The protein database (www.rcsb.org) is a wonderful resource that provides plenty of ideas for structural biologists and chemists alike. While the value of pdb is self-explanatory when it comes to structural biologists, it is not that clear-cut as far as organic chemists is concerned. I do think we should all use this important and searchable resource. There are many reasons to like it and one of them is that you can get some interesting ideas about molecular design. Today I will discuss torsional preferences in protein-bound small molecules. Entropic tricks used by chemists allow one to narrow down conformational freedom that is accessible for a given small molecule and, regardless of your view on the entropy/enthalpy compensation (this should be a separate topic for a post), small molecules likely engage their targets by involving a “quantized” set of functional group orientations. But do these correspond to energy minima? For instance, what if you look at the diarylamine fragment shown below and ask a question about statistical analysis of all pdb instances where it appears bound to a protein? Might there be a preferred abc dihedral angle? If yes, what is it?
In this regard, I will post two important papers today. The first one is rather controversial. It has seen a fair amount of criticism in the literature. This work by Perola shows that over 60% of the ligands do not bind to their targets in a local minimum conformation (http://pubs.acs.org/doi/abs/10.1021/jm030563w). The criticism directed against this work questions the validity of geometrical parameters of small molecules determined by protein crystallographers (remember my old posts – these guys are a different breed when compared to chemistry-oriented crystallographers). The controversy here is connected to the age-old problem of resolution, I suppose. The second paper is by Hao and co-workers (http://pubs.acs.org/doi/abs/10.1021/ci700189s ). This work is about accessible ranges of geometrical parameters for functional groups in small molecule ligands. Figure 1 of this paper is really cool: it shows the conformational histogram derived from PDB X-ray structures as a bar diagram. The authors superimpose the number of occurrences of a given torsional motif with the potential energy calculated using DFT. This paper suggests that the most probable values of the torsion angles agree well with the calculated global energy minima. I think there are creative ways of using these kinds of findings in efforts to design constrained ligands.