Full text of Rod’s bi-monthly newsletters, which include med chem literature reviews in addition to the more general papers included here, can be found on his website at http://rodporterconsultancy.com/newsletter/.
A useful review of the various aromatic descriptors that have emerged over the past few years has just published from a couple of the primary exponents of the negative role of too many aromatic rings has on developability.Interesting observations of marketed versus patented compounds included the fact that only 14% of marketed drugs compared with 63% of patented compounds had more than two aromatic rings. Furthermore Ar- sp3 (the number of aromatic atoms – the number of sp3 atoms in a molecule) stays fairly constant over time with marketed drugs – varying with the ionisation state. Namely acidics had a (high) value of 4.05, bases 0.73, neutal molecules -0.22 and zwitterions -1.48. Possibly this reflects issues around solubiity for neutral and zwitterionic compounds. Amongst the other developability characteristics of a compound solubility does seem to be influenced significantly by aromaticity (above and beyond any change in lipophilicity there may be) – cf a following discussion on predicting solubility.A useful addition to the supplementary material is a .csv spreadsheet which can be used to calculate a range AROM, Fsp3, Ar-sp3, Ar/HA, and PFI from any SMILES string molecule input if JChem for Excel is installed. Apart from this aid, this is a useful review of the various ways that have so far emerged for assessing impact of aromaticity on drug design and the developability of compounds and as such is well worth a read. Bottom line really seems to be keep control of aromatic rings and don’t use an aromatic when an sp3 motif will do – the odd chiral centre is no bad thing by all accounts in helping survive development.
Finding binding sites
An interesting attempt to try to unravel binding sites which caught my eye 1 although perhaps I am not best placed to say if this adds real benefit but it brings to mind some of the recent work I have been reading for personal benefit relating to some epigenetic targets where “gold standard” crystallographic data is failing to identify intact binding sites due to the crystallisation of protein constructs rather than whole proteins. Which of course then brings to mind the dangers of trying to assess druggability of a binding site based solely on published (often incomplete) protein structures or indeed protein complexes. Perhaps looking more holistically so called poorly druggable sites are more druggable than initially thought if whole protein or whole protein complexes were considered. This was I believe some of the point behind the work Cellzome was involved with in assessing target protein in a whole cell environment rather that as expressed proteins – HDAC as a target springs to mind where selectivity of compounds in isolated enzyme systems was very different to that in the whole cell due to the impact of other proteins on the target enzyme.
While predicting/calculating any property is tricky predicting solubility has proved particularly problematic over time. It has been argued that the problem is really due to the relatively poor quality of the data sued to generate models. However a new report suggests that the real problem is that the most appropriate properties to include in models have not yet been identified. The approach used was to compare models built with a collation of literature solubility data versus a set of compounds for which solubility as carefully measured to a high degree of confidence. The conclusion was that no matter which set of data was used the models generated were equally unreliable – back to the drawing board for the modellers!.
Perhaps there is a bit of a tendency to think of proteins as generally reasonably well ordered perhaps driven by our looking at elegant crystal structures – forgetting the omitted loops and other sections too mobile to give useful co-ordinates. A full issue of Chemical reviews 1 has been dedicated to Intrinsically disordered proteins with sections introducing the field and systems of classification. There are also, amongst others, sections on looking at free energy landscapes using NMR, multisteric regulation by structural disorder in modular signalling proteins 2, short linear motifs as diverse protein interaction modules involved in cell regulation 3, looking at conditionally and transiently disordered proteins and a new ‘omic – pathological unfoldomics of uncontrolled chaos. Perhaps this really highlights the dangers of over interpreting crystallographic data certainly during target evaluation and at least in some cases when using structure based lead optimisation.
P-gp substrates and translation to man
In an extension of earlier work in the mouse 1 the Terasaki group have reported 2 on reconstruction of distribution of P-gp substrates in the cynomolgus monkey brain to help increase confidence in translation of CNS drugs into man. “this study experimentally demonstrated that the Kp brain and Kp,uu,brain values of P-gp substrates and non-substrate can be reconstructed by integrating in-vitro P-gp transport activity, P-gp protein expression levels, and the unbound fractions in plasma and brain based on BBB PPx. These results also demonstrate that in-vivo P-gp transport function at the BBB can be reconstructed based on in-vitro P-gp transport activity and P-gp protein expression levels. The key points is that expression (and activity) levels of P-gp are much closer (ratio <1.5) between cynomolgus monkey and human relative to rodent/human and that Fuplasma varies substantially across species. The team also use pH adjusted Fubrain 3 to try and more accurately reflect the in vivo situation. This gives an in vitro method for helping to give an improved estimate of effective exposure in man particularly when working with compounds with evidence for efflux liabilityFu brain was broadly similar for mouse and monkey general differing by only two fold although indinaivr did show a three fold variation which is consistent with earlier data showing a broad conservation of Fubrain while Fu plasma is more variable across species.