Deregulation of secondary metabolism in a histone deacetylase mutant of Penicillium chrysogenum

Abstract The Pc21 g14570 gene of Penicillium chrysogenum encodes an ortholog of a class 2 histone deacetylase termed HdaA which may play a role in epigenetic regulation of secondary metabolism. Deletion of the hdaA gene induces a significant pleiotropic effect on the expression of a set of polyketide synthase (PKS) and nonribosomal peptide synthetase (NRPS)‐encoding genes. The deletion mutant exhibits a decreased conidial pigmentation that is related to a reduced expression of the PKS gene Pc21 g16000 (pks17) responsible for the production of the pigment precursor naphtha‐γ‐pyrone. Moreover, the hdaA deletion caused decreased levels of the yellow pigment chrysogine that is associated with the downregulation of the NRPS‐encoding gene Pc21 g12630 and associated biosynthetic gene cluster. In contrast, transcriptional activation of the sorbicillinoids biosynthetic gene cluster occurred concomitantly with the overproduction of associated compounds . A new compound was detected in the deletion strain that was observed only under conditions of sorbicillinoids production, suggesting crosstalk between biosynthetic gene clusters. Our present results show that an epigenomic approach can be successfully applied for the activation of secondary metabolism in industrial strains of P. chrysogenum.


| INTRODUCTION
During the last decades, the filamentous fungus Penicillium chrysogenum has been used extensively in industry for the production of the β-lactam antibiotic penicillin (Fleming, 1929). The biosynthetic pathway and the corresponding genes involved have been well described and current production strains are generated for the high-level production of penicillins through the implementation of an intense classical strain improvement program. However, the full potential of secondary metabolism of P. chrysogenum largely remained unknown till the genomic sequence became available (Van Den Berg et al., 2008). The genome specifies multiple genes for secondary metabolite formation including 20 polyketide synthases (PKSs), 10 nonribosomal peptide synthetase (NRPSs), 2 hybrids (PKS-NRPS), and 1 dimethylallyltryptophan synthase. The function of most of these genes remains unknown (Samol, Salo, Lankhorst, Bovenberg & Driessen, 2016;Van Den Berg et al., 2008). Recently, a genome-based identification and analysis of the roquefortine meleagrin NRPS gene cluster was performed for P. chrysogenum (Ali et al., 2013;Garcia-Estrada et al., 2011;Shang, Zhang, & Zheng, 2013;Veiga et al., 2012).
However, unlike the roquefortine gene cluster, the expression level of the majority of the secondary metabolite genes under laboratory conditions is low (Van Den Berg et al., 2008). Therefore, more elaborate methods other than gene inactivation are required for identification and further analysis of these so-called "silent" secondary metabolite genes.
New approaches have evolved during the postgenomic era to activate gene clusters such as interference with cluster specific regulatory genes or even of pleiotropic regulator of chromatin structure like LaeA. This has triggered the research on the cryptic potential of fungal secondary metabolism (Brakhage & Schroeckh, 2011). A potential powerful approach is the epigenetic regulation of gene expression. In eukaryotic cells, DNA is compacted into a complex chromatin structure. The histone proteins H2A, H2B, H3, and H4 form the core histone octamer complex with DNA called nucleosome, the structural and functional unit of the chromatin (Luger, 2003). The formation of the nucleosomes may interfere with the recognition of the bound DNA by various transcriptional elements causing gene silencing (Lee, Hayes, Pruss, & Wolffe, 1993). Thus, remodeling of the chromatin by the histone modifications is a trigger that influences transcription, replication, and DNA repair (Yu, Teng, Waters, & Reed, 2011;Zhu et al., 2011). The histone acetylation status is controlled by the balanced activity of histone acetylases (HATs) and deacetylases (HDACs) (Brosch, Loidl, & Graessle, 2008). Hyperacetylation of the chromatin induced by deletion or chemical inhibition of HDACs leads to euchromatin formation and transcriptional activation of silent chromosomal regions (Gacek & Strauss, 2012). Cladochromes and calphostin B in Cladosporium cladosporioides and nygerone A from Aspergillus niger are secondary metabolites that have recently been identified with this strategy using the HDAC inhibitor suberoylanilide hydroxamic acid (SAHA) (Carafa, Miceli, Altucci, & Nebbioso, 2013;Fisch et al., 2009;Henrikson, Hoover, Joyner, & Cichewicz, 2009). An altered secondary metabolite profile was also reported for Alternaria alternata and Penicillium expansum treated with HDAC inhibitor Trichostatin A (TSA) (Shwab et al., 2007).
Histone deacetylases are represented by two protein families: the ''classical'' HDACs and the recently established group of NAD +dependent sirtuins (de Ruijter, van Gennip, Caron, Kemp, & van Kuilenburg, 2003). Members of both families were initially described in S. cerevisiae and subsequently identified in filamentous fungi and humans (Taunton, Hassig, & Schreiber, 1996). The ortholog of the RPD3 (reduced potassium dependency) transcription factor and HDA1 of S. cerevisiae belong to the major classes 1 and 2 of the ''classical'' HDACs, respectively. Recently, multiple effects of the inactivation of hda1 ortholog on the expression of secondary metabolite genes has been reported for a number of fungal species (Lee et al., 2009;Shwab et al., 2007;Tribus et al., 2005).
Here, we have demonstrated that ortholog of the class 2 histone deacetylase hda1 of S. cerevisiae (Pc21 g14570) is a key regulator of the secondary metabolism in the filamentous fungus P. chrysogenum.
By means of the transcriptional and metabolite profiling of the individual gene deletion mutants, the role of HdaA in production of the new metabolite, conidial pigmentation, as well as the broad influence was used for preculturing the conidia for 24 hr before inoculation into secondary metabolite production medium (SMP; (Ali et al., 2013)).

