By Franck Chevalier
During the last 20 years, a great interest has been shown for fragment in order to answer one of the major issue faced by traditional drug discovery program: by its properties, it allows to reduce the number of molecules tested whatever the method used, series are directly related to fragment and optimization of hit is more intuitive. Fragment based drug discovery (FBDD) by molecular dynamics offers now an awesome tool for fragment hit binding.
How has FBDD evolved?
A brief analysis of articles published within last two decades shows that main techniques used include bioassay (inhibition assay, radiolabeled binding assay…) and biophysical methods (Crystallography, NMR, SPR, MS, ITC…).
Low affinity showed by most fragment for their target limits the range of detection for most methods and are not necessarily the high affinity binder that are detected. Most articles in FBDD present the use of a single method, mainly in vitro bioassay. The rate of new articles is about 150/year. Expertise needed, throughput and cost of main biophysics methods still limit their access.
Within the same period, we observed an increase of publications on virtual methods. Molecular dynamics simulations have been published in more than tens of thousands articles. Interestingly less than 5% of these publications deal about discovery at a yield of 300 publications/year within the last few years.
When FBDD is performed, about 1 to 2 techniques are used. If more than 2 techniques are used, 50% of publications mention the use of virtual methods. About 30% only used virtual methods and tendency has been increased. The large dataset of structures generated by crystallography empowered the use of docking. The important role played by protein flexibility has been answered with molecular dynamics.
Discovery or design?
At this stage, most work focused on discovery, meaning the detection of molecules able to bind, inhibit or activate the target. Few articles yet deals about linking, merging, growing fragment.
Favorite targets are kinase, enzymes but more complex system involving ion channel, GPCR are now reported.
Reliability of molecular dynamics
In silico FBDD’s impact on lead development has been limited to date. Inability to describe kinetics and conformational changes upon binding are some of the reasons. In the following reference, a comprehensive computational assay for fragment hit characterization has been performed: 2.1 ms of unbiased all-atom high-throughput molecular dynamics (HTMD) data were obtained and analyzed to show the potential of this technique for FBDD.
Reference: Noelia Ferruz,Matthew J. Harvey, Jordi Mestres, and Gianni De Fabritiis J. Chem. Inf. Model. 2015, 55, 2200−2205