|Home | About | Journals | Submit | Contact Us | Français|
A unified synthetic approach to diverse polycyclic scaffolds has been developed using transition metal-mediated cycloaddition and cyclization reactions of enynes and diynes. The t-butylsulfinamide group has been identified as a particularly versatile lynchpin in these reactions, with a reactivity profile uniquely suited for efficient, stereoselective substrate synthesis and downstream transformations. This approach provides ten distinct, functionalized scaffold classes related to common core structures in alkaloid and terpenoid natural products.
Polycyclic alkaloid and terpenoid natural products exhibit a tremendous array of chemical scaffolds and biological activities.1,2 Accordingly, these structures are attractive targets for the synthesis of natural product-based libraries.3 Ideally, a concise, unified synthetic route would provide an array of distinct, polycyclic scaffolds for use in discovery screening against a wide range of targets. The synthesis of such multiscaffold libraries remains a major challenge in diversity-oriented synthesis.4,5 We envisioned that the modern arsenal of transition-metal mediated cycloaddition and cyclization reactions6,7 would provide a powerful means to generate such libraries from simple tethered enyne and diyne substrates. We report herein the development of a unified synthetic approach leading to ten classes of polycyclic scaffolds, and the emergence of the t-butylsulfinamide moiety8 as a versatile lynchpin for these reactions, affording uniquely suited reactivity and a novel motif for biological evaluation.
To identify transition metal-mediated cycloaddition and cyclization reactions suitable for diversity-oriented synthesis, we set out to evaluate candidate reactions systematically across a panel of substrates having electronically and sterically distinct groups at sites expected to influence reactivity. This would also provide broad insights into the scope and efficiency of these reactions.
To assemble the requisite enyne and diyne substrates, we initially investigated ether, carbamate (N-Boc), and sulfonamide (N-Ts, N-Ns) tethers. After extensive experimentation, the t-butylsulfinamide8 emerged as a uniquely suited lynchpin. This group provides asymmetric induction during substrate assembly, can be readily deprotonated and N-alkylated, does not exhibit rotamers on the NMR timescale, and can be deprotected or oxidized under mild conditions. The t-butylsulfinamide is also a novel motif for biological evaluation, related to sulfonamides in synthetic drugs and natural products. Notably, although a picture of compatibility of the t-butylsulfinamide with metal catalysts is beginning to emerge,9 its stability and reactivity in transition metal-mediated cycloadditions and cyclizations has not yet been explored in detail.
Thus, synthesis of enynes 4 and diynes 5 began with condensation of aldehyde 1 and (R)-t-butylsulfinamide (Scheme 1). The R1 sidechain was designed with a TPDPS-protected alcohol as a potential handle for later functionalization and as a mimic of our reported TBDAS linker for future solid phase syntheses.10 Diastereoselective addition of terminal alkynes afforded sulfinamides 3a–c.11 N-Alkylation with allyl and propargyl bromide was achieved efficiently using n-BuLi/ HMPA to afford enynes 4a–c and diynes 5a–c.12 C-desilylation of TMS-alkynes 4c and 5c then provided terminal alkynes 4d and 5d. Additional functionalized alkenes and alkynes can be envisioned to provide an even broader assessment of reactivity trends in the future.
With gram quantities of enynes 4 and diynes 5 in hand, we evaluated their reactivities in various transition metal-mediated reactions. Initial experiments with Au and Ag π-acids (Group 11) commonly used in such reactions13 resulted in decomposition, possibly initiated by sulfinamide cleavage. In contrast, these substrates were compatible with a variety of Ru, Co, Rh, and Ni catalysts (Groups 8–10), and, after investigation of over 25 different reactions, eight were identified as having suitable selectivity and efficiency for use in library synthesis (Scheme 2).
