3-bromopyruvate is an interesting and controversial compound. It’s been reported to be an active chemotherapy agent, apparently acting via covalent inhibition of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and subsequent metabolic effects via loss of pyruvate itself. Several years ago, you could come across numerous web pages touting it as a wonderful anticancer agent that Big Pharma was keeping from patients, and although that seems to have calmed down a little bit, such stories never really go away. The deaths of three patients in Germany a couple of years ago might have something to do with that.
Any chemist who looks at that molecule, though, will tell you that there’s surely more going on than just the GAPDH story. It’s such a small species, and has a reactive primary halide in it; it has to be hitting other targets. Indeed, it turns out that cells treated with the compound cannot be rescued by giving them back excess pyruvate, suggesting that this isn’t the only pathway that’s being messed up. Here’s a new paper that does just what needs to be done in this case: a full proteomics workup.
The authors do that in straightforward fashion – they synthesized a version of the compound with an acetylene tag on it and showed that it seems to recapitulate the effects of 3-bromopyruvate in cell assays. This was done by making a substituted oxime of the keto-carbonyl; the group refers to that as the amino-oxy (AO) probe. They then exposed cells to this version, lyse them, and use the alkyne as a way to functionalize the now-covalently-modified targets with either a fluorescent tag (to see them in a gel) or a biotin tag (to immobilize them on a solid support for later analysis). There are several controls that you want to run with these sorts of proteomics experiments – for example, competition with an unmodified ligand – but all of this is in order.
When you combine that with mass-spec based proteomic identification, you can assemble a plot like the one on the right. Chemical biology types will immediately recognize the “volcano plot” format, which you see a lot in proteomic and RNA-seq experiments. Along the X-axis you plot the change in the amount of protein that you captured in the AO probe experiments versus the ones where that were being competed with plain 3-bromopyruvate, and along the Y axis you plot the significance of that change. The stuff you truly don’t care about is down there in the caldera (not much change, not much significance), while the things of interest get sprayed up towards the upper left and upper right corners of the plot.
GAPDH is definitely one of those, which is a good sign that the experiment worked the way that it was supposed to. But there are plenty of others. This is, in fact, a fairly sprayed-out plot for its type, and the authors identify over 60 proteins that are affected by the bromopyruvate group in general (the significance-bearing upper part of the plot). The authors refer to that reactivity as “rather broad”, and they have a point. The proteins highlighted in that plot are already known to have reactive cysteine residues in them, and these are almost certainly what are being picked up by the covalent compound (alpha-halo ketones being well known for just that sort of thing). And if you wanted to bear down on it, you could probably find more – the paper shows some results of competition experiments versus both 3-bromopyruvate and an oxime derivative that doesn’t have the acetylene tag, and while they’re broadly similar, they don’t overlap perfectly. So a suite of different competitive reagents could provide an even larger picture.
What this says to me is that while GAPDH inhibition might well be a metabolic target for some tumor types, bromopyruvic acid is a pretty crude tool to accomplish this. You would worry that a molecule this small and this reactive would be accompanied by severe toxicity, and that’s just what you get. The volcano plot above is a first step towards understanding how that toxicity develops, but it’s also enough to serve as a warning. Giving this drug to patients was. . .a borderline call, to be sure, considering that targeting the Warburg effect by itself might well not be curative, rat studies aside.
On the other hand, nitrogen mustards are still chemotherapy agents, and it would be quite interesting to see what some of those volcano plots would look like. They are known to target DNA, but surely they hit a chunk of the proteome as well (and in fact, take a number of nuclear proteins and crosslink them to DNA). But in 2018, I have to think that “not necessarily worse than nitrogen mustard agents” is not a strong case. I can see the potential in topical application, but systemic seems a bit much. You’re going to want a better GAPDH inhibitor, and that may be in the works.