25 in the extracts of WT, and the activity level was significantly higher in the supernatant extract (Fig. 3e and f). This activity was later confirmed to be from galbonolide A by HPLC-MS analysis. Note that the TLC plates were developed three times for better separation. Notably absent was the activity of galbonolide A in the SK-galI-5 extracts (Fig. 3e and f). The mycelia extracts of WT, as well as SK-galI-5, exhibited another antifungal activity that did not migrate under Proteasome inhibitor the elution conditions (Fig. 3f). Next, we used HPLC-MS

analysis to identify galbonolides A and B from the extracts. Extracted ion chromatograms (EICs) of m/z 381 ([M+H]+ for galbonolide A), m/z 365 ([M+H]+ for galbonolide B), m/z 379 ([M−H]− find more for galbonolide A), and m/z 363 ([M−H]− for galbonolide B) revealed

the presence of galbonolides A and B, at 7.2 and 8.7 min, respectively, in the supernatant extract of WT (Fig. 4a). As expected, SK-galI-5 produced galbonolide B, but not galbonolide A (Fig. 4b). In a separate HPLC experiment, elution fractions were collected at 7.0–8.0 min [fraction (fr.) 1)] and 8.0–9.0 min (fr. 2), concentrated, and applied to the antifungal activity assay after TLC separation (Fig. 4c). The antifungal assay demonstrated that the WT fractions retained the high antifungal activity at an Rf value of approximately 0.25, with higher activity in fr. 1. This activity is clearly absent in the SK-galI-5 fractions. Elution fractions from SK-galI-5 had low activity in fr. 2 at an Rf value of approximately 0.35. Although this activity is too low to be reproducibly observed, the Rf value is comparable to the published value for galbonolide B (Abe et al., 1985). Overall, these experiments demonstrate that SK-galI-5 Thiamet G produces galbonolide B, but does not synthesize galbonolide A. The HPLC-MS analysis with gradient elution further supported that SK-galI-5 lost the ability to synthesize galbonolide A (Fig. S2). The proximity of the KAS-related genes (orf3, 4, and 5) to galGHIJK suggests the possibility that these genes are involved in the biosynthesis of galbonolides. Thus, an orf4-disruption mutant was generated and the genotype

of the resulting mutant was confirmed by Southern analysis using the 1.4-kb EcoRV–BamHI fragment as a probe (Fig. 5a and b). A 3.1-kb PstI–NotI fragment was evident in the WT chromosome and it was replaced by 2.8- and 1.7-kb fragments in two progeny of an orf4-disruption mutant (dKS-6 and -7). The 1.4-kb fragment seen in dKS-6 and -7 likely originated from the disruption plasmid, pSK1-dKS. The antifungal activity assay indicated that dKS strains produced a trace level of galbonolide A, while the production of the unknown antifungal compound (the nonmigrating one in TLC) was slightly reduced (Fig. S3). It is certain that the galbonolide A biosynthesis is severely impaired in the dKS mutant, but it is unclear whether a reduction of the unknown compound is associated with a disruption of orf4 or not.