Biosynthesis of 3-Acetyldeoxynivalenol and Sambucinol

The first two oxygenation steps post-trichodiene in the biosyntheses of the trichothecenes 3-acetyldeoxynivalenol and sambucinol were investigated. The plausible intermediates 2-hydroxytrichodiene (2α- and 2β-) and 12,13-epoxytrichodiene and the dioxygenated compounds 12,13-epoxy-9,10-trichoene-2-ol (2α- and 2β-) were prepared specifically labeled with stable isotopes. They were then fed separately and/or together to Fusarium culmorum cultures, and the derived trichothecenes were isolated, purified, and analyzed. The stable isotopes enable easy localization of the labels in the products by 2H NMR, 13C NMR, and mass spectrometry. We found that 2α-hydroxytrichodiene is the first oxygenated step in the biosynthesis of both 3-acetyldeoxynivalenol and sambucinol. The stereoisomer 2β-hydroxytrichodiene and 12,13-epoxytrichodiene are not biosynthetic intermediates and have not been isolated as metabolites. We also demonstrated that the dioxygenated 12,13-epoxy-9,10-trichoene-2α-ol is a biosynthetic precursor to trichothecenes as had been suggested in a preliminary work. Its stereoisomer was not found in the pathway. A further confirmation of our results was the isolation of both oxygenated trichodiene derivatives 2α-hydroxytrichodiene and 12,13-epoxy-9,10-trichoene-2α-ol as natural metabolites in F. culmorum cultures.

Trichothecenes are toxic secondary metabolites produced by fungi, in particular by Fusarium spp. They infect mostly wheat, grains, and corn and therefore affect human and animal health (1)(2)(3)(4). Fusarium culmorum (HLX-1503) produces two major trichothecene metabolites (5): 3-acetyldeoxynivalenol (3-ADN) 1 and sambucinol (SOL) (Fig. 1). These mycotoxins have been known for a long time (6); however, there is not yet an efficient detoxification procedure. Since trichodiene (TDN; 1, Fig. 1), the biosynthetic precursor to trichothecenes, is not toxic, knowledge of its first oxidized metabolite will enable the design of a potent inhibitor to trichothecenes. We therefore decided to focus on the identification of the oxygenation steps after the hydrocarbon trichodiene. The first oxygenated trichodiene derivatives isolated from F. culmorum were 9,10-trichoene-12,13diol (2d) and 12,13-epoxy-9,10-trichoene-2␣-ol (3a) (Fig. 1) (8). These compounds had been detected by the kinetic pulse-labeling method (7) and could accumulate when the inhibitor ancymidol was used (8). In addition, they had been isolated with a radiolabel after feeding (3RS)-[2-14 C]mevalonate to F. culmorum cultures and isolating and purifying the derived radiolabeled 2d and 3a. These compounds were separately fed to F. culmorum cultures, and the derived trichothecenes 3-ADN and SOL were recovered and analyzed. Metabolite 3a was found to be incorporated in both 3-ADN and SOL, whereas the trichothecenes obtained after the feeding of 2d were unlabeled. This preliminary result suggests that 3a might be a biosynthetic precursor to 3-ADN and SOL, whereas 2d is a dead end metabolite (8). This result is not extremely rigorous, since no degradation could be done on the radiolabeled trichothecenes obtained from the feeding of compound 3a to prove the labeled site. Indeed, there is a possibility that the metabolite 3a was degraded and then resynthesized into the trichothecenes. Since we wanted to determine with certainty the sequence of oxygenation post-trichodiene, our first goal was to determine unambiguously if 3a is the dioxygenated metabolite and then what is the first oxygenated precursor: 2a, 2b, or 2c. The isolated 2d was shown to be a dead end product and probably derives from opening of the epoxide 2c (8).
In this paper, we have rigorously proven, for the first time, the two oxygenation steps of trichodiene and their sequence. A biosynthetic scheme for all of the oxidations leading to 3-ADN and SOL is proposed.

