Gibepyrone Biosynthesis in the Rice Pathogen Fusarium fujikuroi Is Facilitated by a Small Polyketide Synthase Gene Cluster*

The 2H-pyran-2-one gibepyrone A and its oxidized derivatives gibepyrones B–F have been isolated from the rice pathogenic fungus Fusarium fujikuroi already more than 20 years ago. However, these products have not been linked to the respective biosynthetic genes, and therefore, their biosynthesis has not yet been analyzed on a molecular level. Feeding experiments with isotopically labeled precursors clearly supported a polyketide origin for the formal monoterpenoid gibepyrone A, whereas the terpenoid pathway could be excluded. Targeted gene deletion verified that the F. fujikuroi polyketide synthase PKS13, designated Gpy1, is responsible for gibepyrone A biosynthesis. Next to Gpy1, the ATP-binding cassette transporter Gpy2 is encoded by the gibepyrone gene cluster. Gpy2 was shown to have only a minor impact on the actual efflux of gibepyrone A out of the cell. Instead, we obtained evidence that Gpy2 is involved in gene regulation as it represses GPY1 gene expression. Thus, GPY1 was up-regulated and gibepyrone A production was enhanced both extra- and intracellularly in Δgpy2 mutants. Furthermore, expression of GPY genes is strictly repressed by members of the fungus-specific velvet complex, Vel1, Vel2, and Lae1, whereas Sge1, a major regulator of secondary metabolism in F. fujikuroi, affects gibepyrone biosynthesis in a positive manner. The gibepyrone A derivatives gibepyrones B and D were shown to be produced by cluster-independent P450 monooxygenases, probably to protect the fungus from the toxic product. In contrast, the formation of gibepyrones E and F from gibepyrone A is a spontaneous process and independent of enzymatic activity.

Metabolites from F. fujikuroi with a 2H-pyran-2-one structure, gibepyrone A and its formal oxidation products gibepyrones B-F (Fig. 1D), were first reported by Barrero et al. (31). Gibepyrone A was also recently detected in F. fujikuroi headspace extracts (32,33). Gibepyrones A and B exhibit a moderate antimicrobial activity against Gram-positive bacteria (Bacillus subtilis and Staphylococcus aureus) and yeasts (Saccharomyces cerevisiae and Candida albicans) at a minimal inhibitory concentration of 100 -200 g/ml (31). Besides F. fujikuroi, gibepyrone A production was also reported for natural isolates of Fusarium keratoplasticum and Fusarium petroliphilum, two members of the Fusarium solani species complex (34). Gibepyrone D was found in cultures of endophytic isolates of Fusarium oxysporum and Fusarium proliferatum (35,36), whereas gibepyrone F was detected in co-cultures of Fusarium sp./Aspergillus clavatus (37). Moreover, gibepyrone F was isolated from extracts of the South China Sea sponge Jaspis stellifera (38) which possibly originates also from a fungal symbiont, as described for other marine sponge-derived compounds (39,40).
The (Z)-stereoisomer of gibepyrone A, fusalanipyrone (Fig.  1E), has been reported from F. solani and has antifungal and phytotoxic activities (41,42). Based on its chemical structure, this C 10 compound was first suggested to be of monoterpenoid origin, but subsequent feeding experiments with deuterated methionine provided evidence that fusalanipyrone could also represent a methylated polyketide product (41,43). The phytotoxic methoxy derivatives nectriapyrone and pestalopyrone (Fig. 1F) were first isolated from Gyrostroma missouriensis and Pestalotiopsis oenotherae, respectively (44 -47). Due to incor-poration of radioactive labeling from [2-14 C]mevalonic acid, nectriapyrone was suggested to be a monoterpenoid, but this result is doubtful, because the "incorporation" of labeling into the widespread plasticizer bis- (2-ethylhexyl)phthalate, mistakenly identified as a natural product, was also reported in the same study (44). In an independent analysis, the incorporation of labeling from [1,2-13 C 2 ]acetate and [methyl- 13 C]methionine was described, providing evidence for a polyketide origin of this compound (48).
Although gibepyrones A-F are well known SMs of F. fujikuroi IMI58289 (31), the genes and enzymes involved in their biosynthesis have not been identified so far. Here we present feeding experiments with isotopically labeled precursors that validated the polyketide nature of the gibepyrones as well as the identification and molecular characterization of their small PKS gene cluster. Gibepyrone A biosynthesis is regulated by members of the fungus-specific velvet complex (Vel1, Vel2, and Lae1) and by Sge1, a major regulator of secondary metabolism in F. fujikuroi. Finally, the mechanisms for the formation of the oxidized gibepyrones B-F were elucidated, using a combination of chemical synthesis and gene knock-out experiments.

Results
Chemical Synthesis of Gibepyrone A and Its Production by F. fujikuroi-Analysis of liquid cultures of F. fujikuroi IMI58289 by HPLC-high resolution mass spectrometry (HRMS) showed the presence of an SM with the exact mass of gibepyrone A. The same compound was also detected via GS-MS in headspace extracts from F. fujikuroi agar plate cultures as reported previously (32,33). For unambiguous identification of gibepyrone A, the compound was synthesized via a known procedure by treating tigloyl chloride (1) with triethyl amine in dichloromethane (31). Its isomer 6-(but-3-en-2-yl)-3-methyl-2H-pyran-2-one (2) was also obtained in this reaction (Fig. 2). The synthetic gibepyrone A was identical to the detected SM in F. fujikuroi with respect to the HPLC-MS and GC-MS retention times (9 and 28 min, respectively) and mass spectra (supplemental Fig.  S1).

Feeding Experiments on the Biosynthesis of Gibepyrone A-
To exclude a terpenoid origin for gibepyrone A, a feeding experiment with [methyl-2 H 3 ]mevalonolactone was performed. No incorporation of labeling into gibepyrone A was detected (supplemental Fig. S2A), which contradicts a terpene biosynthetic pathway. Alternative feeding experiments were carried out to test for a polyketide origin of gibepyrone A. The formation of the carbon backbone from four acetate-derived C 2 building blocks and two S-adenosyl-L-methionine (SAM)-derived C 1 units seemed to be most likely, but the potential formation from two propionate-derived C 3 building blocks and two acetate-derived C 2 building blocks, although atypical for fungi, was addressed as well. Additionally, the possibility of a biosynthetic pathway via an isoleucine-derived (E)-2-methylbut-2-enoyl-CoA starter unit was evaluated. Feeding of [methyl-13 C]methionine resulted in the incorporation of up to two C 1 units with high rates (50%), as indicated by a 2-mass unit shift of the molecular ion of gibepyrone A, whereas no incorporation of labeling from [ 2 H 5 ]propionate was observed (supplemental Fig. S2A). Up to four deuterium atoms were found in gibepyrone A in the feeding experiment with [ Fig. S2B), but not with a hypothetical isoleucine-derived (E)-2-methylbut-2-enoyl-CoA starter unit that would require incorporation of seven deuterium atoms. Indeed, feeding of [2-13 C]acetate resulted in an incorporation of up to four units with a low incorporation rate (supplemental Fig. S2A). Taken together, these results confirmed that gibepyrone A is a tetraketide composed of four acetate building blocks and two SAM-derived methyl groups.

