Acetyl-CoA carboxylase inhibition by ND-630 reduces hepatic steatosis, improves insulin sensitivity, and modulates dyslipidemia in rats

Harriman, G., Greenwood, J., Bhat, S., Huang, X., Wang, R., Paul, D., Tong, L., Saha, A.K., Westlin, W.F., Kapeller, R. and Harwood, H.J., (2016)
Contributed by Jan Jensen

This paper describes the development of ND-630 (aka NDI-010976) which is currently in Phase 2 clinical trials and could help cure a serious liver disease called non-alcoholic steatohepatitis and potentially other diseases. I am highlighting it here because computational chemistry had a lot to do with its discovery both directly and indirectly.

The development of ND-630 is spearheaded by Nimbus Therapeutics, which is basically an off-shoot of Schrödinger, i.e. a company that uses Schrödinger's software to discover new drugs. One of the co-founders (at the VC company Atlas) writes:

Back in the spring of 2009, Atlas (where I'm a partner) founded the company with Schrödinger, a leading computational chemistry software company, after almost a year-long dialogue between myself and Ramy Farid, Schrödinger’s president. At this time, Schrödinger was launching a novel computational tool called WaterMap, an apt name for a technology that maps the energetics of water sites at the receptor-ligand interface, providing a potential roadmap for efficient ligand-receptor interactions. As this cutting-edge technology catalyzed some of our initial thinking, we called it Project Troubled Water Inc (PTW) for the first year or so. 

So in a way, this is also highlight of this article. To summarise: the company was founded because these people believed in computational chemistry as the main driving force behind drug discovery. Did the success of ND-630 prove them right?

Here's how they discovered ND-630 according to the article. They started with the crystal structure of Acetyl-CoA carboxylase with the natural product Soraphen A bound and identified two pockets with high-energy hydration sites using SiteMap and then WaterMap. Then they did a structure-based virtual screen of commercially available compounds using GlideXP and kept only compounds that hit the high-energy hydration sites in both pockets. Soraphen A and these compounds where then used to build two pharmacophore models, which, in turn, where used for a ligand-based virtual screen with hits further refined with GlideXP. "A combined virtual hit-list of a few thousand compounds was clustered to maximize diversity, and 300 representatives were chosen after visualization of the poses. This process led to the identification of ND-022 ... Subsequently, lead optimization proceeded rapidly, guided by WaterMap and Prime/MM-GBSA v. 2.2 estimates of binding free energy." Which finally led to ND-630.

So not exactly Derek Lowe's unicorn dream come true, but I think it's fair to call this computer aided drug design.

Thanks to Victor Guallar for bringing the article to my attention.


This work is licensed under a Creative Commons Attribution 4.0