Identification of a Polyketide Synthase Gene Responsible for Ascochitine Biosynthesis in Ascochyta fabae and Its Abrogation in Sister Taxa

Fungi produce a diverse array of secondary metabolites, many of which are of pharmacological importance whereas many others are noted for mycotoxins, such as aflatoxin and citrinin, that can threaten human and animal health. The polyketide-derived compound ascochitine, which is structurally similar to citrinin mycotoxin, has been considered to be important for pathogenicity of legume-associated Ascochyta species. Here, we identified the ascochitine polyketide synthase (PKS) gene in Ascochyta fabae and its neighboring genes that may be involved in ascochitine biosynthesis. Interestingly, the ascochitine PKS genes in other legume-associated Ascochyta species have been mutated, encoding truncated PKSs. This indicated that point mutations may have contributed to genetic diversity for secondary metabolite production in these fungi. We also demonstrated that ascochitine is not a pathogenicity factor in A. fabae. The antifungal activities and production of ascochitine during sporulation suggested that it may play a role in competition with other saprobic fungi in nature.

IMPORTANCE Fungi produce a diverse array of secondary metabolites, many of which are of pharmacological importance whereas many others are noted for mycotoxins, such as aflatoxin and citrinin, that can threaten human and animal health. The polyketide-derived compound ascochitine, which is structurally similar to citrinin mycotoxin, has been considered to be important for pathogenicity of legumeassociated Ascochyta species. Here, we identified the ascochitine polyketide synthase (PKS) gene in Ascochyta fabae and its neighboring genes that may be involved in ascochitine biosynthesis. Interestingly, the ascochitine PKS genes in other legumeassociated Ascochyta species have been mutated, encoding truncated PKSs. This indicated that point mutations may have contributed to genetic diversity for secondary metabolite production in these fungi. We also demonstrated that ascochitine is not a pathogenicity factor in A. fabae. The antifungal activities and production of ascochitine during sporulation suggested that it may play a role in competition with other saprobic fungi in nature. pathogenicity and competition with other microbes, primarily due to lack of information about the genetic control of biosynthesis. Thus, the objectives of this study were (i) to identify genes responsible for ascochitine biosynthesis and the associated gene cluster in A. fabae, (ii) to examine the roles of ascochitine in the fungal biology through targeted gene knockout study, and (iii) to investigate genetic diversity in ascochitine biosynthetic genes among legume-associated Ascochyta species.

RESULTS
Homology-based search for putative ascochitine PKS gene. To search for a PKS gene responsible for ascochitine biosynthesis, we identified a total of nine putative type I PKS genes from the genome sequence of A. fabae isolate AF1, using the antiSMASH program (28). Then, the KS domain of the citrinin PKS gene (pksCT) in Monascus purpureus was queried against the putative PKS genes found in A. fabae. We found a PKS gene (here referred to as pksAC) similar to the pksCT gene, showing 67% identity in the KS domains at the amino acid sequence level. The PKS domain architectures of the pksAC and pksCT genes were identical (SAT-KS-AT-PT-ACP-MeT-R). To predict a PKS function in an evolutionary framework, we reconstructed the genealogy of fungal PKSs, using the deduced amino acid sequences of the KS domains of the pksAC gene and 51 PKS genes that have been definitively linked to the biosynthesis of specific compounds (see Text S1 in the supplemental material). It seems reasonable to believe that the resulting polyketides produced by phylogenetic orthologs are likely more similar to each other in their chemical structures. As expected, the pksAC gene was placed next to the pksCT genes in the phylogenetic tree (Fig. 1A), providing additional evidence of the link between the pksAC gene and ascochitine biosynthesis. The polyketide products within the subclade that includes the pksAC and pksCT genes share a common characteristic, the heavily oxygenated heterocyclic core, which is observed most often in pyranoquinones, such as ascochitine, azanigerones, citrinin, and mitorubrin (Fig. 1B).
