The Total Synthesis of Irciniastatin A (Psymberin)

Abstract

The total synthesis of (+)-iriciniastatin A (psymberin) is reported in 19 steps and 6% overall yield. Key reactions include a highly convergent enolsilane-oxocarbenium ion union to generate the C8-C25 fragment and a late stage coupling of a hemiaminal and acid chloride to complete the synthesis.

In 2004, Pettit and Crews independently reported the isolation of two remarkably potent and selective cancer cell growth inhibitors from the extracts of deep water sponges. Irciniastatin A (1) was obtained from Ircinia ramosa by Pettit,¹ and psymberin from Psammocina sp. by Crews.² The two secondary metabolites were shown to be structurally related based on spectroscopic studies, however, incomplete and conflicting stereochemical analyses precluded a definitive assignment of absolute configuration. Each displayed incredibly powerful and selective cancer cell growth inhibition. Irciniastatin A was reported to exhibit a greater than 10⁴ differential in activity across related cell lines. The limited natural abundance and lack of an absolute stereochemical assignment, as well as interest in further exploration of the unique chemotherapeutic profile and mode of action, sparked immediate interest within the synthetic community.

Early reports shed light on the stereochemical ambiguities through fragment synthesis³ and natural product degradation.⁴ Ultimately, it was the seminal total synthesis by DeBrabander⁵ that confirmed the absolute configuration and demonstrated that irciniastatin A (1) and psymberin were identical. Subsequent reports, including fragment synthesis,⁶ a formal synthesis,⁷ two total syntheses,^8,9 and two reports of analog syntheses,^10,11 all offered creative and distinct approaches toward the unique structural elements of irciniastatin A.

Our interest in (+)-irciniastatin A (psymberin) focused on developing a highly diastereoselective and modular synthesis that might be suitable for analog preparation. Our efforts toward (+)-1, recently culminated in a completed total synthesis which is reported here.

Retrosynthetically (Scheme 1), (+)-irciniastatin A (1) was envisioned to arise from two key disconnections. A late stage attachment of the psymberic acid chain would be accomplished by coupling of acid chloride 2 with hemiaminal 3, the product of a Curtius rearrangement of the carboxylic acid derived from benzyl ether 4. This tactic would allow for a highly stereocontrolled entry to the C8 hemiaminal and efficient incorporation of the side chain. Tetrahydropyran 4 would arise from the stereoselective addition of enolsilane 6 to the oxocarbenium-ion derived from acetate 5. These two key disconnections segregate the three major subunits of irciniastatin A (1); each of similar size and complexity and accessible in a highly stereocontrolled fashion from standard aldol synthons.

Scheme 1.

Retrosynthetic Analysis of Irciniastatin A

All reported syntheses of the psymberate side chain have relied on functionalizing commercially available chiral pools or enzymatic resolutions. In addition, analog studies^10,11 have shown the side chain to be required for high activity. Therefore, an enantioselective and highly tunable synthesis of the side chain would be ideal for further investigation of side chain function. We reasoned that an oxazolidinethione asymmetric glycolate aldol reaction¹² would allow for enantioselective entry to the psymberate side chain as well as provide sufficient opportunities for derivitization and congener synthesis. The synthesis of acid chloride 2 (Scheme 2) began with the known¹³ anti-aldol^12a reaction of glycolate 9 and 3-methyl-but-3-enal to deliver aldol adduct 10 that was converted to known TIPS ether 11.¹³ Methylation of the alcohol was followed by removal of the allyl protecting group under Kulinkovich conditions^13,14 to give alcohol 12. Protection of alcohol 12 as the SEM ether was followed by selective removal of the primary TIPS ether. The resultant primary alcohol 13 was oxidized under Smith's conditions giving the corresponding acid over two steps. The acid (9 steps, 28% overall) was converted to acid chloride 2.

Scheme 2.

Synthesis of the Psymberic Acid Side Chain

The synthesis of the acetate 5 (Scheme 3) began from known p-methoxybenzylidine acetal 14¹⁵ available from 2-deoxy-D-ribose in two steps. Methylation of alcohol 14 was followed by a dihydroxylation-oxidative cleavage sequence to reveal aldehyde 16. A catalyst controlled Kiyooka¹⁶ aldol reaction of aldehyde 16 and enolsilane 17¹⁷ provided carbinol 18 in 84% yield and 9:1 dr. Protection of alcohol 18 to give TBS ether 19 was followed by acid-catalyzed cyclization to provide a 10:1 mixture of lactone 20 and the corresponding diol, which was converted to lactone 20 by exposure to CF₃CO₂H. Protection of the primary alcohol gave benzyl ether 21 and subsequent one-pot reductive acetylation¹⁸ afforded acetate 5 in quantitative yield (9 steps, 34% overall from 2-deoxy-D-ribose).

