In 2015 I wrote a post with the deliberately provocative title “The End of Synthesis” about some automated organic chemistry work being done at the Burke lab at Illinois. People occasionally send me reminders that several years have passed and that organic synthesis has not (as yet) ended. But although it’s a slow process, I think I’m going to stick with my overall worldview from 2015. Which is (1) that a great number of common synthetic manipulations – up to and including carbon-carbon bond formation – are capable of being run in a largely automated fashion, and that this is becoming more feasible all the time. And (2) that this opens the way to having libraries of automation-compatible synthetic building blocks, which will greatly simplify the construction of a large number of molecules. I should go on to (3), the prediction that this process will gradually erode the “bespoke” nature of a lot of synthesis work as it is done at present, and that (4) organic chemists themselves are going to have to adjust to this, because it changes the nature of the field from what it’s been since the 1800s.
You can see the beginnings of this in things like the Enamine library I was referring to the other day. The catalogs are full of synthetic building blocks with easily coupled functional groups (such as amines and carboxylic acids), and Enamine has taken this to the point of having very large collections of these, with an offer to synthesize any of the combinations thereof on demand. You end up with billions of potential compounds, far more than have ever been characterized in the actual chemical literature, and their synthesis is designed to be as straightforward and uneventful as possible – an afterthought, really. And that’s what more and more organic synthesis might slowly become as we develop more reliable reactions for the most common methods of assembly.
These thoughts are prompted by this new paper from Burke and his co-workers. They’ve modified the MIDA boronates used in the earlier work to make them compatible with a wider range of reaction conditions, specifically C-C bond formation with sp3 carbons (including stereospecific variations), while maintaining the properties (selective sovlent-switched adsorption and elution) that make them useful for multistep automated synthesis. Several examples are shown in the paper, among them the natural product Sch725674 at right. The structure is color-coded so that you can see which parts of it come from which building blocks (the cyclization is the last step). Now, if you’re going to apply this sort of thing across a wider range of organic synthesis then you’re looking at an awful lot of boronate building blocks – but you know, there are an awful lot of amine and carboxylic acid building blocks out there already. It’s just that there are huge numbers of known or desired compounds that can’t be snapped together only by amide couplings – but if you extend the “snapped together” concept to carbon-carbon bonds in general then you have extended it to, well, pretty much everything you’d ever want to synthesize.
And that’s exactly my point. This stick-the-pieces-together approach took over long ago in protein and nucleic acid chemistry work, for obvious reasons: the bonds being formed are generally the same, over and over, and the number of building blocks is far more constrained. Automation in these areas was absolutely inevitable – and strongly desired, too, if you’d ever had to make a large peptide or oligonucleotide by hand in the old-fashioned way. Organic synthesis is a much larger problem, of course, but transforming it into “carbon-carbon bond formation space” reduces it conceptually to its essentials. Add in the other functional-group-based methods we have for bond formation (amides, ethers, amines, sulfonamides), and if C-C bonds can end up in as general and realiable a shape as most of those are, then you’re looking at the slow emergence of a new world.
What happens to synthetic organic chemistry when more and more of it becomes a Lego-like exercise in sticking pieces together? Well, just as in every other field where the playing field changes location on you, synthesis will spend more of its time on the things that can’t be done in this manner. In fact, the field might find itself deliberately trying to find the failure points in these sorts of modular carbon-framework methods in order to have something to work on. Inevitably, such work will lead to some of these gaps being filled in by new chemistry that can be adapted into the modular format – in other words, once you’ve started down this road I don’t think you can avoid staying on it. I have more thoughts along those lines here. More broadly, as I say there, as synthesis itself becomes more routine it will push chemists to confront the harder problems, such as, well, what to synthesize and why. The “how” question that goes along with these has occupied the field for over a century now, but we’re eventually going to have to learn how to say goodbye to it, old friend that it is.