PKS13 Is Responsible for Gibepyrone A Biosynthesis-After
demonstrating the polyketide origin of gibepyrone A by the feeding experiments described above, we confirmed the PKS origin also on a genetic and enzymatic level. Therefore, the F. fujikuroi ⌬ppt1 deletion mutant as well as the complemented strain PPT1 C were analyzed for gibepyrone A production. The phosphopantetheinyl transferase encoded by PPT1 is essential for the activation of the main proportion of fungal PKSs (and NRPSs) through post-translational attachment of the phosphopantetheinyl prosthetic group (49). Analysis of culture fluids by HPLC-MS/MS revealed the production of gibepyrone A by the wild-type (WT) strain and PPT1 C but not by the ⌬ppt1 mutant ( Fig. 3A), confirming the polyketide origin.
In the framework of an ongoing project to identify novel SMs produced by F. fujikuroi, deletion strains for all 18 PKS-encoding genes have been generated. Targeted replacement of one of these genes (PKS13, FFUJ_12020) with the hygromycin B resistance cassette (supplemental Fig. S3) resulted in total loss of gibepyrone A biosynthesis in three independent deletion mutants. In loco complementation of one deletion mutant with the full-length PKS13 gene resulted in full restoration of gibepyrone A formation ( Fig. 3B and supplemental Fig. S4). Therefore, PKS13 is the key enzyme of gibepyrone biosynthesis, designated Gpy1.
The Gibepyrone A Biosynthetic Gene Cluster Includes a Transporter-encoding Gene-In addition to the key enzymeencoding gene, SM gene clusters often comprise genes encoding pathway-specific transcription factors (TFs); several tailoring enzymes, such as methyltransferases, oxidoreductases, or P450 monooxygenases; and efflux transporters for the respective SM (50).
Upstream and downstream of GPY1, several genes were identified that might belong to a putative gibepyrone biosynthetic gene cluster (Fig. 4A). As summarized in Table 1, a meth-yltransferase-encoding gene (FFUJ_12019) was found upstream of GPY1, whereas two genes encoding an ATP-binding cassette (ABC) transporter (FFUJ_12021) and a fungus-specific Zn(II) 2 Cys 6 -type TF (FFUJ_12023) were detected downstream. The gene FFUJ_12022 does not harbor any domains or motifs that might confer a specific function. Moreover, a genomic alignment of different Fusarium spp. (F. proliferatum, Fusarium mangiferae, Fusarium verticillioides, and F. oxyspo-rum) underlined that GPY1 and further adjacent genes are conserved in these species, whereas the gene of unknown function, FFUJ_12022, is not always clustered with GPY1 (supplemental Fig. S5). To assess the involvement of the other indicated genes in gibepyrone A biosynthesis, single deletion mutants were generated (supplemental Figs. S6 -S8). Deletion of the methyltransferase-and TF-encoding genes FFUJ_12019 and FFUJ_ 12023, respectively, did not alter gibepyrone A yields, whereas , and PPT1 C . The strains were grown in the presence of 6 mM glutamine for 7 days. The gibepyrone A peak was verified through comparison with the synthesized chemical standard. B, the WT, the ⌬gpy1 deletion mutant, and the complemented strain GPY1 C were grown and analyzed as described above. The strains were grown in triplicate in the presence of 6 mM glutamine for 7 days with product formation of the WT being set to 100%. C, HPLC-MS/MS and Northern blot expressional analysis of the WT and the overexpression mutants of GPY1 and FFUJ_12023 (TF). The strains were grown and analyzed as described above for gibepyrone A quantification, whereas cell harvest was carried out after 3 days for expressional analysis. In the latter case, GPY1 and FFUJ_12023 were used as probes. Error bars, S.D.
deletion of the ABC transporter-encoding gene FFUJ_12021 resulted in a significantly enhanced level of extracellular gibepyrone A (Fig. 4B). Thus, adjacent to GPY1, a second gene was identified with impact on gibepyrone A production, designated GPY2.
Next, overexpression mutants were generated for GPY1 and the TF-encoding gene FFUJ_12023 (supplemental Fig. S9, A and B). Whereas the OE::GPY1 overexpression mutant showed significantly higher gibepyrone A production, overexpression of FFUJ_12023 did not have any impact (Fig. 4C), indicating that the encoded putative TF does not represent a pathwayspecific TF for gibepyrone biosynthesis.
The ABC Transporter Gpy2 Represses PKS Gene Expression-In addition to the enhanced extracellular gibepyrone A levels upon deletion of the transporter-encoding gene GPY2 (Fig. 4B), significantly elevated intracellular gibepyrone A levels were also observed in the mutant compared with the WT (Fig. 5A). These increased product levels were in accordance with the up-regulated expression of the PKS-encoding gene GPY1 in ⌬gpy2 mutants (Fig. 5B). GPY1 gene expression and gibepyrone A formation were elevated throughout the whole cultivation period of the ⌬gpy2 mutant, as shown for days 1-3 in Fig. 5, C and D. The low producing WT phenotype could be restored through in loco complementation with the full-length GPY2 gene (supplemental Figs. S10 and S11).
To study the potential regulatory role of GPY2 in more detail, this ABC transporter-encoding gene was overexpressed in the WT and GPY1 overexpression background (supplemental Fig.  S9C). Interestingly, the overexpression of GPY2 led to repression of GPY1 gene expression and gibepyrone A product formation in the WT (native GPY1 promoter) but not in the OE::GPY1 background (OE::GPY1/OE::GPY2) (Fig. 5, A and B). These data indicate that GPY1 regulation by Gpy2 occurs on the level of gene expression.
Whereas GPY2 was equally expressed in the WT and ⌬gpy1 background (supplemental Fig. S12A), the addition of 1 or 10 g/ml synthetic gibepyrone A induced GPY2 expression both in the WT and ⌬gpy1 (supplemental Fig. S12B). Interestingly, the PKS-encoding gene GPY1 itself was also up-regulated upon the addition of gibepyrone A to the WT. In contrast, expression of the adjacent gene FFUJ_12019 was not affected (supplemental Fig. S12B). Taken together, these data indicate that gibepy-rone A is able to induce its own production by an elevated GPY1 gene expression in a positive feedback loop ( Fig. 5D and supplemental Fig. S12B). At the same time, expression of the ABC transporter-encoding gene GPY2 is also induced (supplemental Fig. S12B), causing a down-regulation of GPY1 expression in a negative feedback loop.
Gibepyrone A Is Toxic to F. fujikuroi-The highly sophisticated regulatory mechanism described above led to the assumption that gibepyrone A might have a biological activity not only against bacterial and yeast strains (31) but also against the producing fungus itself. A plate assay using synthetic gibepyrone A against the WT strain and the mutants ⌬gpy1 and ⌬gpy2 was performed. Indeed, the growth of all three strains was equally inhibited, as indicated by the reduced colony diameters to ϳ50 and 15% in the presence of 100 and 200 g/ml gibepyrone A, respectively (supplemental Fig. S13).