Characterization of pksAC in A. fabae. To validate the involvement of the pksAC gene in ascochitine production in A. fabae isolate AF1 (wild type [WT]), we conducted targeted gene replacement of the pksAC gene with a hygromycin phosphotransferase gene cassette. Successful replacement in putative transformants was verified with PCR using a primer pair that annealed to flanking regions of the introduced cassette ( Fig. 2A). The resulting knockout mutants (ΔpksAC) were similar to the WT in growth rate ( Fig. 2B; top panel). In the aged culture, the colony of the ΔpksAC gene became highly melanized (Fig. 2B), which was likely caused by the excess of acetates that otherwise should have been used in ascochitine biosynthesis. Similar phenomena have been observed in solanapyrone-negative mutants in Ascochyta rabiei (29) and in a mutant defective in a red pigment production in Fusarium neocosmosporiellum (30).
The ΔpksAC gene lost the ability to produce ascochitine and its derivatives, as evidenced by the chemical profiles of the culture extracts (Fig. 2C). Liquid chromatography-mass spectrometry (LC-MS) analysis of a culture extract of the WT confirmed the production of ascochitine (eluted at 4.28 min) and ascochital (another end product from the ascochitine biosynthetic pathway; eluted at 3.32 min) (9) (Fig. 2C; top panels). In the ΔpksAC gene, there was no detectable production of ascochitine and ascochital, indicating that the pksAC gene is required for ascochitine biosynthesis ( Fig. 2C; bottom panels).
Identification of ascochitine biosynthesis gene cluster. The identity and function of each gene in the citrinin biosynthesis gene cluster in Monascus ruber were recently reported (21). On the basis of the structural similarity between citrinin and ascochitine ( Fig. 1B), we hypothesized that the contents of the citrinin and ascochitine gene clusters are conserved. Since the genome assembly of A. fabae isolate AF1 around the pksAC locus was not complete, we used the genome sequences of another A. fabae isolate (isolate AF247/15) and an A. viciae-villosae isolate (isolate AV22) that were retrieved from the NCBI database to search for ascochitine biosynthesis gene cluster. There was one nucleotide difference in the pksAC sequence in the AF247/15 isolate (G to A at position 5437) that caused a nonsynonymous substitution (valine to isoleucine) relative to the pksAC sequences in the AF1 and AF55/01 isolates. By comparing genes neighboring the pksAC gene to the gene members of the citrinin gene cluster, we identified a minimal set of conserved genes (orf1 to orf11; Table 1) that are likely involved in ascochitine biosynthesis in the genomes of the AF247/15 and AV22 isolates. The gene content and order of the ascochitine biosynthesis gene cluster in the two ascochitine producers were identical, and genes further downstream of orf1 or upstream of orf11 did not have significant sequence similarities with the members of the citrinin gene cluster (not shown).
The orf1, orf3, and orf9 genes were homologous to the citrinin biosynthesis genes encoding NAD(P) ϩ -dependent aldehyde dehydrogenase (mrl4), nonheme Fe(II)dependent oxygenase (mrl2), and NAD(P) ϩ -dependent oxidoreductase (mrl7), respectively. These enzymes were involved in subsequent oxidations of methyl groups to the carboxylic acid of the heterocyclic ring of citrinin (21). The mrl6 gene, encoding an enzyme involved in the generation of carbonyl groups in the heterocyclic ring of citrinin, was missing in the putative ascochitine gene cluster in both A. fabae and A. viciae-villosae. An additional PKS gene (orf7) was found adjacent to the pksAC gene (Fig. 3A). This PKS gene is classified as a highly reducing PKS gene. As such, it may be involved in the addition of the sec-butyl group in ascochitine via mechanisms analogous to the biosynthesis of asperfuranone and azanigerones in Aspergillus spp. (31). The gene cluster also includes a transporter gene (orf6) and a gene encoding a putative transcription factor (orf11). Overall, despite the observed structural similarity in ascochitine and citrinin, the responsible gene clusters appeared to have diverged through gene order rearrangement and through gene gains and losses (Fig. 3A).
