This blog article refers to the very recent work of András Perczel and colleagues in the paper Four Faces of the Interaction between Ions and Aromatic Rings (D. Papp, P. Rovó, I. Jákli, A. G. Császár, A. Perczel J. Comput. Chem. 2017, DOI: 10.1002/jcc.24816). This work is particularly interesting as it uses a mixture of data driven approaches from crystallography and structural biology as well as high level Quantum Mechanical (QM) calculations to answer a question that is raised fairly regularly in molecular design in structurally enabled projects – that of how do we optimise interactions between ionically charged species and aromatic systems.
Biology uses ionic-to-aromatic (IAr) interactions to stabilise macrostructure of proteins and other biological ensembles. Often aromatic residues such as phenylalanine (PHE), tryptophan (TRP) and tyrosine (TYR) interact with charged residues (e.g. negative charged residues (asparagine (ASP) and glutamate (GLU)) or positively charged residues (arginine (ARG) and Lysine (LYS)) to energetically stabilise proteins and peptides. Fundamentally this is the interaction of the charge of the ion and the quadrupole moment of the ring. If we understand this, and the correct vectors and applications of electron density, then we can use it to improve the interactions of aromatic rings in our drug molecules versus charged residues in a target. Take, for instance, a kinase; There are charged catalytic residues in the pocket which are key to activity. Can we use the understanding of these interactions to better get our aromatic rings in our inhibitors to bind to them / disrupt them?
Fig 1: The interaction preferences of a cation (CP), or an anion (AP) either co-planar (ǁ) or perpendicular (┴) to the ring. The darker green represents the most favoured vectors.
The authors investigated the Protein Databank (PDB) and the Cambridge Structural Database (CSD) to pull information on evidence-based interaction vectors, before engaging in ab initio calculations using Quantum Chemical approaches to attempt to quantify the kinds of energies involved. Below you can see the typical angles and distances of interaction between various ions and aromatic residues from the PDB.
Fig 2: Occurrences in the PDB vs. the plane angles of interactions between various residues. Plots on the right demonstrate also the distances of these interactions.
This crystallographic information can help demonstrate which vectors and distances are preferred when designing interaction partnerships in your ligands.
The authors also use high level computational methods (FPA, NBO Hartree-Fock) to demonstrate complex electronics situations of electron-rich and deficient-rings in both small molecule and single point ions to give a semi-quantitative value of interactions (in kCalmol-1):
CP┴ (23–37) > AP┴ (14–21) > CPǁ (9–22) > APǁ (6–16)
Notes from the blogger (who’s thoughts are his own)
Aside from the computational chemistry calculations, the authors have demonstrated how a simple search of available databases such as the CSD and PDB can be used to mine meaningful incidental information for drug design. There are implications of using PDB data however in that the mass of crystallography was shot using various conditions, including salt and pH variations between structures. This may weaken the interaction strength between solvent accessible residues across the structures – this wholesale big data approach should be taken with slight caution for this reason.
The information gathered is quite intuitive to the med chemist, but helps to cement in ideas when designing ligands – either how to enable their rings to better make use of charged interactions, or, more subtly, if the rotamers of an aromatic ring is stabilised by one such charge, how best to use the stabilised vectors to go after other things in the pocket.
Their calculations help set up a semi-quantitative design rules, which may help drive interaction priorities, but as for the actual values, well, they may need to be taken with a pinch of something ionic…
Blog written by Ben Wahab