by RTT109, a HAT of p300/CBP family (Fillingham et al., 2008;Das et al., 2009;Burgess et al., 2010;Da Rosa et al., 2010). In yeast, H3K56ac regulates expression homeostasis and resistance to DNA-damaging agents (Masumoto et al., 2005;Ozdemir et al., 2005;Han et al., 2007;Voichek et al., 2016). Both Sas3 and Esa1 belong to MYST family, which function as the subunits of Nucleosome Acetyltransferase of histone H3 (NuA3) and H4 (NuA4) complexes, respectively (Lee and Workman, 2007). Studies revealed that Esa1 is essential for cell cycle progression in yeast (Clarke et al., 1999). In S. cerevisiae, the transcription of FLO1 was significantly reduced in the absence of both Sas3 and Ada2 (Church et al., 2017). Rpd3 is a founding member of class I and comprises an N-terminal deacetylase domain (Vidal and Gaber, 1991). In yeast, Rpd3 functions as a catalytic subunit in sin3 complexes and has been linked to many biological functions, such as transcriptional regulation, DNA repair, replication, ROS resistance, or azole resistance (Vogelauer et al., 2002;Jazayeri et al., 2004;Baker et al., 2013;Li et al., 2015;Bosio et al., 2017). Hos2 and Hda1 belong to class II, and both of them also contain an N-terminal deacetylase domain (Yang and Seto, 2008). In yeast, Hda1 and Rpd3L contribute to the repair of replicationborn DSBs (double strand breaks) by facilitating cohesin loading thus preventing genome instability (Ortega et al., 2019). Hos2 acts as a key subunit of Set3 complex which deacetylates H4 and H3 by counteracting Esa1 (Pijnappel et al., 2001;Torres-Machorro et al., 2015). The yeast Hos2 is involved in various biological functions, such as these involved in cell integrity pathway, cell resistance to multiple stresses, DNA damage repair, and expression of growth-related genes (Pijnappel et al., 2001;Wirén et al., 2005;Cohen et al., 2008;Kim et al., 2012;Tkach et al., 2012). The best-characterized member in Class III is the NAD-dependent histone deacetylase (HDAC) Sir2. Yeast Sir2 acts as anti-aging gene required for cellular lifespan regulation by caloric restriction (Lin et al., 2000;Dang et al., 2009).
In filamentous fungi, HATs and HDACs have been found to be involved in growth, development, virulence, synthesis of secondary metabolites and multi-stress resistance. Studies in Aspergillus nidulans showed that GcnE (ortholog of Gcn5) controls asexual reproduction, and the biosynthesis of sterigmatocystin, penicillin and terrequinone A (Nützmann et al., 2011;Cánovas et al., 2014). Genetic deletion of Gcn5 leads to severe defects in virulence, sexual or asexual development, and/or virulence in Ustilago maydis, Beauveria bassiana, and Trichoderma reesei (Xin et al., 2013;González-Prieto et al., 2014;Cai et al., 2018b). In B. bassiana, RTT109 deletion mutant shows attenuated virulence, and increased sensitivity to the stress of oxidation and DNA-damaging agents (Cai et al., 2018c). Many other HATs also have been found to be involved in cell growth and pathogenicity in M. oryzae (Sas3), F. graminearum (Elp3), and so on Dubey et al., 2019). HDAC Hos2 acts as a vital subunit of Tig1 complex crucial for conidiation and infectious growth in M. oryzae, and is required for pathogenesis and dimorphic switch in U. maydis (Ding et al., 2010;Elías-Villalobos et al., 2015). Although many HATs and HDACs have been identified to be critical for an array of biological processes in many fungi, the specific function of each gene is not completely the same among different fungi.
Alternaria alternata tangerine pathotype is a disastrous fungal pathogen which causes Citrus Brown Spot on young leaves, shoots and fruits of many citrus cultivars. Previous studies revealed that ROS detoxification-related genes such as Ap1 and Skn7 (Lin et al., 2009;Chen et al., 2012) or host-selective ACT-toxin biosynthesis genes such as ACTT5 and ACTT6 (Miyamoto et al., 2009), are critical for A. alternata to cause lesions on host tissues. Although numerous studies of HATs or HDACs indicate that many of these genes are involved in ROS detoxification, synthesis of secondary metabolites, and various biological functions in many pathogenic fungi, the roles of these genes in A. alternata have not been elucidated. In this study, we identified all the HATs and HDACs orthologs in A. alternata and clarified their roles in growth and development, carbon source utilization, multiple stresses resistance, DNA damage repair and virulence using genetic and biological analyses. We also performed a transcriptome analysis to explore the regulatory role of Gcn5 and Hos2 in this important citrus pathogen.

