Extracellular domain of TREM2 possess two distinct ligand recognition sites: Insights from machine-learning guided docking and all-atoms molecular dynamics simulations.

The myeloid-specific triggering receptors expressed on myeloid cells 2 (TREM2) is a group of class I receptors expressed in brain microglia plays a decisive role in neurodegenerative diseases such as Alzheimer's disease (AD) and Nasu Hakola disease (NHD). The extracellular domain (ECD) of TREM2 interacts with a wide-range of ligands, yet the molecular mechanism underlying recognition of such ligands to this class I receptor remains underexplored. Herein, we undertook a systematic investigation for exploring the mode of ligand recognition in immunoglobulin-like ectodomain by employing both knowledge-based and machine-learning guided molecular docking approach followed by the state-of-the-art all atoms molecular dynamics (MD) simulations. Besides the known binding site formed by complementarity-determining regions (CDR) 1 and CDR2 loops, which enables the binding of different anionic ligands, our study identifies the presence of second binding site formed by β-strands towards the C-terminal end. We observe a dense network of hydrophobic contacts formed between the explored ligands and CDR loops and β-strands, specifically CDR1, CDR2, β-strand C', loop connecting β-strand D and E, and loop connecting β-strand E and F. Ligand binding in immunoglobulin-like ectodomain increases the conformational flexibility of CDR2 loop, thus most frequently observed pathogenic variants i.e. R47H and R62H in TREM2 may affect the development and progression of AD. Our knowledge-based and machine-learning guided docking and physics-based simulations study unveils deep insights into the endogenous ligand recognition by the positive surface ligand binding site and distant core site pave the way for exploration of other small molecules towards development of novel therapeutics against Alzheimer's disease.