Synthesis of a Potent Vinblastine: Rationally Designed Added Benign Complexity

Vinblastine (1) and vincristine (2) were originally isolated from Catharanthus roseus (L.) G. Don^1,2 and are the most widely recognized members of the Vinca alkaloids (Figure 1). They have a rich history and were among the first natural products introduced clinically to treat cancer.³ Their discovery defined an important antineoplastic mechanism of action, entailing the inhibition of mitosis via tubulin binding and inhibition of microtubule polymerization,⁴ that today is regarded as one of the most successful drug targets in oncology. Although additional natural products have been shown subsequently to act on tubulin (e.g., taxol, epothilones, eleutherobin, dolastatins), vinblastine and vincristine differ in that they bind at the tubulin–tubulin head-to-tail dimer interface.⁵ Because of their clinical importance, complex structures, and unique mechanism of action, vinblastine and vincristine have attracted extensive synthetic and mechanistic efforts since their original discovery.^3,6,7

In recent studies, we described concise total syntheses of vinblastine and related natural products based in part on introduction of a powerful single-pot two-step diastereoselective Fe(III)-promoted vindoline/catharanthine coupling and subsequent Fe(III)-NaBH₄/air mediated in situ C20′ oxidation.^8,9 These developments have permitted systematic studies of the effects of deep-seated changes in the natural product structures within either the lower vindoline or upper catharanthine-derived velbanamine subunits.⁹ Among the most significant of the observations made to date in these studies is that while removal of individual substituents or key structural components of the natural products typically results in reductions in biological activity,^8–14 addition of structural features can substantially improve biological properties.^15,16 As a result of our demonstration of the importance of the addition of a key indole C10′ substituent (10′-fluorovinblasine)¹⁵ and with the discovery of the remarkable impact of select C20′ alcohol replacements (C20′ ureas),¹⁶ it appeared that the spatial placement of the indole at one end of velbanamine and the C20′ ethyl group at the other are two especially important features of the structure. The X-ray structure of tubulin-bound vinblastine⁵ (Figure 2) indicates both fit into well-defined protein pockets on the tubulin α and β subunits, respectively, deeply embedded in the tubulin binding site with each occupying corners of a T-shaped bound conformation of vinblastine. The core of the velbanamine subunit fills the intervening space and serves as a rigid scaffold that fixes the placement of these two anchoring groups.

Herein, we report a compound, representative of a strategy rarely considered with complex natural products, where increasing the complexity of the core structure provided a stunning 10-fold enhancement in the biological potency. The results are sufficiently remarkable to suggest that even in instances where Nature appears to have optimized a natural product structure for activity, addition of even seemingly benign complexity to its underlying core structure may be possible that further and substantially improves on what Nature only started. For vinblastine and the work herein, this entailed incorporation of the C20′ ethyl group into an added C15′–C20′ cis-fused 6-membered ring, that further fills the C20′ ethyl binding pocket. Unlike molecular editing of structures by diverted synthesis,¹⁷ the addition of a key functional group, derivatization of the natural product, tactics of simplification of complex biologically active natural products,¹⁸ or introduction of conformational constraints,¹⁹ the design of 7 entails introduction of added benign complexity (ABC) to the underlying core structure of vinblastine (two added sp³ methylenes and a stereocenter) to more favorably occupy and fill a protein binding site without changing the intrinsic conformational or structural features of the natural product. Compound 7 represents a member of an analogue class inaccessible by semi-synthetic derivatization of the natural product, by its late-stage functionalization, or by biosynthetic means, but was prepared herein by late-stage divergent synthesis²⁰ based on the penultimate single-step vindoline-catharanthine coupling strategy. Its synthesis used 6, bearing an alkene exocyclic (vs endocyclic) to the catharanthine skeleton, that incorporates the added fused six-membered ring and additional C15′ stereochemistry (Figure 2). This change in the substrate also substantially improved the diastereoselectivity (>6:1 vs 2:1) of the in situ alkene oxidation reaction used for introduction of the C20′ alcohol while maintaining the exclusive generation of the natural C16′ stereochemistry at the critical coupling site.

Following a known 3-step synthesis,²¹ intermediate 11 was prepared from the free base of catharanthine (4, $16/g) by indole protection (2.5 equiv KH, THF, 0 °C; 2.5 equiv ClCO₂Me, 0–23 °C, 15 h, 90%), single-step conversion of 9 to the lactam 10 (4.5 equiv I₂, 9 equiv Na₂CO₃, THF/H₂O, 0–23 °C, 22 h, 82%) and subsequent allylic oxidation (2 equiv SeO₂, THF, 65 °C, 18 h, 60%) under modified reaction conditions (THF vs EtOH) that substantially reduced the amount of required SeO₂ (2 vs 11 equiv) (Scheme 1). Starting with 11, acid-catalyzed elimination of the secondary allylic alcohol (0.6 equiv TsOH, toluene, 110 °C, Dean–Stark trap, 2.5 h, 97%) cleanly provided the diene 12 poised for introduction of the fused unsaturated six-membered ring by a Diels–Alder reaction. Treatment of 12 with phenyl vinylsulfone (2 equiv, toluene, 110 °C, 45 h, 61%) provided 13 as an inconsequential mixture of cycloadduct isomers derived from β-face (top face) addition of the dienophile with installation of the C15′ stereochemistry. Mild reductive removal of the phenyl sulfone with Mg (10 equiv, MeOH, 23 °C, 4 h) and deliberate completion of the in situ methanolysis of the indole carbamate (6.5 equiv K₂CO₃, MeOH, 23 °C, 1 h) provided 14 (65% for 2 steps), whose structure and stereochemistry were confirmed in a single-crystal X-ray structure determination.²² Final reduction of the lactam 14 (5 equiv 9-BBN, THF, 23 °C, 8 h, 59%) afforded the key substrate 6 (5 steps from known intermediate 11, 8 steps from catharanthine) for coupling with vindoline.

