Implicit model to capture electrostatic features of membrane environment.

Area of Science:

• Biophysics
• Computational Biology
• Structural Biology

Background:

• Membrane protein structure prediction and design are computationally intensive, hindered by complex lipid-protein interactions and electrostatic calculations.
• Existing methods like Poisson-Boltzmann are accurate but not scalable for large-scale design tasks.
• Developing efficient and accurate energy functions is crucial for advancing membrane protein research.

Approach:

• Developed a fast, implicit energy function (Franklin2023, F23) that incorporates realistic lipid bilayer properties.
• F23 models lipid head group impact using a mean-field approach and employs a depth-dependent dielectric constant.
• Built upon the Franklin2019 (F19) model, F23 leverages experimentally derived hydrophobicity scales.

Key Points:

• F23 demonstrated improved prediction of membrane protein tilt angles for WALP peptides (90%), TM-peptides (15%), and adsorbed peptides (25%) compared to F19.
• Stability and sequence recovery performance were comparable between F19 and F23.
• The F23 model's speed and calibration enable exploration of biophysical phenomena at longer time and length scales.

Conclusions:

• The F23 energy function offers a computationally tractable solution for membrane protein structure prediction and design.
• Its efficiency and improved accuracy in modeling membrane environments accelerate the design pipeline.
• This advancement facilitates deeper investigation into membrane protein biophysics and enables more effective protein engineering.