Formal Synthesis of Fostriecin via Asymmetric Alcohol-Mediated Carbonyl Allylation

Abstract

A formal synthesis of fostriecin via convergent assembly of two fragments prepared via asymmetric alcohol-mediated C-C coupling is described. One fragment is made by enantioselective iridium-catalyzed allylation of an allylic alcohol mediated by allyl acetate. The other fragment is made via enantioselective ruthenium-catalyzed reductive syn-(α-alkoxy)allylation of an aldehyde mediated by an alkoxyallene (where 2-propanol is the hydrogen source), representing the first use of this method in target-oriented synthesis. Metathetic fragment union enables interception of a late-stage compound that previously required a 25 step (LLS) synthesis in only 7 steps (LLS).

Graphical Abstract

Fostriecin (CI-920) is a secondary metabolite isolated from the fermentation broth of the Brazilian soil bacteria Streptomyces pulveraceus (subsp. fostreus) in 1983.¹ Fostriecin is the founding member of a broad family of phosphorylated polyketides isolated from various Streptomyces strains that have attracted significant interest from the scientific community due to their cytotoxicity, antifungal activity, and ability to affect signal transduction pathways (Figure 1A).² Fostriecin is a known inhibitor of topoisomerase II (TOP2, IC₅₀ = 40 μM);³ however, its antitumor activity is attributed to its more potent inhibition of protein phosphatase type 2A (PP2A, IC₅₀ = 1.5 nM)⁴ and protein phosphatase type 4 (PP4, IC₅₀ = 3.0 nM),⁵ which results in mitotic entry checkpoint inhibition.^4–6,11c Moreover, fostriecin displays remarkable selectivity for PP2A/PP4 over the closely related protein phosphatase 1 (PP1, IC₅₀ = 40 μM).⁴ This 10⁴ difference in selectivity is attributed to conjugate addition of a cysteine residue not found in PP1 to the α,β-unsaturated lactone of fostriecin.⁷ Saturation of the α,β-alkene results in a 200-fold loss of potency, suggesting that the lactone participates in active site binding.^7b Discovery of a phoslactomycin A-PP2A adduct bound to the Cys-269 residue further supports this model for selectivity.⁸ Fostriecin’s selectivity for PP2A over related protein phosphatase family members made it an attractive clinical candidate;⁹ however, clinical trials were halted in stage IA due to purity and stability concerns associated with naturally sourced material.¹⁰ De novo chemical synthesis potentially offers access to material of consistent purity and analogs with more ideal pharmacokinetic properties. Indeed, as demonstrated in 19 reported total and formal syntheses that range from 16 to 34 steps in length (LLS), great efforts have been put forth by synthetic chemists to meet this challenge (Figure 1B).^11,12 Here, we report a convergent 16 step (LLS) formal synthesis of fostriecin via transfer hydrogenative C-C coupling.^13–15 Specifically, using an enantioselective iridium-catalyzed allylation of an allylic alcohol mediated by allyl acetate¹⁶ and an enantioselective ruthenium-catalyzed syn-(α-alkoxy)allylation of an aldehyde mediated by an alkoxyallene (with 2-propanol as reductant),^17,18 a late-stage intermediate that previously required a 25 step (LLS) preparation^12k,l was intercepted in only 7 steps (LLS), nearly doubling the proportion of skeletal assembly events compared to the parent total synthesis (38% vs 21%) (Figure 1C). This work represents the first use of the enantioselective ruthenium-catalyzed syn-(α-alkoxy)allylation in target-oriented synthesis.

Figure 1.

The fostriecin family of phosphorylated polyketide natural products. Total and formal syntheses of fostriecin (LLS = Longest linear sequence, TS = Total Steps). Retrosynthetic analysis of Hayashi’s late-stage intermediate.^a

Retrosynthetically, a convergent formal synthesis of fostriecin was envisioned in which the previously reported late-stage intermediate, Fragment C, was formed via cross-metathesis of Fragment B and Fragment A. Precedent for cross-metatheses of secondary allylic alcohols or ethers with tertiary allylic alcohols is scant,¹⁹ and corresponding cross-metatheses of secondary allylic esters/lactones with tertiary allylic ethers is, to our knowledge, unknown. Hence, the feasibility of engaging Fragment A in efficient cross-metathesis was uncertain. Access to Fragment B was deemed possible via transfer hydrogenative allylation of allylic alcohol 4 mediated by allyl acetate¹⁶ followed by ring-closing metathesis using the TMS moiety as a regiocontrol element.²⁰ Similarly, it was posited that access to Fragment A could be achieved using our recently developed 2-propanol-mediated reductive syn-(α-alkoxy)allylation of aldehyde 2 mediated by the indicated O-benzhydryl alkoxyallene 1.^17,18

The synthesis of Fragment A begins with the ruthenium-BINAP-catalyzed reductive coupling of allylation of O-benzhydryl alkoxyallene 1 with aldehyde 2 mediated by 2-propanol (Scheme 1).^17,18 The product of syn-(α-alkoxy)allylation 3 was formed in 68% yield on 3 mmol scale with good control of syn-diastereo- and enantioselectivity (10:1 dr, 92% ee). Sodium naphthalenide-mediated cleavage of the benzhydryl and p-methoxybenzyl ethers provided a triol (not shown),²¹ which was converted to the acetonide Fragment A.

