INTRODUCTION

AF¹ and ST are related polyketide compounds that are produced by several Aspergillus species (1). These compounds are acutely toxic and carcinogenic and are a serious concern from the human and animal health perspective (2). Biochemical and genetic studies of AF biosynthesis in Aspergillus flavus and Aspergillus parasiticus have culminated in formulation of a biochemical logic that explains the steps involved in this complex pathway (partially depicted in Scheme I) (3). Aspergillus nidulans and many other members of the genus Aspergillus produce ST, the second-to-last intermediate in AF biosynthesis. We have shown that the genes required for ST biosynthesis by A. nidulans are conserved at the functional, regulatory, and physical levels with the AF pathway genes from A. parasiticus and A. flavus (4-6). ST and AF biosynthetic and regulatory genes are clustered in all Aspergillus spp. examined, and to date six out of the 25 genes defined within the A. nidulans ST gene cluster have characterized homologs in A. flavus and A parasiticus (4-6). Mutations in closely related genes from A. nidulans (stcU (7), aflR (8), stcA (9), stcE,² stcF,² and stcJ (10)); A. flavus (aflR (11)), and A. parasiticus (pksA (12)), nor-1 (13, 14), ver-1 (15), fas-1a (fatty acid synthase) (16), apa-2 (17), and avnA (18)) in each case have disrupted (or implicated to disrupt) the ST/AF pathway at the same step. We have also shown that mutations in some other genes stcP (19), stcK (10), and stcS (20) that have so far been studied only in A. nidulans resulted in the accumulation of intermediates that correlate well with the biochemical logic proposed for AF biosynthesis.

A particularly important step in the AF/ST pathway is conversion of VER B to VER A (Scheme I). VER A is derived through desaturation of the bisfuran ring present in VER B. The critical nature of this step comes from the fact that the unsaturated bisfuran ring can be oxidized by a liver specific P-450 monooxygenase to form an epoxide that can then covalently link to DNA (21). It is this property of ST, AFB₁, and AFG₁ that provides their carcinogenic potential. In contrast, tetrahydrobisfuran-containing compounds AFB₂ and AFG₂ (and presumably DHST) have been found to be less carcinogenic in animal models (22). Current understanding of the origin of the unsaturated bisfuran moiety is that VER B is converted to VER A by a desaturase activity (23, 24). Both VER A and VER B are then converted to AFs by common enzymes as depicted in Scheme I (3, 25).

In this communication we demonstrate the desaturase activity required to convert VER B to VER A is most likely encoded by the A. nidulans stcL gene, which encodes a putative P-450 monooxygenase. Disruption of stcL led to loss of ST production and accumulation of DHST. By analyzing the effects of stcL disruption in combination with a mutation in a gene required for VER A metabolism (e.g. stcU), we show that the same enzyme is needed to convert saturated and unsaturated bisfuran intermediates in DHST and ST biosynthesis.

EXPERIMENTAL PROCEDURES

A. nidulans PW1 (biA1; argB2; methG1; veA1) was obtained from FGSC (Fungal Genetics Stock Center, Kansas City, KS). A. nidulans strain TAHK54.17 (biA1; methG1; veA1; stcL::argB) was created in the present study by transformation of A. nidulans strain PW1 with pAHK54 and TAHK78.36 (biA1; methG1; veA1; stcL::argB; stcU::hph) by transformation of TAHK54.17 with pAHK78. Construction of TJK6 (biA1; pabaA1; veA1; stcU::argB) has been described before (stcU was originally named verA) (7). These strains were grown on standard minimal medium supplemented with the appropriate nutrients (26). Cultures were incubated at 37 °C unless otherwise stated.