EXPERIMENTAL PROCEDURES
Instrumentation-Analytical high performance liquid chromatography (HPLC) was performed on the same instrument used previously (9) or on a Perkin-Elmer Binary LC pump 250 coupled to a Waters 990 photodiode array detector or on a Waters 600 pump coupled to a Waters 996 photodiode array detector. Preparative and semipreparative HPLC were performed on a Waters Delta Prep 3000 instrument coupled to a Lambda-Max model 481 LC spectrophotometric detector. All ultraviolet detectors were set at 204 nm. Infrared spectra were measured in chloroform with a Perkin-Elmer model 683 infrared spectrophotometer. Flash chromatography was performed on silica gel 60, 230 -400 mesh (EM Science). Thin layer chromatography was conducted on Silica Gel 60 F 254 -precoated TLC plates, 0.25 mm (EM Science). High performance TLC was run with LHP-KF, 0.2-mm plates (Whatman). Radiolabeled compounds were analyzed with a Bioscan Imaging Scanner System 200 (10). A Tracor Analytic Delta 300 instrument was used for liquid scintillation counting. F. culmorum slant cultures were homogenized with a Polytron homogenizer (10).
Strain and Cultivation Conditions-An F. culmorum strain (HLX-1503) was grown as described previously (7). Seed cultures were grown for either 2 or 3 days. Production cultures (25 ml of production medium in a 125-ml Erlenmeyer flask) were incubated for 1, 2, or 3 days.
NMR and Mass Spectrometric Measurements-All of the NMR spectra were obtained at room temperature with a UNITY-500 spectrometer operating at 499.843 MHz for 1 H, 125.697 MHz for 13 C, and 76.735 MHz for 2 H. The NMR spectra were obtained on 1-2 mg (natural product) and 5-10 mg (synthetic compounds) dissolved in 0.3-0.4 ml of deuterochloroform. The nuclear Overhauser effect two-dimensional spectroscopy (NOESY) experiment (hypercomplex phase mode) was obtained using a mixing time of 0.3 s and a relaxation delay of 1 s. For the compounds studied, when the diagonal signals were phased negative, the cross-peaks resulting from dipolar relaxation (nuclear Overhauser effect (NOE)) were phased positive. Fast atom bombardment (FAB) mass spectra were obtained with a V G ZAB-HS double-focusing instrument using a xenon beam having 8 kV energy at 1 mA equivalent current. The ratio between enriched and nonenriched metabolites was obtained by comparing, in the proton NMR, the integration of the doublet derived from the coupling of 13 C-H and the singlet corresponding to 12 C-H. The extent of 13 C (or 2 H) obtained from labeled feedings was determined by correction of the appropriate protonated or sodiated quasimolecular ion clusters present in a computer average of 10 scans in low resolution FAB. The intensity of each ion of the cluster was reduced by the calculated contribution at that mass by natural heavy isotope substitution in all lighter ions of the cluster. The remaining intensities then represent the relative abundances of each labeled species. The percentage of isotope labeling estimated by NMR and by mass spectrometry (MS) agreed reasonably well.
High Performance Liquid Chromatography-Analytical, semipreparative, and preparative HPLC was performed with the same columns as previously recorded (9). HPLC gradient A consisted of a linear gradient lasting 50 min with an initial concentration of 15% methanol, 85% water and a final concentration of 75% methanol, 25% water, which was held for 30 min and then increased linearly for 10 min to 100% methanol. The concentration was held for 20 min at 100% methanol before equilibrating to 15% methanol, 85% water. HPLC gradient B consisted of a linear gradient lasting 50 min with an initial concentration of 40% methanol, 60% water and a final concentration of 100% methanol, which was then held for 20 min. The metabolites isolated from the feedings were purified on analytical HPLC following isocratic conditions at a flow rate of 1 ml/min as follows: for 3-ADN, 35% methanol, 65% water; for SOL, 50% methanol, 50% water; for diacyl-SOL, 62% methanol, 38% water.
