I wanted to take a few minutes today to point out this paper, because although it tells a story that many have experienced, it’s a story that more people still need to hear. It’s about a reported small-molecule inhibitor of the “ten-eleven translocation dioxygenase” (TET) enzymes, which oxidize the methyl groups on 5-methylcytosine bases in DNA. That alters transcription in these sequences considerably, as you would imagine, and this sort of DNA demethylation has already been shown to be crucial in embryonic development, tissue development in adult animals, neuronal functions, and surely many more processes.
So an inhibitor of these enzymes could be a very useful tool to study all these important steps, and the paper under discussion focuses on Bobcat339, which is reported to be a low-micromolar inhibitor of TET1 and TET2. It came out of a chemistry effort at Bates College, and since this is exactly the sort of undergraduate research work that gave me my start at Hendrix, I recall being quite happy when I heard about this program a few years ago. You can purchase the compound itself from a number of the chemical supply houses, and its use is noted in several research papers since it became available. The other TET inhibitors tend to have poor selectivity for ketoglutarate-dependent enzymes or have possible structural liabilities, so this one could offer a better look into the whole system.
This new paper (from a team at Minnesota) unfortunately finds that resynthesized and purified Bobcat339 seems to lack TET inhibitory activity, though, while commercial material was still active. A closer look showed that this activity was completely explained by the level of residual copper(II) from the synthesis (which involved a Chan-Lam coupling step): the more pure your compound batch gets, the less active it is. The authors report that the material from MedKoo (the most active sample) was “seafoam green”, while samples from Sigma-Aldrich and the in-house material were both off-white. The Sigma-Aldrich material was also inactive in the enzyme assay, but both it and the freshly prepared material showed activity when spiked with copper(II) sulfate to take it to the levels of the MedKoo material (3 mole percent!) The Bobcat339-copper mixture is a better inhibitor than just copper sulfate alone, although the latter does show activity. The Cu-contaminated material needed more than one chromatography to clean it up, apparently, so it could well be a copper complex that’s causing the trouble. The TET enzymes have an active-site iron atom, and delivering copper in place of it would certainly inhibit them – but it will do a lot of other things to cells, too, and you really cannot trust cell assay data from something like that.
Well, experienced medicinal chemists will not be surprised by this sort of thing. Copper and other metals have long been recognized to cause just this sort of problem, and as you’ll see from this post I’ve been burned by it myself. It happens again and again. So in one way, I feel as if I shouldn’t bring this topic up, because everyone knows about it, right? But that’s the blind spot that you develop after years of experience, because it is obvious that no, not everyone knows about it after all. So whoever might come across this post, do yourself a favor: check your chemical probes. Check your screening hits. Check your batch of lead compound. Don’t assume that someone else has already done it – go check it yourself. Run some sort of metals analysis on it if you can (inductively-coupled plasma/optical emission is good), because many other analytical techniques might miss these things. If you don’t have access to that, then maybe take an aliquot and run it past some metal-chelating solid support, or if it’s a sample straight out of the archives, heck, just run it through a plug of silica gel and see if it looks different (and tests different in the assay!) after it comes out. And be immediately, strongly suspicious of any green or blue colors in small-molecule compounds, because those are quite rare in “ordinary” organic-chemistry structures. You should also be suspicious of noticeable shades of yellow, oranges, red, or brown in your small molecules, naturally, but those can be intrinsic to their structures a bit more easily (although they are of course often signs that there’s crud in there, too, and even if not such molecules can cause trouble with any sort of optical-detection-based assay or fluorescent system and should be checked for that as well). Green/blue, though, is almost invariably a sign of trouble, very likely copper due to its usefulness in many synthetic reactions.
But don’t think you’re in the clear if your material is merely a fashionable shade of off-white. Check it. Check it now before you make the same mistake that so many others have.
