|Home | About | Journals | Submit | Contact Us | Français|
An asymmetric Diels–Alder reaction between acrolein and 1-benzyloxymethyl-1,3-cyclohexadiene affords a bicyclic aldehyde that was elaborated in 11 steps to nor-platencin. nor-Platencin is 4–16 times less active than platencin against several resistant strains of Staphylococcus aureus, macrolide-resistant Enterococcus faecalis, and vancomycin-resistant Enterococcus faecium.
Scientists at Merck recently isolated the novel antibiotics platensimycin (1)1,2 and platencin (2)3 (see Figure 1). These compounds are the first potent inhibitors of bacterial fatty acid biosynthesis (Fab), which is needed for the survival of bacterial pathogens. This pathway is highly conserved among bacteria and is distinct from the mammalian pathway. Platensimycin blocks the fatty acid condensing enzyme β-ketoacyl-[acyl carrier protein] synthase II (FabF) selectively, whereas platencin inhibits both enzymes FabF and β-ketoacyl-[acyl carrier protein] synthase III (FabH) with IC50 values of 1.95 and 3.91 µg/mL in Staphylococcus aureus. These antibiotics do not exhibit cross-resistance to key antibiotic resistant strains, including methicillin-resistant S. aureus, vancomycin-intermediate S. aureus and vancomycin-resistant Enterococcus faecium. Platencin shows potent in vivo efficacy without any observed toxicity.
As expected from the novel structures and potent antibacterial activity, both platensimycin (1)4,5 and platencin (2)6 have been the object of intense synthetic interest. Numerous analogues of platensimycin have also been prepared, establishing that the aromatic moiety is crucial for activity, but that there is some room for structural modification of the tricyclic diterpenoid moiety.7 No close analogues of platencin have been prepared. We speculated that the exo-methylene group of platencin is not needed for biological activity, but is simply present because it is a structural feature of the terpenoid precursor. The exo-methylene group is acid sensitive and may also decrease the metabolic stability of platencin. Finally, the methylene group complicates the synthesis by introducing both reactive functionality and an additional chiral center at the carbon marked by an asterisk. Our goal therefore was to develop a short synthesis of nor-platencin (3) which lacks the exo-methylene group of platencin (2).
Our retrosynthesis is outlined in Scheme 1. We planned to prepare nor-platencin (3) from enone 4, R = H or Me, using methods developed for the synthesis of platencin (2). Enone 4 will be prepared by an intramolecular aldol reaction of keto aldehyde 5, R = H or Me, which can be prepared by homologation of Diels–Alder adduct 6, which will be synthesized from the readily available 1,3-cyclohexadienemethanol (7a, R = H)8 and acrolein (8a, R = H) or methyl vinyl ketone (8b, R = Me). Use of methyl vinyl ketone would introduce the methyl group early in the synthesis. Use of acrolein would require methylation of 4a, as in the platencin syntheses.7
The Diels–Alder reaction of 7a, R1 = H, with methyl vinyl ketone (8b) proceeded poorly either thermally or with Lewis acid catalysis, but gave a reasonable yield of racemic Diels–Alder adduct 6a and stereo- and regioisomers by reaction "on water".9 Unfortunately, the Diels–Alder adducts were difficult to work with because they exist as a mixture of open and hemiketal tautomers. We were able to prepare 6b, R1 = TBS and R2 = Me, but all attempts to homologate this by a Wittig reaction resulted in enolization of the hindered methyl ketone. For these reasons we turned to acrolein (8a) as the dienophile and readily available 1,3-cyclohexadienylmethyl benzyl ether (7b)10 as the diene. The protecting group will prevent formation of hemiacetals and will be removed without an additional step during hydrogenation of the double bond. Enolization should not occur during homologation of 6c because the carbonyl group is an aldehyde rather than a methyl ketone.
In the reaction of 7b with acrolein (8a) as the dienophile, we were able to take advantage of MacMillan's asymmetric Diels–Alder reaction using 10% of imidazolone 911 as the catalyst in 19:1 CH3CN/H2O for 5 days (see Scheme 2). This afforded a 9:1 mixture of the desired endo adduct 6c and exo adduct 10, from which 6c (32%) and 10 (4%) were isolated in pure form along with an additional 10% of impure 6c. Other conditions, including MeOH/H2O and CH3NO2/H2O, were less successful. Chiral HPLC established that the ee of the major product 6c is 87%. Our yield does not compare favorably with that reported by MacMillan for the reaction of 1,3-cyclohexadiene with 8a catalyzed by 5% of 9 which gave a 14:1 mixture of endo and exo isomers in 82% yield with 94% ee for the endo isomer.11a However, in his synthesis of hapalaindole Q, Kerr carefully optimized the Diels–Alder reaction of 1,3-dimethyl-1,3-cyclohexadiene with 3-(3-(N-tosyl)in-dolyl)acrolein catalyzed by 40% of 9 to obtain a maximum yield of 35% with 85:15 endo/exo selectivity and 93% ee for the endo isomer.11c As Kerr also noted in his synthesis,11c the rapid and enantioselective assembly of the key intermediate 6c makes this route attractive despite the modest yield.
