Sunday, December 27, 2015

The Most Concise Enantioselective Approach to Fully Synthetic Steroids Reported To Date

            This is the 100th post at Modern Steroid Science (begun in late 2011) and, likely, the last of 2015. So it is fitting that we will provide an overview of one of the most interesting organic synthesis papers from this year. It is a report from Pavel Nagorny’s group at the University of Michigan describing an asymmetric approach to fully synthetic steroids that proceeds in just two key steps (4 – 5 total steps, if we account for the preparation of the building blocks from essentially earth, wind, fire and water). To put this into context, optically active steroids of a similar level of complexity have been recently prepared in more like ~20 linear operations. Importantly, the methodology provides access to unnatural polycyclic molecules that could not be obtained from semisynthetic derivatization of an abundant sterol and the chemistry is practical. No aspect of this expedient, catalytic sequence (to my eye) seems like it would be limiting with regard to scalability. Indeed, the route has already been demonstrated on gram-scale by the academic team.
            Back in the 1980s, the laboratory of Pierre Deslongchamps at the University of Sherbrooke (Quebec) was working on a rapid approach to fully synthetic steroids. He came up with a one-step synthesis starting from a monosubstituted cyclohexenone (eventual A-ring) and an a,b-unsaturated b-keto ester (containing the D-ring) that zips up an androstane system with six contiguous stereogenic centers (see Scheme above). The chemistry, while highly expedient, has seen limited practical applications in an industrial or academic setting. A variant of this technology was used in the first total synthesis of ouabain. The anionic polycyclization of Deslongchamps furnishes racemic products, unless the starting materials are derived from chiral pool reagents, as was the case for ouabain. Control over relative stereochemistry at the C/D ring junction positions (C13/C14) in this complex annulation process is not highly predictable and seems to involve subtle structural features embedded within the starting materials. Moreover, the steroidal products thus obtained necessarily bear an ester moiety at the C6 position of the B-ring and a decarboxylation step is usually required to excise the unneeded functionality. Almost thirty years later, the anionic polycyclization chemistry first reported by Deslongchamps in 1988 was in need of a bit of a refurbishment.
            Nagorny’s two-step steroid synthesis (shown above) hinges on a challenging asymmetric, diastereoselective Michael addition between a 2-substituted b-ketoester and a b-substituted enone that delivers a conjugate addition product with vicinal quaternary and tertiary stereocenters. The reaction employs catalytic levels of inexpensive copper(II) salts with a noncoordinating counterion in combination with an optically active bis(4,5-dihydrooxazole) (Box) ligand under solvent-free (neat) conditions. The substrate scope of the enantioselective Michael addition is outstanding, generally producing enantiomeric excesses (ee’s) in the low- to mid-nineties. 
            Mechanistically, Deslongchamps’ anionic polycyclization protocol first closes the steroid B-ring by a sequential double-Michael addition. A subsequent intramolecular aldol reaction between the newly formed cesium enolate and a tethered D-ring cyclopentadione moiety yields the intact steroid skeleton. Nagorny has cleverly designed his new Michael addition such that the products obtained from the process competently participate in a sequenced intramolecular aldol reaction to construct the B-ring and produce an intermediate that is similar to the penultimate steroid precursor of Deslongchamps. Nagorny’s approach then benefits from the wealth of precedent in the steroid literature regarding final C-ring formation via an additional tandem intramolecular aldol reaction (for example), this one between the eventual C8 and C14 positions. In the event, the double Michael addition developed at the University of Michigan proceeds with outstanding efficiency and stereocontrol. Alternate reaction conditions were demonstrated that effectively control the C/D-ring junction stereocenters, providing divergent, diastereoselective formation of either a 13b- or 13a-androstanedione system. The latter annulation product possesses a C5 b-oriented hydroxyl group and the former a D5,6 unit of unsaturation derived from elimination of water. The now readily available structures obtained in this fashion may be suitable for eventual conversion to bioactive cardenolide or limonoid natural products, classes of steroids that contain differing configurations of the angular substituent appended to C13. Notably, the modular process allows for introduction of structural variation at the C13 position, as well as alteration of the A/D ring size, which could not be achieved by semisynthetic methods.
            During my initial read of the Nagorny JACS manuscript, I was wondering if the substrate scope that is tolerated by the process would provide access to derivatives that are functionalized at the C3 position of the steroid A-ring. Nearly all bioactive sterols and synthetic derivatives thereof possess oxidative functionality at C3. Indeed, in the course of the first total synthesis of ouabain, Deslongchamps et al implemented an alkylsilane group appended to the pro-C3 position that required a late-stage Tamao-Fleming oxidation to unmask the latent C3-hydroxyl functionality. Nagorny provides a clue to the careful reader in this regard by disclosing the asymmetric Michael example depicted above. Introduction of a vinyl chloride moiety within the A-ring ketoester building block was shown to be well-tolerated. The pro-C3 vinyl chloride-containing Michael adduct is obtained in excellent yield and stereoselectivity, albeit in combination with a simplified b-substituted a,b-unsaturated ketone. It will be exciting to see if this highly expedient approach to pharmacologically privileged steroidal scaffolds will be applied in the coming new year to the production of architecturally ornate natural products or to new and improved drug candidates.


  1. A small nitpicking detail - the catalyst is not as inexpensive, because the ligand takes 3 steps to make, and even though copper is cheap you need 2 equivs of AgSbF6 to turn the dichlochloro complex into cationic (and highly Lewis-acidic) hexafluoroantimonate salt. Also, the catalyst is dramatically attenuated by moisture, so the anhydrous catalyst needs to be weighted out in a glovebox. (There are Cu(II)-Box catalyzed reactions where the hydrated version works and is even preferable, and those are easy to set up without a glovebox, but I don't think this is one of them)

  2. Your point about the cost of the catalyst is well taken. Copper may be inexpensive, but there are other equally important factors, including effort for ligand synthesis, that contribute to the overall cost of a catalyst. As far as sensitivity to residual moisture, that parameter tends to become less important as a process is scaled up because of the increasing volume/surface area parameter. Also, the SI of the Nagorny manuscript makes no mention of the use of glove boxes or Schlenk apparatus. If glove boxes are required for handling Cu(II)-Box and/or running the actual Michael additions, this would be a practical limitation of the technology. It would also be regrettable if this detail has been excluded from the SI.

  3. I don't think you need glovebox to do wet chemistry (running the actual reaction) but you may need one for storing and weighting out the catalyst - because the complexation, anion exchange, drying is labor and time consuming, so you may want to make a gram-sized batch of catalyst and spoon out 20-30mg for individual reaction trial. My own experience with these catalysts is it takes just a little sloppy technique on humid Boston summer and the the hydration happens right away - the pale green color changes to bright blue and you know you hydrated. (In my case, the hydration saved my ass, as the anhydrous catalyst was ripping apart my substrate - in anhydrous form the complex is so hot lewis acid that it is not compatible with MTBE or even with chlorofom. (As you have noticed, they used simplified substrates in several instances, withe fewer carbonyls - each extra Lewis-basic group probably slows down the reaction, by attenuating the catalyst. This is one limitation of this Evans asymmetric system - anything that strongly coordinates to copper (i.e. nitrile, amide) interferes with the activation through bidentate coordination.)

  4. Thanks for the info! Your comments are insightful, as always. I have to admit that I've grown to really enjoy the recent trend of including lots of pictures and detailed step-by-step protocols in SI documents for organic synthesis papers. It removes so much of the guess work from the science and makes procedures generally more reproducible. At the end of the day, reproducibility is everything in science.