Sequence diversity in pksAC orthologs. Genetic diversity in secondary metabolite gene clusters can be introduced by different mechanisms in fungi (32). We investigated how ascochitine gene clusters have diverged during the evolution of legumeassociated Ascochyta species. First, the presence of pksAC orthologous genes was checked in A. lentis and A. lentis var. lathyri, which are closely related to ascochitine producers such as A. fabae and A. viciae-villosae (33). We surveyed genome sequences of three A. lentis isolates, namely, isolates AL1, AL4, and Kewell, and one A. lentis var. lathyri isolate, namely, isolate ER1415, and found orthologous pksAC genes largely as complete copies in the four isolates. However, many single nucleotide polymorphisms (SNPs) were observed in the pksAC orthologs that included nonsense mutations introducing a premature stop codon that resulted in truncated PKS gene products (Fig. 3B). Interestingly, the nucleotide positions of the first nonsense mutation differed in A. lentis and A. lentis var. lathyri: G to T at position 828 in A. lentis and C to T at position 4116 in A. lentis var. lathyri ( Table 2). The A. lentis genome sequences were all from Australian isolates, so we examined additional isolates of different geographical origins, including Asia, Europe, and South America ( Table 2). Since A. lentis var. lathyri is an species endemic to Italy, we checked one more isolate collected in a different year ( Table 2). The SNPs that caused the nonsense mutation were confirmed in all the tested isolates by Sanger sequencing.
To investigate possible selection acting on ascochitine biosynthesis genes, we estimated synonymous and nonsynonymous substitution rates between pksAC orthologs. Although the degrees of sequence divergence between pksAC orthologs were similar at both the DNA and deduced amino acid sequence levels, the ratio of synonymous to nonsynonymous substitution rates between A. fabae and A. viciaevillosae (ascochitine producers) was slightly higher than between A. fabae and the two ascochitine-negative species (Table 3), which could be ascribed to some degree of purifying selection acting on the pksAC genes in the ascochitine producers.
Possible roles of ascochitine. Ascochitine has been postulated to represent a pathogenicity factor (4). Thus, the role of ascochitine in the A. fabae and faba bean pathosystem was evaluated by comparing the WT and the ΔpksAC mutant with respect to their disease-causing abilities. Faba bean strains with accession numbers PI 358266  (moderately resistant) and PI 667234 (susceptible) served as hosts. The ΔpksAC mutants maintained the ability to cause disease on faba bean, and there was no significant difference between the WT and the ΔpksAC mutant with respect to disease incidence or severity (one-way analysis of variance [ANOVA] P Ͼ 0.05; Fig. 4A). These results suggested that ascochitine production is not essential for pathogenicity in A. fabae on faba bean. matrix-assisted laser desorption ionization (MALDI) imaging mass spectrometry enabled the detection of ascochitine on the surface of the fungal colony, and ascochitine accumulated in aged hyphae where pycnidia (the asexual fruiting bodies) formed (Fig. 4B), suggesting that ascochitine was associated with sporulation and might play a protective role against competing microbes in nature.

DISCUSSION
With the advent of high-throughput sequencing technologies, fungal genomes are becoming available in exponentially increasing numbers (34). Nevertheless, relatively few secondary metabolite pathways have been linked to their respective biosynthetic gene clusters. Although classes of compounds can be made through completely distinct biochemical pathways, phylogenetic orthologs that encode enzymes with identical or nearly identical biochemical functions are expected to produce similar compounds. Therefore, homology-based prediction within an evolutionary framework can facilitate linking of genes to molecules, and, ultimately, the genetic information can be used for developing heterologous systems for expression of compounds of pharmaceutical and industrial interest in the postgenomic era (31,35).

