I think that most medicinal chemists look at the structure of ebselen and say “That’s not a drug”. Selenium atoms don’t belong in drugs, we figure, and Se-N bonds most certainly don’t. But it stands as a rebuke to our intuitions, because it’s been kicking around in the clinic for some time (and has certainly looked more interesting there than a lot of stuff that I’ve worked on over the years).
One recent application has been treatment of Clostridium infection. In 2015 it was reported that ebselen could be an antivirulence agent, and a mechanism was proposed where it affected the cysteine protease activity in the autoprocessing of the Toxin B virulence factor. That’s quite plausible, because (as you might well expect) ebselen is known for hitting Cys residues and inactivating them, which accounts for the weird variety of conditions and targets it’s been studied against over the years. The screen was for autoprocessing inhibitors, and ebselen bubbled to the top quickly.
Now comes word that further study has made things a bit more complicated. This new work took advantage of a Toxin B construct that had all of its Cys residues mutated, and you would expect ebselen to be pretty much useless in that situation. But it was equipotent versus wild-type and “de-Cys-ified” protein, which takes care of that theory. Moving to another domain of Toxin B, though, it appears that ebselen may be doing its work on the toxin’s glucosyltransferase activity. The human target of that activity is a GTPase protein called Rac1, and ebselen covalently modifies its three cysteines, which makes it no longer a substrate.
So far, so good: but the 2015 paper had specifically reported that ebselen didn’t have an effect on the glucosyltransferase function. The authors dug down into the details:
To reconcile these conflicting findings, we evaluated all aspects of the two studies to find any differences between our methodologies that might account for the discrepancies in our results. The answer turned out to be that the reducing agent dithiothreitol (DTT), which removes ebselen from Cys side chains, was included by Bender et al. in their assay of GTD activity, but not in their assay of APD activity (1). To demonstrate this, we tested inhibition of APD by ebselen in the absence and presence of DTT and saw that inclusion of DTT completely abrogated inhibition by ebselen (Fig. 1G). Similarly, addition of DTT to ebselen-labeled Rac1 relieved the inhibition of GTD activity (fig. S1). Thus, in the absence of DTT, ebselen can covalently inhibit both the APD and the GTD; however, as outlined above, on the basis of its ability to protect cells from intoxication by wild-type and Cys-less TcdB equally, it is not the APD inhibition that is responsible for the protective effects of ebselen.
There you go – if your mechanism of action depends on messing with Cys residues, the temptation is to make sure that you’ve added DTT to ensure that they’re all reduced and ready, but in this case, that same reagent is also wiping out the effects of the compound under study. This would seem to nail down ebselen’s activity against C. dificile toxin, but given the way that it hits proteins from every direction, it’s going to be hard to say when this story is over.
As everyone who’s worked on the compound knows, that also complicates its potential clinical development. Ebselen has been in clinical trials for stroke, and it’s being looked at now for bipolar disorder, suggested for use against osteoporosis, as an antifungal, and many more applications. Normally, one would get a bad feeling about a compound that seems to hit so many things (witness the curcumin story from the other day), but ebselen appears to be more real, although still a very hard candidate to develop. We’ll see if it ever finds a home!