Designing useful catalysts is one of the most challenging frontiers in chemistry, and it stretches across the whole field. Inorganic, synthetic organic, analytical, computational – you name it, and there’s a challenge there to pitch in on. The rewards are substantial. Without catalytic reactions, the modern world economy would come to a halt – you can start with what has to be done to petroleum to turn it into useful materials and go on from there. And life itself, at least anywhere near as we know it, would not be possible without catalysis by enzymes. There’s a very long list of crucial biochemical reactions that are sped up by huge amounts just when and where they need to be.
How does this voodoo get worked? A lot of it has to do with getting molecules into the right positions, holding particular functional groups at the angles and distances needed. Then you add in some local improvements in charge distribution and the like to make a carbonyl more open to attack, a proton more likely to be pulled off, a nucleophile more reactive. If you get these things just right, suddenly a slow reaction can start turning over at blinding speed, or a reaction that you’d never see at all will become useful.
But getting this to work means not only holding the target molecule in the right position, but holding the catalyst that way, too. Many catalytic sites are actually a suite of different functionalities, some of which would be incompatible if they were brought together in the wrong way. For example, you might need a basic group over here and an acidic one just up the way from it, but if that acid has a chance to just protonate that base, the catalytic site will be short-circuited and everything will just sit there. Engineering bifunctionality of this type can be quite tricky.
All this leads me up to this new paper, though, which presents an ingenious way to make large ring compounds. Such macrocycles are of great interest, since they can be quite complex molecules that have a much less complex conformational space open to them than the corresponding open-chain analogs. You find a lot of biomolecules and natural products taking advantage of that effect, which lets them achieve a useful spatial arrangement without paying a big cost in entropy as they get into and hold that position. But synthesizing macrocycles from open-chain precursors means that you have to prepay that thermodynamic penalty yourself, and it can be a real challenge. The two ends of the chain may have no real reason to see each other without something to direct them into place.
And that’s what this paper (from the Gellman group at Wisconsin) does. They’re using a “foldamer”, a polyamide chain that’s roughly like a protein and similarly can adopt a helical conformation. The side chains poke out at regular intervals, which gives you a change to arrange functional groups so that they’ll be next to each other – in this case, there’s a primary amine and a secondary one. These can react with long-chain dialdehydes so that one of them forms an enamine and the other an imine, setting things up for an aldol reaction as shown at right. And their foldamer does quite well at it. The paper shows a variety of 14- to 22-membered ring formations – if you take the precursors and just add, say, pyrrolidine and ethylamine to them instead (same functionalities but not spatially arranged), then you get very low yields of the macrocycles along with a lot of intermolecular aldol products, as you would expect. But the foldamer stitches them right up into rings in yields of 55 to 95%. You get an unsaturated aldehyde, as usual from the elimination of the aldol, which can be reduced or functionalized in turn.
There are some interesting SAR effects seen in the foldamer catalyst itself, and I can only imagine how many combinations have been tried out. If you space the two amine-containing residues differently along the chain, catalysis drops off hard (as well it should). Even if the residues are in the right position on the helix, the spacing of the primary amine coming out from it is also crucial. Swapping the positions of the primary and secondary amines on the effective catalyst still gives you product, but at lower levels.
These results are a strong argument for the mechanism shown; that sort of spatial dependence is just what you’d expect if that bifunctional route is operating. All sorts of subtle effects can come into play – conformational stability, modifications of the amine reactivity depending on its environment on the helical foldamer, availability for the reaction of the first or second aldehyde, preferred spatial arrangement of the imine/enamine pair, and more. Predicting these would be a major challenge; it’s frankly far easier to take your best guess, make the foldamers and related analogs, and see what they do. The modular construction of these polyamide chains makes that quite feasible.
What we’re seeing here is just the sort of thing that an enzyme does. There’s no natural pyrrolidine-side-chain amino acid that I’ve ever heard of, but if one were common this is one of the reactions you might see evolution finding a use for. For the full enzymatic effect, you’d probably have the rest of a longer type of foldamer molecule forming a reversible cup or cage for the substrate to crank the rate up even more, but going from 3% to 95% yield is going to have to be enough for us mortals. Nice stuff!