Fungal Strains and Culture Conditions
The wild-type Z7 strain of A. alternata (Fr.) Keissler used in the mutagenesis experiments was isolated from an infected citrus (Citrus suavissima Hort. Ex Tanaka) in Zhejiang, China (Huang et al., 2015;. Fungal strains were cultured on potato dextrose agar (PDA) at 26 • C and conidia were collected after incubating on V8 medium for 8 days. Mycelium cultured in liquid potato dextrose broth (PDB) was collected by passing through cheesecloth and used for purification of DNA or RNA.

Microscopy
Morphological observation of conidia and hyphae was carried using a Nikon microscope equipped with a LV100ND image system (Nikon, Japan).

Virulence Assays
Fungal virulence was assessed on detached Dancy (Citrus reticulata Blanco) leaves inoculated by placing a 5-mm dia. agar plug covered with fungal mycelium on each spot, and those inoculated leaves were kept in a plastic box at 26 • C for 2-4 days for lesion development.

Transcriptome Analysis
Wild-type, Gcn5 and Hos2 each with two replicates were cultured in PDB (120 rpm, 28 • C) and were harvested after 36 h. RNA was extracted using Axygen RNA purification kit (Capital Scientific, Union City, CA, United States). The libraries were produced using an IlluminaTruSeq RNA Sample Preparation Kit, and they were sequenced on an Illumina HIseq 2500 platform, generating 150 bp paired-end reads. Trimmomatic Ver 0.36 (Bolger et al., 2014) was used to remove adaptors and lowquality reads. TopHat2 (Kim et al., 2013) was used to map sequences to the reference genome and the mapped reads on each gene were determined by HTSeq (Anders et al., 2015). Differential expression analysis was determined using DESEQ2 (Anders and Huber, 2012) based on the overall transcript counts after normalization. Transcripts with an adjusted FDR less than 0.01 and the absolute value of log2FC (log2 fold change) greater than 2 were considered to be differentially expressed genes (DEGs). DEGs were annotated by searching against the NCBI nr databases. GO and KEGG pathway were performed using clusterProfiler v3.6 . anti-SMASH 4.0 (Blin et al., 2017) was used to predict gene clusters associated with secondary metabolites. Protein domains were predicted in SMART database available at http://smart. embl-heidelberg.de/. Carbohydrate-active enzymes (CAZymes) were predicted using dbCAN meta server. 1 The transcriptomes raw data (SRS6969981, SRS6969980, SRS6969978, SRS6969979, SRS6969977, SRS6969976) have been deposited in NCBI's Sequence Read Archive.

Whole-Genome Resequencing Analysis
A paired-end library of each sample was constructed and sequenced by Hiseq2500, with insert sizes of approximately 350 bp. Trimmomatic was used to remove adaptors and lowquality reads. The high-quality resequencing reads was mapped to the Alternaria alternata reference genome with Bowtie2 (Langmead and Salzberg, 2012) using the default parameters. The aligned results were further filtered for unmapped reads and duplicated reads using SAMtools (Li et al., 2009) and Picard packages, respectively. SNP calling was carried out using Freebayes. Genomic variation analysis was performed using VCFtools (Danecek et al., 2011). IGV (Helga et al., 2013) was used to visualize reads on the genome. MapChart (Voorrips, 2002) was used to visualize SNPs on the genome. The raw data of whole-genome resequencing (SRS6971885, SRS6971887, SRS6971884, SRS6971886) have been deposited in NCBI's Sequence Read Archive.

Gene Expression Analyses
Quantitative Real-time PCR (qRT-PCR) was carried out on a 7300 Real Time PCR system (Applied Biosystems, Carlsbad, CA, United States) to validate the transcriptome data. RNA was extracted with an Axygen RNA purification kit (Capital Scientific) and the cDNA was synthesized from RNA using a PrimeScript RT regent kit (Takara, Shiga, Japan). Actin-coding gene (KP341672) was used as an internal control and the resulting data were normalized using the comparative Cτ method as described previously (Sun et al., 2011).

