How do Antimicrobial Peptides Interact with the Outer Membrane of Gram-Negative Bacteria? Role of Lipopolysaccharides in Peptide Binding, Anchoring, and Penetration

Gram-negative bacteria possess a complex structural cell envelope that constitutes a barrier for antimicrobial peptides that neutralize the microbes by disrupting their cell membranes. Computational and experimental approaches were used to study a model outer membrane interaction with an antimicrobial peptide, melittin. The investigated membrane included di[3-deoxy-d-manno-octulosonyl]-lipid A (KLA) in the outer leaflet and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) in the inner leaflet. Molecular dynamics simulations revealed that the positively charged helical C-terminus of melittin anchors rapidly into the hydrophilic headgroup region of KLA, while the flexible N-terminus makes contacts with the phosphate groups of KLA, supporting melittin penetration into the boundary between the hydrophilic and hydrophobic regions of the lipids. Electrochemical techniques confirmed the binding of melittin to the model membrane. To probe the peptide conformation and orientation during interaction with the membrane, polarization modulation infrared reflection absorption spectroscopy was used. The measurements revealed conformational changes in the peptide, accompanied by reorientation and translocation of the peptide at the membrane surface. The study suggests that melittin insertion into the outer membrane affects its permeability and capacitance but does not disturb the membrane’s bilayer structure, indicating a distinct mechanism of the peptide action on the outer membrane of Gram-negative bacteria.


S1. Formation of the asymmetric model outer membrane of Gram-negative bacteria
Langmuir-Blodgett (LB) and Langmuir-Schaefer (LS) transfers were used to prepare asymmetric KLA-POPE bilayers on a gold surface.First, a POPE (or d31-POPE) monolayer was transferred from the aqueous subphase by a vertical LB withdrawal, see Fig. S1A.
Withdrawal of a hydrophilic gold substrate from the aqueous subphase through the air|water interface covered by the phospholipid monolayer gave the inner leaflet of the model outer membrane.Next, a monolayer of KLA on a 0.1 M KClO4 and 5 mM Mg(ClO4)2 aqueous subphase was compressed to the surface pressure Π = 30 mN m −1 and a horizontal LS transfer was used to fabricate the second leaflet of the model outer membrane, see Fig. S1B.During this transfer, the hydrophobic surface of the POPE modified gold surface and was exposed to the hydrophobic hydrocarbon chains in the KLA monolayer present at the aqueous electrolyte|air interface.The gold substrate was covered by a Y-type lipid bilayer as shown schematically in Fig. S1C.

S2. Area per lipid of the outer membrane and volume of the simulation box during the equilibration simulation
During the equilibration process of the model outer membrane, the volume of the simulation box and the average area per lipid attributed to KLA and POPE molecules decreased.Figure S2A shows the average area per KLA lipid in the outer leaflet of the membrane over simulation time, while Fig. S2B shows the average area per POPE phospholipid in the inner leaflet of the outer membrane.The average areas per lipid AKLA and APOPE converge to the values 1.87 nm 2 and 0.58 nm 2 , respectively.These values agree well with the LB-LS transfer conditions at with the average area per AKLA was 1.96 nm 2 and APOPE − 0.67 nm 2 . 1 The time evolution of the total volume of the simulation box VS is shown in Fig. S2C.The capacitance of a film modifying an electrode surface is where ε0 is the permeability of vacuum, ε is the dielectric constant of the film at the electrode surface, d is the thickness of the film, and A is the surface area of the electrode.The only purpose of presenting this equation is due to the description of the dependence of the measured capacitance on the dielectric constants of molecules present in the film.
Our previous studies 2 demonstrated that the tilt of the acyl chains in the KLA-POPE bilayer with bound melittin do not change during the potential scan indicating that the OM thickness remains constant.Therefore, changes in the OM do not contribute to the measured capacitance.
Adsorption of a protein on top of the OM, could potentially lead to a decrease in the capacitance, however the effect of the dielectric constant is stronger than of the membrane thickness changes.