The lower ee in our case could result from an uncatalyzed background reaction. Reaction of 7b and 8a for 5 days without catalyst 9 afforded Diels–Alder adducts in 10% yield. The background reaction will be much less significant in the presence of catalyst 9 because the diene is consumed in the more rapid catalyzed reaction. The ee can probably be improved somewhat by increasing the catalyst loading. Equilibration of both 6c and 10 with aqueous NaOH in EtOH afforded the identical 3:1 mixture of 6c and 10. The catalyst controls the stereochemistry adjacent to the aldehyde center so that epimerization of exo adduct 10 will give ent-6c, providing another possible explanation for the lower ee. However, the isolation of a 9:1 mixture of endo isomer 6c and exo isomer 10 suggests that epimerization is not a major issue.
Homologation of 6c by a Horner–Wittig reaction12 now proceeded smoothly to give 11 as a 3:2 mixture of stereoisomers in 67% yield (see Scheme 3). Hydrolysis of 11 in a two phase system13 with aqueous HCl in CH2Cl2 and THF provided the homologated aldehyde 12 in 90% yield. Addition of MeMgBr to 12 afforded secondary alcohol 13 in 99% yield as a mixture of stereoisomers. Buffered PCC oxidation of 13 gave ketone 14 in >90% yield. To our surprise, hydrogenation of 14 afforded mainly tetrahydropyran 15. Hydrogenation and hydrogenolysis provided the desired saturated hydroxy ketone which cyclized to form the hemiketal. Further reduction, probably by dehydration to the enol ether and hydrogenation, afforded 15.
We therefore decided to carry out the hydrogenation and hydrogenolysis before formation of the ketone. Hydrogenation of 13 afforded a mixture of 1,5-diols 16 in 77% yield. Swern oxidation appears to be the method of choice for the oxidation of 1,5-diols to keto aldehydes,14 because other procedures give the δ-lactone as a byproduct.15 Swern oxidation of 16 using i-Pr2EtN,16 rather than Et3N, gave the unstable keto aldehyde 5a, which was treated with NaOH in EtOH to give the desired cyclohexenone 4a17 in 70% overall yield from 16.
Alkylation of 4a with LHMDS and MeI in 10:1 THF/HMPA afforded 87% of 4b containing a few percent of the stereoisomer (see Scheme 4).6b Elaboration of 4b to carboxylic acid 18 was carried out using procedures developed by Corey for the preparation of a platensimycin analogue and by Mulzer for the synthesis of platencin.6h,7b Treatment of 4b with t-BuOK and methyl acrylate in ether/t-BuOH for 30 minutes at 0 °C provided a 5:1 mixture of ester 17 and the diastereomer from which the desired isomer 17 was isolated in 39% yield by preparative HPLC. Hydrolysis of 17 afforded acid 18 in 71% yield. Amide formation was accomplished by reaction of acid 18 with protected aniline 195d,6a,18 using HATU (O-(7-azabenzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate) in DMF for 38 h to give trimethylsilylethyl ester 20 (58%), which was hydrolyzed with TASF in DMF to provide (–) nor-platencin (3) in 56% yield.
The antibiotic acitivity of nor-platencin (3) was deter-mined against several resistant strains of Staphylococcus aureus, macrolide-resistant Enterococcus faecalis, and vancomycin-resistant Enterococcus faecium. The minimum inhibitory concentration (MIC) data shown in Table 1 indicate that nor-platencin (3) is 4–16 times less potent than platencin (2) as an antibiotic. As with platencin (2), the best activity is shown against vancomycin-resistant Enterococcus faecium. The reduced potency of 3 indicates that the exo methylene group of platencin (2) contributes modestly to the antibiotic activity.
In conclusion, an asymmetric Diels–Alder reaction between acrolein (8a) and 1-benzyloxymethyl-1,3-cyclohexadiene (7b) affords bicyclic aldehyde 6c that was elaborated in 11 steps to nor-platencin (3). nor-Platencin (3) is 4–16 times less potent than platencin (2) against several bacterial strains indicating that the exo methylene group of platencin (2) contributes modestly to the antibiotic activity.
We are grateful to the National Institutes of Health (GM-50151) for support of this work.
Supporting Information Available: Complete experimental procedures, copies of 1H and 13C NMR spectral data. This material is available free of charge via the Internet at http://pubs.acs.org.