C-Methylation Mechanism in the Biosynthesis of Gibepyrone A-As underlined by feeding experiments with isotopically labeled precursors, gibepyrone A consists of a polyketide backbone that is modified by two methylation steps (supplemental Fig. S2A). Because deletion of the methyltransferase-encoding gene FFUJ_12019 did not have an impact on gibepyrone A production ( Fig. 4B), the C-methyltransferase domain within the highly reducing PKS Gpy1 might solely be responsible for attaching both methyl groups to the nascent polyketide backbone. To verify this hypothesis, an amino acid substitution was introduced into the catalytic center of the C-methyltransferase domain of Gpy1. Glycine at position 1443 was exchanged for valine (G1443V) because corresponding mutations in the F. verticillioides fumonisin synthase (Fum1) and Phoma sp. C2932 squalestatin tetraketide synthase (SQTKS) were described as sterically inhibiting SAM substrate binding (51,52). An amino acid alignment of the C-methyltransferase domains of Gpy1, Fum1, and SQTKS underlined the conserved position of this active site glycine (supplemental Fig. S14A). The ⌬gpy1 strain was complemented with this mutated GPY1 variant, yielding independent transformants of GPY1 G1443V . Although expression of GPY1 G1443V was similar to GPY1 C , re-introduction of this point-mutated gene copy could not restore gibepyrone A production (supplemental Fig. S14, B and C), suggesting that the PKS alone is able to condense the methylated polyketide gibepyrone A. Single-methylated or completely non-methylated gibepyrone A derivatives could not be identified in GPY1 G1443V mutants by HPLC-HRMS analysis (supplemental Fig. S15).
Production of Gibepyrone A Derivatives-The initial report on gibepyrone A by Barrero et al. (31) also mentioned the presence of a series of oxygenated derivatives, gibepyrones B-F, in F. fujikuroi (Fig. 1D). All of these compounds were assumed to have a common biosynthetic origin. In fact, gibepyrones B-D are characterized through oxidations at the methyl group C8, whereas gibepyrone E carries a 6,7-epoxide function in the side chain. Furthermore, gibepyrone F was hypothesized to be generated through oxidative cleavage of the C6ϭC7 double bond (31). No genes encoding oxidizing proteins were identified in close proximity to GPY1 ( Table 1), suggesting that independent enzymes encoded outside of the gene cluster might facilitate these reactions.
The oxidation of a single carbon atom (e.g. the conversion of gibepyrone A (methyl group) to gibepyrone D (acid group) via gibepyrone B (hydroxyl group) and gibepyrone C (aldehyde group)) is a typical reaction sequence catalyzed by P450 monooxygenases. To test this hypothesis, WT cultures were compared with those of the ⌬cpr mutant. CPR encodes the major NADPH-cytochrome P450 reductase in F. fujikuroi that is associated with P450 monooxygenases, being essential for the electron transfer from NADPH onto the prosthetic heme group of the P450s (53). Thus, in the ⌬cpr mutant, the majority of all P450s is inhibited, which has been first demonstrated for the P450s encoded by the gibberellic acid gene cluster (53). To further examine the involvement of CPR in the biosynthesis of the oxidized gibepyrones, a chemical synthesis of gibepyrones B, C, and E was performed (Fig. 2). Gibepyrone E was synthesized via epoxidation of gibepyrone A using meta-chlorobenzoic acid, whereas 2 was used as starting material for the synthesis of 6-(2-hydroxybut-3-en-2-yl)-3-methyl-2H-pyran-2-one (3) using modified Riley conditions (54). Compound 3 was converted into gibepyrone B via acidic rearrangement of the tertiary allylic hydroxy function using methanesulfonic acid or to gibepyrone C via oxidative rearrangement using pyridinium chlorochromate (Fig. 2).
The synthetic compounds were used as standards to search for oxidized gibepyrones in the F. fujikuroi WT and mutant strains. Gibepyrones B and D were identified by HPLC-HRMS The WT and indicated deletion and overexpression mutants were grown in the presence of 6 mM glutamine for 7 days. Culture fluids were directly measured for extracellular gibepyrone A content, whereas metabolite extraction from washed mycelium was applied before quantifying intracellular levels. The cultivation was carried out in triplicate, and product formation of the WT was set to 100%. B, qRT-PCR analysis of the relative expression (RE) of GPY1 and GPY2 using the ⌬⌬Ct method. The WT, ⌬gpy2, and OE::GPY2 mutants were cultivated for 3 days, whereupon total RNA was isolated from the harvested mycelium. Error bars (S.D.) originate from a technical replicate, and expression of the WT was arbitrarily set to 1. C, HPLC-MS/MS analysis of extracellular gibepyrone A. The WT and ⌬gpy2 were grown in the presence of 6 mM glutamine for 1-3 days (d). The cultivation was carried out in triplicate, and product formation of the WT at 3 days was set to 100%. D, qRT-PCR analysis of the relative expression of GPY1 using the ⌬⌬Ct method. The WT and ⌬gpy2 were cultivated for 1-3 days, whereupon total RNA was isolated from the harvested mycelium. Error bars (S.D.) originate from a technical replicate, and expression of the WT at 3 days was arbitrarily set to 1. analysis in WT cultures and in increased amounts in the OE::GPY1 strain, but not in the ⌬gpy1 mutant, linking the production of these metabolites to the gibepyrone PKS ( Fig. 6 and supplemental Fig. S16). In contrast, gibepyrone C was detected in neither WT cultures nor the OE::GPY1 mutant (supplemental Fig. S16). Comparison of WT and ⌬cpr cultures revealed that gibepyrones B and D were missing in the mutant, whereas an accumulation of gibepyrone A was observed (Fig. 6). These data are in full agreement with the conversion of gibepyrone A to gibepyrones B and D by one or multiple CPR-dependent P450 monooxygenases encoded outside of the gibepyrone biosynthetic gene cluster.
The HPLC-HRMS analysis of an aged synthetic gibepyrone E standard showed a slow compound degradation in aqueous solution. Furthermore, gibepyrone E could not be detected in culture fluids, whereas two of its degradation products were present in cultures of the WT and the ⌬cpr mutant (Fig. 6). Presumably, these two compounds represent diastereomeric diols that are formed from gibepyrone E by hydrolysis of its epoxide function in the presence of water (supplemental Fig.  S17A). Because the two compounds were also found in the ⌬cpr mutant, the conversion of gibepyrone A into gibepyrone E must be independent of the P450 monooxygenase(s) involved in the formation of the other oxidized gibepyrones. Instead, gibepyrone E formation and its hydrolysis were also observed in an aqueous solution of synthetic gibepyrone A (supplemental Fig.  S17B), indicating its autoxidation in the presence of molecular oxygen.
Gibepyrone F was only detected in headspace extracts of the F. fujikuroi WT via GC-MS and not in liquid cultures via HPLC-HRMS. Similar to gibepyrone E, this compound was also formed spontaneously from synthetic gibepyrone A (supplemental Fig. S17C) and was also detectable in headspace extracts from the ⌬cpr mutant strain (supplemental Fig. S18), indicating its formation independent of enzymatic activity.