S4. IR spectra of KLA : melittin-POPE bilayer
Figure S4 shows PM IRRA spectra of a freshly transferred KLA : Mel (9:1 mole ratio)-POPE bilayer and after its immersion into the electrolyte solution.The PM IRRA spectra in Fig. S4 feature the ν(C=O) stretching modes in the ester carbonyl groups in lipids, amide I' vibration mode mainly in melittin and νas(COO − ) stretching mode in KLA.The PM IRRA spectrum of a freshly LB-LS transferred KLA:Mel (9:1 mole ratio)-POPE bilayer gives a strong signal in 1700 -1600 cm −1 , which arises from the amide I' vibrational mode in melittin, see gray line

S6
of a significant fraction of melittin from the membrane surface into the electrolyte phase.
Note, that the ν(C=O) absorption band in lipids underwent a down-shift, due to a better hydration of the ester carbonyl groups in the KLA and POPE molecules upon immersion in D2O. 3,4 owever, the integral intensity of the ν(C=O) absorption band did not change, indicating no loss of the lipid molecules upon immersion of the bilayer into the electrolyte solution.

S5. Calculation of electrostatic and disperse (van der Waals) contributions to the interaction energy
The electrostatic contribution to the interaction energy of a residue x in the peptide with the membrane is calculated as: where rij is the distance between two charges qi and qj, where the first summation goes over all the N atoms of the residue of interest, while the second summation goes over all the M atoms of the membrane.The dispersive (van der Waals) contribution to the interaction energy could be calculated using the Lennard-Jones potential as where rij is the distance between two atoms i and j, where the first summation goes over all the N atoms of the residue of interest, while the second summation goes over all the M atoms of the membrane.  and   are the equilibrium van der Waals energy and distance for a given pair of atoms.

S6. Secondary structure analysis of melittin interacting with the outer membrane
The secondary structure of the peptide was determined using the STRIDE algorithm in VMD. 5,6 Heicity is defined as the amount of α-, 310-and π-helices in the secondary structure of the peptide.The time dependency of the helicity is plotted in Fig. S5A

S8. Deconvolution of the amide I' band in melittin
The second derivative and Fourier self-deconvolution were used to deconvolute the measured IR spectra.Fourier self-deconvolution was done to fit the measured amide I' band with Gauss    Once the angle between the µ  and E  vectors equals 90 º (Fig. S9B) the integral intensity of the amide I' band equals zero.In this case there is no coupling of the transition dipole and the electric field vectors.
For a peptide where the α-helical fraction is known, one can calculate the θ between the transition dipole moment µ  of the deconvoluted amide I' band of α-helices relatively to the surface normal, being colinear with the direction of the electric field vector Finally, the order parameter of the long axis of the α-helix (Shelix) can be calculated as where α is the angle between the long axis of the α-helix and the transition dipole moment of the amide I' band of α-helices.In an α-helical protein fragment the transition dipole vector of the amide I' mode µ  makes an angle of 34 ° -38 ° vs. the long axis of the α-helix 10,11 , see Fig. S9.The order parameter Shelix was used to calculate the tilt of the helix with respect to the surface normal (Tilthelix).

Figure S1 .
Figure S1.Illustration of the fabrication procedure of the asymmetric KLA-POPE model outer membrane of Gram-negative bacteria A: transfer of the inner POPE leaflet by Langmuir-Blodgett vertical withdrawal, B: transfer of the outer KLA leaflet by Langmuir-Schaefer method and C: molecular scale order in the LB-LS transferred bilayer.

Figure
Figure S2.A: Time evolution of the average area per KLA lipid in the outer leaflet of the model outer membrane; B: Time evolution of the average area per POPE phospholipid on the inner leaflet of the model outer membrane.The values were computed as an average over all lipids present in the simulation box.C: Volume of the simulation box VS over simulation time.

