Antimicrobial mechanisms due to hyperpolarisation induced by nanoporous Au

Nanomaterials such as nanoparticles exhibit remarkable antimicrobial activities. Nanoparticles directly disturb the cell membrane or cytoplasmic proteins because they pass through the cell wall. Nanoporous Au (NPG) is another antimicrobial nanomaterial, which cannot pass through the cell wall of bacteria but can still kill bacteria, utilising interactions between the surface of NPG and cell wall of bacteria. The origins of antimicrobial activities without direct interactions are unknown. It is necessary to elucidate these mechanisms to ensure safe usage. Here we show that the antimicrobial mechanism of NPG consists of two interactions: between the surface of NPG and cell wall, and between the cell wall and cell membrane. Fluorescent experiments showed that the cell wall was negatively hyperpolarised by NPG, and molecular dynamics simulations and first-principles calculations suggested that the hyperpolarisation of the cell wall leads to delicate structural changes in the membrane proteins, rendering them bactericidal. Thus, the hyperpolarisation induced by NPG plays a critical role in both interactions. The combination of molecular dynamics simulations and first-principles calculations allows a deeper understanding of the interactions between metallic surfaces and biomolecules, because charge transfer and exchange interactions are calculated exactly.


Construction of peptidoglycan model
A bacterial cell wall consists of a network of peptidoglycan. The structural characteristics of peptidoglycan have been studied by electron microscopy 1 and molecular dynamics (MD) simulations 2,3 . Peptidoglycan is composed of repeating units consisting of a disaccharide (i.e., N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc)) and a cross-link peptide. In the present study, repeating units of a stem (L-Ala-D-iso-Gln-L-Lys-D-Ala-D-Ala) and a bridge (Gly1-Gly2-Gly3 -Gly4 -Gly5) were selected as a component of cross-link peptide in accordance with the previous study 4 .
The 3-dimensional structure of peptidoglycan is still not known although many studies tried to clarify the architecture of peptidoglycan 2,3 . Two major candidates of peptidoglycan structure have been proposed: the layered model 5 and the scaffold model 2,3 .
It has been shown that the scaffold model well represents the mechanical properties of cell wall, compared with the layered model 2,3 . Thus, the scaffold model was used in the present study.
The components (i.e., a disaccharide and a cross-link peptide) of peptidoglycan were constructed and the geometry optimizations were performed using the Gaussian program package. Four GlcNAc-MurNAc disaccharides were connected each other and the pentapeptides and the pentaglycine were attached to the glycan chains to construct the cross-link structure. Thereafter, a terminal of pentapeptides was connected to another GlcNAc-MurNAc strand, and then MurNAc molecule, which was not still connected to the cross-link structure in GlcNAc-MurNAc strand, was connected to other pentapeptides and pentaglycine. By repeating this process, a scaffold model of peptidoglycan was constructed.
Energy minimizations and MD calculations were performed to obtain a stabilized structure of peptidoglycan by the Discovery studio 4.0 software (Biovia Inc, San Diego, CA), using the CHARMM forcefield 6 . The peptidoglycan was immersed in a spherical water solvation. The center of water solvent was positioned at the mass center of peptidoglycan and the diameter of spherical solvent water was 50.0 nm. Counter ions of 43 Na + and 43 Clwere added to neutralize the system. The system was energy-minimized using the steepest decent algorism (200,000 steps) and the conjugate gradient algorism (100,000 steps). MD simulations were performed with the time step of 2.0 fs. The system was gradually heated from 5 to 300K for 4ps to activate thermal motion in the system.

Interactions between nanoporous Au and peptidoglycan
An interaction between MurNAc, which is a part of peptidoglycan, and a surface of lattice strains exist at its surface. Thus, a lattice strain of +5% or -5% was loaded into the (111) layers of the FG model and two NPG models were constructed: NPG (+5% strain) and NPG (-5% strain).
The geometry optimization calculations were performed on the FG and the NPG models by first principles calculations using the Dmol3 code 7,8 . In the DMol3 method, the physical wave functions were expanded in terms of the accurate numerical basis sets.

First principles tensile tests
First principles tensile tests with the Dmol3 code 7,8 were performed to investigate effects of the hyperpolarization of peptidoglycan on the elastic modulus. Four parts of peptidoglycan shown by arrows in Fig. S2, which were absorbed on NPG, were investigated. The distance between two oxygen atoms shown by arrows in Fig. S3 was increased by 1% strain without relaxation. This operation was repeated to 8% strain. The exchange-correlation energies were treated according to the generalized gradient approximation (GGA) with the Perdew-Wang 1991 (PW91) approximation 9 to deal with the core (DNP). The ultrasoft pseudopotentials 10

Lipid membrane interacting with hyperpolarized cell wall
The 24 POPE lipid were solvated in 1,412 water molecules using CHARMM-GUI web site. A part of peptidoglycan interacting with FG or NPG was positioned on lipid membrane (Fig. S4). Periodic boundary conditions were applied. MD simulations were performed using the CHARMM force field 6 with gromacs 5.1.1 code 14,15 . The position of peptidoglycan was fixed during the simulations. The system was energy-minimized using the steepest decent algorism (500,000 steps). MD simulations were performed with the time step of 1.0 fs. The system was equilibrated for 10 ns to obtain a stable structure of peptidoglycan with the constant number of particles, volume and temperature (NVT) ensemble. Finally, the 100 ns NVT simulations were performed.

Potassium channel interacting with hyperpolarized cell wall
Initial coordinates for potassium channel were taken from the crystal structures 1K4C.
The channel was embedded in a bilayer of 2,182 POPE lipid and solvated in 15,078 water molecules and 15 Clions using CHARMM-GUI web site. A part of peptidoglycan interacting with FG or NPG was positioned on the ion channel (Fig. S5). Periodic boundary conditions were applied. MD simulations were performed using the CHARMM force field 6 with gromacs 5.1.1 code 14,15 . The position of peptidoglycan was fixed during the simulations. The system was energy-minimized using the steepest decent algorism (500,000steps). MD simulations were performed with the time step of 1.0 fs. The system was equilibrated for 10 ns to obtain a stable structure of peptidoglycan with the constant number of particles, volume and temperature (NVT) ensemble. Finally, the 100 ns NVT simulations were performed.