Protein-ligand interactions are driven predominantly by strong hydrogen-bonding interactions, de-solvation energetics, and other entropic considerations including ligand lipophilicity. However, the combination of many weaker interactions can contribute to potency, and are often necessary to obtain selectivity when targets and anti-targets have very similar binding pockets. Weak interactions such as Dunitz interactions, polarized C-H hydrogen bonding interactions, halogen bonds, and sulfonyl hydrogen bonds have frequently been observed crystallographically, but how relevant these observed interactions truly are to ligand binding has been debated. (For an excellent review summarizing molecular interactions relevant to the medicinal chemist, see this classic paper from Roche: C. Bissantz, Kuhn, B., and Stahl, M. “A Medicinal Chemist’s Guide to Molecular Interactions.” J. Med. Chem. 2010, 53, 5061-5084. DOI: 10.1021/jm100112j)
Recently, Bernd Kuhn, Erik Gilberg, and Oliver Korb from Roche and collaborators at the CDD published an elegant study in J. Med. Chem. addressing this question. (Kuhn, B., Gilberg, E., Taylor, R., Cole, J., Korb, O. “How Significant Are Unusual Protein-Ligand Interactions? Insights from Database Mining.” J. Med. Chem. 2019, DOI: 10.1021/acs.jmedchem.9b01545)
In short, the Roche team searched the PDB database for contacts between ligand atoms and protein binding sites, keeping only contacts where interacting atoms were in “line-of-sight” to one another. They then measured the frequency of these interactions appearing at appropriate interaction angles, normalizing against the probability that these interactions could appear by chance due to relative surface area. Some of the interactions that they picked out to be most significant in the PDB are summarized below:
Their study empirically supports the idea that these weak interactions can contribute to protein-ligand binding, and be highly competitive with respect to all other possible favorable interactions that might form in a binding site. Interestingly, they were not able to support many other interactions in this study. Most notably, the often-proposed C-F···H-N/O hydrogen bonds were not observed to a level achieving statistical significance, likely due to the fact that strong hydrogen bond donors often have many more favorable alternatives to bind to (such as water, or carbonyl groups within the protein itself). They were also not able to detect with significance aromatic sulfur groups bonding intermolecularly with electron-rich motifs through the sulfur sigma-hole. This may be due to the limited number of crystal structures capturing aromatic sulfur heterocycles, and warrants further investigation. All in all an excellent article and great team-brainstorming resource.
Happy hunting! Explore drughunter.com for more.