Friday, December 21, 2018

Simply Strained: Total Synthesis of Herqulines

Our recent syntheses of herqulines B and C have been published today in JACS. While the paper describes the direct nature of our final approach towards these highly strained alkaloids and even hints at some of the difficulties encountered along the way, the truth is that the journey towards the herqulines was a two-year campaign plagued by failure and confusion. There were many lows along the way, but there were also some highs that made it all worth it. In the end, I learned many lessons about chemistry, but more broadly about science, problem solving, time management, and the formula for a truly winning team (huge shoutout to my partner in crime, Chi He). As I’m sure many of you know, what is reported in a synthesis paper is most likely the end result of many more lessons learned that are not visible in a short communication. Thus, I hope to convey a sense of this in the space below.

I was first introduced to the herqulines in the fall of 2016 as part of a thought experiment that we lovingly refer to in the Baran Lab as synthesis “boot camp.” In contrast to some of the other targets pursued in our lab, these tiny alkaloids don’t look like more than a simple reduced dipeptide upon initial inspection of the two-dimensional structure. It wasn’t until I built a plastic model of herquline B that I realized what a beast this thing truly is (late-stage x-ray structure pictured, gives one a decent idea what we’re dealing with). Strained is an understatement when describing the herquline core. The piperazine ring sits at the base of the macrocycle, with two six-membered rings flanking each side of the heterocycle in a
bowl shape. Throw in a pair of weird skipped enones and a 1,4-dicarbonyl at the strained ring juncture and you’ve got yourself one unhappy customer. The structural features, it turns out, were just the beginning of challenges involved when dealing with these alkaloids. As I would come to appreciate in short order, strained diamines are not only very challenging to handle from analytical and purification standpoints, but are also exquisitely unstable when the wrong transformation is applied at the wrong stage of the synthesis. In short, this was a proper adventure in total synthesis. So let’s get down to the nitty gritty. Let’s get the show on the road...

Our first approaches were predicated on the (spoiler alert: incorrect) precedent that Birch reduction of a strained biaryl was not a viable strategy towards our target. I’ll spare you from the gory details, but well over a year was spent perusing a host of strategies that ultimately led us back to the strategy in the paper, in which the macrocycle is built as a biaryl DKP.
Our task was simply to defy all precedent and reduce two aromatic rings and a DKP all within a highly strained, unstable core. Initially, I set out to reduce the DKP, while Chi fearlessly embarked on a quest to perform the first Birch reduction. For the DKP reduction, most common hydride reducing conditions resulted in either decomposition or monoreduction of the tertiary amide. One standout example, however, was Brookhart’s Ir-catalyzed amide reduction mediated by diethylsilane. However, the resulting biaryl diamine was highly unstable in our hands, and could not be isolated. It seemed that the strain was just too much and thus one of the aromatic rings needed to be reduced first.

To that end, Chi had screened a range of Birch conditions and protecting group swaps and was beginning to see some potential (see figure).

While the regioselectivity was incorrect, it nonetheless proved that this strategy could potentially be fruitful. The true complexity of the situation was soon realized when we discovered that when the DKP nitrogens are both N– H, the regioselectivity flipped and it matched our product’s requirements! This was a truly exciting day but much work still needed to be done, as the natural product of course has one methylated nitrogen.

We quickly prepared the asymmetric DKP macrocycle found in the herqulines (up until that pointN-methylation was somewhat of an afterthought) only to find that the regioselectivity was again incorrect. To complicate matters further, the resulting styrenyl position readily was reduced under most Birch conditions screened, and we even saw some elimination of methanol and additional styrene reduction.

screening all known metals, solvents, temperatures, concentrations, and 
additives, we had only proton sources to screen before giving up on yet another strategy. We screened a bunch of proton sources which all gave the same negative result, with the exception of 2,2,2-trifluoroethanol, which cleanly delivered the desired regioisomer! This breakthrough was the result of empirical observation and to this day the mechanistic underpinnings of this selectivity are at best conjecture.

Recalling our above efforts to effect DKP reduction, Brookhart’s Ir-catalyzed amide reduction again reduced both amides to the corresponding piperazine. We were now one Birch reduction away from the promised land and excitement was brewing.

Of course, the synthesis gods were not ready to smile upon 
us yet, as every Birch condition screened resulted in either no reaction, or reduction of either homobenzylic position prior to any reduction of the second aromatic ring. We were back in the familiar position of despair, beginning to think that at this advanced stage we would ultimately fail once again, when it occurred to us that ketalization may be a viable workaround, given its ability to relieve strain (two more spcarbons), alter the conformation, and remove a double bond (improving selectivity).

After some screening, the ethylene glycol ketal proved a competent substrate and the second Birch proceeded cleanly, 
with the correct regioselectivity. Simple hydrolysis delivered herquline C, which, after a few days in CDCl3, equilibrated to an approximately 1:1 mixture with its more stable congener, herquline B. Treatment of herquline C (or this mixture) with DBU in toluene brought about clean conversion to pure herquline B, thus wrapping up an adventure that we will not soon forget. In the end, the simple, direct approach panned out. When I look at this route now, this pathway even seems obvious, but if you’ve read this far (sorry if I got a bit long-winded), you’ll know that this route was the result of much experimentation and intimate understanding of the reactivity and stability of every intermediate along the way. In short, this was a hell of an adventure in total synthesis. Thanks for reading!

Tom Stratton
La Jolla, CA 

December, 2018

p.s. See also the Wood group's elegant synthesis of the same molecules!