(53e) Implicit Model Captures Electrostatic Features of Cell Membrane Environment | AIChE

(53e) Implicit Model Captures Electrostatic Features of Cell Membrane Environment


Samanta, R. - Presenter, The University of Texas at Austin
Gray, J. J., John Hopkins University
Membrane proteins play a critical role in our body, constituting about a third of all human protein. They are targets for more than half of all drugs. However, prediction and design of membrane proteins are challenging due to the presence of diverse and complex lipid environment. Relative to soluble proteins, the model development of membrane proteins particularly lag behind due to the sparse and low-quality data, leading to overfit tools and specific models.

Implicit models accelerate this complex biomolecular problem by representing the solvent and the lipid environment as a continuum medium. Additionally, to overcome the challenge of sparse dataset, we assembled a suite of 12 tests on independent datasets ranging from predicting structural property, stability to protein-protein docking and design to test the model. Most implicit models often do not consider the effect of pH, lipid head group, or dielectric constant of membrane environment. In this work, we propose to develop an implicit approach that captures the crucial electrostatic interactions due to the membrane, such as the effect of lipid head groups the influence of pH and dielectric variations inside the membrane layer. Our energy function franklin2022 is built upon franklin2019, an existing energy function based on experimentally derived hydrophobicity scales that could capture the anisotropic structure, the shape of water-filled pores, and nano-scale dimensions of membranes with different lipid compositions. Our new method uses a constant-pH algorithm to sample the protonated and de-protonated states of protein residues. Further, it captures the effect of lipid head group using a mean-field based approach and uses a depth-dependent dielectric constant to characterize the membrane environment. Relative to franklin2019, this model has the ability to capture the effect of pH and improved the calculation of ΔΔGpH of low pH insertion peptides (pHLIP) in extracellular acid environments, important biomarkers of cancer cells. Further, after including the effect of lipid head groups, franklin2022 have improved the prediction of tilt angles of adsorbed peptides relative to franklin2019. We will further test the performance of franklin2022 on the benchmark suite to evaluate its ability to predict the stability, structure, and design membrane proteins.

The speed of such implicit models and the model calibration based on diverse tests will help access biophysical phenomena at different time and length scales to accelerate the design pipeline for membrane proteins.