# Martini 3 as a Transferable Force Field for Lipopolysaccharide Parametrization

## TL;DR

The paper introduces transferable parameters for simulating lipopolysaccharides in bacterial membranes using the Martini 3 model, enabling efficient large-scale simulations.

## Contribution

The paper introduces a new transferable parametrization for LPS in Martini 3, expanding its use to bacterial outer membrane studies.

## Key findings

- CG simulations using Martini 3 accurately reproduce structural and dynamic properties of LPS.
- Parameters were validated for multiple medically relevant pathogens including E. coli and P. aeruginosa.
- The approach combines automated and manual refinement for accurate LPS modeling.

## Abstract

Lipopolysaccharides
(LPS), as critical components of the outer
membrane (OM) of Gram-negative bacteria, play essential roles in maintaining
bacterial integrity and mediating environmental interactions. All-atom
molecular dynamics (AA-MD) simulations provide detailed insights into
LPS behavior at atomic resolution, but they remain computationally
limited in accessing biologically relevant time scales. Coarse-grained
(CG) models, such as Martini 3, offer a computationally efficient
alternative while retaining sufficient accuracy for key biophysical
properties. Although Martini 3 has been widely applied to proteins
and phospholipids, only a few LPS models have been developed within
this framework, limiting its utility for bacterial OM studies. To
address this gap, we developed and validated CG parameters for LPS
from multiple medically relevant pathogens, including Escherichia
coli and Salmonella enterica, as well as
two ESKAPE pathogens, Klebsiella pneumoniae and Pseudomonas aeruginosa. Our approach leverages the transferability
of Martini parameters: we parametrized 57 unique disaccharide units
using the Bartender tool, which automates CG-to-AA mapping and parametrization.
These parameters were then combined and manually refined to accurately
reproduce the complex dynamics of complete LPS molecules. We conducted
extensive AA and CG simulations of asymmetric bilayers composed of
phospholipids in the inner leaflet and LPS in the outer leaflet allowing
detailed comparisons between the two for key structural and dynamic
properties. The close agreement between the CG and AA simulations
demonstrates the accuracy and robustness of our transferable parameter
set, providing a valuable tool for simulating Gram-negative bacterial
OMs at larger scales and longer time scales.