Feeding of 12,13-Epoxy-  C]9,10-trichoene-2␣-ol 3a and Purification and Characterization of the Derived 3-ADN, SOL, pre-SOL, and 11␣-2␤,13␤-Apotrichodiol-The 12,13-epoxy-[16-13 C]9,10-trichoene-2␣ol, 38 mg in total, was dissolved in ether and equally distributed among eight sterile 125-ml Erlenmeyer flasks. The ether was allowed to evaporate from the stoppered flasks overnight. To each of the eight flasks was added 0.5 ml of a 5% Brij 35 solution and one 24-h production culture (previously prepared from a 2-day-old seed culture). Three controls were prepared by adding 0.5 ml of a 5% Brij 35 solution to each of three 24-h production cultures. The cultures were incubated for 5 days at 25°C and 220 rpm. After 5 days, the three controls and the eight samples were filtered separately. The filtrates were saturated with NaCl and extracted with ethyl acetate. The organic extracts were dried over MgSO 4 , filtered, and concentrated in vacuo. This crude extract was fractionated by analytical HPLC using gradient A at 1 ml/min. Fraction 1, corresponding to 3-ADN (t R ϭ 38 -43 min) was further purified by HPLC (t R ϭ 39 min). After evaporation, the structure of 3-ADN was confirmed by high resolution mass spectrometry. . Fraction 5, corresponding to pre-SOL (t R ϭ 62-64 min) was further purified by analytical HPLC using 55% methanol, 45% water at 1 ml/min (t R ϭ 64 min). After evaporation, the structure was confirmed by high resolution mass spectrometry. FAB-MS: (M ϩ Na) ϩ : 273.1466; C 15 H 22 O 3 Na ϩ requires 273.1467. The 13 C NMR spectra of pre-SOL show incorporation of 13 C at 23.0 ppm, corresponding to C-16. The percentage of 13 C incorporation calculated by low resolution mass spectrometry was 51.7%. From NMR, we obtained the ratio of 44:56 between 13 C-enriched and 12 C-nonenriched metabolites by comparing the integration peaks of the doublet derived from the coupling of 13 C-H-16 and the singlet 12 C-H-16 at 1.84 ppm (J C-H ϭ 126.0 Hz). Fraction 6 (t R ϭ 64 -69 min), from HPLC gradient A, was acetylated with 145 l of deuterated acetic anhydride and 135 l of pyridine at 25°C overnight. The acetic anhydride and pyridine were evaporated under a stream of nitrogen, and the acetylated fraction was purified by analytical HPLC using 65% methanol, 35% water at 1 ml/min (t R ϭ 56 min). NMR characterization identified this compound as 11␣-2␤,13␤apotrichodiol, which has previously been isolated from cultures of F. culmorum (14). The 13 C NMR of 11␣-2␤,13␤-apotrichodiacetate shows incorporation of 13  tone was allowed to evaporate, and to each of the 11 flasks was added 0.5 ml of a 5% Brij 35 solution and one 48-h production culture (previously prepared from a 3-day-old seed culture). Two controls were prepared by adding 0.5 ml of a 5% Brij 35 solution to each of two 48-h production cultures. The cultures were incubated for 5 days at 25°C and 220 rpm. After 5 days, the two controls and the samples from each individual feeding experiment were filtered separately. The filtrates were saturated with NaCl and extracted with ethyl acetate. The organic extracts were dried over MgSO 4 , filtered, and concentrated in vacuo. The crude extracts were fractionated by analytical HPLC, and 3-ADN and SOL were isolated and purified as described for the feeding of 3a. When 2a or a mixture of 2a and 2c was fed, the 3-ADN structure was confirmed by high resolution mass spectrometry. of 3a) was acetylated by incubation at 25°C with 120 l of acetic anhydride and 90 l of pyridine. The diacyl-SOL was purified by analytical HPLC using 65% methanol, 35% water at a flow rate of 1 ml/min (t R ϭ 25.8 min), and the structure was confirmed by high resolution mass spectrometry. FAB-MS: (M ϩ Na) ϩ : 373.1627; C 19 H 26 O 6 Na ϩ requires 373.1627. The 13 C NMR spectra of the diacyl-SOL derived from the feeding of 2a give the exact site of incorporation at C-16 at 23.0 ppm. The percentage of 13 C incorporation calculated by low resolution mass spectrometry was 34.2%. From NMR, we obtained the ratio of 39:61 between 13 C-enriched and 12 C-nonenriched metabolites by comparing the integration peaks of the doublet derived from the coupling of 13 C-H-16 and the singlet 12 C-H-16 at 1.76 ppm (J C-H ϭ 126.2 Hz). When a mixture of 2a and 2c was fed there was not enough sambucinol to isolate.