Statistical Analysis
The statistical significance of treatments was determined by analysis of variance and means separated by Duncan's test (P < 0.05).

Identification and Genetic Deletion of Histone Acetyltransferases and Histone Deacetylases in Alternaria alternata
Seven HATs and six HDACs were identified combining the prediction results of dbHiMo 2 and hmm analysis, and the families to which these genes belong are also listed in Table 1. Nucleotide sequence length of corresponding predicted genes ranged from 654 to 3,450 bp, the number of introns contained varied from 0 to 8, and the length of proteins varied from 193 to 1,082 aa ( Figures 1A,B). Functional domain prediction of amino acid sequences encoded by these genes revealed that all 13 proteins contain functional domains unique to HAT or HDAC, indicating these gene predictions are highly reliable. Mapping these HATs and HDACs to chromosomes revealed that all 13 genes are localized in the essential chromosomes, whereas no HATs or HDACs were identified in conditionally dispensable chromosomes (CDC) (Supplementary Figure 1). Phylogenetic analysis using the neighbor-joining method revealed that all HATs and HDACs predicted in A. alternata were highly similar to their corresponding genes in other fungi ( Figure 1C).

Histone Acetyltransferases and Histone Deacetylases Are Required for Growth and Cell Development
In HATs disrupted mutants, vegetative growth rate of Gcn5, Elp3, RTT109, Nat3, and Sas3 was reduced by 66, 44, 12, 71, and 41% compared to wild-type on potato dextrose agar (PDA), indicating that most HATs play a vital role on the growth of A. alternata (Figure 2A and Supplementary Figure 3). In addition, Gcn5 and Nat3 produced fewer aerial hyphae than wild-type on PDA. Microscopic examination revealed that the impacts of different HATs on the development of A. alternata varied greatly. The Gcn5 strain produced swollen and stubby hyphae and more hyphae branches, but failed to produce any spores (Figures 2B,C and Supplementary Figure 4). No spores were observed in Nat3 after 6 days of incubation, but a few spores appeared as incubation time increased. The hyphal tip of Nat3 was irregularly twisted and the hyphae were frequently branched. The conidiation of Elp3, Sas3, and RTT109 decreased significantly, but the mycelium morphology remained the same as the wild-type level. Only Sas2 displayed wild-type levels of vegetative growth, hyphal development, and conidiation.

Histone Acetyltransferases and Histone Deacetylases Are Required for Fungal Virulence
To determine the effect of HATs and HDACs on the pathogenicity of A. alternata, virulence assays were performed on detached   Supplementary Table 10.
Frontiers in Microbiology | www.frontiersin.org FIGURE 3 | HATs and HDACs were required for A. alternata pathogenicity. Pathogenicity was assayed on detached Dancy leaves by placing 5 mm agar plug covering fungal mycelium on the bottom of leaves (left, mutants; right, Z7). The leaves were incubated in a plastic box for lesion development and the necrotic lesions was observed at 2 days post inoculation (dpi). The percent changes of necrotic lesions appearing on the leaves calculated in relation to those induced by Z7 are also indicated.
Dancy leaves. For HATs deletion mutants, Gcn5, Sas3, and Nat3 failed to induce visible lesion on citrus leaves. RTT109 and Elp3 induced smaller necrotic lesions than those induced by the wild type 2 days post inoculation (dpi) (Figure 3 and Supplementary Figure 5). Only Sas2 induced necrotic lesions at a rate and magnitude comparable to those for the wildtype on Dancy leaves 2 dpi. For HDACs deletion mutants, Rpd3 and Hos2 induced smaller necrotic lesions compared to those induced by the wild type at 2 dpi. Hos3, Sir2, and Hda1 produced necrotic lesions similar to those induced by Z7 (Figure 3 and Supplementary Figure 5). Quantitative analysis revealed that the area of lesion induced by the RTT109, Elp3, Sas2, Hos2, Rpd3, Hda1, Hos3, Hst2, and Sir2 were about 79,32,92,18,11,97,104,96, and 103%, respectively, of those induced by Z7.