Scheme 1.

Synthesis of Fragment A via enantioselective ruthenium-catalyzed. syn-(α-alkoxy)allylation aldehyde 2.^a

^aYields of material isolated by silica gel chromatography. Diastereoselectivities were determined by ¹H NMR analysis of purified material. Enantioselectivities were determined by chiral stationary phase HPLC analysis. See Supporting Information for further details.

The synthesis of Fragment B begins with the redoxneutral p-allyliridium-C, O-benzoate-catalyzed coupling of allyl acetate with allylic alcohol 4²² via hydrogen auto-transfer (Scheme 2).¹⁶ The product of allylation 5 was formed in 74% yield on 4 mmol scale with good control of enantioselectivity (90% ee). As previously documented,²³ the π-allyliridium-C, O-benzoate is the catalyst resting state and this stable 18-electron complex is readily recovered and recycled without any erosion in performance. With allylic alcohol 5 in hand, conversion to Fragment B is readily achieved via successive acryloylation-ring-closing metathesis to form the dihydropyranone, followed by protonolytic cleavage of the vinylic trimethylsilyl moiety,²⁴ which was required to suppress competing dihydrofuranone formation.²⁰ Notably, the corresponding dihydropyranone bearing a SiMe₂Ph moiety was prepared through an analogous route, but was resistant to protonolysis, presumably due to inductive destabilization of the carbocation intermediate. Fragment B prepared in this manner is identical to material previously prepared.²⁵

Scheme 2.

Synthesis of Fragment B via enantioselective iridium-catalyzed allylation of alcohol 4.^a

^aYields of material isolated by silica gel chromatography. Enantioselectivities were determined by chiral stationary phase HPLC analysis. See Supporting Information for further details.

As anticipated, the metathetic union of Fragment A and Fragment B proved quite challenging and required extensive optimization (Scheme 3). An excess of Fragment A was employed to mitigate dimerization of Fragment B and bias the reaction toward formation of the desired cross-metathesis product 6. Fragment A is itself stable with respect to dimerization and is readily recovered and recycled. Slow addition of the catalyst as a solution via syringe pump improved yields.²⁶ However, these initial reaction conditions^19f provided 6 in only 30% yield (Scheme 3, Entry 1). Performing the reaction at elevated temperature improved the yield of 6 (Scheme 3, Entries 2,3), as did conducting the reaction under a stream of nitrogen vs an argon balloon (Scheme 3, Entry 4), presumably due to the removal of ethylene.²⁷ Remarkably, with the aforesaid measures taken into account, the cross-metathesis of Fragment A and Fragment B provided adduct 6 in 88% yield as a single geometrical isomer, along with an 88% recovery of Fragment A. Finally, TEMPO-catalyzed oxidation of primary alcohol 6 provided Fragment C in 7 steps (LLS); a synthetic intermediate en route to fostriecin that had previously been prepared in 25 steps (LLS).^12k,l Fragment C prepared in this manner is identical in all respects to the reported material.^12k,l The stereogenic centers at C5, C8, and C9 of Fragment C were each generated via asymmetric alcohol-mediated C-C coupling, and were assigned as described in the parent methodological reports (based on single crystal X-ray diffraction analyses).^16,17

Scheme 3.

Metathetic union of Fragment A and Fragment B to form Fragment C.^a

^aYields of material isolated by silica gel chromatography. See Supporting Information for further details.

In summary, we report a formal synthesis of fostriecin using two asymmetric alcohol-mediated C-C couplings: an enantioselective iridium-catalyzed allylation of an allylic alcohol mediated by allyl acetate, and an enantioselective ruthenium-catalyzed reductive syn-(α-alkoxy)allylation of an aldehyde mediated by an alkoxyallene (employing 2-propanol as the hydrogen source). Metathetic union of these fragments enabled interception of a late-stage intermediate that previously required a 25 step (LLS) synthesis in only 7 steps (LLS) with control of relative and absolute stereochemistry.

Supplementary Material

Experimental procedures, spectroscopic and chromatographic data for all new compounds (¹H NMR, ¹³C NMR, IR, HRMS). This material is available free of charge via the internet at http://pubs.acs.org.

Data Availability Statement

The data underlying this study are available in the published article and its Supporting Information.