The nucleotide sequence of stcL is available as a part of the ST gene cluster under the GenBank^TM accession no. U34740[GenBank] (6), and the coordinants of the stcL coding region within the cluster are from 37984 to 36416, with a putative intron from 36820 to 36753. The plasmid pAHK54 was used for the disruption of the genomic stcL open reading frame (Fig. 1B). pAHK47 is a modified pBluescript KS() vector, in which the polylinker EcoRI site was eliminated by a fill-in reaction using standard methods. A 4.8-kb XhoI fragment from the cosmid pL11CO9 (6) was cloned into pAHK47 generating the plasmid pAHK48. An EcoRI fragment from pSalArgB (27) was ligated to EcoRI digested pAHK48 to create pAHK54, which resulted in the disruption of the stcL open reading frame (Fig. 1A). pAHK54 was used to transform A. nidulans strain PW1 using standard methods to obtain arginine prototrophs that were analyzed by Southern blot analyses to identify a stcL disruption strain.

Plasmid pAHK78 (Fig. 1C) was created for insertional disruption of the stcU open reading frame using the hygromycin phosphotransferase (hph) gene. Plasmid pJK1 (7) was digested with EcoRI, and the overhangs were filled-in using standard methods. A SalI fragment from pDH25 (28) containing the hph gene was also treated similarly and then ligated to pJK1 generating the plasmid pAHK78. pAHK78 was used to transform TAHK54.17 (biA1; methG1; veA1; stcL::argB) to obtain hygromycin B (2000 units/plate, Calbiochem-Novabiochem Corp., La Jolla, CA)-resistant transformants using an agar overlay.

ST (Sigma, VER A (prepared as described previously (7)), VER B (generous gift from Dr. K. Yabe, Institute of Animal Health, Japan and Prof. T. Hamasaki, Tottori University, Japan) were from the sources indicated.

Oat flake medium (3 g of oat flakes + 3 ml of water, autoclaved at 121 °C for 20 min) was inoculated with 3 × 10⁸ spores of the appropriate A. nidulans strain, and the cultures were grown at 30 °C for 6 days. Metabolites were extracted using 30 ml of acetone:chloroform (1:1, v/v) mixture. The organic extract was filtered through anhydrous sodium sulfate to remove residual aqueous content and then dried at room temperature. Samples were resuspended in 1 ml of acetone, and 10 µl of the extract was analyzed on TLC plates (silica gel, 250 µm thick, 20 × 20 cm, Analtech) using benzene:acetic acid (95:5, v/v) or toluene:ethyl acetate:acetic acid (80:10:10 v/v) with appropriate standards. Compounds were visualized after spraying the TLC plates with a 20% (w/v) aluminum chloride solution in ethanol (95% v/v) and heating in a 100 °C oven for 5 min (29). Plates were photographed using Mineralight UV-25 UV light (UVP Inc., San Gabriel, CA) and Polaroid 667 instant film (Fig. 2).

Large scale preparation of the metabolites for NMR analyses was achieved by scaling up the oat flake cultures 10-fold followed by separation using preparative TLC (silica gel preparative plates, 1000 µm thick, 20 × 20 cm, Analtech) employing the above solvent systems. The band containing the metabolite of interest was identified by UV light and then scraped off the plates with a spatula. The metabolite was extracted from the scrapings using 5 volumes of acetone. Further purification was achieved by rechromatography on the preparative TLC plates using benzene:acetic acid (95:5, v/v) or toluene:ethyl acetate:acetic acid (80:10:10, v/v) until apparent homogeneity was achieved.

Approximately 5 mg of extract from the large scale preparation of metabolites was isolated and used for NMR analysis. The sample was dissolved in 0.75 ml of deuterated chloroform (Cambridge Isotope Laboratories, Andover, MA) containing 0.01% tetramethylsilane (Aldrich) as an internal reference (0.0 ppm) and was transferred into a 5-mm NMR tube (Wilmad Glass, Buena, NJ). A standard ¹H spectrum was obtained on a Varian Associates Unity Plus-500 NMR spectrometer at a field of 11.75 tesla (Fig. 3). 64K complex data points were acquired at a sweep width of 8000 Hz and transformed without apodization. Homonuclear decoupling and two-dimensional COSY experiments were performed on a Varian Unity Plus-300 spectrometer operating at 7.1 tesla. Single dimensional spectra were acquired with 32K complex points at 4000 Hz sweep width. The two-dimensional correlation spectrum was acquired with 1024 × 256 complex points at 2413 Hz sweep width in both directions (Fig. 4). A control sample was taken from a blank location on the developed TLC purification plates and used to identify ¹H NMR impurities in the spectra.