The venerable Pauson–Khand reaction12b,14 was effective for all four enynes 4a–d, providing [5,5]-bicyclic cyclopentapyrrolidinone scaffolds 6a–d (Table 1). Krische’s Rh-catalyzed reductive enyne cyclization15 provided excellent yields of exo-pyrroline scaffolds 7a,b,d, with reagent-controlled diastereoselectivity for the internal alkynes 4a,b, while the TMS-alkyne 4c was unreactive under these conditions. Evans’ Rh-catalyzed butadiene [4+2+2] cycloaddition also proved useful;16 while enynes 4 underwent AgOTf-induced sulfinamide cleavage under the reaction conditions, consistent with our earlier findings, the reaction proceeded effectively after oxidation to the corresponding t-butylsulfonamides, affording [5,8]-bicyclic cyclooctapyrrolidine scaffolds 8a–d in moderate yields but complete diastereoselectivity. Enyne metathesis of 4 with Grubbs’ 2nd generation catalyst17 led to vinylpyrrolines 9a,b,d; the TMS-alkyne 4c was again unreactive. Interestingly, the diene products 9 proved recalcitrant to subsequent Diels–Alder reactions with numerous dienophiles. 18,19 However, these reactions could be achieved after oxidation to the corresponding t-butylsulfonamides; subtle conformational effects may account for this reactivity difference. Thus, reactions with N-phenylmaleimide provided [5,6,5]-tricyclic benzodipyrrolidine scaffolds 10a,b,d.1 Reactions with dimethylacetylene dicarboxylate (DMAD) afforded diastereomeric mixtures that converged conveniently to [5,6]-bicyclic isoindoline dicarboxylate scaffolds 11a,b,d upon oxidation with DDQ.1
Several effective transition metal-mediated cycloaddition reactions were also identified for diynes 5a–d. While [2+2+2]-cyclotrimerization with various alkynes using reported Rh(I), Ni(0), or Ir(I) catalysts20 suffered from poor regioselectivity and competing dimerization, treatment of 5a–d with Grubbs’ 1st generation catalyst21 and propargyl alcohol yielded [5,6]-bicyclic isoindoline scaffolds 12a–d efficiently (Table 2). The reactions were regioselective, except in the case of the pseudosymmetric substrate 5d, and regioisomers were readily separated in all cases. Diynes 5a–d also cyclotrimerized with benzyl isocyanate under the agency of Yamamoto’s Ru(II) catalyst22 to provide [5,6]-bicyclic pyrrolopyridone scaffolds 13a–d, and regioisomeric products were again readily separated. No other transition metals were found to catalyze this reaction. Cyclotrimerizations of 5a–d with ethyl cyanoformate proceeded effectively using Tanaka’s Rh(I) catalyst23 to afford [5,6]-bicyclic pyrrolopyridine carboxylate scaffolds 14a,b,d. Notably, the reaction was regioselective even for the pseudosymmetric substrate 5d, while the TMS-alkyne 5c was unreactive. In contrast, reactions with aryl or alkyl nitriles proceeded with low to moderate regioselectivity, while the use of alternative Ni(0), Co(I), or Ru(II) catalysts24 gave no reaction or poor regioselectivity. Saito’s Ni(0)-catalyzed [3+2+2]-cyclotrimerization 25 with ethyl cyclopropylideneacetate provided larger [5,7]-bicyclic cycloheptapyrrolidine scaffolds 15a–c as single regioisomers and inseparable E:Z mixtures. While the E:Z ratios could not be improved using alternative cyclopropylidenes, solvents, or phosphine ligands, conversion to the corresponding Weinreb amides26 16a–c allowed convenient separation of the isomers.
To probe the potential influence of the remote sulfinamide stereogenic center on the diastereoselectivity of the enyne cycloaddition reactions, we also carried out Pauson–Khand reactions with anti-enyne substrates 17a–d having the opposite stereochemistry relative to syn-enynes 4a–d (Scheme 3).1 Increased diastereoselectivities were observed for 18a,b versus 6a,b, suggesting a matched/mismatched scenario.12b,27 While these effects have not yet been studied in other reactions, access to both syn- and anti- diastereomers would also provide increased stereochemical diversity in resulting libraries.
Overall, we identified 27 reaction/substrate combinations suitable for use in multiscaffold library synthesis (≥84:16 dr or isolable as single regioisomer), leading to a total of 32 different scaffolds related to alkaloid and terpenoid natural products (after in silico desilylation; includes six additional Diels–Alder products).1 To assess the structural diversity of the multiscaffold library accessible using our synthetic strategy, and its relationship in chemical space to synthetic drugs, natural products, and drug-like libraries, we evaluated these compounds for 20 structural and physicochemical properties in the context of our reported principal component analysis. (Figure 1).1,3b,28 As expected, our scaffolds sample a distinct region of chemical space compared to drugs and drug-like libraries, and overlap with alkaloids and terpenoids. Analysis of parameter loadings indicates that aromatic ring content and stereochemical complexity are two major factors that distinguish the natural products and the multiscaffold library from drugs and drug-like libraries.1
In conclusion, we have used a systematic approach to analyze the effectiveness of transition-metal catalyzed cycloaddition and cyclization reactions across a range of enyne and diynes, resulting in the identification of eight reactions suitable for use in multiscaffold library synthesis. More broadly, our results provide valuable insights into the effective scope of such reactions across a panel of differentially substituted substrates and into the reactivity of the t-butylsulfinamide lynchpin. Synthesis of discovery libraries using this approach is ongoing and will set the stage for quantitative comparison of the abilities of drug-like and natural product-like libraries to address distinct regions of biological space through screening against a broad range of biological targets.
We thank Dr. George Sukenick, Dr. Hui Liu, Hui Fang, Dr. Sylvi Rusli, and Anna Dudkina for expert mass spectral analyses. D.S.T. is an Alfred P. Sloan Research Fellow. Financial support from the NIH (R21 GM104685, P41 GM076267, T32 CA062948-Gudas), NYSTAR Watson Investigator Program, William H. and Alice P. Goodwin and the Commonweath Foundation for Cancer Research, and MSKCC Experimental Therapeutics Center is gratefully acknowledged.
Supporting Information Available: Detailed experimental procedures and analytical data for all new compounds. This material is available free of charge via the Internet at http://pubs.acs.org.