Histone Acetyltransferases and Histone Deacetylases Coordinate Carbon Sources Utilization
To examine whether HATs or HDACs were required for the utilization of carbon sources, mycelial radical growth assays were performed on MM medium supplemented with sucrose, starch, glucose, or lactose as sole carbon sources (Figure 4). In HATs disruption mutants, the vegetative growth inhibition rate of Gcn5 on all tested sole carbon source media for 4 days was 100%. At the same time, the growth inhibition ratio of Sas3 on the MM medium supplemented with glucose, sucrose, lactose, and starch was increased by 28, 41, 39, and 41%, compared to the wild type. In contrast, Sas2, Elp3, and RTT109 displayed wild-type sensitivity to all tested sole carbon source media. Most HDACs disruption mutants, including Sir2, Hos3, Hda1, and Hst2, exhibited wild-type sensitivity to all test medium. Only Hos2 showed slightly increased sensitivity on carbon sources of glucose and starch.
To prove further that the severe growth defects of Gcn5 on sole carbon source medium was not caused by its slow vegetative growth, the incubation time was extended to 10 days. Growth examination indicated that the growth inhibition ratio of Gcn5 on the medium with lactose as sole carbon source was still 100%, and only a few hyphae of Gcn5 appeared on carbon sources of sucrose, starch, or glucose (the growth inhibition ratio is close to 100%), confirming that the Gcn5 was required for the utilization of carbon sources (Supplementary Figure 6).

Histone Acetyltransferases and Histone Deacetylases Are Involved in Resistance to Multiple Stresses
In mutants defective for HATs, both Elp3 and Gcn5 were more sensitive to H 2 O 2 and CHP (cumyl-H 2 O 2 ), while Sas3 and RTT109 just showed increased sensitivity to H 2 O 2 , compared with wild-type. In addition, Sas3 showed elevated sensitivity to cell wall-targeting agents such as Congo red (CR) and sodium dodecyl sulfonate (SDS). At the same time, Elp3, Gcn5, and RTT109 just displayed increased sensitivity to CR. Sas3, Gcn5, and RTT109 displayed increased sensitivity to NaCl, KCl, and CuSO 4 . Only RTT109 displayed increased sensitivity to CaCl 2 . However, Sas3, Elp3, and Gcn5 displayed enhanced resistance to CaCl 2 compared to wild-type. Both Sas3 and Elp3 also showed increased tolerance to V8 medium compared to wild-type. Only Sas2 showed wild-type resistance to all test chemicals (Figure 5).
Unlike HATs which were widely involved in the adaption to various stresses, most of HDACs such as Hos3, Hda1, Hst2, and Sir2 were not involved in the resistance to most tested chemicals. Sensitivity assays revealed that Hos2 showed increased sensitivity to H 2 O 2 , CHP, CR, SDS, NaCl, KCl, CaCl 2 , and V8 medium, but not sorbitol and CuSO 4 (Figure 6).
Assays for fungicide sensitivity indicated that both Gcn5 and Hos2 displayed increased sensitivity to chlorothalonil, but not mancozeb and boscalid. All the other HATs or HDACs mutants displayed wild-type sensitivity to the test fungicides (Supplementary Figure 7).