Single-step incorporation of 6 into the targeted vinblastine 7 by room temperature Fe(III)-promoted single electron oxidative coupling with vindoline (3, $36/g) in aqueous 0.05 N HCl/trifluoroethanol (TFE, 10:1, 5 equiv FeCl₃, 23 °C, 3 h) followed by in situ Fe(III)-promoted free radical oxidation of the trisubstituted alkene found in 8 (Fe₂(ox)₃, NaBH₄, air, 0 °C, 30 min) proceeded with complete control of the C16′ stereochemistry and good diastereoselectivity for C20′ alcohol introduction (>6:1) (Figure 3). The latter represents an increased diastereoselectivity for the C20′ alcohol introduction relative to that observed with vinblastine itself (2:1)⁸ and may be attributed to the added α-face C15′ steric hindrance for reaction of O₂ with the intermediate tertiary radical derived from Fe(III)/NaBH₄-mediated hydrogen atom addition to intermediate 8. Significantly, the fused six-membered ring in 6 did not impact the exclusive α-face Fe(III)-promoted coupling with vindoline.^8,23 This reaction is thought to proceed by single electron oxidation of 6 to the indole-centered radical cation and its reversible fragmentation to an intermediate charge-separated cation radical that undergoes radical addition to vindoline.⁸ It is of special note that this initial coupling still occurs with exclusive control of the C16′ stereochemistry, with clean stereochemical inversion at the reacting C16′ center, despite the increased steric hindrance due to the added C15′ substitution. This observation is supportive of the suggested intermediacy of an intramolecular one-electron-two-center bonding interaction between the putative C16′ radical site and the fragmented iminium carbon that may stabilize a conformation dictating backside addition of vindoline.^8c

Compound 7 was assessed alongside vinblastine as a direct comparison in cell growth inhibition assays against both a mouse leukemia (L1210) and human colon cancer (HCT116) cell line that have been used traditionally to initially examine vinblastine analogues (Figure 3).²⁴ Consistent with the design but stunning nonetheless, compound 7 proved to be 10-fold more potent than vinblastine, displaying an IC₅₀ of 600–700 pM in the cell growth inhibition assays.

To confirm that the improved activity of 7 is derived from tubulin binding effects, compound 7 was compared alongside vinblastine in a tubulin binding assay, measuring their relative ability to competitively displace BODIPY-vinblastine (Figure 4).²⁵ Compound 7 was found to displace BODIPY-vinblastine bound to tubulin much more effectively than vinblastine itself, confirming that it binds the same site and with a much higher affinity. Although it is not possible to rule out the impact of other features,²⁶ this direct correlation of functional cell growth inhibition activity with target tubulin binding affinity, the relative magnitude of the effects, and the small benign structural differences between 1 and 7 suggest that the improved potency of 7 is derived predominately, if not exclusively, from on target effects on tubulin. The modeled binding of 7 with tubulin is presented in Supporting Information Figure S1.

Approaches to improving the properties of natural products typically involve efforts to simplify or edit their structures through semi-synthetic modifications, diverted total synthesis, or late-stage (C–H) functionalization. Such efforts often strive to remove, introduce, identify or replace key functional groups to enhance target binding affinity, improve physiochemical properties, incorporate stabilizing modifications, or introduce conformational constraints. Rarely, if ever, does one consider adding benign molecular complexity to the underlying core structure.²⁷ In part, this may be attributed to the perceived added challenge intrinsic in the requisite compound synthesis. Herein, we describe the synthesis of a rationally designed ABC (added benign complexity)²⁷ modification to vinblastine that was as accessible as the natural product or its simplified analogues, that enhanced the stereochemical control of the underlying synthesis, and that provided an analogue 10-fold more potent than the natural product. The compound represents a vinblastine analogue accessible only by chemical synthesis and is presently inaccessible by natural product derivatization, late-stage functionalization, or biosynthetic methods. Moreover, in instances when the productive properties of a natural product are directly related to its emergence in Nature and has undergone continued optimization by natural selection as is likely the case with vinblastine, it may not be easily subjected to structural simplifications. In such cases, it is possible that added molecular complexity^15,16 may be used to enhance target binding and functional biological activity, and approaches as simple as ABC illustrated herein may represent a rational conceptual approach to achieving such objectives.²⁸