This article was originally published August 28, 2017.
Do you want to find new drugs treat human bacterial infections? (If you don’t, you’d better hope that someone else does!) The standard view in the field has long been that you have to target some essential pathway in those bacteria, one that doesn’t have a counterpart in human biology (or whose sequence and structure is different enough so that there’s plenty of therapeutic room). That makes a lot of sense, because then you’re maximizing the chances of killing off the infectious agent while minimizing the chances of human side effects. The problem is that this approach is leaving us with an increasingly limited field in which to work. The distinction in the field is between bactericidal compounds (the killers) and bacteriostatic ones (which just slow or stop growth), and in general, the feeling has been that since ars longa, vita brevis, you’re better off working on the former.
This new paper, though, a multicenter collaboration between the Broad Institute, Harvard, MGH, Chicago, and Argonne, suggests that we might be able to open up more. The authors describe an inhibitor of Mycobacterium tuberculosis tryptophan synthase, which is not an enzyme that one would normally think of targeting (it’s not essential for growth of the bacterium as long as there’s sufficient tryptophan available). To be fair, the team wasn’t targeting it – their lead compound came out of a flat-out screen of growth inhibition against one of the Broad’s chemical libraries. A lot of the time, when you try that sort of screen, you end up discovering things that have been discovered before, and a lot of them are toxic to human cells, too. But in this case, BRD4592 emerged and turned out not only to be an inhibitor of the tryptophan synthase, but to bind at an allosteric site.
That’s an interesting feature, too. Allosteric inhibition has long been on the list of “Yeah, that’s really interesting, we ought to find some way to target that more often”. For those outside the field, an allosteric inhibitor works not by binding right at a protein’s site of action, but at some other region entirely. This binding event shifts the conformation of the whole protein around to the point where its activity is affected. Allosteric sites are often part of regulatory pathways, responding to separate signaling molecules to change the activity of the underlying protein. Targeting them deliberately is not so easy; I’d guess that the great majority of allosteric inhibitors have been found by more or less random screening, in the sense of “Hmm, that doesn’t seem to bind at the active site, must be some allosteric thing”. There are probably more of them known for receptor proteins than for enzymes, but it’s a field with a lot of interest.
The tryptophan synthase enzyme is a complex one, and the authors go into detail on how BRD4592 seems to work by several different mechanisms simultaneously (there’s a crystal structure to help). It’s a weird system: the compound increases the stability of the enzyme structure, and increases its affinity for its substrate, so you’d think it would be some sort of activator. But it’s a pretty strong inhibitor of activity. It seems to both keep an indole intermediate from moving between two active sites in the protein complex, and to stabilize the enzyme/product state (which keeps the enzyme from turning over and starting a new reaction). The authors believe that it might be the most mechanistically complex allosteric inhibitor yet reported – you would not have designed this one from first principles, that’s for sure.
The enzyme, for its part, seems to be conditionally essential: it’s apparently more important during infection of host organisms than it is in the culture dish. It’s long been realized that there are such targets out there in bacteria, but they’re understandably harder to track down, since these differences only show up under specific conditions, some of which may be hard to replicate. But this means that there are compounds that are bactericidal at times and bacteriostatic at others, and since we’re going to need all the help we can get against resistant bacteria, we’d better come to grips with these things.
BRD4592 itself, though, is probably not going to be one of those compounds. As a screening hit, it’s unoptimized for the properties needed for an actual drug (and indeed, it’s not active in mice because it’s so rapidly cleared by metabolism). But that’s what drug development does – turns screening hits into drugs – and this looks like it could be a good starting point. That azetidine ring with the alpha nitrile, essential for binding, might be tricky to deal with, but a valid hit is a valid hit. Let’s see if someone can make something of this chemical series, or of the target in general.