Feeding of 2␤-Hydroxy-[16-13 C]trichodiene 2b and Purification and Characterization of the Derived 3-ADN and SOL-The 2␤-hydroxy-[16-
13 C]trichodiene, 50 mg in total, was dissolved in methanol and equally distributed among 10 sterile 125-ml Erlenmeyer flasks. The methanol was allowed to evaporate from the stoppered flasks overnight. To each of the 10 flasks was added 0.1 ml of a 5% Brij 35 solution and one 48-h production culture (previously prepared from a 3-day-old seed culture). Two controls were prepared by adding 0.1 ml of a 5% Brij 35 solution to each of two 48-h production cultures. The cultures were incubated for 5 days at 25°C and 220 rpm. After 5 days, the two controls and the 10 samples were filtered separately. The filtrates were saturated with NaCl and extracted with ethyl acetate. The organic extracts were dried over MgSO 4 , filtered, and concentrated in vacuo. The crude extract was fractionated by semipreparative HPLC, using gradient A, at 3 ml/min. The 3-ADN that was isolated, as described for the feeding of 3a, showed no incorporation of 13 C at C-16 in the 13 C NMR. There was not enough SOL to isolate due to inhibition during the feeding.
Isolation, Purification, and Characterization of 2␣-Hydroxytrichodiene 2a and 12,13-Epoxy-9,10-trichoene-2␣-ol 3a from F. culmorum Cultures-Three-day-old production cultures of F. culmorum were filtered through miracloth (Calbiochem) to remove the mycelia. Each liter of the filtrate (5 liters in total) was extracted with ethyl acetate (3 ϫ 500 ml), the organic layer was washed with a saturated solution of NaCl (3 ϫ 500 ml), dried over anhydrous MgSO 4 , filtered, and evaporated under reduced pressure to give a yellow oil (1.4 g). This crude extract was dissolved in methanol and fractionated on preparative HPLC using program B at 18 ml/min. The region with a retention time between 49 and 60 min was collected. The combined fractions were evaporated to a 10-ml volume under reduced pressure and extracted with pentane (3 ϫ 5 ml). Evaporation of the pentane was done with a stream of nitrogen, and the extract was dissolved in methanol to be repurified by fractionation on semipreparative HPLC using program B at 3 ml/min. The peaks with a retention time corresponding to standard 12,13-epoxy-9,10-trichoene-2␣-ol (3a, t R ϭ 52.5 min) and 2␣-hydroxytrichodiene (2a, t R ϭ 55.8 min) were collected. Each fraction was extracted with pentane, which was then removed under nitrogen. There was enough of metabolite 3a to be characterized by NMR, and indeed its proton NMR was identical to the standard compound synthesized in our laboratory as well as the NMR published from its accumulation in the media with ancymidol (8). In order to rigorously identify the metabolite that had the same retention time on HPLC as 2a, we acetylated it with radiolabeled acetic anhydride. The compound was dissolved in a mixture of 120 l of pyridine and 25 l (12.5 Ci) of [1Ј-14 C]acetic anhydride (100 mCi/mmol). The mixture was incubated for 24 h at 25°C, and then 145 l of unlabeled acetic anhydride was added, and the mixture was incubated for an additional 24 h at 25°C. Following the removal of acetic anhydride and pyridine, methanol was added, and the solution was fractionated by analytical HPLC using 80% methanol and 20% water at a flow rate of 1 ml/min. The peak observed at t R ϭ 56 min was collected and extracted using a ChemElut tube (Extube ® , Varian) with 4 ϫ 10 ml of pentane and evaporated under a stream of nitrogen to 20 l and spotted on a high performance TLC plate. The plate was developed with ethyl acetate/hexane (1:4) and then analyzed with the Bioscan imaging scanner system. One symmetrical radioactive peak was detected with an R F of 0.81. In order to ensure that the compound was pure and was identical to synthesized 2␣-acetyl-trichodienol, it was purified on HPLC using three different conditions of elution at 1 ml/ min, and its retention time coincided with the standard. The conditions were as follows: 80% methanol, 20% water, t R ϭ 56 min; 85% methanol, 15% water, t R ϭ 32 min; and HPLC program B, t R ϭ 61 min.