Gibepyrone A Production by F. solani-The reported structure of fusalanipyrone from F. solani DSM 62416 (41) represents the (Z)-stereoisomer of gibepyrone A. However, genome sequencing of the strain F. solani 77-13-4 (55) revealed that this organism encodes a PKS with the same domain organization and 84% amino acid identity compared with Gpy1 (17) (accession number NECHADRAFT_123282; supplemental Fig. S5). Although a homolog of the ABC transporter-encoding gene GPY2 cannot be found adjacent to the F. solani GPY1 homolog, the methyltransferase-encoding border gene (a homolog of FFUJ_12019) is conserved in F. solani, strongly suggesting that NECHADRAFT_123282 is the true Gpy1 homolog (supplemental Fig. S5). This bioinformatic analysis underlined that F. solani may also produce gibepyrone A, whereas the production of the stereoisomer fusalanipyrone by Gpy1 would be difficult to understand. Comparison of the NMR data reported for fusalanipyrone (41) with the NMR data of synthetic gibepyrone A (see "Experimental Procedures") showed the identity of these compounds. Their identity was additionally verified by comparison of liquid culture extracts from F. solani DSM 62416 and F. fujikuroi IMI58289 with the synthetic gibepyrone A standard by GC-MS (supplemental Fig. S19). Therefore, the structure of fusalanipyrone must be revised to that of gibepyrone A.
Regulation of Gibepyrone A Biosynthesis-F. fujikuroi SM biosynthesis was shown to be differentially regulated by global regulators, such as members of the fungus-specific velvet complex (Vel1, Vel2, and Lae1) (56) as well as Sge1, a recently char-   (57). To study their impact on gibepyrone A production, the yields were compared between the WT and the regulatory mutants. Gibepyrone A levels were elevated in deletion mutants of the velvet complex, ⌬vel1, ⌬vel2, and ⌬lae1, but down-regulated in ⌬sge1 (Fig. 7A). Analysis of GPY1 and GPY2 expression by quantitative RT-PCR (qRT-PCR) verified these results (Fig. 7B). Besides GPY1 and GPY2, no further adjacent genes were up-regulated in the ⌬vel1 mutant (Fig. 7C), underlining that these two genes represent a co-regulated small biosynthetic gene cluster.

Discussion
Before this study, the biosynthetic origin of gibepyrone A and its oxidized derivatives was unknown. Therefore, this study focused on the biosynthetic pathway of gibepyrones A-F, the elucidation of their biosynthetic gene cluster, and the regulation of these genes in F. fujikuroi.
Biosynthesis of the Polyketide Gibepyrone A-In this work, we provide evidence for gibepyrone A biosynthesis by F. fujikuroi PKS13 (FFUJ_12020), designated Gpy1. Feeding experiments with isotopically labeled precursors and analysis of the ⌬ppt1 mutant ruled out a terpenoid origin for this compound.
Although a polyketide biosynthetic pathway was suggested for the structurally related compounds fusalanipyrone and nectriapyrone (43,48), unambiguous proof on a molecular level was lacking for this group of metabolites. By deleting GPY1, the responsible key gene of gibepyrone biosynthesis was identified, encoding a highly reducing PKS. A proposed biosynthetic pathway to gibepyrone A by Gpy1 uses an acetyl-CoA starter unit, three malonyl-CoA extender units, and two SAM-dependent methylations to establish the gibepyrone A carbon backbone, followed by product release upon intramolecular cyclization (Fig. 8). Introduction of a loss-of-function mutation into the C-methyltransferase domain of Gpy1 and the non-effective deletion of a methyltransferase encoding gene adjacent to GPY1 (FFUJ_12019) verified that both methylation steps are catalyzed by the PKS itself. It is unclear whether PKS biosynthesis in the GPY G1443V mutant is blocked or results in an unidentified shunt product that may be unstable or quickly metabolized. The latter was suggested for Phoma sp. C2932 SQTKS (52), whereas for other systems, including the lovastatin PKS, the formation of defined non-methylated shunt products was shown by in vitro experiments in the absence of SAM (58). The strains were grown in triplicate in the presence of 6 mM glutamine for 7 days with product formation of the WT being set to 100%. B, qRT-PCR analysis of the relative expression (RE) of GPY1 and GPY2 using the ⌬⌬Ct method. The strains were cultivated for 3 days, whereupon total RNA was isolated from the harvested mycelium. Error bars (S.D.) originate from a technical replicate, and expression of the WT was arbitrarily set to 1. C, qRT-PCR analysis of the relative expression of the indicated genes using the ⌬⌬Ct method. The WT and ⌬vel1 mutant were cultivated for 3 days, whereupon total RNA was isolated from the harvested mycelium. Error bars (S.D.) originate from a technical replicate, and expression of each gene in the WT was arbitrarily set to 1.

Distribution of Putative Gibepyrone A Biosynthetic
Genes among the Genus Fusarium-Our bioinformatic analysis verified that GPY1 is conserved among members of the F. fujikuroi species complex and can also be found in the closely related F. oxysporum and the distantly related F. solani species complexes (supplemental Fig. S5). These data are in agreement with the isolation of gibepyrone A from members of the F. solani species complex (34) and with our structural revision for fusalanipyrone from F. solani DSM 62416 that is identical to gibepyrone A. Conclusively, GPY1 orthologs seem to be present within the whole genus Fusarium.
Moreover, we suggest that homologous PKSs are responsible for the biosynthesis of the related compounds nectriapyrone and pestalopyrone (Fig. 1, E and F) from G. missouriensis and P. oenotherae, respectively (44,45). Both nectria-and pestalopyrone harbor a methoxy group that is most likely introduced as hydroxyl group by the responsible PKS and modified in a second step by a cluster-encoded O-methyltransferase, as reported recently for fusarubin and bikaverin pigment biosyntheses by F. fujikuroi (8,59). Furthermore, the demethyl derivative pestalopyrone is most likely produced with the use of only one SAM unit by a PKS that is very similar to the gibepyrone PKS, which provides another example of the highly programmed nature of tailoring domains in fungal iterative PKSs (60).
Gibepyrone A Derivatives Have Both an Enzymatic and a Non-enzymatic Origin-In addition to gibepyrone A, we analyzed the production of the derived compounds gibepyrones B-F. Gibepyrones B and D were shown to be produced from gibepyrone A by one or multiple CPR-dependent P450 monooxygenases encoded by genes not clustered with GPY1 and GPY2. A derivatization by P450s, encoded outside of the respective gene cluster, was also described recently for the F. fujikuroi SM fusaric acid. In this case, the conversion into the less toxic derivatives dehydrofusaric acid and fusarinolic acid represents a detoxification mechanism for the producing fungus (11). Whereas gibepyrone A was shown to be moderately active against the Gram-positive bacteria B. subtilis and S. aureus as well as against the yeasts S. cerevisiae and C. albicans, gibepyrone B was only active against S. aureus and S. cerevisiae (31). Because gibepyrone A has a toxic effect on the producing fungus itself, it is tempting to hypothesize that its derivatization into the oxidized compounds does represent a detoxification mechanism. In a comparable manner, incubation of the plant pathogen B. cinerea with the toxic compound 6-pentyl-2H-pyran-2-one (Fig. 1C) from Trichoderma spp. led to the formation of oxidized derivatives harboring hydroxyl and acid groups. These metabolized products were not toxic in a germination assay, whereas 6-pentyl-2H-pyran-2-one clearly was (61). The fact that gibepyrone D was also isolated from cultures of F. oxysporum and F. proliferatum indicates that this detoxification mechanism would not be specific to F. fujikuroi (35,36).
