“It’s not the end of the earth. But you can see it from there.” That was Lou Holtz, talking about coaching football in Fayetteville, Arkansas. But today I’m talking about a new paper from Marty Burke and his group at Illinois, and although it isn’t the end of organic synthesis, you can see it from there.
Now, that sounds a bit frightening, or a bit idiotic, or maybe a bit of both. But have a look at the paper. I had a chance to see him talk about this work a few months ago – I found it fascinating and startling, and I’ve been thinking about the implications ever since. This paper is the perfect opportunity to talk about it all (here’s a commentary at Science). It’s a summary of a lot of work that the Burke lab has been publishing over the last few years, and when you put it all together, there are some far-reaching consequences. On one level, it’s about assembling sets of molecules from modular building blocks, each containing MIDA boronates and bromides. That’s been a worthwhile reaction to study, since these boronates are very easy to handle and shelf-stable. What Burke’s group has found, though, is that the MIDA complexes have an unusual property: they stick to silica, even when eluted with MeOH/ether. But THF moves them right off.
This trick allows something very useful indeed. It’s a universal catch-and-release for organic intermediates. And that, as the paper shows, opens the door to a lot of automated synthesis. You take a MIDA boronate intermediate, and deprotect it to the free boronic acid. You then couple it to another intermediate, which has a reactive bromide (or what have you) at one end, and another MIDA boronate at the other. The solvent switch lets you purify the crude reaction by loading it onto silica, washing everything else off with MeOH/ether, and then eluting the MIDA-containing product with THF. Then you do it again. And again.
The paper shows a wide range of products produced in just this fashion. Yields are decent, although varied, but there’s always product coming out the other end at some level. The number of possible compounds that can be made in this way is limited, at the first level, by the number of MIDA-boronate containing intermediates that you can synthesize, and you can certainly make a heap. At the second level, it’s limited by the sorts of couplings that boronic acids can do, and we still don’t have general methods to make them do bond formation between saturated carbons very well. But that’s an area of intensive research, and it looks like a solvable problem, eventually. I would go so far as to suggest that this paper makes a good case for trying to get this to work with boronic acids (as opposed to alkylboranes, etc.), because of the immediate application of the catch-and-release purification, but we’ll see what happens.
What gets me about this current paper, though, is the concept behind it. This has the potential to take a large part of organic synthesis into the realm now occupied by peptide and nucleotide synthesis. Those two are certainly easier problems – you have one kind of bond between every subunit, and a limited number of subunits themselves. But the advent of solid-phase iterative methods to synthesize these sorts of molecules was still a huge advance. It took making such things out of the realm of every-one-a-new-individual-challenge, and into the world of “Sure, we should be able to make that. Fire up the machine.”
That first category, we should note, is where total synthesis of natural products has traditionally been. And proudly so. I’ve had a lot to say about that over the years around here, going back to 2002, but I’ll summarize: I think that total synthesis was, at one time, one of the most vital and important parts of organic chemistry. But that day is past. Modern analytical methods have largely (although not quite totally) eroded the structure determination reasons for doing it, and modern synthetic techniques have put a vast number of molecules within theoretical reach. “Theoretical”, in this case, meaning “Given enough postdocs, enough grant money, and enough time”. That certainly wasn’t always the case. When Woodward, Stork, or (fill in your favorite here) started out to synthesize some complex molecule back fifty years ago, it was often not very clear at all how one might go about it. Just coming up with a semi-plausible synthetic route was a real intellectual accomplishment, and dealing with what happened when these ideas met the real world was another. Total synthesis took all the brainpower and all the skill that could be brought to bear on it.
It’s still not easy. But it’s sure not the same. It’s much harder to draw a molecule that’s truly a stumper these days. We have so many reactions and approaches that you can generally come up with at least a paper synthesis – mind you, it may not be a very nice paper synthesis, but in the old days you probably couldn’t even come up with that much. So if fewer and fewer molecules really are an adventure – or really promise to advance human knowledge in the course of making them – what’s left?