Scheme 3.

Synthesis of Acetate 5

The synthesis of enolsilane 6 (Scheme 4) began from known catechol 24^6,19 prepared by cycloaddition of allene 23²⁰ and diene 22.²¹ Protection of catechol 24 as the bis-TIPS ether was followed by selective ester reduction to give aldehyde 25 in 78% yield over 2 steps. An asymmetric propionate aldol²² reaction of aldehyde 25 and propionyl thiazolidinethione 26 afforded the Evans-syn-aldol adduct 27 in 94% yield and >20:1 dr. Direct displacement of the chiral auxiliary with Weinreb's amine was followed by protection of the secondary alcohol as a TBS ether to deliver silyl ether 28. The methyl ketone prepared from Weinreb's amide 28, was then readily converted to desired enolsilane 6 (7 steps; 66% overall from catechol 24).

Scheme 4.

Synthesis of Enolsilane 6

Having devised highly stereocontrolled routes to all three key fragments, the union of enolsilane 6 and acetate 5, was investigated (Scheme 5). Rigorous experimentation revealed that BF₃●OEt₂ was the optimum Lewis acid and that TBS enolsilane 6 performed better than the corresponding TMS enolsilane for the assembly of 5 and 6. Most intriguing, however, was that addition of a solution of enolsilane 6 to a premixed solution of 5 and BF₃●OEt₂ at -40 °C was required for efficient coupling. The union proceeded with high diastereoselectivity, giving tetrahydropyran 4 as the only detectable diastereomer. The high diastereoselectivity can be rationalized by well precedented²³ pseudoaxial addition of the nucleophile to the oxocarbenium conformer 29, proceeding through a favorable chair-like conformation. Conformer 29 would also be expected to be favored as a result of through-space stereoelectronic stabilization²³ of the oxocarbenium ion by the axially positioned C11 ether.

Scheme 5.

Coupling of Acetate 5 with Enolsilane 6

With the C8-C25 carbon framework in place, conditions to set the C15 stereocenter were investigated (Scheme 6). Standard achiral reducing agents proved to be non-selective or completely selective for the undesired diasteromer,²⁴ however, the use of the chiral (R)-CBS²⁵ agent exclusively provided desired diastereomer 30²⁶ in 82% yield. After protection of the secondary alcohol as its TBS ether and cleavage of the benzyl ether, carbinol 31 was oxidized over 2 steps to carboxylic acid 32. Initial studies showed the one-pot Curtius reaction with (O,O)-diphenylphosphoryl azide²⁷ to be ineffective, providing significant quantities of inseparable carbamoyl azide impurity.²⁸ Therefore, Weinstock's procedure^9,29 was employed to generate the intermediate isocyanate, which under mild conditions using copper (I) chloride afforded Teoc-protected hemiaminal 33.³⁰ Much to our dismay, a screen of reported conditions³¹ for acylation of stucturally related intermediates never yielded N-acyl hemiaminal 34. A survey of the previous syntheses of irciniastatin A finds that all feature a late stage aminal and side chain incorporation with a preformed lactone of the dihydroisocoumarin. Albeit unlikely that such remote functionality should have any effect on the hemiaminal coupling, the limited number of available options warranted pursuit of this lead.³²

Scheme 6.

Completion of the Synthesis of (+)-Irciniastatin A

Strategically, after the CBS reduction, the lactone was formed concomitantly with removal of the TBS ether to give 35 adding only one synthetic step over the original plan (bottom of Scheme 6). An analogous sequence of TBS protection of the C15 carbinol and hydrogenolysis of the benzyl ether gave carbinol 36, which was subjected to a two step oxidation to acid 37. Acid 37 was subjected to the previously established Curtius conditions to give hemiaminal 3. With hemiaminal 3 featuring the lactonized dihydroisocoumarin, it was found that i-PrMgCl, a base not previously used for this transformation, successfully effected reaction with acid chloride 2 in a remarkable 87% yield. Finally, global deprotection with TASF in DMF at 50 °C provided (+)-irciniastatin A (1) in 94% yield.

In summary, the total synthesis of (+)-irciniastatin A (1) was completed in 19 steps with a 6% overall yield from 2-deoxy-D-ribose. The successful application of this strategy allowed for rapid assembly of the three key fragments with high diastereocontrol. Studies directed toward analog synthesis and further evaluation of the antitumor activity of irciniastatn A are underway in collaboration with the Lineberger Comprehensive Cancer Center.