Gibepyrone C, which represents the intermediate between gibepyrones B and D, was identified by Barrero et al. (31) but was not found in our liquid cultures compared with the synthetic gibepyrone C. Reasons for this contrasting result could be different cultivation procedures or the fact that gibepyrone C, harboring an aldehyde group, is a highly reactive metabolite that might be quickly processed into gibepyrone D, maybe even by the same P450 catalyzing the conversion to gibepyrone B.
Furthermore, we could establish that gibepyrones E and F are generated spontaneously, in the absence of enzymatic activity, based on spontaneous oxidation of synthetic gibepyrone A in the presence of air. For the formation of these compounds, a reaction pathway can be proposed, starting with the addition of molecular oxygen to the C6ϭC7 double bond of gibepyrone A (supplemental Fig. S20). Under neat conditions, the resulting dioxetane may decompose to gibepyrone F and acetaldehyde, whereas the dioxetane can be attacked by water in aqueous solution to decompose to gibepyrone E and hydrogen peroxide (supplemental Fig. S20). No degradation of clean synthetic gibepyrone A was observed over several months, when stored at Ϫ20°C in the dark.
The Special Role of the ABC Transporter Gpy2-In F. fujikuroi, an ABC transporter-encoding gene, GPY2, belongs to the gibepyrone gene cluster in addition to GPY1. This gene was found to be co-regulated with GPY1 in the regulatory mutants. Furthermore, the generation of GPY2 deletion and overexpression mutants clearly verified its impact on gibepyrone A production. Our data suggest that Gpy2 plays only a minor or no role at all in the actual transport of gibepyrone A out of the cell. We propose that at least one other transporter is involved in metabolite efflux, because gibepyrone A was found both in the culture fluid and mycelium of ⌬gpy2 mutants. Our data indicate that Gpy2 is able to regulate the expression of the PKS-encoding gene GPY1. GPY1 expression and gibepyrone A production were significantly elevated in the ⌬gpy2 mutant but down-regulated in the OE::GPY2 mutants. This highly sophisticated Gpy2-mediated regulatory mechanism allows a fine-tuning of gibepyrone A biosynthesis, probably protecting the producing fungus from the toxic effect of its own product.
There are several examples where deletion of a transporterencoding gene negatively affected SM production, and in all of Gibepyrone Biosynthesis in F. fujikuroi DECEMBER 30, 2016 • VOLUME 291 • NUMBER 53 these cases, it was highly anticipated that the SMs are toxic to the producing fungi. Such SM-specific transporters conferring self-protection are major facilitator superfamily transporters as encoded by the Aspergillus fumigatus gliotoxin cluster (GliA) (62), the Fusarium graminearum trichothecene cluster (Tri12) (63), and the Fusarium spp. fusaric acid cluster (Fub11) (11,64). Deletion of gliA and FUB11 resulted in a reduced intra-and extracellular accumulation of gliotoxin and fusaric acid, respectively, most likely due to the detoxification of these compounds via derivatization, whereas cluster gene expression was not affected (11,62,64). Similarly, deletion of TRI12 led to diminished trichothecene levels via a yet unknown mechanism (63).
Far fewer SM gene clusters harbor ABC transporter-encoding genes. Examples are the Fusarium verticillioides fumonisin cluster (Fum19) (65), the Leptosphaeria maculans sirodesmin cluster (SirA) (66), and the F. fujikuroi beauvericin cluster (14). Upon disruption of FUM19, little impact was detected on extracellular fumonisin contents, and the authors assumed that other transporters are able to compensate for Fum19 (65). To the best of our knowledge, only two reports on enhanced extracellular SM contents upon deletion of a transporter-encoding gene have been described before, a phenotype that would be comparable with the one observed for F. fujikuroi ⌬gpy2. Thus, the L. maculans sirodesmin NRPS-encoding gene sirP was upregulated in ⌬sirA ABC transporter mutants. Unfortunately, the authors did not quantify intracellular sirodesmin levels, but they did find enhanced extracellular sirodesmin in ⌬sirA mutants (66). Additionally, an up-regulation of NRPS gene expression and SM product formation was observed upon deletion of F. fujikuroi BEA3, encoding the beauvericin ABC transporter that we analyzed recently (14).
It is tempting to hypothesize that this regulatory role of SM transporters is specific to ABC transporters; however, further research is needed to prove this. A transporter that has been implicated in having a sensing function in addition to its transport function is the S. cerevisiae general amino acid permease Gap1 (67). The authors were able to introduce C-terminal truncations that only affected the sensing function and not the transport efficiency of Gap1, establishing that the signaling pathway is mediated via protein kinase A (67). A potential signal transduction pathway that might be responsible for GPY1 repression via the ABC transporter Gpy2 as well as its biological function remain to be elucidated.
Global Regulatory Mechanisms Controlling Gibepyrone A Biosynthesis-Because no cluster-specific TF is encoded by the small gene cluster, we evaluated the impact of global regulators on gibepyrone A biosynthesis. Thus, members of the fungusspecific velvet complex, Vel1, Vel2, and Lae1, represent repressors of gibepyrone A biosynthesis, whereas Sge1 acts as positive regulator.
In addition to the pigment bikaverin, gibepyrone A represents the second SM that is influenced in a negative manner by members of the velvet complex in F. fujikuroi. In the case of bikaverin, however, only Vel1 and Vel2 act as repressors, whereas Lae1 does not (56). The mode of action of the velvet complex is still not fully understood. Whereas homologs of the velvet domain proteins Vel1 and Vel2 have been described as having DNA-binding properties (68), the putative histone methyltransferase Lae1 has been implicated to act on the highest hierarchical level of chromatin remodeling (69). Furthermore, the complex may vary in its conformation when regulating different aspects of differentiation and secondary metabolism (70). Notably, this is the first example in which all three members of the velvet complex seem to work together to repress gibepyrone A biosynthesis in F. fujikuroi.
Sge1 was shown to be a major positive regulator of secondary metabolism in F. fujikuroi (57), which also held true for gibepyrone A biosynthesis. Also for Sge1, the mode of action is not fully resolved. Although the yeast Sge1-homologs in C. albicans and Histoplasma capsulatum have been described to bind the promoters of their target genes (71,72), no common DNAbinding motif has been identified in filamentous fungi so far. In fact, Sge1 in F. fujikuroi was suggested to act on a high hierarchical level, also regulating genes encoding histone-modifying enzymes on a transcriptional level (57).
