Formal Synthesis of (±)-Platensimycin

Abstract

Reductive alkylation of 5-methoxy-1-tetralone (6) with 2,3-dibromopropene gave an equilibrium mixture of bicyclic diones 7 (51%) and 8 (35%). Radical cyclization of 7 afforded tricyclic dione 5 (84%), which was reduced, cyclized and dehydrated to give tetracyclic alkene 13 in 63% yield. Allylic oxidation of 13 with SeO₂ and activated MnO₂ afforded enone 2 in 85% yield, thereby completing a short formal synthesis of (±)-platensimycin.

The broad spectrum antibiotic platensimycin (1) (see Scheme 1) was recently isolated by a Merck group from Streptomyces platensis as part of a screening program designed to isolate inhibitors of bacterial fatty acid biosynthesis by the highly conserved condensing enzyme FabF.¹ Only the weak antibiotics cerulenin and thiolactomycin were known to act by this mechanism. Potent inhibitors of this enzyme are expected to be antibiotics with no cross-resistance to existing drugs. Platensimycin acts by specific binding with the acyl-enzyme intermediate of FabF. The structure and absolute stereochemistry of platensimycin were determined by a combination of spectroscopic methods and X-ray crystallography of a bromo derivative.¹

Scheme 1.

Retrosynthesis of Platensimycin

We thought that the acyl portion of platensimycin should be readily accessible by introduction of a methyl group and a propanoic acid side chain onto enone 2. Nicolaou recently reported the first synthesis of platensimycin (1) in which he prepared 2 in 10 steps and elaborated it to (±)-platensimycin (1).² We planned to prepare 2 by dehydration of the alcohol of 3 and allylic oxidation. Acid-catalyzed cyclization of unsaturated diol 4 should afford the ether linkage of 3. L-Selectride reduction of dione 5 should provide the bis axial alcohol 4 (see Scheme 1).

This approach was attractive because Marinovic reported a two-step synthesis of dione 5 in 1983.³ Reductive alkylation of 5-methoxy-1-tetralone (6) with 2,3-dibromopropene by Narisada’s procedure⁴ afforded bicyclic diones 7 and 8 in 68% yield with unspecified stereochemistry (see Scheme 2). Radical cyclization of this mixture of 7 and 8 with n-Bu₃SnH in benzene at reflux afforded the tricyclic diones 5 and 9 in 85% yield, again with unspecified stereochemistry.

Scheme 2.

Synthesis of Tricyclic Diones 5 and 9

Although this route is very short, it is only attractive if the desired tricyclic dione 5 can be prepared cleanly and in good yield. Unfortunately, molecular mechanics calculations⁵ suggested that the desired tricyclic dione 5 is 1.6 kcal/mol less stable than epimeric dione 9. However, calculations also suggested that the desired bicyclic dione 7 is 0.1 kcal/mol more stable than epimeric dione 8. Therefore, it might be possible to isolate 7 in acceptable yield and convert it to 5 if the radical cyclization can be carried out without epimerization.

In our hands, the reduction of 6 was best carried out with potassium in NH₃/Et₂O at -78 °C.⁶ Addition of LiBr and then 2,3-dibromopropene effected alkylation. Hydrolysis of the enol ether with conc HCl in THF for 30 min afforded a readily separable mixture from which the desired bicyclic dione 7 was isolated 51% yield and the epimer 8 was obtained in 35% yield. The structures of 7 and 8 could not be assigned at this point so both compounds were carried on. The HCl hydrolysis step produced close to an equilibrium mixture. Acid catalyzed equilibration of either 7 or 8 provided a 4:3 mixture of 7 and 8. Equilibration of 8 afforded additional 7 (19%), which was therefore isolated in 70% overall yield from 5-methoxy-1-tetralone (6).

Radical cyclization of 7 with n-Bu₃SnH and catalytic AIBN in benzene at reflux afforded the desired tricyclic dione 5 in 84% yield without any epimerization. This practical two-step route to 5 proceeds in 59% overall yield. A similar sequence converted the undesired bicyclic dione 8 to tricyclic dione 9 in 81% yield. Equilibration of either 5 or 9 with KOH in MeOH gave a 1:4 mixture of 5 and the more stable tricyclic dione 9. The equilibration of both bicyclic diones 7 and 8 and tricyclic diones 5 and 9 thus gave results close to those expected from molecular mechanics calculations.