Transcriptome Analysis Defines the Global Regulatory Role of Gcn5 and Hos2
The results described above indicated that both Gcn5 and Hos2 played vital roles in vegetative growth, development, multiple stresses resistance, carbon sources utilization, ROS detoxification, and virulence. To explore the transcriptional FIGURE 4 | HATs and HDACs were involved in carbon source utilization. Z7, and mutants with impaired HATs or HDACs cultured on MM medium amended with sucrose, glucose, lactose, or starch as sole carbon source. The percentage of growth reduction determined by comparing a cumulative percentage of the growth of Z7 and mutants grown on same medium is also shown (*, p < 0.01). regulatory mechanism of Gcn5 and Hos2, transcriptome analyses of Gcn5 and Hos2 compared to wild type were performed. Overall, 1,503 up-regulated and 1,051 down-regulated genes were identified in Gcn5 (Supplementary Table 2). In Hos2, 469 up-regulated and 536 down-regulated genes were identified (Supplementary Table 3). Gcn5 and Hos2 were one of the genes with lowest transcription level in Gcn5 and Hos2, respectively, indicating the genetic deletion of Gcn5 and Hos2 was reliable (Supplementary Tables 2, 3). In addition, RT-qPCR results of 14 randomly selected differentially expressed genes (DEGs) was consistent with transcriptome data of Gcn5 and Hos2, indicating that the transcriptome data were reliable. Although the DEGs in Gcn5 and Hos2 accounted for up to 20 and 8% of all genes in the genome, respectively, the transcription levels of the other 12 HATs and HDACs in Gcn5 or Hos2 did not change significantly, indicating that Gcn5 and Hos2 are not required for the transcription of the other HATs and HDACs.
COG (Clusters of Orthologous Groups of proteins) analysis of Gcn5 and Hos2 revealed that many DEGs were enriched in the metabolism category, such as Carbohydrate transport and metabolism (G), Amino acid transport and metabolism (E), Lipid transport and metabolism (I), and Inorganic ion transport and metabolism (P). There are also many DEGs in Gcn5 or Hos2 enriched in Transcription (K) and Replication, recombination and repair (L) (Supplementary  Figures 8, 9). KEGG enrichment analysis revealed that the most affected metabolism-related pathways in Gcn5 or Hos2 were mainly related to carbon and nitrogen sources metabolism, such as starch and sucrose metabolism, galactose metabolism, Fructose and mannose metabolism, Methane metabolism, and Nitrogen metabolism. In addition, many DEGs in Gcn5 or Hos2 were enriched in genetic information processing, such as DNA damage repair, RNA transport, and tRNA biosynthesis (Supplementary Figures 8, 9). Although many DEGs in Gcn5 and Hos2 were enriched in the same pathways, the amount and expression level of those genes varied greatly.
In addition, deletion of Gcn5 or Hos2 also affected the transcription of genes involved in Carbohydrate-active enzymes (CAZymes), secondary metabolite cluster, CYP450, and conidiation (Supplementary Tables 5-8 and Supplementary  Figure 12). The transcription of all ACT toxin synthesisrelated genes did not change significantly in Gcn5 and Hos2 compared to wild-type. However, AALT_g11758 (CYP450), located in the ACT toxin gene cluster and required for pathogenicity but plays no role for ACT toxin synthesis (Wang et al., 2019), was significantly down-regulated in both Gcn5 (FC = −25) and Hos2 (FC = −6).

Histone Acetyltransferases and Histone Deacetylases Are Required for the Resistance to DNA-Damaging Agents
Previous studies showed that Gcn5, Hos2, and RTT109 are required for DNA damage repair in B. bassiana and S. cerevisiae (Burgess et al., 2010;Cai et al., 2018a,c). Therefore, we tested whether RTT109, Gcn5, and Hos2 were involved in the resistance to DNA-damaging agents such as camptothecin (CPT), hydroxyurea (HU) and methyl methanesulfonate (MMS) in A. alternata. Compared with wild-type, growth of RTT109 was significantly suppressed by 40 and 89% on PDA containing 5 µM CPT or 0.1% MMS, respectively (Figure 8). However, RTT109 displayed wild-type radial growth on PDA amended with 5 mM HU. Unlike RTT109, the growth of Hos2 on HU-added PDA was inhibited by 20% compared to wildtype. At the same time, Hos2 showed wild-type resistance to 5 µM CPT or 0.1% MMS. In addition, Gcn5 displayed increased sensitivity to 5 mM HU and 0.1% MMS (the growth inhibition rate is 29 and 19%, respectively, compared FIGURE 8 | HATs and HDACs were involved in the resistance to DNA-damaging agents. Z7, Hos2, RTT109, and Gcn5 were grown on PDA with or without the indicated chemicals [5 mM hydroxyurea (HU), 0.1% methyl methanesulfonate (MMS), or 5 µM camptothecin (CPT)]. The percentage of growth reduction determined by comparing a cumulative percentage of the growth of Z7 and mutants grown on same medium is also shown (*, p < 0.01).
to wild-type) but displayed wild-type resistance to 5 µM CPT (Figure 8).