RESULTS AND DISCUSSION
Putative Post-trichodiene Oxygenated Trichothecene Precursors (Fig. 1)-Putative post-trichodiene oxygenated precursors were synthesized with a stable isotope label ( 13 C or 2 H) at a nonlabile position and fed to F. culmorum cultures. The derived metabolites were analyzed by 13 C or 2 H NMR and mass spectrometry to determine if the precursor was incorporated and the site and extent of incorporation. In order to confirm the preliminary data (8) suggesting that 12,13-epoxy-9,10-trichoene-2␣-ol (3a, Fig. 1) is a post-trichodiene biosynthetic precursor to trichothecenes, we synthesized it with a 13 C label at C-16 (Fig. 2). Its stereoisomer 12,13-epoxy-[16-13 C]9,10-trichoene-2␤-ol (3b, Fig. 2) was also prepared with a 13 C label at C-16. We wanted to determine if both stereoisomers were precursors to trichothecenes, implying a 2-keto-12,13-epoxy-9,10trichoene intermediacy. The syntheses of two of the three plausible monooxygenated trichodiene intermediates (2a and 2b, Figs. 2 and 3B) are closely related to that of 3a and 3b. We therefore decided to synthesize the different stereoisomers of the putative monooxygenated precursors (2a-2c, Figs. 2 and 3) as well as 3a and 3b, specifically labeled with stable isotopes ( 13 C or 2 H). The dioxygenated trichodiene intermediates 3a and 3b (Fig. 2) and the monooxygenated stereoisomers of 2-hydroxytrichodiene (2␣-and 2␤-) were labeled with a 13 C label at C-16 (2a and 2b, Figs. 2 and 3B), while 12,13-epoxytrichodiene was labeled with a deuterium at C-15 (2c, Fig. 3A). We have prepared only the ␤-epoxide (2c), since all natural trichothecenes have this stereochemistry. The rationale for using a different isotope for 2a, 2b, and 2c is to enable the simultaneous feeding of two plausible monooxygenated precursors (2a and 2c; 2b and 2c) under the exact same conditions. The analysis of the derived products will therefore determine the relative incorporations. We confirmed these results by separate feedings of each compound. Below, the synthetic methods for specifically labeling 3a and 3b and 2a, 2b, and 2c will be described.
Syntheses of 12,13-Epoxy-    (Figs. 2 and 3)-This is the first time these labeled compounds were prepared. The 13 C label was introduced via a Wittig reaction of a ketone with Ph 3 P 13 CH 3 I. No unlabeled reagent was added in order to ensure 100% incorporation of 13 C. The 2 H label was derived from reduction of an aldehyde group at C-15 with NaB 2 H 4 , thereby introducing one deuterium in the resulting primary alcohol in the two prochiral hydrogens, prior to reducing it to a methyl group. The reactions involved in the syntheses of 3a, 3b, and 2a are the following. The tricarbonyl [4-methoxy-1-methyl-1-(1methyl-2-methylenecyclopentyl)-(2-5-)-cyclohexa-2,4-dienyl] iron (4) was prepared according to Pearson's method (11,25) and was oxidized using selenium dioxide and a 90% solution of tert-butyl hydroperoxide (16). Compound 5 was obtained with an allylic alcohol at C-2 with a ␤-orientation as confirmed by NOESY experiments. Indeed, the NOE correlations of 5 show a strong interaction between H-2 and Me-15, thereby confirming the existence of a ␤-OH on C-2. Decomplexation of compound 5, using CuCl 2 in ethanol (17) produced enone 6. Subsequent epoxidation of enone 6 using vanadyl acetylacetonate with a 90% solution of tert-butyl hydroperoxide (16) and silylation of the epoxide 7 with t-butyldimethylsilyl chloride produced compound 8. Wittig olefination of compound 8 with Ph 3 P 13 CH 3 I (12) enabled the introduction of the 13 C label at C-16 (trichothecene numbering) to afford compound 9, which was subsequently reduced with sodium in liquid ammonia to give compound 10 (11,25). Desilylation of compound 10 with tetrabutylammo-nium fluoride produced 12,13-epoxy-[16-13 C]9,10-trichoene-2␤ol (3b). This compound was acetylated (Ac-3b), and its NOE correlations determined by the NOESY experiment show a strong interaction between H-2 and Me-15, thereby confirming the existence of a ␤-OH on C-2.