What’s left, I’d say, is for organic synthesis to get braced to take the next step. That is, it needs to stop being an end in itself, and start becoming a means to other ends. That’s already what we use it for in drug research – the only reason we do organic chemistry is that we don’t know any other ways to make small-molecule drug candidates. In the earlier stages of a project, we don’t much care about way we make things, just so long as they get made. As I’m fond of saying, in discovery med-chem, there are only two yields: enough and not enough. Did you make a sample of the compound that can be tested in the assay? That’s enough. And that’s the primary concern – how you made it is secondary. This is sometimes a bit of a surprise for people coming from high-powered academic synthesis groups, because you can do an awful lot of good med-chem using just reactions from the first semester of sophomore organic chemistry, and you can do an awful lot of good med-chem while putting up with reaction yields that no academic group would stand for. But one adjusts.
We may all need to adjust. What if this MIDA boronate protocol, or some later variant of it, starts turning big swaths of organic synthesis into a process of stick-the-pieces-together? Like peptide synthesis? These routes may not be the most elegant and highest-yielding things ever seen, especially not at first. But that leads to the question of why you’re making these molecules in the first place. Are you making them so that you can do something with them – test them as drugs, use them as nanotech building blocks, make a new battery or solar cell, investigate a new kind of material? Then fine – you probably have enough now to get started on the next phase of that idea, thanks to this Synth-O-Matic over here. Or are you trying to make the best possible synthesis of your molecules (fewest steps, highest yield, etc.)? In that case, you need to be careful. That’s a very worthy goal if you already know that this is a valuable molecule, which is what the process chemists do in industry. But if it’s just another new molecule, then why are you optimizing its synthesis? If along the way you’re discovering new and better synthetic reactions and protocols, then good for you – but I would define “better” as “better able to be used to crank out new molecules for other purposes”, not “done in five fewer steps than the last group had to use to make the same molecule”. Not that alone. Not any more.
If organic synthesis become modular, then the new chemistry and new reactions are going to go more into making new modules. All our problems are still there – tricky functionality, multiple chiral centers, quaternary carbons. But if we end up making large molecules mostly by looking for boronate disconnections and stitching the pieces together, then we’re on a hunt to make the pieces, not to make the whole molecules.
But what about the art? What about the elegance? Well, we’re going to have to say goodbye to some of it. The printing press drove fine hand copy from the world – you don’t see so many gold-leaf illuminated letters any more. More recently, and in our own field, the advent of modern analytical chemistry drove out the classic methods of structure determination. Now there was a puzzle worthy of the finest thinking that could be thrown at it. Old-fashioned degradation and derivatization was a fiendishly difficult challenge, like playing chess with the lights off and the moves called out in a language you don’t know. But that kind of chemistry is gone, totally gone, and it’ll never come back. No one does it like that any more. There were chemists who just couldn’t face that, when it happened back in the 1960s and into the 1970s, when they found that what they were really good at was no longer of value. It was hard. But organic synthesis may have to face up to the same sort of realization, that time has overtaken it and that arts gratia artis is no longer a fit slogan to work by. This paper today is the first one that’s really made me think that this transition is in sight. For me, organic synthesis is never quite going to be the same.
But in science, when something dies it’s because something else is being born. The idea, the hope, is that if the field does become modular and mechanized, that it frees us up to do things that we couldn’t do before. Think about biomolecules: if peptides and oligonucleotides still had to be synthesized as if they were huge natural products, by human-wave-attack teams of day-and-night grad students, how far do you think biology would have gotten by now? Synthesizing such things was Nobel-worthy at first, then worth a PhD all by themselves, but now it’s a routine part of everyday work. Organic synthesis is heading down the exact same road – more slowly, because it’a a much harder problem, but (I think inexorably). Get ready for it. We’re going to need to stop being so focused on just making molecules, and start to think more about what we do with them.
Note: for previous (and partly superseded) thoughts here on automated organic synthesis, see this post.
Update: for more thoughts on this, see here.