Gcn5 Is Required for DNA Damage Repair
In order to analyze the DNA damage repair ability of Gcn5, we performed genome resequencing using ultraviolet ray (UV) irradiated Gcn5 and wild-type. In the Gcn5 coding region, no reads were detected in Gcn5, but many reads were found in wild-type, which proves once again that the knock-out experiment of Gcn5 was successful ( Figure 9B). In Gcn5, 949 unique SNPs were detected after UV irradiation. However, in UV irradiated wild-type, only 298 unique SNPs were found, which accounted for 31% of Gcn5, indicating that Gcn5 is involved in the repair of DNA damage caused by UV ( Figure 9A).

Histone Acetylation Targets of Gcn5
To test whether histone acetylation levels were altered in Sas3 and Gcn5, western blot was carried out using specific antibodies, directed against H3K4ac, while antibody against H3 was used as a loading control. As shown in Figure 10, the level of H3K4ac was significantly decreased (P < 0.01) in the Gcn5 compared to Z7. No reduction of the signal for H3K4ac was observed in Sas3.

DISCUSSION
Histone acetylation, which is important for multiple cellular processes, occurs in different organisms ranging from fungi to mammals, and the acetylation level manifested by HATs and HDACs dynamically changes over time during development and FIGURE 9 | SNPs analysis of Gcn5 and wild-type irradiated by UV. Gcn5 and Z7 grown on PDA were irradiated with UV for 10 s every 24 h, and repeated 4 times. Hyphae were taken using sterilized toothpicks and incubated in PDB (150 rpm, 2 days). DNA was extracted from mycelium collected from PDB for whole-genome resequencing. The protocol for SNPs analysis is described in Experimental procedures. (A) Statistical analysis of SNP number in Gcn5 and Z7. (B) The mapping result of Gcn5 and Z7 genome sequencing data on the reference genome.
FIGURE 10 | Western blotting analysis of proteins extracted from Z7, Sas3, and Gcn5, respectively. The respective strains were grown for 36 h in PDB. The anti-acetyl H2K4 (H3K4ac) and anti-acetyl H3K18 (H3K18ac) were used for detecting alteration of acetylation levels. Antibody H3 was used as a loading reference. (*, p < 0.01).
differentiation (Lee and Workman, 2007;Kong et al., 2018). In plant pathogenic fungi, several HATs and HDACs have been found to be required for growth, virulence and adaptation to environmental stress (Maeda et al., 2017;Kong et al., 2018;Dubey et al., 2019). Although HATs and HDACs are conserved in cells, the functions of these genes are various in different species.
In this study, a total of 13 HATs and HDACs were identified in A. alternata. To systematically elucidate the functions of HATs and HDACs in A. alternata, we created 6 HATs and 6 HDACs deletion mutants. Our studies showed that HATs and HDACs were widely involved in biological processes of mycelial growth and development, conidial production, multiple stresses resistance, carbon source utilization, DNA damage repair, and pathogenicity in A. alternata.
As shown in the present study, HATs and HDACs are involved in developmental processes because Gcn5, Hos2, Rpd3, Nat3, Sas3, and RTT109 produce no or fewer conidia. In addition, Gcn5, Rpd3, and Nat2 produces more hyphal branches than wild-type. In some filamentous fungi like N. crassa and A. nidulans, the central development pathway relies upon the key activator genes BrlA, AbaA, and WetA (Etxebeste et al., 2010;Park and Yu, 2012). Six upstream developmental activators (FlbA, FlbB, FlbC, FlbD, FlbE, and FluG) are required for the activation of BrlA and the initiation of conidiation (Adams et al., 1988;Wieser et al., 1994). Our transcriptome data revealed that FlbA, FlbC, and FlbD were down-regulated in Gcn5 (Supplementary Table 8). Previous studies have revealed the conidiation of A. alternata is closely regulated by the calcium-mediated signaling pathway, Fus3 and Slt2 MAPK signaling pathway, as well as the cAMP dependent protein kinase A (PKA) Wang et al., 2010;Yago et al., 2011;Tsai et al., 2013). Other factors including NADPH oxidase (Nox), thioredoxin reductase (Trr1), glutathione reductase (Glr1), nascent polypeptide associated complex α subunit (Nac1), and skn7 response regulator are also required for the developmental processes in A. alternata (Chen et al., 2012;Yang and Chung, 2013;Ma et al., 2018;Wang et al., 2020). Whether or not those regulators interact with HATs or HDACs remains to be determined. In addition to conidia formation, HATs and HDACs are required for vegetative growth.
