Application Note - Dr Okamoto Masako
Mitsubishi Tanabe Pharma Corporation
Structural Development of a hA2AAR Antagonist considering ADME profiles
There is a growing interest in hA2AAR as a drug target because hA2AAR antagonists can potentially be developed for the treatment of neurodegenerative disorders, such as Parkinson's disease. A number of hA2AAR antagonists have recently been reported by research groups and, with the availability of X-ray crystal structures, it is now possible to better understand their SAR. This report explores the possibility of developing the strucuture of a known hA2AAR antagonist by considering ADME Scoring Profiles using StarDrop.
As a starting point I chose an hA2AAR antagonist (comp 16) from the literature (ACS Med. Chem. Lett., 2012, 3 (9), pp 715-720) and drew its structure in StarDrop's Design module (Figure 1). The central part of comp 16, containing a triazole and amide, is thought to be essential to maintain the antagonist activity. To develop the structure of comp 16, I used the Nova module with strict masking (to preserve this central scaffold) as shown in Figure 2, and specified that Nova run for 2 generations.
As a result, I obtained 7,674 derivatives by Nova, and confirmed the structural expansion using a chemical space plot drawn using StarDrop's Visualization tools (Figure 3). In Figure 3-1, compounds generated by Nova are shown in green and the parent compound (comp 16) is highlighted in the chemical space with a light blue dot. Figure 3-2 highlights the first generation of compounds in dark blue.
I selected compounds using a multi-parameter-optimisation approach on the basis of having a better score than the parent compound (comp 16) using an Oral CNS scoring profile . The scoring profile, shown in Figure 4, highlighted that the most important issue for comp 16 is blood-brain barrier penetration as indicated by StarDrop's models. Therefore, further selection was performed using StarDrop's continuous model for predicting blood-brain barrier penetration.
The data set of selected compounds was exported as an SD file to perform docking simulations using an X-ray crystal structure of hA2AAR (3EML).
A good example with an appropriate binding mode is shown in Figure 5 (in this figure, the predicted binding mode and its predicted ADME profiles are shown). The compound has a better Oral CNS score than comp 16; in particular, the predicted value for BBB log has been improved. However the predicted binding mode remains the same as reported for comp 16 in the literature. As the next step, I'd like to synthesize some of the selected compounds in order to perform the associated biological assays. This will enable us to determine the relationship between predictive ADME profiling data and experimental data. If it is possible to obtain compounds with a promising profile then the aim will be to design further compounds with good predictions when tested against StarDrop's BBB models, in a stepwise manner.