The ¹H NMR spectra of 5 and 7-9 were hard to analyze because of extensive overlap. Fortunately, all the hydrogens of 5 could be resolved in C₆D₆ at 800 MHz and the stereochemistry of 5 was tentatively assigned based on an NOE between the ring fusion hydrogen and one of the allylic methylene hydrogens (see Figure 1). The stereochemical assignments of 5 and 7-9 were unam-biguously established by X-ray crystal structure determination of both tricyclic diones 5 and 9 (see Figure 1).

Figure 1.

Three Dimensional Representations and Molecular Structures of 5 and 9 Established by X-ray Structure Determination.

L-Selectride reduction of the unhindered ketone of 5 occurred readily at -78 °C, but gave a 1:1 mixture of equatorial and axial alcohols. The other ketone was reduced at 25 °C affording a 12:1 mixture favoring the desired axial alcohol. This resulted in the formation of an inseparable 1:1 mixture of 4 and 11 in 90% yield (see Scheme 3).⁷ This mixture was treated with TFA and CH₂Cl₂ to effect formation of the ether linkage as described by Nicolaou² for a related substrate with different functionality in the isolated ring. This two-step sequence afforded axial alcohol 3 in 39% yield and the equatorial alcohol 12 in 42% yield from the mixture of diols. Treatment of the axial alcohol 3 with Tf₂O and pyridine in CH₂Cl₂ afforded the triflate, which eliminated readily to give alkene 13 in 90% yield. Similar treatment of the equatorial alcohol 12 afforded triflate 14, which did not undergo E2 elimination readily because there is no β-hydrogen anti to the triflate.⁸ Eventually, we found that treatment of crude triflate 14 with either silica gel or hydrochloric acid resulted in clean elimination, possibly by an E1 mechanism, to form alkene 13 in 84% yield from 12.⁹ This three-step sequence converts tricyclic dione 5 to tetracyclic alkene 13 in 63% overall yield, but diols 4 and 11 cannot be characterized.

Scheme 3.

Reduction, Cyclization and Dehydration of 5

Alternatively, reduction of dione 5 with NaBH₄ in EtOH at -78 to 0 °C afforded equatorial keto alcohol 10. Reduction of 10 with L-Selectride afforded diol 11 contaminated with a few percent of the equatorial alcohol⁷ in 83% yield. Acid catalyzed cyclization afforded 12 (90%), which was elaborated to 13 as previously described. This four-step sequence converts tricyclic dione 5 to tetracyclic alkene 13 in 55% overall yield.

Allylic oxidation of alkene 13 with CrO₃·3,5-dimethylpyrazole¹⁰ in CH₂Cl₂ at -25 °C provided an inseparable 4:1 mixture of the desired enone 2 and the regioisomer 15 in 75% yield (see Scheme 4). Oxidation of 13 with CrO₃·pyridine was slower, but gave the same ratio of products. The formation of mixtures of products was expected¹¹ because both ends of the intermediate allylic cation are secondary.

Scheme 4.

Allylic Oxidation of 13 to Give Enone 2

We therefore turned to SeO₂ oxidation, which should be regiospecific because the oxygen is introduced by an ene reaction followed by a [2,3]-sigmatropic rearrangement.¹² Oxidation of 13 with 8 equiv of SeO₂ in dioxane at 140 °C in a microwave reactor for 30 min afforded allylic alcohol 17 in only 27% yield. The major product was dienone 16 (59%). Oxidation of alkene 13 with SeO₂ should give allylic alcohol 17 and enone 2, which is apparently oxidized further to dienone 16 as has been observed in related systems.¹³ Oxidation of alkene 13 with only 3 equiv of SeO₂ in dioxane at 110 °C in a microwave reactor for only 10 min afforded allylic alcohol 17 in 83% yield and enone 2 in 7% yield. Oxidation of alcohol 17 with activated MnO₂ provided enone 2 in 94% yield. This two-step sequence converts alkene 13 to enone 2 in 85% yield. The ¹H and ¹³C NMR spectra of 2 are identical to those reported by Nicolaou.

In conclusion, we have developed an efficient route (seven steps, 32% overall yield) to tetracyclic enone 2, a late intermediate in Nicolaou’s (±)-platensimycin (1) synthesis.