Metabolite feeding of A. nidulans strains used minimal medium supplemented with the appropriate nutrients (26) and oat flakes (1% w/v). Cultures (50 ml, 10⁵ spores/ml) were incubated at 30 °C for 48 h shaking at 200 rpm. At this point, 100 µg of appropriate metabolite in 50 µl of acetone were added to the culture, and growth was continued for the time indicated. Metabolites were then extracted and analyzed by TLC.

RESULTS

The stcL gene was described previously as encoding a member of the P-450 monooxygenase Family CYP60 (6) and its predicted amino acid sequence is shown in Fig. 1A. To determine if stcL is required for ST biosynthesis, A. nidulans strain PW1 was transformed with the stcL disruption plasmid pAHK54 (Fig. 1B) and the arginine prototrophs recovered were analyzed by Southern blots to identify stcL disruptants. Genomic DNA was isolated from 36 of the transformants, restricted with XhoI, and probed with the ~4.8-kb insert from pAHK48. Eight strains (TAHK 54.5, 54.6, 54.8, 54.11, 54.16, 54.17, 54.29, and 54.31) had the ~6.8-kb XhoI fragment, indicating replacement of genomic stcL by the disrupted copy (data not shown).

stcL mutant strains were grown on oat flake medium, and extracts were analyzed for the production of ST. None of the mutant strains produced ST, but instead accumulated a different orange-red UV fluorescent compound that migrated slightly slower than ST on each of the different solvent systems employed (Fig. 2). HPLC analyses using a C₁₈ column (acetonitrile:water, 80:20, v/v) coupled with photodiode array detection indicated that the UV absorbance spectrum of the compound was also similar to ST (data not shown). One stcL mutant (TAHK54.17) was selected for further studies, and large scale oat flake cultures of this strain were used to isolate the ST-like compound in milligram quantities. This was further purified to apparent homogeneity through additional rounds of preparative TLC (see "Experimental Procedures" for details).

Fig. 3 shows the 500-MHz ¹H NMR spectrum of the isolated metabolite. Previous NMR studies of ST and derivatives compared with our spectra confirm the structure of the metabolite as DHST³ (30, 31). Homonuclear decoupling and two-dimensional correlation spectroscopy were used to determine the chemical shifts and measure scalar couplings of each hydrogen on the molecule. Table I summarizes the assignments of chemical shifts and scalar coupling measurements for DHST. The singlets appearing for the -OH, -OCH₃, and isolated aromatic hydrogen (H5) were easily assigned from peak intensity data. The isolated, scalar coupled system of aromatic hydrogens H7-H9 resulted in well defined classical multiplet patterns and subsequent assignment was straightforward. The six hydrogens H1-H4 on the bifuranyl group were more difficult to assign. Integrated peak intensities suggested that each hydrogen had a different chemical shift except for two at 2.3 ppm. At high field the multiplets at 4.2 ppm were resolved and it was noticed that they were each highly symmetric. The simplicity of these patterns indicated that the two hydrogens ~4.2 ppm were not strongly coupled together and therefore could not be bonded to the same carbon as previously assigned (32). Further analysis was necessary to determine the correct chemical shifts for H1-H4.