| Plasmids construction
All the plasmids in this study were constructed using the modified Gateway cloning protocol (Invitrogen, California, USA) published previously . 5-′ and 3-′ fragments for the dele-

| Fungal transformation
For all the transformations, 5 μg of plasmid were digested with the suitable restriction enzymes. The linearized plasmid was used to transform protoplasts isolated from P. chrysogenum as described previously   Figure S1).
All the primers used are described in Table S1.

| Southern blot analysis
The 3′ downstream region of the hdaA gene was used as a probe and amplified by PCR with primer set listed in Table S1.

| Genomic DNA extraction
The total genomic DNA (gDNA) was isolated after 96 hr of cultivation in YGG liquid medium using an adapted yeast genomic DNA isolation protocol (Harju, Fedosyuk, & Peterson, 2004). The mycelium was broken in a FastPrep FP120 system (Qbiogene, Cedex, France).

| Total RNA extraction and cDNA synthesis
Total RNA was isolated from colonies of fungal mycelium that was grown on solid R-agar medium and SMP medium for 7 and 3 days,  Pc21 g12600 (chyC), Pc21 g12610 (chyM), Pc21 g12620 (chyD), Pc21 g12630 (nrps 9; chyA), Pc21 g12640 (chyR) (Viggiano et al., 2017)] are shown in the Table S1. Primers were designed at both sides of the introns in order to be able to discriminate between the amplification on gDNA and cDNA. For expression analysis, the γ-actin gene (Pc20 g11630) was used as a control for normalization (Nijland et al., 2010). A negative reverse transcriptase (RT) control was used to determine the gDNA contamination in the isolated total RNA. The expression levels were analyzed, in duplicate, with a MiniOpticon system (Bio-Rad) using the Bio-Rad CFX ™ manager software, with which the threshold cycle (ct) values were determined automatically by regression. The SensiMix SYBR Hi-ROX kit (Bioline, Australia) was used as a master mix for qPCR. The following thermocycler conditions were applied: 95°C for 10 min, followed by 40 cycles of 95°C for 15 s, 55°C

| qPCR analysis
for 30 s, and 72°C for 30 s. Subsequently, a melting curve was generated to determine the specificity of the qPCRs (Nijland et al., 2010;Weber, Polli, Boer, Bovenberg, & Driessen, 2012). The expression analysis was performed for two biological samples with two technical replicates. The analysis of the relative gene expression was performed through the 2 −ΔΔCT method (Livak & Schmittgen, 2001).

| Secondary metabolite analysis
The extraction of secondary metabolites from solid R-agar medium for HPLC and MS analysis was done by the modified microscale extraction procedure for standardized screening of fungal metabolite production in cultures (Smedsgaard, 1997). A plug of the agar medium (5 mm in diameter) was taken for extraction from the middle of the colony obtained after 10 days of growth. The extraction mixture (PDA) and it was performed as described previously .
Metabolite analysis was performed with two biological samples with two technical duplicates. Reserpine was used as internal standard.

| Scanning electron microscopy
Conidia were immobilized on glass cover slips and fixed with 2% glutaraldehyde for 1 hr followed by washing with cacodylate buffer (pH 7.4). Samples were incubated with 1% OsO 4 in 0.1 mol/L cacodylate buffer during 1 hr and washed with MQ water. The immobilized spores were dehydrated with a concentration gradient of 30, 50, and 70% of ethanol within 30 min followed by three steps of final dehydration with 96% ethanol within 45 min. Next, the samples were incubated in 100% ethanol/tetramethylsilane (TMS) 1:1(v/v) for 10 min followed by 15-min incubation with pure TMS and air-dried. Dried samples were coated with 2 nm Pd/Au using Leica EM SCD050 sputter coater and analyzed with SUPRA 55 FE-SEM (Carl Zeiss, Jena Germany) at 2 kV.

