Asymmetric Synthesis of the Core of AMPTD, the Key Amino Acid of Microsclerodermins F-I

Abstract

We report a stereoselective synthesis of the five consecutive stereocenters of AMPTD in seven steps. Highlights include an Evans glycolate aldol reaction, the use of a Weinreb amide as an aldehyde masking group, and a Mannich reaction with an Ellman-type chiral sulfimine.

Microsclerodermins F–I (1-4, Figure 1), members of a family of antitumor and antifungal cyclic peptides, were isolated in 2000 from a deep-water species of the marine sponge Microscleroderma.¹ A biological assay of 1 revealed cytotoxicity against the human carcinoma cell line HCT-116 (IC₅₀ = 2.7 µM for 2) and antifungal activity against Candida albicans. The biological activity of these molecules, coupled with their poor availability from natural sources (0.001 wt% from the sponge), attracted our synthetic interest.

Figure 1.

Structures of microsclerodermins F–I (1–4).

Microsclerodermins F and H contain a D-tryptophan residue, while G and I feature a dehydrotryptophan residue. The molecules offer several synthetic challenges, including a pyrrolidinone with a β-amido hemiaminal and a γ-amino-β-hydroxy amino acid; however, the most intimidating is 3-amino-6-methyl-12-phenyl-2,4,5-trihydroxydodeca-7,9,11-trienoic acid (AMPTD, 5, Figure 2). AMPTD is a polyhydroxylated β-amino acid with five contiguous stereogenic centers and an all-trans phenyltriene; 3 and 4 also feature a methyl substituent on the triene portion. Shioiri reported a synthesis of AMMTD, a constituent of microsclerodermins A and B with the same stereochemistry as AMPTD but a different side chain, in seventeen steps without the side chain² and thirty-one steps with the side chain.³ We report herein a concise synthesis of the stereochemical core of 5 adaptable to the synthesis of 1–4.

Figure 2.

Retrosynthetic analysis of 5.

Our retrosynthesis of the protected AMPTD moiety 5 introduces four stereogenic centers via aldol methodology (Figure 2). We envisioned preparation of 5 by condensation of the known diene phosphonate 6⁴ and aldehyde 7, prepared by replacement of the chiral auxiliary in 8 with the methyl ester, followed by a deprotection-oxidation sequence. The amino alcohol 8 would come from a syn-selective aldol reaction between a chiral glycolate equivalent and imine 9, which in turn could be prepared from alcohol 10 by protection of the 1,2-diol and conversion of the chiral auxiliary to the corresponding imine. The alcohol 10 would come from another syn-selective aldol reaction between a chiral glycolate and an α-chiral aldehyde 11, prepared in two steps from the commercially available Roche ester.

We first explored aldol reaction of the cis-lactone 13; Andrus had previously synthesized the trans diastereomer.⁵ We found that treatment of the open form 12 with TFA in the presence of triethylsilane as a cation scavenger yielded cis O-lactone 13 in much higher yield than our previously reported two-step procedure from 12.⁶ Aldol reaction of 13 with aldehyde 14 (obtained by reduction of the corresponding ester in situ to avoid epimerization of the methyl group upon standing)² gave the desired adduct 15 in good yield; unfortunately, we obtained 15 as a 1:1 mixture of diastereomers (Scheme 1). Our attempts to remove the biphenyl failed to yield diol 16, and we were thus unable to ascertain the stereochemistry of the product mixture; we anticipate that the use of higher hydrogen pressure would allow template removal. Studies towards the use of 13 as a template for the synthesis of chiral α-substituted-α-hydroxyacids (via alkylation) and β-substituted-α,β-dihydroxyacids (via aldol reaction) are ongoing.

Scheme 1.

Application of O-lactone 13.

We next probed aldol reactions of Andrus’ modified Masamune norephedrine template 17 (Scheme 2)⁷ with 14. Compound 17 gave a moderate yield of addition product 18a, but attempted benzylation of the secondary alcohol returned 18a unchanged. Basic hydrolysis of the chiral template removed the TBDPS ether to give alcohol 19.

Scheme 2.

Application of chiral template 17.

Crimmins’ oxazolidinethione 20 (Scheme 3)⁸ gave a moderate yield of and diastereoselectivity for the corresponding aldol product 21a. We wished to produce the diol and form the acetonide; unfortunately, the sulfur prevented hydrogenolysis to 21b. Removal of the chiral template gave diol 22a quantitatively, but hydrogenation again failed, presumably due to traces of sulfur contamination. Birch debenzylation was not pursued, as the resultant triol would be difficult to purify from the reaction.

Scheme 3.

Application of chiral template 20.

We next turned to Evans’ chiral oxazolidinone 23a,⁹ which underwent syn-aldol reaction with 14 to give adduct 24. The reaction was not clean at room temperature, but we obtained a single diastereomer of 24 at 0° C (Scheme 4). Hydrogenation of the benzyl ether was blocked by the bulky oxazolidinone and TBDPS group; thus we converted the oxazolidinone to Weinreb amide 25. With the steric hindrance reduced, the diol was obtained and protected as acetonide 26. Reduction of the Weinreb amide to the aldehyde and Wittig reaction produced trans-α,β-unsaturated ester 27; attempted Sharpless amino-hydroxylation of 27 (in analogy with a previous synthesis of ours)¹⁰ failed, yielding not the desired amino alcohol (28) but only loss of the TBDPS residue.

Scheme 4.

Synthesis and attempted aminohydroxylation of 27.

We then switched protecting groups: aldol reaction of the corresponding para-methoxybenzyl (PMB) glycolate 23b and benzyl-protected aldehyde 29 gave adduct 30 in moderate yield, and conversion to the Weinreb amide 31 allowed direct oxidation to the p-methoxyphenyl acetal 32 (Scheme 5). While we preferred this protecting-group scheme, both the aldol reaction and acetal formation proved to be capricious, so we abandoned this route.

Scheme 5.

Use of alternate protecting groups to give 32.

Returning to our original route, we converted acetonide 26 via reduction and condensation to the chiral sulfimine 33 (Scheme 6).¹¹ Addition of Boc-protected methyl glycolate gave ester 34, bearing all of the stereogenic centers of AMPTD. While the published work did not discuss deprotection of the adduct, we discovered that treatment of 34 with a dry solution of HCl in dioxane allowed selective removal of the sulfimine group to yield amine salt 35. TBAF-mediated removal of the TBDPS group from 34 also proceeded smoothly to give 36; conversion of the alcohol to bromide 37 was unsuccessful under various conditions, presumably due to deprotection of the acetonide by HBr formed in the course of the reaction. (Our original retrosynthesis envisioned conversion of 37 to a phosphonate and reaction with the corresponding aldehyde analogue of 6.) Studies on the oxidation of 36 to 7 and reaction with 6 are ongoing.

Scheme 6.

Completion of the five chiral centers and deprotection.

In summary, we have synthesized 34, containing the stereochemical core of AMPTD, in seven steps from known starting materials. We are exploring use of the alcohol of 36 as a key intermediate for the total synthesis of the microsclerodermins. We have also shown that the protecting groups can be selectively manipulated, thus allowing us to explore structure-activity relationships of the phenyltriene side chain in microsclerodermins F–I. Studies toward the total synthesis of 1–4 are ongoing in our laboratories.