Understanding and predicting the specificity of small molecule protein interactions
Final Report Abstract
This project consisted of three main parts, namely i) the development of a docking-based system to predict the selectivity profiles of ligands against multiple receptors; ii) the structurebased derivatization of compounds (in particular fragments) in order to obtain more selective and/or more potent ligands; and iii) the development of an easy-to-compute estimation of rotatable bond entropy in order to improve scoring functions. The selectivity prediction system has been successfully applied to the chemokine receptors CXCR3 and CXCR4, and the results have already been published. For the kinase combinations +EGFR/+ErbB2/-B-Raf and +EGFR/+VEGFR/-B-Raf, the molecules have been picked and assayed. As we have added a third (anti)target, it did not come unexpectedly that our hit rates were not quite as high as in the case of the GPCR pair. In the case of the β2 AR, we have successfully optimized fragment-sized ligands towards higher affinity, with a top improvement of ≈40-fold. We have now expanded this to selectivity considerations, developing newly identified fragments in both directions: first, towards higher selectivity for the β1 AR, and second towards higher selectivity for the β2 AR. In parallel, we have been screening our database of novel, easily synthesizable compounds, SCUBIDOO, for chemical matter that can later be derivatized in a straightforward manner. Our efforts to directly correlate a graph measure, such as betweenness centrality, with rotatable bond entropy were not leading to the desired positive relationship. However, we have established a data set of calculated MD trajectories for a set of 186 molecules and are currently attempting to provide a rule-based estimator for probable entropy values depening on the precise type of bond. Such a decision tree can then be implemented in a software in order to allow high-throughput applications.
Publications
- Limits of ligand selectivity from docking to models: In silico screening for A1 adenosine receptor antagonists. PLOS Med. 7(11), e49910 (2012)
Kolb, P., Phan, K., Gao, Z.-G., Marko, A. C., Sali, A., and Jacobson, K. A.
- Computer-aided design of selective ligands binding to G proteincoupled receptors. Dtsch. Med. Wochenschr. 138(44), 2260–2264 (2013)
Schmidt, D. and Kolb, P.
- Identifying modulators of CXC receptors 3 and 4 with tailored selectivity using multi-target docking. ACS Chem. Biol. 10(3), 715–724 (2015)
Schmidt, D., Bernat, V., Brox, R., Tschammer, N., and Kolb, P.
(See online at https://doi.org/10.1021/cb500577) - SCUBIDOO: A large yet screenable and easily searchable database of computationally created chemical compounds optimized toward high likelihood of synthetic tractability. J. Chem. Inf. Model. 55(9), 1824–1835 (2015)
Chevillard, F. and Kolb, P.
(See online at https://doi.org/10.1021/acs.jcim.5b00203)
