The immune system represents a key line of defense against pathogens and cancer cells and is the subject of intensive research due to its therapeutic potential. Cancer and senescent cells express inhibitory ligands such as HLA-E on their surface which suppress the cytotoxic activity of NK cells through interaction with the transmembrane receptor NKG2A/CD94. Transduction of the inhibitory signal occurs through ITIM regions in the intracellular part of the receptor, which are phosphorylated by Src family kinases.
Using AlphaFold2, we built the first complete model of the NKG2A/CD94 receptor. Molecular dynamics simulations showed that binding of HLA-E increases the exposure of ITIM regions to water molecules, enabling their phosphorylation and signal transduction. We confirmed the presence of a hydrogen bond network in the extracellular region and identified the formation of new bonds between connecting regions of the receptor. These changes were associated with a tilting of the transmembrane helices into a more crossed conformation, suggesting a possible mechanism of signal transduction.
We demonstrated that the lipid bilayer composition strongly affects receptor dynamics and inhibitory signal transmission. The presence of negative charge on the membrane surface reduced the flexibility and exposure of intracellular regions. A thinner lipid bilayer induced the tilting of transmembrane helices and influenced neighboring domains, while membrane curvature affected the exposure of one of the ITIM regions. These results highlight the importance of carefully selecting the lipid environment when simulating transmembrane proteins.
Using Src kinases Fyn and Lyn, we developed a new virtual screening protocol based on water pharmacophore models. We confirmed its applicability and identified two novel potential inhibitors, whose binding was further evaluated by molecular dynamics simulations. We showed that water pharmacophore models are less effective at capturing interactions with flexible protein regions. Therefore, combining them with information from known ligands and experimental structures can enhance predictive power and enable more effective inhibitor design.
|