In order to synthesize compound 3a containing the C-2 hydroxy with an ␣ configuration, the allylic alcohol of compound 3b had to be inverted. Alcohol 3b was inverted via its paratoluenesulfonyloxy intermediate 1 with cesium acetate in dimethylformamide (18) to produce acetate 12, which was then hydrolyzed to obtain 12,13-epoxy-     C coupling with no singlet in the middle corresponding to H-16-12 C: 3a is 100% labeled with 13 C at C-16. B, 13 C NMR of the isolated 3-ADN with a large enriched 13 C-16; the natural abundance 13 C-13 is shown to emphasize the incorporation at C-16. C, 13 C NMR of SOL also with a large enriched 13 C-16 peak. D, 13 C NMR of pre-SOL with a large enriched 13 C-16; the natural abundance of 13 C-13 is shown to emphasize the incorporation at C-16. E, 13 C NMR of 11␣-2␤,13␤-apotrichodiacetate with a large enriched 13 C-16; the natural abundance of 13 C-4 is shown to emphasize the incorporation at C-16. 13 C-H-16 and no singlet that would arise from 12 C-H-16. This constitutes a proof that we succeeded in synthesizing 3a practically 100% with 13 C at position 16.
2␤-Hydroxy-[16-13 C]trichodiene 2b was also prepared from enone 6 using the same procedure as for the preparation of compound 3b except that 6 was not epoxidized. The NOE correlations of 2b determined by the NOESY experiment show a strong interaction between H-2 and Me-15, thereby confirming the existence of a ␤-OH on C-2.
On the other hand, the 3-ADN derived from the feeding of 3a under the exact same conditions showed a peak very clearly in the 13 C NMR spectra at C-16 (15.3 ppm) (Fig. 4). Also, the average percentage of 13 C incorporation (ratio between 13 Cenriched and nonenriched 3-ADN) as calculated by mass spectrometry and 1 H NMR data is 38%. Three more compounds isolated from the feeding of 3a showed in the 13 C NMR spectra a large peak at C-16 (Fig. 4): SOL, pre-SOL (a precursor to SOL (21)), and a dead end metabolite, 11␣-2␤,13␤-apotrichodiol, previously isolated from F. culmorum (14). This last metabolite was diacetylated after isolation to facilitate its purification (Fig. 4). The structure of this apotrichothecene was rigorously proven by independent synthesis (14) (Fig. 4). No mass spectral data are available for 11␣-2␤,13␤-apotrichodiol, because it decomposed prior to analysis. We have, however, the detailed 1 H and 13 C NMR characterization, and they are identical to the ones reported except for the enrichment in 13 C. It is interesting to note that the conversion of 3a to these three metabolites  13 C NMR of the derived 3-ADN from both feedings; no deuterated 3-ADN was detected. c, 13 C NMR of the derived SOL from the feeding of 2a.
We also succeeded in isolating 12,13-epoxy-9,10-trichoene-2␣-ol from a large amount of crude extract prepared from F. culmorum. There was enough metabolite to characterize by NMR spectroscopy, and the data were identical to those of the unlabeled synthesized compound.
