|Home | About | Journals | Submit | Contact Us | Français|
A total synthesis of (±)-phomactin A is described to highlight the final completion of a complex natural product target that had commenced with an intramolecular oxa-[3 + 3] annulation strategy in the construction of the ABD-tricycle. These efforts reveal structural intricacies of this ABD-tricycle with an illustrative example being the conformational analysis that was ultimately critical for the C5a-homolgation.
In the last decade, phomactin A,1–4 a structurally unique natural product isolated from the culture filtrate of a parasitic fungus Phoma sp. [SANK 11486] found on the shell of Chinoecetes opilio, has captured an impressive array of synthetic efforts.5,6 Although possessing only modest inhibition against platelet-activating factor7 [PAF] induced platelet aggregation [IC50 = 10 μM], phomactin A embodies a new class of PAF antagonists. The most active member is D and was synthesized by Yamada.8 Phomactin A is the only known tetracycle [discounting epoxides] in the phomactin family with all other members lacking either the B-ring or C-ring. Thus, phomactin A represents structurally the most complex member. To date, two elegant total syntheses of (±)- and (+)-phomactin A were completed by Pattenden9 and Halcomb,10 respectively. Recently, Wulff11 reported the synthesis of phomactin B2.
We approached12 (±)-phomactin A with an intent to feature our intramolecular oxa-[3 + 3] annulation strategy13–16 en route to ABD-tricycle 2, which possesses a unique structural topology [Scheme 1].17 We recently designed a 12-step asymmetric synthesis of the annulation precursor 318 entailing Suzuki-Miyaura coupling of vinyl bromide 4 with the derivative of A-ring 5, which was assembled from an asymmetric Diels-Alder cycloaddition19,20 of 6. We communicate here our total synthesis of (±)-phomactin A.
An immense amount of effort17 was exerted to succeed in oxidizing the C3-3a olefin of ABD-tricycle 2 into its proper oxidation states. Ultimately only a singlet-oxygen Diels-Alder cycloaddition could be achieved selectively to give endo-peroxide 821 [Scheme 2] without significant competition from [2 + 2] cycloaddition or ene reaction with the C3′-4′ olefin in D-ring or on the “belt.” After failing an array of reductive protocols [i.e., Lindlar’s [H], thiourea, or Ph3P] to cleave the weak endo-peroxide bond, KOAc and 18-c-6 successfully opened the endo-peroxide bridge via a deprotonation pathway to give ene-dione 9. Treatment of 9 with p-TsOH in MeOH isomerized the lactol motif to methyl ether 10, proceeding through a vinyl oxocarbenium intermediate that was trapped by MeOH at C3a position.
We recognized that while the singlet-oxygen Diels-Alder cycloaddition sets up the desired stereochemistry for the C3-OH group, ring-opening of the endo-peroxide bond through the deprotonation pathway effectively destroyed this valuable stereochemical information. Consequently, with the knowledge of the “belt” blocking the bottom face, we chose a small hydride source such as NaBH4 to reduce the C3 ketone in 10, but only to attain a mixture of isomers with a 4:1 ratio in favor of the wrong alcohol diastereomer 11-α. Surprisingly, when using L-Selectride™, we isolated only the desired isomer 11-β. In hindsight, by examining the model of 10 [Figure 1– left side] , it would appear that the pseudo-axial C3a-OMe group likely plays a bigger role than the “belt” in the facial differentiation of the reduction. With a more bulky hydride, the C3a-OMe group was able to better prevent the hydride approaching from the top face.
Reduction at C8a was relatively less eventful. As shown in Scheme 3, capping of C3-OH in 11-β with TESCl followed by demethylation with BBr3 led to vinylogous ester 12. However, no condition that we screened [i.e., L-Selectride™, CuI/LAH, or Na/IPA] was capable of reducing the vinylogous ester motif in a 1,4-manner. Realizing that vinylogous ester 12 may not be sufficiently electron deficient, we oxidized C3a-OH using Dess-Martin periodinane reagent, and an ensuing reduction effectively gave diketone 13, which is isolable, but with extended reduction time at temperatures slightly greater than rt, hydroxy ketone 14 was obtained in 96% overall yield with completely selective reduction at C3a.