In conclusion, this work provides a comprehensive overview of the biosynthesis and regulation of gibepyrone A and its derivatives in F. fujikuroi, applying feeding experiments, chemical synthesis of gibepyrone standards, gene knock-out, and overexpression strategies as well as regulatory mutants. As summarized in Fig. 9, the PKS Gpy1 is responsible for the production of gibepyrone A, whereas one or several cluster-independent P450 monooxygenases account for the production of gibepyrones B and D. Furthermore, gibepyrones E and F are gained in the absence of enzymatic activity. We found evidence that the ABC transporter Gpy2 represses expression of the PKS-encoding gene GPY1 via a yet unknown mechanism. Finally, gibepyrone biosynthesis underlies global regulatory mechanisms, as verified for Sge1 and the velvet complex that affect gibepyrone A biosynthesis in a positive and negative manner, respectively.
For all experiments in liquid culture, the fungal strains were preincubated for 3 days in 300-ml Erlenmeyer flasks containing 100 ml of Darken medium (73) on a rotary shaker at 180 rpm and 28°C in the dark. 500 l of this preculture were transferred into 100 ml of synthetic ICI medium (Imperial Chemical Industries Ltd., London, UK) (74) supplemented with 6 mM glutamine as the sole nitrogen source. This main culture was incubated under the indicated conditions for an additional 1-7 days. Feeding experiments in liquid culture were performed after the addition of 0, 1, or 10 g/ml synthetic gibepyrone A to a 2-day-old 30-ml ICI culture. After 2 h of induction, the mycelium was harvested and used for the isolation of RNA. For protoplasting F. fujikuroi, 500 l of the preculture were transferred into 100 ml of ICI medium with 10 g/liter fructose instead of glucose as well as 0.5 g/liter (NH 4 ) 2 SO 4 as a nitrogen source and shaken for no longer than 16 h. The general maintenance of the fungal strains was carried out on solidified complete medium (CM) (75), whereas, before DNA isolation, the strains were incubated on CM covered with cellophane for 2-4 days at 28°C in the dark. Hyphal growth of fungal strains was assessed on solidified CM supplemented with 0 -200 g/ml synthetic gibepyrone A. Plates were incubated for 4 days at 28°C in the dark. The solvent (ethanol absolute) was added as a control, not resulting in an inhibition of fungal growth (data not shown).

Plasmid Constructions
The cloning of deletion, complementation, and overexpression constructs was accomplished with the use of yeast recombinational cloning (76,77). For targeted gene deletion, ϳ0.6 -1-kb large upstream (5Ј) and downstream (3Ј) regions of the gene of interest were amplified with the primer pairs 5F/5R and 3F/3R, respectively (supplemental Table S1). The hygromycin resistance cassette hphR, consisting of the hygromycin B phosphotransferase gene hph and the strong trpC promoter from Aspergillus nidulans, was amplified with hph_F/hph_R from the template pCSN44 (78). S. cerevisiae FY834 (79) was transformed with the obtained fragments as well as with the EcoRI/ XhoI-restricted shuttle vector pRS426 (80), resulting in the deletion vectors p⌬12019, p⌬gpy1, p⌬gpy2, and p⌬12023.
For in loco complementation of the PKS13-encoding gene GPY1 (FFUJ_12020), the full-length gene was amplified in three overlapping fragments using a proofreading polymerase and primer pairs 12020_c_F1/12020_c_R1, 12020_c_F2/12020_ c_R2, and 12020_c_F3/12020_c_R3 (supplemental Table S2). It was cloned upstream of the glucanase terminator Tgluc from B. cinerea B05.10, which was amplified with BcGlu_Term_F2/ Tgluc_Nat1_R. The nourseothricin resistance cassette natR, including the strong A. nidulans trpC promoter, served as a resistance marker and was amplified with hph_R/hph_R2 from the template pZPnat1 (GenBank TM accession number AY631958.1). GPY1 5Ј and 3Ј sequences were gained as indicated above. S. cerevisiae FY834 was transformed with the obtained fragments as well as with the EcoRI/XhoI-restricted vector pRS426, yielding the complementation vector pGPY1 C (supplemental Fig. S4A). For introducing the G1443V mutation into the Gpy1 C-methyltransferase domain (51, 52), a modified in loco complementation vector was cloned, pGPY1 G1443V . Using overlapping primers that harbor the nucleotide exchange, GPY1 was amplified in four overlapping fragments, two of which were based on the novel primer combinations 12020_c_F2/12020_G1443V_R and 12020_G1443V_F/12020_ c_R2 (supplemental Table S2). Furthermore, for in loco complementation of the ABC transporter-encoding gene GPY2 (FFUJ_12021), the following gene-specific fragments were amplified: 12021_5F/12021_c_R1, 12021_c_F2/12021_c_R2, and 12021_3F/12021_3R (supplemental Table S2). S. cerevisiae FY834 was transformed with the obtained fragments, Tgluc and natR, as well as with the EcoRI/XhoI-restricted vector pRS426, FIGURE 9. Biosynthesis and regulation of gibepyrone A and its derivatives in F. fujikuroi. The PKS Gpy1 facilitates the condensation of the C 10 compound gibepyrone A. Beside the PKS, an ABC transporter-encoding gene, GPY2, belongs to the cluster. However, Gpy2 seems to have a minor impact on the efflux of gibepyrone A out of the cell, but it was shown to repress expression of the key gene GPY1. Furthermore, members of the velvet complex, Vel1, Vel2, and Lae1, negatively affect cluster gene expression and gibepyrone A product formation, whereas Sge1 represents a positive regulator of gibepyrone biosynthesis. The oxidized compounds gibepyrones B and D are derived from gibepyrone A through the activity of cluster-independent, CPR-dependent P450 monooxygenases, whereas gibepyrones E and F are formed in the absence of enzymatic activity. DECEMBER 30, 2016 • VOLUME 291 • NUMBER 53 yielding the complementation vector pGPY2 C (supplemental Fig. S10A).

Gibepyrone Biosynthesis in F. fujikuroi
Overexpression of GPY1 was achieved through fusion to the strong oliC promoter from A. nidulans. Thus, GPY1 was again amplified in three overlapping fragments using a proofreading polymerase, in this case with 12020_OE_F/12020_c_R1, 12020_c_F2/12020_c_R2, and 12020_c_F3/12020_OE_R (supplemental Table S2). The latter primer combination included ϳ300 bp of the GPY1 terminator sequence. S. cerevisiae was transformed with these fragments as well as with the NcoI/ SacII-restricted plasmid pNDN-OGG (77), resulting in pOE:: GPY1 (supplemental Fig. S9A). Similarly, for overexpressing FFUJ_12023 via PoliC, the gene of interest was amplified with the primer pair 12023_OE_F/12023_OE_R (supplemental Table S2). The NcoI/NotI-restricted plasmid pNAH-OGG (77) served as the plasmid backbone, yielding pOE::12023 (supplemental Fig. S9B), and for overexpression of GPY2 in the WT and OE::GPY1 overexpression backgrounds, the first 1.4 kb of GPY2 were amplified with the primer pair 12021_OE_F/ 12021_OE_R (supplemental Table S2). The fragment was fused to the NcoI/SacII-restricted plasmid pNAH-GGT harboring the strong glnA promoter from F. fujikuroi (11), resulting in pOE::GPY2 (supplemental Fig. S9C). The correct assembly of all complementation and overexpression vectors was verified by sequencing with primers listed in supplemental Table S2.