| Oxidative stress assay
Fungal conidia of 7-day grown mycelium were resuspended in 1 ml water containing 0.05% of Tween-20 to prevent aggregation. The equal amount of the spores (3 × 10 3 spores per ml) in solution were adjusted by series of dilutions and measured with Bürker-Türk counting chamber using Olympus CX20 ™ light microscope (Olympus, Hamburg, Germany). A conidial suspension (100 μl) was used for inoculation to obtain approximately 300 germination events per control plate in the assay. R-agar sporulation medium with increasing concentrations of hydrogen peroxide from 0.5 to 3.5 mmol/L was used in this assay. To prepare each plate, the corresponding amounts of hydrogen peroxide have been mixed with 25 ml of R-agar medium before solidification to provide equal distribution of the supplement in the plate. The experiment was performed twice using two sets of the hydrogen peroxide supplemented R-Agar plates as technical replicas.

| Other methods
To study the effect of sorbicillinoids on the secondary metabolism of P. chrysogenum, the feeding experiment has been performed as it was reported previously (Guzman-Chavez et al., 2017). The preculture of the strain DS368530 was grown on YGG medium for 24 hr and subsequently used (3 ml) to supplement 20 ml of fresh SMP medium. The filtered supernatant (2 ml) obtained from the growth medium of the strain DS68530Res13 grown in SPM for 3 days was collected as the source of sorbicillinoids used for the feeding experiment. The control culture was supplemented with the supernatant derived from the non-sorbicillinoids-producing strain DS68530.

| Deletion of the hdaA gene
0.02 0.03 and the correct inactivation of the hdaA gene was validated by sequencing the locus of the insertion ( Figure S1).

| Effect of the hadA deletion on the expression of secondary metabolite genes
To examine the effect of inactivation of hdaA on the transcription of secondary metabolite genes, the expression of all 20 PKS and 11 NRPS genes was examined using quantitative real-time PCR analysis.
RNA was isolated from the mycelium of the deletion and the parental strains grown on SMP medium for 3 days. The related supernatant fractions obtained after 3 and 5 days of culture growth were used for secondary metabolite profiling (see below). The qPCR analysis of the various secondary metabolite genes was performed using primers listed in Table S1. Out of the 31 analyzed secondary metabolite genes, the expression of eight genes was dramatically altered in ∆hdaA mutants from the different genetic backgrounds (sorbicillinoids producer and none producer strains).
An up to 500-fold increase in expression occurred for the PKS enzymes SorB (pks12; Pc21 g05070) and SorA (pks13; Pc21 g05080) in the sorbicillinoids-producing strain, while the transcript levels of the corresponding genes in the ∆hdaA mutant that is not able to produce sorbicillinoids, was only 12-fold higher. Interestingly, the deletion of