Table I. 1H NMR spectral data for DHST Location Chemical shift Type Scalar coupling ppm Hz 1aa 3.69 multiplet 1ba 4.18 multiplet 2a 2.32 multiplet 2b 2.32 multiplet 3 4.23 multiplet 4 6.51 doublet J3,4  = 5.7 5 6.37 singlet OCH3 4.00 singlet 7 6.76 doublet of doublet J7,8  = 8.5 8 7.50 triplet J8,9  = 8.5 9 6.83 doublet of doublet J7,9  = 0.7 OH 13.3 singlet a  Assignment of H1a versus H1b not certain, see text.

A two-dimensional homonuclear correlation analysis, shown in Fig. 4, was performed to trace the arrangement of hydrogens in the DHST structure. The two aromatic hydrogens H7 and H9 were strongly coupled (J = 8.5 Hz) to H8 as evidenced by the cross-peak at A. Weaker coupling between H7 and H9 (J = 0.7 Hz) was not detected in the COSY spectrum. The strong scalar couplings between the six hydrogens of the bifuranyl ring system resulted in numerous cross-peaks describing the sequence of hydrogens around the ring. Starting with H4 at 6.51 ppm, cross-peak B shows a connection to H3 at 4.23 ppm. H3 was also connected through cross-peak D to both H2a and H2b, each with a coincidental resonance at 2.3 ppm. Cross-peak D also reveals a weak coupling (J = ~4 Hz) between H2 and H1b which is not easily seen in Fig. 3. A stronger coupling was found between H2 and H1a at cross-peak E. The cross-peak at C between H1a and H1b was evident as expected for a strongly scalar coupled pair of methylene hydrogens. We were not able to distinguish the chemical shifts between H1a and H1b from the experiments in Figs. 3 and 4.

The finding that an A. nidulans stcL mutant accumulated DHST raised the possibility that A. nidulans, as proposed for A. flavus and A. parasiticus, has a branch in the ST/AF pathway at VER B. If true, then feeding TAHK54.17 with VER A should restore ST production (Scheme I). As expected, when TAHK54.17 cultures were supplemented with VER A, both ST and DHST accumulated by 96 h of growth (data not shown). By contrast, TAHK54.17 cultures fed with NOR, the first stable intermediate in the ST/AF pathway produced only DHST. Finally, when the VER A accumulating stcU mutant strain TJK6 was supplemented with DHST, no ST was produced indicating that stcL is not able to use DHST as a substrate for desaturation (Scheme I). Together these findings indicate that stcL must produce an enzyme activity that is responsible for the conversion of VER B to VER A.

Because common enzymes are thought to be responsible for conversion of VER A and VER B to aflatoxins in A. flavus and A. parasiticus (Scheme I) (25, 33), we examined this possibility in A. nidulans. As previously mentioned, A. nidulans stcU mutants accumulate VER A, indicating that StcU is required to convert VER A to DMST (7). If StcU were also required to convert VER B to DMDHST, we predicted that an stcL stcU double mutant strain should accumulate VER B (Scheme I). A double mutant strain was constructed by transforming TAHK54.17 with pAHK78 (Fig. 1C) to generate hygromycin-resistant colonies. Genomic DNA from 36 of these hygromycin-resistant transformants was isolated and restricted using XbaI and EcoRI. One transformant (TAHK78.36) had the genomic pattern expected if stcU had been disrupted by the hygromycin phosphotransferase (hph) gene (XbaI fragments ~13.5, ~4.5, and ~2.6 kb; EcoRI fragments ~6.8 and ~10.7 kb, data not shown). As shown in Fig. 2, TAHK78.36 did accumulate VER B, but did not produce either VER A or DHST.