Fungal Transformations and Analysis of Transformants
Protoplast transformation of F. fujikuroi was carried out as described previously (81). About 10 7 protoplasts were transformed with the amplified replacement cassettes to generate deletion mutants. Amplification was achieved with the primer combination 5F/3R (supplemental Table S1), whereas the circular deletion vectors served as DNA templates. For in loco complementation of ⌬gpy1 T5 with pGPY1 C (supplemental Fig.  S4A) or pGPY1 G1443V , 30 g of the vectors were linearized with ApaI before transformation. Similarly, for in loco complementation of ⌬gpy2 T9 with pGPY2 C (supplemental Fig. S10A), 30 g of the vector was linearized with AseI. Furthermore, 15 g of the overexpression vectors pOE::GPY1 (supplemental Fig.  S9A), pOE::12023 (supplemental Fig. S9B), and pOE::GPY2 (supplemental Fig. S9C) were introduced in a circular manner. Selection of transformants was achieved under the use of 100 g/ml hygromycin B (Calbiochem) or 100 g/ml nourseothricin (Werner-Bioagents, Jena, Germany), depending on the resistance marker.

Standard Molecular Methods
For DNA isolation from F. fujikuroi, lyophilized mycelium was ground in the presence of liquid nitrogen and extracted following the protocol of Cenis (82). Generated deletion mutants were analyzed via Southern blotting (83) regarding ectopically integrated replacement cassettes. Thus, their genomic DNA was digested with an appropriate restriction enzyme (Fermentas, St. Leon-Rot, Germany), whereupon it was separated in a 1% (w/v) agarose gel and transferred onto a nylon membrane (Nytran TM SPC, Whatman, Sanford, ME) by downward alkali blotting (84). Hybridization was performed with 32 P-labeled probes that were generated with the random oligomer-primer method (85). Amplified 5Ј or 3Ј sequences served as templates for probe generation (supplemental Table  S1). Southern blotting analyses of ⌬gpy1 and ⌬gpy2 can be found in supplemental Figs. S3 and S7, respectively. Plasmid DNA from S. cerevisiae as well as Escherichia coli Top10 FЈ (Invitrogen, Darmstadt, Germany) was isolated with the NucleoSpin plasmid kit (Machery-Nagel, Düren, Germany). Concerning DNA amplification by PCR, BioTherm TM DNA polymerase (GeneCraft, Lüdinghausen, Germany) was chosen for standard applications, especially diagnostic PCR, and handled according to the manufacturer's instructions. For the purpose of amplifying long fragments, TaKaRa LA Taq DNA polymerase (Takara Bio, Saint-Germain-en-Laye, France) was used, whereas Phusion high-fidelity DNA polymerase (Finnzymes, Vantaa, Finland) was applied in case proofreading was essential.
RNA isolation was carried out with the use of TRI Reagent TM (Sigma-Aldrich, Steinheim, Germany). For expressional analyses by Northern blotting (86), 20 g of total RNA were separated in a 1% (w/v) denaturating agarose gel (85). Upon transfer of the separated RNA onto a nylon membrane (Nytran TM SPC, Whatman), it was hybridized with 32 P-labeled probes, as indicated above (85). ϳ1-kb fragments of GPY1 and FFUJ_12023 served as templates for probe generation and were amplified with primer pairs 12020_WT_F/12020_WT_R and 12023_ WT_F/12023_WT_R, respectively (supplemental Table S1). Before expressional analyses by qRT-PCR, 1 g of total RNA was treated with DNase I (Fermentas) and transcribed into cDNA using oligo(dT) primers and SuperScript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. Subsequently, qRT-PCR was performed with iQ SYBR Green Supermix (Bio-Rad, München, Germany) and cDNA as template using an iCycler iQ real-time PCR system (Bio-Rad). The annealing temperature was 60°C, and the primer efficiencies were between 90 and 110%. The ⌬⌬Ct method was used for calculating the results (87). For quantification of mRNA levels of the desired genes (FFUJ_12019, GPY1, GPY2, FFUJ_12022, and FFUJ_12023) and the constitutively expressed reference genes (FFUJ_07710, a GDP mannose transporter gene; FFUJ_05652, a related actin gene; and FFUJ_08398, a ubiquitin gene), the primers listed in supplemental Table S3 were used. The experiments were performed in two technical replicates and at least two biological repeats.

Quantitative Analysis of Gibepyrone A and Its Derivatives by HPLC-MS/MS and HPLC-HRMS
For quantifying gibepyrone A in the supernatant as sensitively as possible, a QTRAP 6500 MS system (SCIEX, Darmstadt, Germany) coupled to a 1260 Infinity LC system (Agilent Technologies, Waldbronn, Germany) was used. Seven-day-old cultures were filtered using 0.45-m membrane filters (BGB Analytik, Schlossböckelheim, Germany). Metabolite extraction from the mycelium was performed with a modified method described by Niehaus et al. (12). Briefly, 0.1 g of harvested, washed, and lyophilized mycelium was mixed thoroughly with 1.5 ml of ethyl acetate/MeOH/dichlormethane (3:2:1, v/v/v) for 2 h. 0.75 ml of the extract was evaporated to dryness and taken up in 0.75 ml of 15% (v/v) MeOH.
