More screening PAIN

Another valuable contribution form the Monash group on PAINS 1 – frequent hitters in screening assays. This time aminothiazoles are in the spot light – those favourites of early parallel synthesis aficionados with the mix of alpha-bromomethylketone and substituted thiourea. The authors observed issues with this motif in fragment libraries and in conventional screening decks – not necessarily that they were really bad just that they did hit frequently and were challenging to attempt to optimise – best steer clear.

In another paper 2 where the authors were screening for HDAC inhibitors using a thiol trap fluorescence read-out several series of compounds were highlighted showing activity due to thiol trapping which had not been detected with their cheminformatics filters prior to running the assay.  Compounds tended to show non-specific thiol trapping, furthermore these same compounds were shown to have activity across a range for targets (Pubchem) and unfortunately had been reported as tools for other targets. This paper is sumamrised nicely by Erlanson 3 with additional commentary on PAINS compounds

  1. S. M. Devine et al J. Med. Chem., Article ASAP DOI: 10.1021/jm501402x Publication Date (Web): January 16, 2015 Copyright © 2015, American Chemical Society
  2. J. L. Dahlin, et al, J. Med. Chem., Article ASAP DOI: 10.1021/jm5019093 Publication Date (Web): February 21, 2015 Copyright © 2015, American Chemical Society
  3. D. A. Erlanson J. Med. Chem., Article ASAP, DOI: 10.1021/acs.jmedchem.5b00294, Publication Date (Web): February 24, 2015 Copyright © 2015, American Chemical Society

Selective versus Specific

Am I the only one who worries about what might be a trivial point except for the ability to mislead when the word specific is used to describe how selective a compound is.  Specific to me implies that said compound interacts with one target and one target only – this we all know is nonsense.  At best a compound shows good selectivity (greater than 100 fold, greater than 1000 fold – whatever you might choose to define selective) against the targets it has been tested against and that of course assumes the assays can really be compared in such a simple way – is radioligand binding  a good measure of selectivity – what about the kinetics at target, functional efficacy etc.  Anyway back to selective v. specific the danger as I see it is when a compound gets described as “specific” any pharmacology seen with that compound is then ascribed to interaction with that target – which may be true but there again may not thus we can end up with a lot of fairly misleading literature.  Of course the worst case is when a compound is shown to have weak interaction against one target, tested against no another target and promptly described as specific – you will have all seen examples.

Perhaps journal editors and referee’s should ban the word “specific” from any article discussing the pharmacological profile of any compound – or am I being just being an unfair nit-picker!

Non-covalent hydrophobic tags and protein degradation

An article that recently caught my eye set me thinking about the liabilities of hydrophobicity – where a ligand binds to a protein, leaving a lipophilic motif exposed to cytosol. Long et al in Chemistry and Biology 2012, 19, 629 (see also Neklesa and Crews, Nature 2012, 487, 308  discuss the role of lipophilic motifs binding non-covalently to proteins, which directs the protein to the proteasome for degradation. Specifically, the authors investigated the binding of (Boc)3Arg conjugated Trimethoprim to DHFR and the resultant reduction of DHFR levels in mammalian cells. The effect was blocked by Trimethoprim or by proteasome inhibitors. Ethacrynic acid, a covalent inhibitor of GST, also induced degradation of GST when tagged with Boc3Arg.

The main thrust of the articles is in the context of a therapeutic strategy, but arguably, in the case of off-target proteins, could unanticipated degradation due to “tagging” by a drug result in toxic events possibly only manifested over the long term, if the lipophilicity enhanced degradation is modest?

While the authors discuss ligands with a Boc3Arg motif, it is not a huge leap to think of, for example, drugs with lipophilic biaryl groups that become exposed, on binding of a polar head group, to an off-target protein, with consequent demolition by the proteasome.

If this is truly physiologically relevant, can the (unwanted) degradation phenomenon be reduced with a more uniform distribution of polarity in a drug molecule? Or, put another way, is it better to have a molecule of lipophilicity X with a reasonably uniform surface distribution of polarity, or a compound with the same lipophilicity but a non-uniform surface distribution of polarity? Intuitively I would prefer the former, but if anyone has seen any analysis of the distribution of polarity and developability, I would love to see it. Given that lipophilicity seems to be discussed as a global phenomenon for a molecule, perhaps considering local lipophilicity may be useful as well.

A couple of additional relevant articles related to covalent binding of a lipophilic tag and proteasome degradation are Neklesa et al Nature Chem. Biol., 2011, 7, 538 and H. S. Tae et al ChemBioChem 2012, 13, 538