Figure S3 .
Figure S3.Schematic structure of the KLA-POPE model outer membrane on a gold electrode surface with possible arrangements of melittin interacting with the membrane from solution A: parallel B: tilted and C: perpendicular orientations.The dielectric constants of polar head groups and hydrocarbon chains in lipids as well as of the peptide are given in the figure.

Fig. S4 .
Fig. S4.Immersion of the bilayer into the electrolyte solution cased a dramatic decrease in the intensity of the amide I' mode, see black line in Fig.S4.This result indicates a dissolution for the three production simulations of the composite systems (Sim. 1 -Sim.3) and the control simulation of melittin in water (Control).The helicity value decreased most notably in the Control simulation from 75% to 35% during the simulation time of 1000 ns.In Sim. 1 the helicity decreased slightly from 75% to 70% during the whole simulation, whereas in Sim. 2 the unfolding of the helical S7 structures is concentrated in the first 100 ns of the simulation time and the helicity remained largely constant at 60% for the rest of the simulation time.The results indicate that the interaction of the peptide with the membrane stabilizes the helical structures.On the other hand, the rapid changes in helicity throughout the simulation time, as shown in the results of Sim. 3 indicate a high conformational flexibility of the peptide.

Figure
Figure S5B-C show the Ramachandran plots of the probability distribution of the backbone dihedral angles ψ and φ computed for Sim. 1, Sim. 2, Sim. 3 and Control, respectively.While in Sim.1-3 the right turning α-helix is the dominant secondary structure followed by the β-

Figure S6 .
Figure S6.Characteristic configurations of melittin bound to the model outer membrane.The N-and C-terminus are shown in red and blue, respectively.The lower leaflet of the outer membrane is shown in gray, while the lipid A part of KLA in the upper leaflet is shown in cyan and the inner core region of KLA is shown in silver.Each panel shows a snapshot of melittin atop the membrane, once from the side and once from the top.The snapshots are taken from Sim. 1 after approximately 30 ns (A/B) and 600 ns (C/D).

Figure S7 .
Figure S7.Upper panel: the second derivative of the measured IR spectra and the bottom panel: corresponding IR spectra (thick color lines) KLA-POPE lipid systems after interaction with melittin.Thin black lines show the band deconvolution results.A: attenuated total reflection IR spectrum of 4.4 × 10 -4 M melittin after 1h of interaction with KLA-POPE vesicles, C-D: PM IRRA IR spectra of the KLA-POPE bilayer at E = 0.0 V vs Ag|AgCl after B: 15 min interaction with1 µM melittin, C: 1h interaction with 1µM melittin and D: 15 min interaction with 10 µM melittin.All spectra were recorded in 50 mM KClO4 and 5 mM Mg(ClO4)2 in D2O.The scale bars show absorbance in arbitrary units.
FigureS8shows the PM IRRA spectra of the KLA-POPE bilayer exposed to 1 µM melittin solution for 60 minutes.

Figure S8 .
Figure S8.IR spectra of KLA-POPE bilayers on Au(111) after 1 h of interaction with 1 µM melittin (dark blue lines) recorded at different electrode potentials.The upper and lower panels show results for the negative and positive scans, respectively.Thin black lines show the band deconvolution results.The bands highlighted in gray show the amide I' mode of α-helices in melittin.Measurements were carried out in 50 mM KClO4 and 5 mM Mg(ClO4)2 in D2O.The absorbance is shown in arbitrary units.

Figure S9 .
Figure S9.Limiting cases for the orientation of an α-helical peptide fragment adsorbed on a solid surface in which the transition dipole moment of the amide I' band is A: parallel and B: normal to the direction of the electric field vector of the reflected IR radiation.The blue lines show the direction of the long axis of the α-helix.The direction of the transition dipole vector of the amide I' mode (red arrow) and the direction of the electric field vector of the p-polarized light at the phase boundary (gold arrow) are shown in the figure.
c corresponds to the percent content of the amide I'α-helix band in the entire amide I' band in a measured PM IRRA spectrum and the R c is the percent content of the α-helical structures in the solution phase (random distribution).The angle θ can be used to calculate the order parameter S as follows9