Identification of Ascochitine Gene Cluster
To date, ascochitine has been found in many species in legume-associated Ascochyta and other genera in the Dothideomycetes, such as Calophoma, Phoma, Pleospora, and Stagonosporopsis (4,(8)(9)(10). Recent efforts to resolve the structure of complex Didymellaceae, including Ascochyta and Phoma taxa in the Dothideomycetes, reclassified P. clematidina into Calophoma, which is a sister genus to the Ascochyta, and A. salicorniae into Stagonosporopsis (11,13,14). Ascochyta hyalospora was also renamed Pleospora chenopodii, which belongs to the Pleosporaceae (12). These widespread distributions of ascochitine production in the Dothideomycetes suggested that the trait of producing ascochitine is ancestral to legume-associated Ascochyta species. Despite the phylogenetic distance between ascochitine-and citrinin-producing fungi, the ascochitine and citrinin biosynthetic genes exhibited high degrees of sequence similarities (up to 62%), suggesting a common origin of ascochitine and citrinin gene clusters. The two gene clusters showed gene content polymorphisms and appear to have experienced multiple gene translocation and inversion events during divergence from the gene cluster that shared the most recent common ancestry. Interestingly, the ascochitine gene cluster included a HR-PKS (orf7) with an unusual PKS domain architecture (KS-DH-ER-KR-ACP), missing AT and MeT domains found in many HR-PKSs, such as lovastatin diketide synthase lovF (KS-AT-DH-MeT-ER-KR-ACP) (36). Most PKS require an AT domain to transfer acyl groups to the KS and ACP components. It is currently unknown whether the orf7 without an AT domain is truly functional. Since the orf7 ortholog was also found in A. viciae-villosae, it is less likely that the unusual PKS architecture of the orf7 was due to genome assembly or annotation errors. Thus, the precise role of the HR-PKS in ascochitine biosynthesis needs to be determined.
It was previously hypothesized that a reduced diketide is used as a starter unit for citrinin biosynthesis by the pksCT gene (37). Recently, however, He and Cox demonstrated that acetate is used as a starter unit for subsequent 4-fold extension by the pksCT gene, yielding an unreduced trimethylated pentaketide without the need of another PKS (21). The only structural difference between ascochitine and citrinin is that ascochitine has a sec-butyl group in the heterocyclic ring. As an alternative to the previously proposed biosynthetic scheme (9), the finding of the orf7 in the ascochitine gene cluster in A. fabae suggested that the HR-PKS makes a diketide starter unit which is passed to the pksAC gene for further extension, producing ascochital and ascochitine via a mechanism analogous to that of asperfuranone biosynthesis (38). Notably, in the NR-PKS subclade that includes ascochitine (Fig. 1B), asperfuranone, azanigerone A, and bisvertinolone (a sorbicillinoid) are synthesized by the combination of an NR-PKS (AfoE, AzaA, and SorB, respectively) and a HR-PKS (AfoG, AzaB, and SorA, respectively) (31,38,39).
The ascochitine gene cluster also included a gene encoding a short peptide with a cupin domain (orf2) that is often found in secondary metabolite gene clusters (30,40). The orf11 gene encodes a fungus-specific Zn(II)2Cys6-type transcription factor and exhibited the least sequence similarity with the counterpart gene (mrl3) in the citrinin gene cluster. Unlike the mrl3 gene, the deduced amino acid sequence of the orf11 gene lacked a DNA-binding domain and contained only the conserved domain (termed "middle homology region") that is commonly found in Zn(II)2Cys6-type transcription factors (41). Interestingly, the sol4 transcription factor with only the middle homology region was shown to regulate the entire solanapyrone gene cluster in A. rabiei (29). This type of transcription factor without an apparent DNA-binding domain may activate genes by interacting with cognate repressor proteins, enabling timely induction of the repressed gene clusters.
When A. lentis and A. lentis var. lathyri are sexually crossed in vitro, they produce viable progeny, and segregation of molecular markers is normal, indicating that no intrinsic mating barrier exists between these taxa (42). However, pathogenic specialization of these two recently diverged species is evident (42). Although molecular phylogenies of protein-coding genes cannot clearly differentiate A. lentis from A. lentis var. lathyri (99.6% to 100% similar) (42), the two SNPs that caused a nonsense mutation in each taxon were consistent within each taxon, suggesting a prezygotic mating barrier and a lack of gene flow between the two probably sympatric species. Despite the presence of the nonsense mutation, transcripts of the pksAC gene in A. lentis isolate AL4 were detected on the basis of transcriptome sequencing (RNA-seq) data (see Fig. S1 in the supplemental material), suggesting that the gene is still active and likely undergoing pseudogenization. Compounds other than ascochitine were identified from cultures of A. lentis and A. lentis var. lathyri (15,16,43). Among the diverse mechanisms that produce genetic diversity in fungal secondary metabolism, nonsense mutation would be the fastest and the most effective, as the whole biosynthetic pathway of a compound can be disrupted by a single point mutation and indels in the responsible gene cluster (32,(44)(45)(46)(47)(48).
