If I had to pick a broad theme about the evolution of medicinal chemistry over the years, I might well choose Things That Don’t Target Active Sites. I don’t mean that we’re abandoning such targets – far from it – but that there’s been a long-term movement into other types of compounds besides these.
It’s understandable that med-chem has spent a lot of effort over the years in the traditional areas, though. A lot of natural products work via these kinds of mechanisms, and natural products are where the vast number of earlier drugs have come from, and still more drug targets and hypotheses. Dropping some sort of inhibitor into an enzyme’s active site is probably the most clear-cut example of a small molecule drug effect that you could imagine, followed closely by dropping some sort of antagonist into the binding site of a receptor. And if your disease hypothesis allow for it, that’s frankly still the way to go, because we know so much about discovering drugs that way. It’s no accident whatsoever that the first bespoke molecules that are emerging against the coronavirus are things like viral protease inhibitors rather than (say) inhibitors of some protein-protein interaction needed for cellular entry.
But those protein-protein targets have been growing in importance over the course of my whole career, moving from “Are you sure that you want to try something like that?” to “OK, a PPI project, let’s see what we can do”. It’s still a lower chance of success than a straight enzyme inhibitor effort, but it’s become a much more normal part of drug discovery. Similarly, allosteric compounds, which don’t work at the site of action of a given protein but rather change its activity by binding somewhere else, cause a bit less stir than they used to cause in target proposals. They also have more complications than active-site compounds, but we’ve seen more of them over the years, and advances in structural biology have put many of these things on a firmer conceptual base as well. (I’m going to avoid for now the topic of ion channels and their multitude of binding sites; this post is long enough already!)
Here’s an example – that paper is looking at inhibition of human herpesvirus proteases, but really via the active site itself. Instead, there’s a non-catalytic cysteine residue nearby that can affect the protease’s activity, because the active form for many of these enzymes is a homodimer and it’s involved in that transition. (The paper itself has numerous references to other groups who have taken a similar approach in the last few years). Targeting HHV protease active sites directly is in fact pretty difficult – these enzymes tend to have a rather odd catalytic mechanism with a serine residue that’s less reactive than usual, and getting good small-molecule inhibitors has lagged. This team took some active-site compounds that weren’t good enough inhibitors on their own and added reactive groups that could reach the nearby Cys residue and form a covalent bond with it.
Now, people have tried this trick before on other proteins in order to make the inhibitors jam up the active site more efficiently, but that doesn’t seem to be what’s going on here, or at least not entirely. The monomeric protein has some disorder in the helices that are important for dimer formation, which settle down into an orderly structure when that occurs. But the site where these inhibitor compounds work is much less accessible in the dimer itself, so they’re probably not coming in at that point. The authors believe that the compounds have their best shot in the monomer-dimer equilibrium system, rather than doing all the work on either the pure monomer or the pure dimer, and that the protease activity is disrupted mainly by the compound’s presence at the interface of the dimer. This mechanism seems to hold across several HHV viral types (cytomegalovirus, herpes simples, Epstein-Barr), because all of them have conserved Cys residues in the right spot. As it stands, though, the compounds still have some off-target reactivity and will need to be optimized if they’re going to become drugs, but the principle is an interesting one to follow up on.
Here’s a 2018 paper that shows another allosteric covalent compound, this one inhibiting an enzyme that SUMOylates proteins. It’s quite an unexpected result – the compound was obtained from a high-throughput screening effort, and turns out to target a Cys residue in a previously unrecognized binding pocket of the protein. Once it sticks there, though, it totally changes the conformation of the active-site region of the protein, throwing that machinery into disarry and making it completely nonfunctional – the analogy of hitting the right exhaust port on the Death Star comes to mind. You’re not going to see this happen every time you go out screening for inhibitors, of course, but that fact that it can happen at all is motivation enough to go looking for some more. Here’s borussertib, a covalent-allosteric inhibitor of the AKT kinase (and more followup on that idea), and here’s an attempt to do something similar with the longtime-undruggable phosphatase enzyme PTP1B.
The big name in this area is of course KRAS, another longtime-undruggable cancer target, and targeting a mutant form of the protein that has an attackable Cys residue has led to compounds which can actually show clinical efficacy, such as Amgen’s sotorasib and Mirati’s adagrasib, among others. This one illustrates, though, that tremendous med-chem efforts and a great deal of biological research do not always translate into instant success: the G12C compounds have not performed as well under real-world conditions as single agents as people had once hoped. The story is in its early days, and there could well be much better outcomes in various combination trials, but hitting KRAS alone is not the knockout punch.
Non-active site mechanisms are not limited to covalent-allosteric compounds, of course. Here’s another set of examples of such ideas, also from the antiviral field. This paper is an overview of work being done on HIV, with attention to the broadly-neutralizing antibodies that have been the subject of a huge amount of research in recent years. These have the potential to be therapies on their own, but there have also been efforts to see if there’s room to develop small molecules that work similarly (these could have significant advantages in cost, storage, and dosing). Studying the binding sites (epitopes) that such antibodies target on the HIV viral particles also tells you where to target your small molecules, if you can find something that will also bind there. That’s a challenge, of course – antibodies can recognize more surface area on a target than you can deal with using a druglike molecule. But structural biology is getting better at working out the details of these binding events, and on the chemistry side, our definitions of what a druglike molecule is have evolved as well.
Fostemsavir is a recently approved example – this targets the gp120 surface glycoprotein of HIV and interferes with its entry into CD4 T cells. There’s another area on that protein (the Phe-43 cavity) that’s been the subject of a lot of work over the last twenty years, but so far no small molecules that bind to it have quite made it through the clinic, although improved ones have been reported. Another region that’s targeted by many antibodies (and is the subject of vaccine research as well, for the same reason) is the gp41 protein, specifically the conserved membrane-proximal external region (MPER). And there have been reports of small molecules binding to the same region. There has even been a paper showing that connecting both gp120- and a gp41-binding compounds to make one larger molecule might be useful – that takes you into some rather large molecular weights, but there are more and more serious drug candidates exploring this sort of structural and property space thanks to mechanisms like targeted protein degradation.
Now, none of these ideas (nor others like them) are simple to execute – indeed, some of these targets have been worked on for years now, with multiple failures along the way. But there was a time when none of these ideas would have gotten off the ground at all. All the upside and downside surprises with them represent hard-won knowledge that is being cycled right back into the next generations of candidates, naturally, and it’s good to see. The universe of small-molecule therapies is larger than the universe of active binding sites, and coming to grips with these things is an essential part of progress in the field.
