Thanks @Wal-Ward, really appreciate the incredibly detailed writeup. I will make sure this gets passed on to everyone on the assay teams. I don’t feel qualified to comment on all of the issues myself, but recognize their importance from what you and others have voiced in this thread.
I can comment on a few issues though:
Yes, great idea. We are working on it! And to some of your other points, we do have crystal structures of some of the amino-pyridine non-covalent hits which are quite ligand efficient. We also have a very new crystal structure (should be released this week) on one of the <10 uM quinolone hits; no covalent bonding is seen in the structure.
These quinolones (2-hydroxyquinolines) look like molecules with strong fluorescence, is it checked that this does not mess up the read-out of the fluorescence based affinity assays?
@Wal-Ward: Is there any precedent for the need to control the buffer redox potential explicitly using defined ratios of GSH:GSSG rather than simply adding GSH and some quantities of DTT/TCEP?
Hi John. My understanding is that the cytoplasm (in liver) typically contains around 5 mM GSH and 0.1 mM GSSG (Kosower & Kosower, 1978, Int Rev Cytol 54, 109). It could be a bit more oxidising in lungs? 5 mM DTT or 1 mM TCEP is typically used in assays of cytoplasmic enzymes (because more stable than GSH in air). The only time I have heard of people using GSSG is when looking at enzymes which reduce it back to GSH. DTT and TCEP may be more reducing than required to mimic the cytoplasm and this sometimes can give problems, especially for inhibitors of enzymes with a catalytic Cys. Even GSH can give issues with compounds which react with Cys. Hope this helps, Wal
I’ve taken a quick look at the output and TRY-UNI-714a760b-6 (IC50 = 25 µM; crystal structure: x2646) particularly caught my eye. in particular, the amide nitrogen looks like it could be used to link a warhead (my best guess would be aldehyde with a methylene linker although I’m not currently in a position to model it). As an aside, an aldehyde would also provide synthetic access to acrylamides and vinylsulfones.
I recall that one of the problems with DTT is that it can lead to hydrogen peroxide formation under assay conditions. TCEP is a phosphine and I believe that it doesn’t lead to hydrogen peroxide formation (at least to the same extent as DTT). Here’s a review on the topic that may be helpful:
I first encountered these issues working on the apopain (Caspase-3) project in Wilmington in 1997-1998 although we didn’t fully recognize the nature of the problem at the time (David Aharaony and colleagues subsequently did some great diagnostic work although I don’t think that it was published). Here’s the general structure of the problem compounds:
I also recall some issues on the PTP1B project (closed around 2004) that were resolved by using TCEP in place of DTT. We were working with nitriles on both Cat K and Cat L projects and we didn’t have redox issues (we had run HTS against Cat L and none of the output was considered to be worth pursuing).
I might be a little late in noticing this, but I just realised that we can now sort compounds by the tags; and one of them allows us to sort by chloroacetamides, for instance.
This is a wonderful way of implementing this suggestion and in allowing us (the med chem scientific community) to quickly sort through the data! Thanks!