With hydroxy ketone 14, we completed our efforts in transforming B-ring into its proper oxidation states, and a single-crystal X-ray structure of 14 [Figure 1 – right side] further affirms our success. However, we were concerned about the reactivity of the C5a carbonyl group because the challenge of C5a-homologation lies ahead. It became obvious that no reduction of the C5a carbonyl group in 10 had occurred when using L-Selectride™, and nor did NaBH4 touch the C5a carbonyl group in 13. While one could concede that reduction in 10 involved a vinylogous ester, lack of reduction in 13 was quite disconcerting.
Upon examination of the minimized Spartan™ model of 10 and X-ray structure of 14, we found some unique conformational elements. In 10, the α-Me group in the Aring [red] is pseudo-equatorial with the β-Me group [blue] being pseudo-axial, thereby blocking any incoming nucleophiles toward the C5a carbonyl group. On the other hand, hydroxy ketone 14 assumes a very different conformation with its AB-ring junction being both sp3-hybridized instead of sp2 as in 10. In this case, the β-Me group [blue] is now pseudo-equatorial with the α-Me group [red] turning to occupy the pseudo-axial position, thereby hindering the attack of the C5a carbonyl group.
In contrast, the AB-ring junction of endo-peroxide 8 consists of sp3-hybridized C8a and sp2-hybridized C8b [Figure 2]. This set of hybridizations leads to yet another conformation in which the β-Me group [blue] remains pseudo-equatorial as in 14, but the α-Me group [red] shifts away versus its respective position in 14. We hoped that this minor shift would provide just enough opening to allow C5a to be accessible for homologation.
To test this hypothesis, we elected to construct enone 15 and epoxy ketone 16 for which both ring-junctions contain one sp3- and one sp2-hybridized carbon.22 As shown in Scheme 4, after failing to eliminate the C3a-OH group via dehydrative protocols, we isolated sulfite 17 during an attempt to chlorinate at C3a. A retro-Diels-Alder process in refluxing toluene would extrude SO2 and afford the desired enone 15. To our relief, we could add various one-carbon nucleophiles such as ROCH2Li [R = PBMB or MOM] to afford ene-diols 18a/b, thereby succeeding what had appeared to be a daunting task in C5a-homologation. However, ene-diols 18a/b were not useful for the total synthesis. Consequently, we prepared epoxy ketone 16, but in three steps, because epoxidation of enone 15 would not take place unless the TES group was removed. Homologation of 16 via addition of MeLi followed by elimination gave vinyl epoxide 19.
Nucleophilic ring-opening of vinyl epoxide 19 at C5 via a SN2′ pathway would have been truly welcome at this stage [Scheme 5]. Instead, a SN1-like process occurred with H2O adding at C8b with retention of stereochemistry when using Mg(OTf)2 in wet CH3CN,23 leading to 1,2-diol 20 after re-silylating C3-OH. Allylic alcohol transposition using Dauben’s protocol24 led to epoxy diol 21.25,26 Subsequent treatment of 21 with Ph3P-I227 followed by Luche reduction of the enal intermediate 22 gave 1,4-diol 23, which was confirmed through its X-ray structure.
After failing a number of approaches for constructing the C-ring using either enal 22 or 1,4-diol 23, mainly due to our inability to consistently oxidize the C3a-OH group, we managed to first acrylate C5-OH. Subsequently, we found that oxidation of the C3a-OH group employing Dess-Martin periodinane reagent at a warmer temperature gave enone 25. An ensuing de-protection sequence allowed for the formation of the lactol C-ring and the final completion of our total synthesis of (±)-phomactin A in 24 steps from ABD-tricycle 2.
We have described a total synthesis of (±)-phomactin A that highlights the final completion of a complex natural product target that had commenced with an intramolecular oxa-[3 + 3] annulation strategy in the construction of an ABD-tricycle. Our efforts reveal structural intricacies of this ABD-tricycle with an illustrative example being the conformational analysis that was ultimately critical for the C5a-homolgation.
We thank NIH [NS38049] for funding.
†This paper is dedicated to Professor Bill Wulff on the very special occasion of his 60th birthday.
Supporting Information Available: Experimental procedures as well as NMR spectra, characterizations, and X-ray structural files are available for all new compounds and free of charge via Internet http://pubs.acs.org.