| Epigenetic activation of the sorbicillinoids biosynthesis gene cluster
The genes belonging to the sorbicillinoids BGC (Guzman-Chavez et al., 2017) were highly upregulated in the ΔhdaA strain (Figure 2a; ∆hdaA_DS68530Res13 strain). The sorC gene showed an increase in the expression levels of more than 100 times, while sorD and sorT were 2500-fold upregulated relative to the DS68530 strain.
Overexpression of this BGC resulted in the high-level production of sorbicillinoids in the supernatant fraction (Figure 2b).    (Figure 4a,b). Likewise, at day 5, also most of the chrysoginerelated compounds were produced at lower levels. Taken together, these results indicate that the chrysogine BGC is not only subjected to epigenetic activation, but also suppressed by the production of sorbicillinoids.
The DHN-melanin biosynthetic pathway was described initially for Verticillium dahliae and Wangiella dermatitidis (Bell, Puhalla, Tolmsoff, & Stipanovic, 1976;Geis, Wheeler, & Szaniszlo, 1984).  The pentaketide origin of fungal melanins is common in other melanized fungi (Langfelder, Streibel, Jahn, Haase, & Brakhage, 2003;Wheeler et al., 2008). In A. fumigatus, the polyketide product of the PKS Alb1p, the heptaketide naphthapyrone YWA1, requires the enzymatic post-PKS conversion into the pentaketide 1,3,6,8-tetrahydroxynaphthalene (T4HN) via hydrolytic polyketide shortening by Ayg1p (Fujii et al., 2004). This enzymatic step is absent in C. lagenarium where the pentaketide T4HN is a direct product of PKS1 (Fujii et al., 1999). Next, it is reduced to scytalone via the T4HN reductase Arp2p, followed by dehydration to 1,3,8-trihydroxynaphthalene (T3HN) by the scytalondehydratase Arp1p. The following reduction to vermelon is Arp2 dependent but the presence of other specific reductase(s) caring this reaction has been also proposed for Aspergilli and other fungi (Tsai, Wheeler, Chang, & Kwon-Chung, 1999;Wang & Breuil, 2002 (Hamilton & Gomez, 2002;Jacobson, 2000;Langfelder et al., 2003). Screening of the genome sequence of P. chrysogenum indicates the presence of the corresponding ortholog genes of the DHN-melanin BGC: abr1 (Pc21 g16380), arp1 (Pc21 g16420), arp2 (Pc21 g16430), ayg1 (P21 g16440), abr2 (P22 g08420) and associated pks17 (pcAlb1, Pc21 g16000) were found partially clustered in the genome. To examine the role of HdaA in the biosynthesis of DHN-melanin in conidial pigmentation, the ∆hdaA mutant and DS68530 strains (no sorbicillinoids producers) were grown on solid R-agar medium for 10 days, which resulted in a major decrease in the green conidial pigmentation in the ∆hdaA mutant as compared to the reference strain. qPCR analysis of the putative DHNmelanin BGC indicated the fourfold downregulation of pks17 in the ∆hdaA mutant, while arp1, arp2, and ayg1 were fourfold upregulated. Expression of abr1 and abr2 was not significantly changed ( Figure 5a,b). To determine if pks17 is involved in conidial pigment biosynthesis, the pks17 gene was deleted and overexpressed in order to identify the related polyketide product. A gene inactivation strain was obtained as described earlier (see materials and methods section) using primers listed in Table S1. The resulting ∆pks17 mutant displayed an albino phenotype of the conidia while grown on sporulating R-agar medium (Figure 6a). For the overex- strains. An up to 50-fold increase was observed in the Pks17 overproducing strain (Figure 4a). YWA1 could only be detected in the culture supernatant of the Pks17-overproducing strain, and was not found in parental strain.

| Role of pks17 in conidia formation and tolerance to oxidative stress
Melanins are important components of the conidial cell wall and its integrity. They play an essential role in physical properties of the spores like surface interaction, hydrophobicity, and virulence in pathogenic fungal species. The effect of hdaA deletion on the conidial surface in P. chrysogenum was examined using scanning electron microscopy.
The conidia of the reference DS68530, ∆hdaA_DS68530, ∆pks17, and overexpression mutant oepks17 were isolated from colonies grown for 7 days on the sporulating R-agar medium. The reference DS68530 exhibited a typical tuberous surface of the conidia, while Aside from the mechanical properties of the pigments (Howard, Ferrari, Roach, & Money, 1991) and their role as pH buffering systems, the scavenging of reactive oxygen species is an important feature supporting UV and thermo-tolerance and pathogenicity of the conidia (Jacobson, Hove, & Emery, 1995;Kawamura et al., 1997;Romero-Martinez, Wheeler, Guerrero-Plata, Rico, & Torres-Guerrero, 2000).The P.chrysogenum ∆hdaA_DS68530 mutant was grown on hydrogen peroxide supplemented medium for 5 days to verify the ability of the conidia to survive oxidative stress conditions in the absence of pigmentation. The survival rate was decreased by 20% when the ∆hdaA strain was exposed to 2 mmol/L of hydrogen peroxide in the media. Under the same conditions, the survival rate of the ∆pks17 (∆alb1) was reduced more than 50%. There was no enhanced survival observed for the oepks17 overexpression mutant ( Figure 6c). 2007; Tribus et al., 2005). Here, we have examined the effect of chromatin modification on the expression of the secondary metabolismassociated genes and products in the fungus P. chrysogenum. In this work, the P. chrysogenum hdaA gene encoding an ortholog of the class 2 histone deacetylase Hda1 of S. cerevisiae was deleted. The deletion mutant showed significant changes of secondary metabolite gene expression including PKS and NRPS genes with known and unknown function (Figure 1). In ΔhdaA mutants, the transcriptional levels of the sorbicillinoids BGC were significantly increased. SorA and sorB genes, which encodes for the two polyketide synthases (highly reducing and nonreducing, respectively) involved in the sorbicillinoids pathway  were overexpressed in the ΔhdaA_DS38530 strain.

