Let’s talk chirality, and chirality of the most annoying kind. I realize that for many chemists that might be a tough competition!
Background for those outside the field: chirality is pretty much what we tend to call “handedness”. There are some objects that can exist in separate right-handed and left-handed forms – common examples are hands themselves (and the gloves that fit them), shoes, and hardware screws. Meanwhile, there are plenty of other objects that exist in a “non-handed” way – think of a dinner plate, a nail, a rubber ball, or a unmarked square plastic container. I’m starting to give the secret away there: the key to chirality is lack of a plane of symmetry. If you can imagine running an infinitely thin shiny reflective surface through the middle of something and the two sides of it are now perfect reflections of each other, that’s a plane of symmetry, and that applies to that second list of objects but not to the first. Chiral objects are “different in all three directions”, as opposed to nonchiral ones with some sort of symmetry to them.
The standard four-single-bonds carbon atoms that occur in organic chemistry can be chiral if they have four different things coming off of them. A carbon with a hydrogen, a fluorine, a chlorine, and another carbon coming off it has chirality, whereas one with a hydrogen, two fluorines, and another carbon doesn’t, because you can now show that there’s an internal plane of symmetry thanks to those two identical F atoms. Chirality is all over biology: all the main amino acids except glycine are chiral molecules, because they have a central carbon with four different substituents (glycine has two hydrogens on it there). The carbohydrates are chiral – glucose and the other simple sugars are almost entirely made up of chiral carbon centers! So proteins and polysaccharides are chiral, and so are a lot of natural products. And a fair number of drug molecules are, too. . .
We medicinal chemists would just as soon avoid chiral compounds when we can – it usually makes the syntheses harder, and it can put you in a position where you make the exact 50/50 mix (a racemate) and have to do a chiral separation to get the correct isomer purified. And what do you do with the other one once you’ve done that? Throw half of all your hard work into the incinerator? No, it’s much better to use chiral reagents and starting materials and make just the isomer (enantiomer) that you need, but that generally takes more time and effort, because there truly is no free lunch on offer in organic synthesis. There are drugs without chirality – aspirin and acetominophen are simple examples, but there are a number of much larger compounds (such as many of the kinase inhibitors) that avoid chiral centers, often by stringing together flat aromatic rings. But there are a lot of extremely important drugs that are unavoidably chiral.
Ibuprofen is a simple example there – as sold, it’s the 50/50 mix of its two enantiomers, and only one of them actually has the anti-inflammatory activity. Omeprazole is another one, which is why the follow-up to it was the single enantiomer that really had the acid-lowering activity. It’s become harder and harder to bring a racemic mixture to market, because you’re really dosing your actual drug at 50% purity, with the other half being a compound that might well go off and do something else that you don’t want. What about those two I just mentioned? Well, they have some special features: the inactive (R) isomer of ibuprofen is actually converted to the active (S) form by a human enzyme, so the problem pretty much takes care of itself. As for omeprazole, some people do the same thing, converting the inactive enantiomer to the active one. It depends on your personal suite of metabolizing enzymes. But as it happens, both enantiomers are converted to reactive achiral intermediates at the site of action in the gut anyway, so that lessens any potential problems as well.
Most chiral drugs, though, don’t have these funky interconversion pathways going; their chiral centers are more or less stable. But at the other end of the scale are compounds whose chiral centers are notoriously unstable, giving you compounds that are going to racemize no matter what you do. The usual way you end up with this is to have a chiral center on a relatively “acidic” carbon, with a hydrogen that is removable by any reasonable strong base to give you an anion that will give you a mix of both enantiomers when it reprotonates. Both carbanion and carbocation centers will generally scramble whatever chiral character they started out with, so if your drug structure can form such intermediates readily you’re in for some complications.
The classic example of this is thalidomide, whose two enantiomers are shown at right. It’s that carbonyl at 11 o’clock that’s the problem: the hydrogen on that carbon where the phthalamide nitrogen comes off is too easily removed, and the chiral center scrambles. This has led to no end of confusion, given the compond’s terrible history of causing birth defects in pregnant women. I wrote about this twenty years ago (!) on the blog – long story short, the tale about how one of the two enantiomers is the toxic one is not really true. The compound racemizes quickly in human dosing, so it’s not like you could even give a pure enantiomer and expect it to stay that way, and it turns out that in most animal species (including us) both enantiomers are tetratogenic. You actually can see different toxicities in mouse models at high doses, but in this case it’s the mice that are the weirdo outliers: they don’t racemize the compound much, and they’re not that sensitive to either enantiomer right from the start! Rats, for their part, don’t show the reproductive toxicity at all, while rabbits (and humans) are very sensitive to the effects.
Thalidomide (famously) made a comeback as a drug some years ago, mostly for multiple myeloma, with very careful monitoring to make sure that no one who could possibly be getting pregnant will be taking it. Investigations into its mechanism led to a real surprise: it binds to a protein called cereblon, which is part of a large ubiquitin ligase complex that’s involved in sticking ubiquitin residues on proteins and thus marking them for destruction in the proteasome. And that mechanism has been (famously) co-opted in recent years for “targeted protein degradation”, where you put something like thalidomide on one end of some sort of chain and a ligand for a target protein at the other end, and let the resulting ubiquitination complex go to town on that target to cause it to disappear from the cell.
TPD is really interesting stuff, not least because it’s a completely new mechanism for small-molecules drugs (OK, small-ish). But if you’re using a thalidomide-like molecule at one end, you are likely to have that same racemization problem, and what exactly does that lead you to? It’s been shown that the (S) enantiomer binds about tenfold better to cereblon than the (R), but there are crystal structures for each binding pair. They’re similar, but the (R) enantiomer has to twist a bit more to fit, and that exact energy difference is likely why it’s a weaker binder right there. The fact that the chiral center racemizes probably makes the (R) just a source of more (S) enantiomer for binding to cereblon in vivo, but what else might it do, in various TPD molecules?
Well, I sure don’t know – and that racemization is going to make it tricky to find out in animal models, too. As it stands, imid-based degrader molecules are indeed in the clinic and being vigorously investigated pre-clinically, but as far as I’m aware, the racemizing center on these is sort of “dark matter” mechanistically. Practically, it’s probably a difference that makes no difference, since you’re going to be dosing both enantiomers no matter what you do!