DISCUSSION

The mutagenic and carcinogenic properties of ST, AFB₁, and AFG₁ are largely a consequence of the mammalian liver P-450 dependent oxidation of the desaturated bisfuran moiety in these compounds (2, 3). Because elimination of the desaturated bond greatly reduces the carcinogenic nature of these molecules (2), identification of the activity responsible for this biochemical step could provide a target for ST/AF control strategies. The results presented here provide genetic evidence that the putative P-450 monooxygenase encoded by stcL is responsible for desaturation of the bisfuran in converting VER B to VER A. As shown in Figs. 2, 3, 4, mutation of the stcL gene results in the utilization of an alternate branch of the ST pathway resulting in synthesis of DHST. Analogous to the preferential production of AFB₁ and AFG₁ in the AF biosynthetic pathway (Scheme I), the branch leading to ST synthesis seems to be the preferred route for the metabolism of the intermediates as DHST is not reported to be accumulated by wild-type cultures of A. nidulans. The molecular properties of ST and DHST are so similar that separation of these compounds by TLC or HPLC is subtle and large quantities of ST could easily hide small amounts of DHST. The lack of detectable DHST may also be a reflection of the high affinity of StcL for VER B that results in preferential accumulation of only ST in A. nidulans. In contrast, Aspergillus versicolor is reported to be capable of accumulating minor amounts of DHST (31) and A. flavus and A. parasiticus elaborate constant, but minor, levels of AFB₂ or AFG₂ (3).

Another enzyme that could be important in the disproportional shunting of intermediates down the two branches of the ST/AF pathway is the ketoreductase, StcU. Keller et al. (7) had previously shown that stcU mutants accumulated VER A and postulated StcU to be the first enzyme required in converting VER A to DMST, which requires among other activities, a Baeyer-Villiger oxidation resulting in the formation of a xanthone from an anthraquinone (3, 7). The stcL stcU double mutant specifically accumulated VER B indicating that StcU, and probably StcS and StcP, must be active in both branches of the ST/DHST pathway in A. nidulans (Scheme I). Our observation that liquid shake cultures of the stcL mutant accumulated detectable amounts of VER B in addition to DHST (Fig. 2) can be attributed to inefficient processing of the saturated bisfuran-containing intermediates, creating a bottleneck at the branch point of the ST/DHST biosynthetic pathway. A similar biochemical observation was made with extracts from A. parasiticus in which common enzymes (presumably homologs of StcU, StcS, and StcP) predicted to function in the metabolism of VER A and VER B reportedly had much lower affinities for the tetrahydro-bisfuran ring containing intermediates including DMDHST, DHST, and OMDHST (33).

Our results are consistent with those of McGuire et al. (34) who proposed that VER B was an intermediate in VER A formation and that VER B served as the branch point between saturated and desaturated bisfuran containing intermediates in the AF pathway. Yabe et al. (23) subsequently described a NADPH-dependent desaturase activity that could convert VER B to VER A, and suggested that this enzyme determines the bifurcation in A. parasiticus AF biosynthesis. The predicted amino acid sequence of StcL contains the hallmark heme (35) binding motif of P-450 monooxygenases (Fig. 1A), which are known to carry out diverse NADPH-dependent reactions including dehydrogenation (3, 36). Because the enzyme activities encoded in the ST gene cluster in A. nidulans appear to be conserved in A. flavus and A. parasiticus, we expect that mutations in the stcL homolog (most likely ord1) (5) in A. flavus and A. parasiticus would result in strains that preferentially accumulate AFB₂. In fact, A. flavus mutants that accumulate only AFB₂ have been described (37, 38), and we predict that these strains likely have mutations in the A. flavus StcL homolog.

Finally StcL represents one of the five unique monooxygenases that are encoded by the ST gene cluster in A. nidulans (stcB, stcF, stcS, stcW). Keller et al. (7) previously showed that stcS is required for ST biosynthesis and that stcS mutants are blocked between VER A and DMST. We have evidence that stcF and stcW are also required for ST biosynthesis and mutants in each gene accumulate as yet uncharacterized intermediates.⁴ Investigations into the specificity of these monooxygenases may ultimately provide insights into their function in Aspergillus AF/ST biosynthesis as well as their versatility in metabolic conversions of natural products in general.