Let’s talk about enzyme envy. That’s what we organic chemists experience when we stop to think about how every complex natural product in the world is synthesized so much more quickly and efficiently than we can do it. All those crazy multiple rings systems, those bizarre heterocycles, huge macrolides, and dense arrays of stereochemistry are cranked out at ambient temperature, under aqueous conditions, on a time scale of minutes to hours. Oh, and they’re made offhand, in the background, as time permits, while the cells and organisms themselves are otherwise occupied with the even more startling business of being living chemical systems. Envy? It’s more like enzyme terror, when you really think about what’s going on.
Now let’s talk about nanotechnology. Not the press-release stuff, I mean hard-core nanotech. The dream of assembling molecules and materials to order, atom by atom, has been around for some years now, and it largely remains just that: a dream. Eric Drexler and others have taken a lot of grief for maintaining that such things are possible, and to be sure, I have trouble myself with the real atomic scale this-carbon-goes-right-here level of the idea. But what about “this acetyl group goes right here?” That is, small-molecular scale versus atomic scale? That is exactly what every living cell on the planet is doing right now – that’s enzymatic chemistry, and I see no reason why we can’t get smart and capable enough to do the same sorts of things ourselves, and more.
That leads up to this new paper, which is a big (but small) example of just that sort of thing. A group from the University of Manchester reports the system shown above, which is basically a molecular-sized stereoselective synthesis machine. You put that unsaturated ester together on the left-hand side, and the first step is an olefin cross-coupling with it and acrolein dimethyl acetal. Now comes that “rotary switch” in the middle. If you treat the molecule with trifluoroacetic acid, it deprotects the acetal, and the acrolein now is set up to do a tandem nucleophile/electrophile addition under the influence of the chiral prolinol group. That reaction is done in solution, as stereoselective synthesis fans will know, but you only really get the syn isomers.
This system will give you both, and both enantiomers, depending on the reaction conditions. Depending on the conditions (time spent treating with trifluoroacetic acid, or with triethylamine afterwards), that hydrazone in the middle can go from E to Z putting the acrolein aldehyde in range of either the (S) or the (R) prolinol, at your choice. That forms a chiral iminium, whereupon you add a thiol nucleophile. The iminium is now released, and you have your choice again of whether you want the hydrazone in the middle to be E or Z. Whichever one you choose, you now form an enamine intermediate and bring in the electrophile (a highly activated Michael acceptor, 1,1 phenylsulfonylethylene). That does the second bond formation, and reduction with LAH releases the ester linkage and frees the product as a primary alcohol. You end up with two chiral centers, and you can choose which of the four diastereomers you want by moving the hydrazone switch around during the reaction sequence.
OK, we’re not all going to be making compounds like this next week – fine. This system is set up with every bias possible to make it succeed (such as a terrific thiol nucleophile and a terrific sulfonyl electrophile), and it doesn’t produce pure stereoisomers out the other end, even with that. Rather, you get ratios from about 80:20 to about 60:40 (although, to be sure, the solution methods to do this reaction don’t always do much better, and they can’t give you all four isomers under any conditions). The point is not that this is a wonderful new stereoselective reaction that we can use to make chiral products; the point is not the chiral products themselves at all. The point, of course, is that this thing has been designed to give you such chiral products, in a programmable fashion, and that it can indeed switch states on command to give them to you. It’s a machine – a very small machine, with limited inputs, but the first computer circuits were also very small machines with limited inputs, too.
The scientific and engineering challenges are similar – how to make these things work more reliably and how to scale them up to greater levels of complexity and utility. It’s going to be harder for chemistry than it was for computation, since our computers, fundamentally, are all about moving 1s and 0s around (throwing on/off switches over and over in increasingly complicated ways). That made the development of processing circuits more straightforward, as difficult as it was in reality to get all that to work. Organic synthesis has a lot more letters in its alphabet than just A and B, and its processes are a lot more irreducibly complex than binary logic is. But that just means that it’s a trickier problem, not that it’s an insoluble one.
I think this work points to one way around the difficulties: try, whenever you can, to reduce the key steps to binary ones. The heart of this molecular machine is that hydrazone, which can point in this direction or in that one, which makes all the difference. If we develop these and other spatiotemporal switches to be faster, more responsive, more reliable, and more orthogonal to other functional groups, we can adapt them (as this paper does) to do some fairly complicated chemistry by positioning our other reagents in the appropriate places. Some of these things are already out there in the literature (azo groups or other double bonds that isomerize photochemically, pH-responsive hydrogen bonding groups, and others). We can take our cues from the biochemical world, which uses both of those just-named processes, and how, and we can design new ones that no living cell has ever gotten around to. Cells, when you get down to it, build incredible synthesis machinery, but it’s bespoke stuff: it does what it does with fantastic speed and precision, but if you don’t want exactly what it does, you’re out of luck. What we’re after is are tougher, more general solutions: machines that will make whole ranges of molecules, the way we tell them do, with (ideally) the ability to be switched around to different functions as needed.
T. Ross Kelly and Marc Snapper have a good commentary on this paper in the same issue of Nature.
It is commonplace to dismiss seemingly impossible ideas, such as Drexler’s molecular assemblers, out of hand, and the use of such devices in chemical synthesis might indeed never find favour. One could further argue that Kassem and colleagues’ “programmable molecular machine” is more contrived than ingenious. But given that the most recent chemistry Nobel prize was awarded for the design and synthesis of molecular machines, those who dismiss the concept of molecular assemblers should heed the lesson of Lord Kelvin’s infamous 1895 pronouncement that “heavier-than-air flying machines are impossible”.
As opposed to some other similar statements that get quotes, Kelvin apparently really did say that, as well as saying (in 1900) that he didn’t think there was anything new to be discovered in physics. Anyone who watched a bird knew that heavier-than-air flight was possible; the question was whether we could figure it out ourselves. The attempt to make molecular assemblers – which is, after all, the attempt to make our own enzymes, when we know that enzymes exist – is the same sort of problem. The molecular-machine folks (last year’s Nobel!), who care about rendering fundamental ideas and processes into molecular form, and the more traditional synthetic organic chemists, who care about what products any new reaction or machine can produce for them, can find common ground in building such things. Bring on the molecular machines!
Note: All opinions, choices of topic, etc. are strictly my own – I don’t in any way speak for my employer