We initiated our project by retrieving the TIGIT protein structure (PDB ID: 3UCR) and searching potential binding pockets using DoGSiteScorer, selecting one with a score of 0.48. Key amino acid residues within this pocket were either selected or deselected after careful consideration of structural and functional data from the literature. Then, known TIGIT inhibitors were docked using SeeSar Software, which provided an interactive environment for visualizing docking results and adjusting docking strategies, generating 500 poses per ligand to estimate binding affinities. Compound A21 exhibited the highest affinity, in the nanomolar range. Finally, a pharmacophore model was developed based on the TIGIT-A21 complex, incorporating 2 aromatic, 1 hydrophobic, 1 hydrogen donor, and 1 hydrogen donor/acceptor features. Parallel efforts involved a virtual screening for compounds “similar to our lead” (40-68.4% of similarity) using infiniSee, identifying 31,879 candidates.
After 3 months, Rodrigo has achieved the following milestones:
- We identified and characterized the binding pocket of TIGIT using SeeSAR, which allowed the progression of our project. We conducted a thorough evaluation of the concordance between experimental data and the docking results obtained from SeeSAR. During this analysis, it was observed that SeeSAR struggled with generating accurate poses free from atomic clashes and faced challenges in reliably estimating binding affinities. Notably, despite limitations in direct comparisons between binding affinities and IC50 values, it is typically expected that inactive compounds exhibit poor binding. However, SeeSAR indicated that the inactive compound A21 was capable of binding to the TIGIT pocket, while it failed to accurately model the binding of known active compounds like A7, which binds in the µM range. Given the challenges encountered with A7 docking, we instead generated a pharmacophore model using compound A21. This decision was based on the inability of SeeSar to effectively dock A7.
- Implementation of Virtual Screening campaigns across multiple databases. We successfully initiated our virtual screening efforts by querying several databases available, including Ambrosia, CHEMriya, eXplore, FreedomSpace, GalaXi, KnowledgeSpace, and REALSpace. Our search utilized ECFP4 Fingerprints to identify compounds with a minimum of 40 % similarity, yielding several potential candidates: 3,574 from Ambrosia, 28,145 from eXplore, and smaller counts from other databases, in a total of 31,879 similar compounds. Notably, the eXplore database contributed the five most similar compounds, with the highest similarity score reaching 68.4%. Moving forward, our strategy involves the detailed molecular docking of the top 1,000 compounds that exhibited the highest similarity scores. This approach will help refine our selection and prioritize candidates with the greatest potential for inhibitory activity against TIGIT.
- NA