Abstract
Cell membrane, enveloping internal organelles and surrounding the entire cell, is an essential structural element in all forms of life. A typical cell membrane is composed of two leaflets of lipids containing hundreds of different types, along with proteins. The interaction between lipids and proteins plays a crucial role in cellular processes such as membrane fusion, protein transport and signal transduction. Molecular dynamics (MD) simulations, acting as a computational microscope, provide insights at the atomic level into the molecular mechanism underlying the function of membrane proteins. Understanding protein-lipid interplay using MD simulations has attracted considerable interest.First, I employed all-atom models to study membrane protein systems at the small length and time scales. ARF proteins are guanine nucleotide-binding proteins extensively involved in membrane trafficking. It has been found that the ARF6 protein binds to membranes in a nucleotide-dependent manner. I revealed the detailed mechanism of ARF6 binding to the membrane using all-atom MD simulations. The results indicate that two key structural elements, MYR and AH, insert into the membrane when ARF6 is in the GTP-bound form, enhancing the strength of membrane binding. Enhanced sampling methods were also utilized. The calculated PMF through umbrella sampling confirms that ARF6 binds more strongly to the membrane in the GTP-bound state. Metadynamics simulations demonstrated the thermodynamic preference of the MYR and AH components for the membrane.
In addition, the effect of nanosheets on the conformation of membrane proteins was also investigated. The activation of integrin proteins is an important regulator in the immune system, with activation primarily manifested in the destabilization of the transmembrane protein region. All-atom MD simulations were performed to elucidate the mechanisms by which three types of nanosheets (GRA, BN and BP) mediate the activation of integrin αvβ8. I observed that BP can unclasp the inner membrane clasp in certain membrane orientations, thereby activating the integrin, while GRA and BN failed. BP exhibited a relatively mild effect on membrane structure and lipid properties.
However, for some large-scale phenomena such as membrane remodeling, the high computational cost of all-atom models is unaffordable. Coarse-grained (CG) models simplify simulated systems by representing multiple atoms with a single interaction bead, enabling simulations on longer time scales. The development and application of CG models in the MD simulations of membrane proteins have attracted widespread attention. Therefore, I conducted CG MD simulations on the SNX1 protein, which belongs to the BAR superfamily that includes many types of membrane-remodeling proteins. Experimental evidence shows that the SNX1 protein can bend membranes, but the molecular mechanism behind this process remains unclear. Our simulation demonstrates that SNX1 bends the membrane through the insertion of the AH and the scaffold effect of the BAR domain, and bending occurs only when these two mechanisms act simultaneously. The presence of negatively charged lipid PI3P in the membrane is also essential for the membrane bending process.
In my research on SNX1, the widely used MARTINI 3.0 force field was employed. However, some benchmark studies have shown that the force field can lead to inaccurate structure and thermodynamic properties when describing lipids. I developed a bottom-up coarse-grained lipid force field based on graph neural networks. I trained models based on all-atom simulations of DOPC, DOPS, and a mixture of DOPC/DOPS lipids. The CG simulations using the models reproduce the thermodynamic properties in the all-atom simulations. Furthermore, the model shows great potential transferability in configurations and temperature.
Overall, I have used various computational methods, including all-atom and coarse-grained MD simulations, umbrella sampling, metadynamics, and machine learning coarse-grained potential, to gain a comprehensive understanding of the interactions between proteins and membranes, and to delve into the molecular mechanisms behind membrane proteins in detail. These efforts have not only deepened our understanding of membrane protein functions in experiments but also advanced the multiscale theory in the field of MD simulations.
| Date of Award | 24 Apr 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Jun FAN (Supervisor) |