Production of ascochitine by P. clematidina was evident during the infection process, and the amount of ascochitine production in axenic culture was positively correlated with the ability to cause the disease among P. clematidina isolates (4). In contrast, there was no such correlation between the aggressiveness of activity on faba bean cultivars and the amount of ascochitine production among A. fabae isolates (49). In this study, we demonstrated that ascochitine was not involved in the pathogenicity of A. fabae. Solanapyrone, another phytotoxic compound produced by A. rabiei, was also proven not to be a pathogenicity factor (29,50). Therefore, caution should be taken in claiming the involvement of secondary metabolites in pathogenicity based on in vitro tests of phytotoxicity.
It is always a challenging task to determine the role of secondary metabolites in an interaction with hosts or in the environment (51). However, the role of secondary metabolites may be deduced from their biological activity and spatiotemporal production patterns during the life cycle of the producing fungi. As with solanapyrone, ascochitine was produced during sporulation, while another polyketide, pinolidoxin, produced by A. pinodes, was detected mainly in young, advanced hyphae (52). Together with the reported antifungal activities, these observations suggest that ascochitine may play a role in competition with other fungi during saprobic growth stages in nature (2,52). Genes involved in secondary metabolite production are generally believed to confer a fitness advantage to the producing organism (51,53). Otherwise, the highly plastic fungal genomes would have purged out costly genes encoding megasize proteins such as PKS by selective pressure over the course of evolution. To end on a speculative note, some lineages of legume-associated Ascochyta may have lost the ability to produce ascochitine, a compound to which some saprobic fungi are insensitive (2). Studying the precise role of ascochitine in fungal ecology will provide an opportunity for tracing the fate of secondary metabolite biosynthesis gene clusters in species divergence.

MATERIALS AND METHODS
KS domain phylogeny. The predicted KS domains of 51 PKS genes with known polyketide products plus a putative ascochitine PKS gene (pksAC) were aligned using the MUSCLE program (v3.8.31) (54) (see Text S1 in the supplemental material). Subsequently, spurious sequences or poorly aligned regions from the 52 KS domains were automatically removed, using the trimAl program (v1.2) (55) with the following option argument: "-gappyout." A maximum likelihood phylogeny based on a general empirical model (WAG) (56) was constructed using the Phangorn R package (v2.4.0) (57). Nodal support was evaluated by 1,000 bootstrap replications using nearest-neighbor interchange rearrangements.
Gene replacement. The pksAC gene in A. fabae isolate AF1 (WT) was deleted, using the split marker method (58). The DNA fragments used to construct the split marker were amplified using double-joint PCR (59). In the first round of PCR, a 756-bp upstream region of the pksAC coding sequence was amplified using primer pair L5 and L3, and a 947-bp downstream region of the pksAC coding sequence was amplified using primer pair R5 and R3. A 1,372-bp hph cassette was amplified from pDWJ5 plasmid (60) using primer pair HYG-F and HYG-R. The L3 and R5 primers carried 27-bp sequence tails that overlapped the 5= and 3= ends of the hph cassette, respectively. In the second round of PCR, the flanking DNA fragments were fused to the hph cassette through PCR by overlap extension. In the third round of PCR, the resulting PCR product was used as a template to generate the split-marker constructs that were amplified using primer pair N5 and HY-R and primer pair N3 and YG-F for the upstream and downstream flanking DNA fragments, respectively. Preparation of fungal protoplasts and transformation were conducted as previously described (61) with a minor modification. Volumes of 1 to 2 g of each upstream and downstream split-marker DNA construct were used for genetic transformation. Putative transformants were subcultured on peptone-dextrose agar (PDA) containing hygromycin B (200 g/ml). The gene replacement was confirmed by PCR with primer pair L5 and R3. Primers used in this study are listed in Table S1 in the supplemental material.