| DISCUSSION
Interestingly, overexpression of the sorbicillinoids BGC was also observed in the DS38530Res13 strain, in which hdaA was not deleted.
This has been attributed to a complex regulation mechanism that involves sorbicillinoids as auto inducers (Guzman-Chavez et al., 2017).
This phenomenon could potentially also involve HdaA altering the chromatin landscape (Brosch et al., 2008), since the deletion of the . Likely, the chrysogine BGS is subjected to eplgenetic activation by HdaA, but at the same time sorbicillinoids production reduces the expression of this gene cluster.
In A. nidulans and A. fumigatus, the homologous hdaA is a main contributor of histone deacetylase activity in these fungi. In A. nidulans, deletion of the hdaA gene stimulated penicillin and sterigmatocystin production but not a telomere-distal gene cluster involved in terraquinone A biosynthesis. It was suggested, that HdaA silences the expression of subtelomeric chromosomal regions (Tribus et al., 2005).
In contrast, the HdaA homolog in A. fumigatus was reported to activate gliotoxin biosynthesis and to repress several NRPSs including one gene of the siderophore BGC. A subtelomeric specificity of HdaA was not apparent for this fungus (Lee et al., 2009 Figure S2). The action of HdaA thus seems mostly to be restricted to the transcriptional coregulation of a particular genomic area rich in BGCs (Van Den Berg et al., 2008).
The production of sorbicillinoids and the deletion of the hdaA gene, causing increased sorbicillinoids production, had similar effects on the expression of other PKS and NRPS genes (Figures 1,2,4) including the BGC that specifies chrysogine. One possible explanation is that sorbicillinoids might act as HdaA inhibitors, since hdaA was transcribed at the same levels in the sorbicillinoids producer and nonproducer strains (data not shown), while feed sorbicillinoids did not alter the transcription of hdaA in the DS68530 strain. Moreover, a novel compound was detected only when hdaA was deleted or when DS68530 was fed with sorbicillinoids ( Figure 3). A similar phenomena occurs when Cladosporium cladosporioides and A. niger are exposed to suberoylanilide hydroxamic acid (SAHA), a HDAC inhibitor, which induces the synthesis of two new compounds, cladochrome and nygerone A, respectively (Rutledge & Challis, 2015). Our work represents the first example of a regulatory crosstalk between BGCs in P. chrysogenum. In A. nidulans, overexpression of a regulator (scpR) was found to activate the expression of two cryptic NRPS genes, belonging to the same cluster, as well as the induction of genes responsible for the production of the polyketide asperfuranone (Bergmann et al., 2010;Brakhage, 2012).
The deletion of the hdaA gene also has a functional effect in P. To identify the polyketide product, pks17 was overexpressed causing the accumulation of the yellow naphthoγ-pyrone (a polyketide precursor of the conidial pigment) into the medium. This indicates that P. chrysogenum uses the DHN-melanin biosynthetic pathway like previously reported for Aspergillus (Hamilton & Gomez, 2002;Jacobson, 2000;Langfelder et al., 2003). The corresponding ∆pks17 strain showed an albino phenotype confirming the primary role of this gene in the conidial pigmentation. In the closely related fungus A. fumigatus at least six genes are required for DHN-melanin biosynthesis, which were found to be partially clustered in the genome of P. chrysogenum. qPCR analysis ( Figure 5) showed the fourfold upregulation of the arp1 (scytalondehydratase), arp2 (T4HN reductase), and ayg1 (enzyme of hydrolytic polyketide chain shortening activity) genes, while the transcript level of abr1 (multicopper oxidase) and abr2 (laccase) which products catalyze the last steps of the polymerization of 1,8-dihydroxynaphthalene were not significantly changed.
Scavenging of reactive oxygen species by fungal melanins provides an important defense mechanism during growth under oxidative stress condition. We examined the effect of hdaA deletion on the ability of the conidia to survive high concentrations of hydrogen peroxide. An increased sensitivity of the HdaA was noted toward hydrogen peroxide, while this effect was even more pronounced for the ∆pks17 albino mutant (Figure 6c). These results suggest that the oxidative stress response in P. chrysogenum involves HdaA and is mediated by the transcriptional regulation of DHN-melanin gene cluster.
In conclusion, our results demonstrate that HdaA has a broad impact on secondary metabolism of P. chrysogenum at the transcriptional level causing marked changes in metabolite production.
Furthermore, HdaA influences conidial pigmentation and the surface structure of spores. This work provides evidence of crosstalk between gene clusters, which impacts secondary metabolism. The presented data suggest that an epigenome approach can be successfully applied for the discovery of novel biosynthetic pathways in P. chrysogenum.