Experiments further demonstrated that Gcn5, Sas3, RTT109, and Hos2 are required for cellular resistance to osmotic stress. In yeast, SAGA (Spt-Ada-Gcn5) is required for the transcription of Hog1-mediated genes under severe osmostress (Zapater et al., 2007). Chip-on-chip experiments revealed that acetylation at lysines 9 and 12 of histone H3 increases in osmostress up-regulated genes and decreases in repressed genes (Magraner-Pardo et al., 2014). The Hog1 deletion mutant in A. alternata displayed reduced vegetative growth rate and increased sensitivity to KCl and NaCl, which is consistent with the phenotype of Gcn5, Sas3, RTT109, and Hos2, indicating that transcription of some genes regulated by Hog1, Gcn5, Sas3, and Hos2 may be similar . Similar results were observed in F. graminearum, in which the Hog1, Gcn5, Sas3, RTT109 are also involved in vegetative growth and resistance to osmotic stress (Nguyen et al., 2012;Kong et al., 2018). In addition, Sas3 and Hos2 are more sensitive to cell wall-disturbing compounds including CR and SDS, indicating both of these genes play a positive role in cell wall assembly.
Carbon source (sugar) metabolism, which is one of the most important biological processes in organisms, is involved in growth, development, and multiple stress resistance (Koch, 2004;Ene et al., 2012). This study found that Sas3, Hos2, and especially Gcn5 displayed severe growth defects on sole carbon source medium, suggesting an involvement of these genes in carbon metabolism in A. alternata. In C. albicans, acetyl-CoA and acetate metabolism play a central role in growth on both glucose and non-glucose carbon sources (Carman et al., 2008). In B. bassiana, deletion of Rpd3 and Gcn5 resulted in growth defects on medium modified with different carbon sources, confirming the important role of histone acetylation in carbon source utilization (Cai et al., 2018b,d). In addition, most DEGs enriched in carbon source metabolism pathways, such as "Glycolysis/Gluconeogensis, " Fructose and Mannose Metabolism, " "Galactose Metabolism, " "Starch and Sucrose Metabolism, " etc., are down-regulated in Gcn5 and Hos2, making it not surprising to see the reduced carbon source utilization ability in Gcn5 and Hos2. In microorganisms, glucose metabolites would lead to carbon catabolite repression (CCR) and inhibit the utilization of other carbon sources (Celenza and Carlson, 1986;Kayikci and Nielsen, 2015). Sucrose non-fermenting 1 (SNF1) protein kinase complex is required for the utilization of non-fermentable carbon sources in fungi and enable them to adapt to the adversity of glucose deficiency (Alepuz et al., 1997). In Fusarium virguliforme, the growth of FvSnf1 deletion mutants on medium amended with galactose as sole carbon source is completely inhibited (Islam et al., 2017). SNF1 complex-related genes (AaSnf1, AaSip2, and AaSnf4) were not significantly differentially expressed in Gcn5 and Hos2 compared to wild-type. However, the transcription of SUC2 (AALT_g2401), which is regulated by Snf1 and is essential for sucrose utilization in S. cerevisiae (Carlson et al., 1981;Oezcan et al., 1997), was down-regulated significantly in Gcn5 and Hos2. Previous studies have shown that Snf1, which regulates Gcn5 occupancy and H3 acetylation at a specific sequence, works in concert with Gcn5 to regulate transcription (Lo et al., 2001;Abate et al., 2012). In addition, similar with Gcn5 and Hos2, Snf1 also plays an important role in the carbon source utilization in A. alternata (Wang et al., 2020). Therefore, we speculate that the interaction relationship between Snf1 and these two genes may also exist.
