Macrocyclic natural products are valuable to drug discovery and development. Through chemical synthesis, many of these often-complex molecules have been synthesized. The goal of our project was to devise a set of protocols that allow for generating, in a controllable and programmable fashion, a variety of precisely altered skeletal analogs. The latter compounds can then be used to screen for better efficacy, reduced toxicity, and enhanced physicochemical properties. Through the use of anti-cancer agent epothilone C, as the model macrocyclic natural product, we show that three different types of catalytic olefin metathesis, ring-opening/cross-metathesis, cross-metathesis and ring-closing metathesis, can be merged with olefin transposition, catalytic stereoselective boryl substitution and catalytic E-selective alkene isomerization to generate virtually any new expanded, contracted or distorted macrocyclic analog. Accordingly, we have been able to present a worksheet for how countless analogs of a large variety of macrocyclic natural products may be prepared in short order and exactly which catalysts and reagents would be needed for doing so.
A recent study by our research group illustrated that through ground-up programmable skeletal alteration, the three dimensional contours a small number of skeletal analogs of a naturally occurring bridged polycyclic indole alkaloid, originally found to be only mildly antimalarial, can result in identification of new leads (~3mM activity) for anticancer drug discovery.
The exciting outcomes of the above project led us to wonder whether we could design a set of practical and efficient strategies through which provide facile access to precisely altered analogs of other important classes of bioactive compounds. This led us to focus on macrocycles, a set of organic molecules that continue to emerge as compelling therapeutic candidates. Macrocycles are particularly relevant to difficult-to-drug or “non-druggable” targets. This is because they can offer the necessary blend of conformational rigidity and flexibility, which allows them to adapt to and interact effectively with a range of biological receptors.
At the heart of our approach, which we referred to as late-stage diverted total synthesis, is a class of compounds that are present towards the end of many total syntheses of macrocycles, namely, linear dienes that are subjected to macrocyclic ring-closing metathesis (MRCM). The macrocyclic alkene thus formed is either the final target and the olefin can be further functionalized (for example, by hydrogenation or epoxidation). Our approach is founded on the principle that such dienes are considerably more valuable realized, that is, merely as precursors to a natural product. Rather, they can also serve as hubs for generating analogs that are expanded or contracted at a specific location by no more than two or three additional steps. What is more, if the macrocyclic alkene can be accessed easily, the diene hub can be obtained simply by a single transformation: ring-opening/cross-metathesis (ROCM) involving ethylene (ethenolysis).
We show that ring-expanded analogs can be obtained by extending the length of a diene hub through catalytic cross-metathesis (CM) followed by a substitution reaction and then performing MRCM to generate either a one- or two-unit expanded macrocycle that contains either a Z- or an E-alkene. The corresponding ring-contracted analogs can be secured by iterative methylene deletion comprised of two catalytic transformations: alkene isomerization and ethenolysis. Again, ensuing MRCM leads to the desired smaller ring analogs in either stereoisomeric form.
As the model platform, we chose epothilone C, a naturally occurring macrocyclic alkene with considerable anti-tumor activity. There were several other reasons why we opted for epothilone C. (1) It is commercially available and can be generated in significant quantities through fermentation. Furthermore, it has been prepared by catalytic stereoselective MRCM of a diene hub that can be accessed by well-established total synthesis routes. (2) Previous investigations have led to the identification of more active epothilone C analogs derived from aziridination of the cyclic alkene (even more so than the epoxide, which is epothilone B). (3) It contains various polar functional units and a heterocyclic ring, thus allowing us to develop a set of protocols that tolerate these frequently occurring moieties.
We targeted six distinct types of remodeling operations. One of our goals was to probe the feasibility of converting the natural Z-disubstituted macrocyclic alkene to its E isomer (Twist). Another aim was to use our approach to expand the macrocycle through site-specific insertion of a methylene unit (Expand-I or Expand-II). Similarly, we targeted contracted analogs such as Contract-I and Contract-II. Finally, methylene insertion and deletion could be merged to generated “distorted” frameworks (Distort).
