Monday, January 26, 2015

A computational drug design success story

Kiss, Gyöngyi N., et al. "Virtual screening for LPA2-specific agonists identifies a nonlipid compound with antiapoptotic actions." Molecular Pharmacology 82.6 (2012): 1162-1173.

Patil, Renukadevi, et al. "Design and Synthesis of Sulfamoyl Benzoic Acid Analogues with Subnanomolar Agonist Activity Specific to the LPA2 Receptor." Journal of Medicinal Chemistry 57.16 (2014): 7136-7140.

Highlighted by +Jan Jensen

Figure 1. Reprinted (adapted) with permission from J. Med. Chem. 57, 7136. Copyright 2014 American Chemical Society

I always listen to podcasts while I vacuum and last Sunday it was this Five Live Science podcast. At 19:48 a new segment starts on a new anti-radiation drug developed by Gábor Tigyi and co-workers (read more about it here).  When asked about the discovery process Tigyi replied
The process began many, many years ago when, with medicinal chemist colleagues, we introduced simple modifications into the natural compound and that got us somewhat towards the ultimate goal of making a drug-like substance, but it didn't work very well. So then using computers to design drugs, we have been able to screen millions of compounds and we have found structures that fit in like a key into the keyhole.
Always interested in a computational drug design success story I dug a little further and here is what I pieced together. In the 2012 paper by Kiss et al. the authors write
In a virtual screen using a structure-based pharmacophore of LPA$_1$ (Perygin, 2010), we serendipitously identified compound NSC12404, which was a weak agonist of LPA$_2$
The reference is a PhD thesis, which I haven't read but, but the word "serendipitously" leads me to suspect that there is an interesting story here. Anyway, NSC12404 was used as basis for a similarity search of the UC-DCC chemical library using Pipeline Pilot from Accelerys and the 225 returned hits were trimmed down, using the diversity subset function in MOE, to 27 compounds for experimental testing.  Of these 3 (Figure 1) proved the most promising.

Patil et al. (2014) then used Autodock Vina to flexibly dock 10 scaffolds based on 3 to a homology model of LPA$_2$ and 4 was selected as the most promising and binding was verified experimentally. Further docking studies suggested that potency could be increased by a modest lengthening of the chain linker region (Figure 1) and SAR analysis helped guide the optimization of the head group. One of the resulting compounds (12a, R$_1$ = COOH, R$_2$ = R$_3$ = H) is the anti-radiation drug mentioned in the podcast

This work is licensed under a Creative Commons Attribution 4.0 International License.