LC-MS analyses. The WT and ΔpksAC mutant were grown on PDA for 14 days. Whole-agar chunks with fungal colonies were excised, soaked in 10 ml of ethyl acetate (EtOAc) in 50-ml Falcon tubes, and incubated at 4°C overnight. An aliquot of 200 l of the extracts was transferred to a microcentrifuge tube, evaporated to dryness, reconstituted with 30 l of MeOH, and subjected to LC-MS analyses. Chromatographic separation was achieved using an Acquity ultraperformance liquid chromatography (UPLC) system (Waters Corp., Milford, MA, USA), as previously described (50). MS analysis was performed on an inline Synapt G2-S high-definition mass spectrometry (HDMS; Waters Corp.) time of flight mass spectrometer using the positive-ion electrospray mode for data collection. The UV data were acquired using a wavelength of 210 to 400 nm, resolution of 1.2 nm, and 20 points/s with an Acquity photodiode array detector (Waters Corp.).
Synonymous-nonsynonymous analysis. SNAP (v2.1.1) (62) was used to estimate synonymous and nonsynonymous substitution rates and to calculate the ratio of synonymous to nonsynonymous substitutions between pksAC homologs in A. fabae AF247/15, A. viciae-villosae AV22, A. lentis AL4, and A. lentis var. lathyri ER1415 isolates. Since the genome assembly in the pksAC gene locus in A. lentis var. lathyri ER1415 was incomplete (with two N stretches [176 bp and 111 bp] of gap sequence in the coding region), the corresponding 96 codons from the other pksAC genes were excluded in this analysis.
Pathogenicity tests. Two faba bean strains (accession no. P1 667234 and PI 358266) were planted in Deepots (Stuewe & Sons, Inc., Corvallis, OR, USA) (6 by 25 cm) in a growth chamber. The inoculum was prepared by harvesting conidia (asexual spores) from 2-week-old fungal colonies on V8 agar media and was adjusted to 2 ϫ 10 5 conidia/ml. Two-week-old plants were sprayed with inoculum until runoff (approximate 2 ml per plant) and were immediately covered with an inverted translucent plastic cup to form a minidome to produce uniformly high levels of relative humidity for 24 h to facilitate infection. Plants were then placed in a growth chamber (Conviron model PGR 15, Winnipeg, MB, Canada) with settings of 12 h day (20°C) and 12 h night (16°C) at 100% relative humidity. Control plants were sprayed with water but were otherwise treated in the same manner as the inoculated plants. At 2 weeks after inoculation, disease scores were determined and numbers of lesions were measured, as previously described (63).
MALDI imaging analysis. Ascochyta fabae isolate AF1 and A. viciae-villosae isolate AV1 (33) was grown on half-strength PDA in which a microscopic slide was embedded. After 2 weeks of growth, the embedded slide with the fungal colony growing on the thin layer of PDA (about 2 mm in diameter) was removed from the plate and dried for 1 h at 65°C for MALDI imaging analysis. The dried specimen on the slide was sprayed with 20 mg/ml of 2,5-dihydroxybenzoic acid (in a mixture that also contained 0.1% trifluoroacetic acid and 50% methanol) as a matrix, using a TM-sprayer system (HTX Technologies, Carrboro, NC, USA) with the following parameters: 12 passes; track spacing, 1.2 mm; nozzle temperature, 70°C; linear velocity, 110 cm/min; flow rate, 60 l/min. A selected area of the specimen was analyzed on a 9.4T Fourier transform ion cyclotron resonance (FTICR) MS instrument (Bruker Daltonics, Billerica, MA, USA) in positive-ion mode. Data were acquired using ftmsControl software (Bruker Daltonics). Approximately 100 shots per spot were acquired with a Smartbeam II Nd:YAG laser using repetition rate of 1 kHz. Image acquisition was carried out using flexImaging 4.1 (Bruker Daltonics), and data were normalized to the total ion current. Ascochitine standard compound purified in our previous study (8) was spotted, and reference MS signals were obtained in a simultaneous run.