Biosynthesis of Trichothecenes: Sequence of the Oxygenations Post-trichodiene (Figs. 5 and 6)-The results shown in Fig. 5 prove conclusively for the first time that the sequence of the two oxygenation steps of the hydrocarbon trichodiene are as follows: first hydroxylation at position 2␣, leading to 2␣-hydroxytrichodiene, followed by the epoxidation at C-12-C-13 to give 12,13-epoxy-9,10-trichoene-2␣-ol, which has been successfully incorporated to both 3-ADN and SOL (see above). The feeding of 2␣-hydroxy-[16-13 C]trichodiene 2a to F. culmorum cultures gave 3-ADN and SOL (which was diacetylated in order to purify it) with a clearly enriched C-16 in the 13 C NMR spectra at 15.2 and 23.0 ppm, respectively (Fig. 5). In addition, the incorporation of 2a into 3-ADN and SOL was significant as calculated by mass spectrometry and 1 H NMR data (experimental): ϳ28% into 3-ADN and ϳ37% into SOL. When the two possible monooxygenated trichodiene precursors, 2␣-hydroxy-[16-13 C]trichodiene 2a and [15-2 H]12,13-epoxytrichodiene 2c, were simultaneously fed to F. culmorum cultures, the produc-tion of 3-ADN and SOL was inhibited to the extent that no SOL could be isolated for characterization. The only trichothecene that was produced, 3-ADN, was highly enriched with 13 C at position 16 (Fig. 5). There was no deuterated 3-ADN produced. In order to ensure that the 2␣-hydroxy is absolutely required, we also synthesized the stereoisomer: 2␤-hydroxy-[16-13 C]trichodiene 2b and fed it to F. culmorum cultures. It seems to inhibit considerably the production of trichothecenes. SOL could not be isolated, and the 3-ADN obtained was unlabeled. We are therefore very confident that the first oxygenation step post-trichodiene is the formation of 2␣-hydroxytrichodiene. In addition, we succeeded in finding 2␣-hydroxytrichodiene as a metabolite by radioactive dilution. A large amount of crude extract was prepared from production cultures of F. culmorum. After successive purifications on HPLC, a very small peak was found with the same retention time as unlabeled 2␣-hydroxytrichodiene. The amount was too small to isolate and characterize by NMR. We therefore decided to acetylate that minute quantity with [1-14 C]acetic anhydride and obtained a single peak on HPLC that had all of the radioactivity transferred to the new peak at a retention time identical to that of standard 2␣-acetyltrichodienol. In one experiment, the compound was purified to constant specific activity under three different analytical HPLC conditions that gave peaks with retention times identical to that of the synthetic acetylated standard. A Bioscan tracing of 2␣-[1Ј-14 C]trichodienol acetate demonstrated the symmetrical distribution.
A trioxygenated derivative of trichodiene (4a, Fig. 6) has been isolated and converted to 3-ADN (22,23). The conversion of 4a to sambucinol has never been shown, but it seems that the fungal strain utilized did not produce sambucinol, since it was not reported. We can therefore postulate that 4a is probably the more plausible trioxygenated intermediate of 3-ADN and SOL. The biosynthetic pathway of 3-ADN and SOL seem to bifurcate at an early stage (21). An attractive metabolite that would be converted to both trichothecenes could be 4a. A fourth hydroxylation could give compound 5a, which is a natural product and has been isolated from Fusarium sporotrichiodes (24). Metabolite 5a has a hydroxyl at position 3 with the correct stereochemistry and would be converted to the tricyclic metabolite isotrichodermin (Fig. 6), a proven biosynthetic precursor to 3-ADN (9,21). The conversion of isotrichodermin to 3-ADN involves four oxidations and six metabolic conversions from 3a. 12,13-Epoxytrichothec-9-ene (Fig. 6, EPT) has been converted to pre-SOL and SOL with no connection to the isotrichodermin metabolic pathway (21). Therefore, we can see (Fig. 6) that more steps are involved from 3a to 3-ADN than from 3a to pre-SOL and SOL, which accounts for the relative incorporations. One conversion that seems particularly efficient is the feeding of 12,13-epoxy-[16-13 C]9,10-trichoene-2␣-ol 3a to F. culmorum cultures, which leads to 90% of the 13 C being incorporated into the apotrichothecene 11␣-[16-13 C]2␤,13␤-apotrichodiol (Fig. 6). One possibility could be that the SN2-type water attack on C-2, the cleavage of the ether bond, and opening of the epoxide of 12,13-epoxytrichothec-9-ene could happen simultaneously on the enzyme surface. On the other hand, the conversion of 12,13-epoxytrichothec-9-ene to pre-SOL and SOL is probably more involved. Knowledge of the sequence of the different oxygenations will be very helpful in understanding these ubiquitous enzymes.