For relative quantification of gibepyrone A via HPLC-MS, methylparaben (MePa) was used as an internal standard, and chromatographic separation was carried out on a 150 ϫ 2.1-mm inner diameter, 5-m, Eclipse XDB-C 18 column (Agilent Technologies) with MeOH ϩ 1% formic acid (FA) as solvent A and H 2 O ϩ 1% FA as solvent B. The column was tempered to 40°C. After isocratic running for 3 min at 15% A, the gradient rose to 100% A in 10 min. With the concentration of solvent A, the flow rate also rose from 400 to 450 l. After holding these conditions for 2 min, the column was equilibrated for 3 min. A divert valve was used to discard polar substances from the medium in the first 3 min of the run. For MS/MS analysis with electrospray ionization (ESI), the following parameters were used. The curtain gas (nitrogen) was set to 30 p.s.i., the nebulizer gas (zero grade air) was set to 35 p.s.i., and the drying gas (zero-grade air) was set to 40 p.s.i. The ion spray voltage was set to ϩ5500 V in the positive and to Ϫ4500 V in the negative mode. Nitrogen was also used as the collision gas in medium mode. Gibepyrone A was analyzed with a declustering potential of 56 V, an entrance potential of 10 V, and a cell exit potential (CXP) of 12 V. The collision energy (CE) was adjusted for each multiple-reaction monitoring, respectively. The proton adduct of gibepyrone A was used as a parent ion for gibepyrone A analysis, and the following transitions were applied. For the identification of gibepyrone A derivatives, HPLC-HRMS measurements were carried out on an HPLC system (Accela LC with Accela pump 60057-60010 and Accela autosampler 60057-60020, Thermo Scientific, Dreieich, Germany) coupled to a Fourier transform mass spectrometer with a heated ESI source (LTQ Orbitrap XL, Thermo Scientific). The HPLC column used was a 150 ϫ 2.00-mm inner diameter, 5-m, Gemini C 18 with a 4 ϫ 2-mm Gemini NX C 18 guard column (Phenomenex, Aschaffenburg, Germany). The MS parameters were set as described elsewhere (17)  General Synthetic and Analytical Methods-Chemicals were purchased from Acros Organics (Geel, Belgium) or Sigma-Aldrich and used without purification. Commercially available isotopically labeled compounds were purchased from Euriso-Top (Saarbrücken, Germany) or Sigma-Aldrich. Thin layer chromatography was performed with 0.2-mm precoated plastic sheets Polygram Sil G/UV254 (Machery-Nagel). Normal phase column chromatography was carried out using Merck silica gel 60 (70 -200 mesh). Reversed phase column chromatography was carried out using Merck Lichroprep RP-18 silica gel (40 -63 m). 1 H NMR and 13 C NMR spectra were recorded on Bruker AV I (300 MHz), AV I (400 MHz), and AV III HD Prodigy (500 MHz) spectrometers and were referenced against CDCl 3 (␦ ϭ 7.26 ppm), C 6 D 6 (␦ ϭ 7.16 ppm), and CD 2 Cl 2 (␦ ϭ 5.32 ppm) for 1 H NMR and CDCl 3 (␦ ϭ 77.01 ppm), C 6 D 6 (␦ ϭ 128.06 ppm), CD 2 Cl 2 (␦ ϭ 53.84 ppm) for 13 C NMR. Analyses via GC-MS were carried out with a HP 7890B gas chromatograph connected to a HP 5977A inert mass detector fitted with a HP5-MS fused silica capillary column (30 m, 0.25-mm inner diameter, 0.50-m film). The instrumental parameters were 1) inlet pressure, 77.1 kilopascals, helium 23.3 ml/min; 2) injection volume, 2 l; 3) transfer line, 250°C; and 4) electron energy, 70 eV. The GC was programmed as follows for synthetic samples: 5 min at 50°C increasing at 10°C/min to 320°C and operated in split mode (50:1, 60-s valve time). For headspace analysis, autoxidation experiments, and analysis of liquid culture extracts, the GC program was as follows: 5 min at 50°C increasing at 5°C/min to 320°C and operated in split mode (10:1, 60-s valve time). The carrier gas was helium at 1 ml/min. Retention indices (I) were determined from a homologous series of n-alkanes (C 8 -C 40 ). IR spectra were recorded on a Bruker Alpha ATR spectrometer. UV-visible spectra were recorded on an Agilent Cary 100 spectrometer.

Synthetic Procedures
Gibepyrone A-A solution of triethyl amine (8.96 g, 88.6 mmol, 1.05 eq) in dichloromethane (10 ml) was added dropwise over 15 min to a solution of tigloyl chloride (10.0 g, 84.3 mmol, 1.0 eq) in dichloromethane (80 ml) under argon atmosphere at room temperature. The solution changed colors immediately from colorless to deep red and was stirred for 18 h at room temperature. The solvent was removed in vacuo, and the solid residue was taken up in Et 2 O. The insoluble material was removed by filtration, and the filtrate was concentrated in vacuo, giving a red oil, which was subjected to column chromatography on reversed phase silica gel (H 2 O/MeOH, 1:1 changing to 2:1). The product-yielding fractions were collected, concentrated to an aqueous suspension, and extracted with dichloromethane (3ϫ). The combined organic layers were dried over MgSO 4 , and the solvent was evaporated, giving gibepyrone A (1.26 g, 7.67 mmol, 18%) as a colorless highly viscous oil and its isomer 6-(but-3-en-2-yl)-3-methyl-2H-pyran-2-one (2, 1.18 g, 7.22 mmol, 17%) as a yellow oil.  3 J H,H ϭ 6.9 Hz, 3H, CH 3 ) ppm. 13 (7), 77 (11), 55 (14), 53 (40). High resolution electron ionization mass spectrometry (HREIMS) calculated for C 10 13  Gibepyrone E-To a solution of gibepyrone A (300 mg, 1.83 mmol, 1.0 eq) in dichloromethane, meta-chloroperbenzoic acid (70% in H 2 O, 541 mg, 2.20 mmol, 1.2 eq) was added. The reaction was quenched after 30 min by adding NaHCO 3 (aqueous, saturated). The mixture was extracted with dichloromethane (3ϫ), and the combined organic layers were dried over MgSO 4 . The solvent was removed in vacuo, and the oily residue was subjected to column chromatography on a reversed phase silica gel (H 2 O/MeOH, 2:1). The product-yielding fractions were collected, concentrated to an aqueous suspension, and extracted with dichloromethane (3ϫ). The combined organic layers were dried over MgSO 4 13 (7), 39 (7). HREIMS calculated for C 10 H 12 O 3 •ϩ : 180.0781, found: 180.0772. Gibepyrone B-A mixture of tetrahydrofuran (3 ml) and H 2 O (0.7 ml) was degassed by an argon stream for 60 min. Hydroxypyrone 3 (110 mg, 0.67 mmol, 1.00 eq) and methanesulfonic acid (118 mg, 1.22 mmol, 2 eq) were added subsequently under argon atmosphere at room temperature, and the resulting mixture was stirred at room temperature for 25 days in the dark. The reaction mixture was diluted with H 2 O and extracted with EtOAc (3ϫ). The combined organic layers were dried with MgSO 4 , the solvent was removed in vacuo, and the crude product was subjected to column chromatography on silica gel (cyclohexane/EtOAc, 1:2), yielding gibepyrone B (35 mg, 0.19 mmol, 29%) as a pale yellow viscous oil.

GC-MS Analysis and Feeding Experiments
For analysis of emitted volatile compounds, the fungal strains were inoculated on solid CM using a small piece of preculture grown on the same medium and grown for 7 days at 28°C in the dark, before collection of the volatiles on charcoal filter traps by use of a closed loop stripping apparatus, followed by analysis of headspace extracts by GC-MS, as described previously (32).
For the comparative analysis of gibepyrone A production by F. fujikuroi IMI58289 and F. solani DSM 62416, both strains were cultivated in liquid ICI medium (6 mM glutamine) and in the medium used for the isolation of fusalanipyrone (41) for 7 days at 28°C and 180 rpm. After filtration through a cotton cloth and centrifugation, 45 ml of the respective culture supernatant was extracted with EtOAc (2 ϫ 5 ml). The combined organic layers were dried over MgSO 4 and subjected directly to GC-MS analysis.
Autoxidation of gibepyrone A to gibepyrone F was performed by solving 0.5 mg of gibepyrone A in Et 2 O (1.5 ml) and analysis by GC-MS. Then the solvent was allowed to evaporate at room temperature, and the dry sample was kept at ambient conditions for 7 days before being solved in Et 2 O (1.5 ml) and analyzed by GC-MS again.