Mutants lacking Gcn5, Sas3, Elp3, RTT109, or Hos2 in A. alternata were defective in ROS detoxification at varying degrees, indicating all of these five genes are involved in resistance to oxidative stress. RTT109, Gcn5, and Hos2 in B. bassiana, Elp3 and Sas3 in F. graminearum also play important roles in the resistance to oxidative stress Cai et al., 2018a,b,c;Kong et al., 2018). However, Gcn5 in Aspergillus flavus is not required for ROS resistance (Lan et al., 2016). In addition, Hos2 plays a negative role in H 2 O 2 resistance in M. oryzae, indicating that the function of specific genes in HATs and HDACs have been differentiated in fungi . In A. alternata, the ROS detoxification ability is critical for pathogenesis and survival (Chung, 2012;Ma et al., 2018;Gai et al., 2019). A. alternata strains lacking Nox, Hog1, Yap1, Skn7, Gpx3, or Nac1 all exhibit increased cellular sensitivity to ROS and decreased pathogenicity on citrus leaves Chen et al., 2012;Yang and Chung, 2012;Chung, 2014;Yang et al., 2016;Wang et al., 2020). Virulence assays revealed that the virulence of Gcn5 and Sas3 was completely lost, which is consistent with the findings in F. graminearum (Kong et al., 2018). Furthermore, the virulence of Elp3, RTT109, and Hos2 decreased significantly, supporting the requirement of ROS detoxification for successful infection by A. alternata. In addition, we observed that Nat3 and Rpd3 were significantly reduced in pathogenicity and vegetative growth rate. The severe growth defects of Nat3 and Rpd3 may contribute to its reduced virulence. Recent research shows that chlorothalonil causes a redox state change leading to oxidative stress generation in Danio rerio (Da Silva Barreto et al., 2020). Interestingly, sensitivity assays revealed that A. alternata strains with impaired Gcn5 or Hos2 displayed high sensitivity to chlorothalonil fungicide. In addition, transcriptome data of Gcn5 revealed that all DEGs in the sterol synthesis pathway were significantly down-regulated, including CYP51 homologous genes, which is closely related to DMI fungicide resistance. In C. albicans, the loss of Hda1 and Rpd3 both caused the pathogen to be significantly more sensitive to azole fungicides, indicating that histone acetylation levels are crucial for pathogens to resist DMI fungicides . Transcriptome analysis also revealed that most DEGs involved in ROS detoxification were up-regulated in Gcn5 and Hos2 even without external oxidative stress, implying that the redox state in cells of these two mutants are changed. In eukaryotic cells, endoplasmic reticulum (ER) is essential for folding and trafficking of proteins that enter the secretory pathway, while the disorder of redox homeostasis of ER could result in protein misfolding (Wang and Kaufman, 2016). In Gcn5 and Hos2, many genes involved in "Protein processing in endoplasmic reticulum, " "Ubiquitin mediated proteolysis, " and "Regulation of autophagy" pathways are up-regulated, which mainly play roles in protein folding and degradation.
It is now clear that a subset of HATs and HDACs regulate the genome stability (Driscoll et al., 2007;Nagarajan et al., 2013). In B. bassiana, mutants lacking HATs (Mst2 and RTT109) or HDACs (Hos2) displayed increased sensitivity to DNA-damaging stresses, indicating that all of these genes are involved in DNA damage repair (Cai et al., 2018a,c;Wang et al., 2018). In our study, many DEGs are enriched in DNA damage repair pathways. Moreover, Gcn5, RTT109, and Hos2 displayed hypersensitivity to a variety of DNA-damaging agents such as MMS, HU, or CPT, all of which induce replication fork collapse or stalling (Interthal and Heyer, 2000;Boddy et al., 2001;Doe et al., 2002), suggesting Gcn5, RTT109, and Hos2 play an important role in DNA damage repair in A. alternata. This speculation was evidenced by the result that more SNPs were generated in Gcn5 in comparison to wild-type when both of them were irradiated by UV.
In summary, our study revealed that HATs and HDACs are required for the vegetative growth, conidiation, pathogenicity, multiple stresses resistance, DNA damage repair, and carbon source utilization in A. alternata. In addition, Gcn5 and Hos2 play direct or indirect roles in the transcriptional regulation of genes involved in carbon source metabolism, DNA damage repair, ROS detoxification, protein processing and degrading, ergosterol synthesis, and secondary metabolite synthesis.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

AUTHOR CONTRIBUTIONS
HM, FG, and HL contributed to the conception of the study. HM, LL, YG, XYZ, YNC, XKZ, and YZC performed the experiments. HM, YG, and CJ performed the data analysis. HM wrote the manuscript. All authors contributed to the article and approved the submitted version.