Variable Non-Gaussian Transport of Nanoplastic on Supported Lipid Bilayers in Saline Conditions

Nanoplastic–lipid interaction is vital to understanding the nanoscale mechanism of plastic adsorption and aggregation on a lipid membrane surface. However, a single-particle mechanistic picture of the nanoplastic transport process on a lipid surface remains unclear. Here, we report a salt-dependent non-Gaussian transport mechanism of polystyrene particles on a supported 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) lipid bilayer surface. Particle stickiness on the POPC surface increases with salt concentration, where the particles stay longer at the surface and diffuse to shorter distances. Additionally, a non-Gaussian diffusion state dominates the transport process at high salt concentrations. Our current study provides insight into the transport mechanism of polystyrene (PS) particles on supported lipid membranes, which is essential to understanding fundamental questions regarding the adsorption mechanisms of nanoplastics on lipid surfaces.

POPC lipid was dissolved in chloroform to make a concentration of 25 mg/ml and diluted to 5 mg/ml with 20 mM HEPES buffer for small unilamellar vesicles (SUV) preparation.

Supported lipid bilayer (SLB) preparation
Supported lipid bilayer was formed through the vesicle fusion process. 1, 2A solution of POPC lipid (Avanti Polar Lipids) in HEPES buffer having a concentration of 5 mg/ml was extruded through a polycarbonate membrane of pore size 100 nm.The extruded SUV solution (~ 50 μL) was then injected into the microfluidic chamber on the hydrophilic surface of the plasma cleaned coverslip.
After 90 minutes, the chamber was washed with copious amount of HEPES buffer (2 times, at leaset 500 μL) and then the supported lipid bilayer (SLB) was ready for single-particle imaging.

Microfluidic device preparation for single particle measurements
We prepared the SLB of POPC lipid on a plasma cleaned cover glass and a microfluidic sample chamber having an inlet as well as an outlet.A very dilute solution (2.2 ×10 8 particles/ml) of carboxy modified PS bead solution is passed through the microfluidic with a speed of 200 nL/min and the transport of PS beads is measured using a total internal reflection fluorescence (TIRF) microscope (Figure S1).

Total Internal Reflection Fluorescence Imaging (TIRF)
A wide-field TIRF microscope (Nikon) was used to capture single-particle fluorescence images of PS beads with video rate imaging.
Using a 561 nm laser for excitation, the beam was focused onto the sample surface using a 100X NA 1.49 oil-immersion objective.Data acquisition was conducted over a 512×512pixel area on the sample, with fluorescence emission collected by an EMCCD camera (Andor, iXon 897).The camera operated using Andor Solis software in frame transfer mode, with an integration time set to 30 ms.

Single Particle Tracking (SPT) analysis
We have used a localization based single particle tracking algorithm (Troika 3 ) to identify, localize, and track individual PS particles over multiple frames.Troika performs three key sequential steps to analyze the raw video files; first the signal-to-noise is enhanced by multiplication with a pixel averaging matrix, then each particle on each frames are identified by local intensity maximum and radial symmetry fitting, 4 and finally localized particle positions are linked using a nearest neighbor algorithm to generate particle trajectories at different conditions.Here we note that, we have used a first frame filtering Matlab script to reject any stuck particle that have been identified on the first frame to reduce redundancy and bias in our data analysis.We performed the first frame filtering as we have observed already stuck particles on the surface for some salt conditions.Statistical analysis is performed on all the trajectories at different conditions as explained in the following sections.

Quantifying the number of adsorbed particles on the POPC surface at various conditions
We counted the total number of trajectories under each condition for multiple data sets and presented them in Table S1.The average and standard deviation are calculated in each case and compared with the blank coverslip.We find that there are a few particles that we could identify on black coverslip and under buffer conditions, however, the number of stuck particles under these conditions are negligible.However, with increasing salt concentration, total number of individual particle trajectories as well as the number of stuck particles increased.

Control experiment on glass coverslip
Interaction of blank glass coverslip with the PS particles are negligible as compared to the SLB at both low and high salt concentrations.We showed some representative frames from raw data on a blank coverslip and compared to the SLB surface images in buffer conditions and at 1000 µM salt concentration.More numbers of active particles are observed at high salt concentration.

Markov Chain Monte Carlo (MCMC) analysis on single frame displacement (SFD)
distributions for all conditions SFD distributions and and MCMC analysis 5 for all trajectories are presented in Figure S2.
Generally, we observe two populations in the SFD distribution and the average and standard deviations for each populations are presented in Table S1.For 1000 μM NaCl, we only observe a short population in SFD with a broader full width at half maxima.We also see a small shoulder at large displacements, however, the MCMC analysis was unable to resolve this shoulder peak.
Hence we approximated that the short population have the 100% contribution to overall displacements albeit with a broader distribution, consistent with our raw data.S10

SRT fitting parameters
Table S3: Three component fitting parameters for SRT graphs for particles in buffer, 10 μM NaCl, 100 μM NaCl, and 1000 μM NaCl, and washing in buffer after experiment. .

van-Hove correlation analysis
The distribution of particle displacements was quantified using the self-part of the van Hove correlation function, 6   (Δ, where  is the Dirac delta function, Δ is the displacement of all particles at a time of Δ, and the angle brackets denote an ensemble average.The van Hove function was estimated with a normalized histogram of one hundred bins binning all particle displacements.
The van Hove correlation distribution in all our datasets are separated into two parts: a center peak region around Δ = 0, and a broad "tail" region.These two regions of the van Hove correlation function can be fitted as two Gaussians, where both large and small displacements are common, or in cases where large displacements are rare, a Gaussian head with an exponential tail. 7 Time dependent van-Hove graphs.
Figure S5: Time dependent van Hove graphs for particles in buffer, 10 μM salt, 100 μM salt, 1000 μM salt, and after washing with buffer.The insets shows a zoomed in section for the short displacements.

Non-Gaussian parameter analysis
The gaussianity of the van Hove correlation graphs were quantified through the calculation of the non-Gaussian parameter, 8 where Δ() is the displacement of all particles at a time of t.An α2 value of zero represents Gaussianity, while a value greater than zero represents non-Gaussianity.The magnitude of α2 signifies the non-Gaussianity of the van Hove graph, with larger α2 values meaning greater non-Gaussian behavior in particle transport.

Localization precision of our analysis
Figure S6: Localization precision of the X and Y axis of our TIRF microscope, calculated from localizing stationary beads using the same algorithm.

System Preparations for molecular dynamics (MD) simulations
A polystyrene molecule comprising a linear chain of 25 styrene molecules (PS) immersed in water and ions was built using CHARMM-GUI. 9A 400 ns of MD simulations were carried out for this system.Similarly, a POPC membrane (in water and ions) was built using CHARMM-GUI.Each leaflet of the membrane contains 200 POPC molecules.This system was equilibrated for 200 ns.
The last structure of PS derived from the 400 ns simulation of PS in water was inserted into the last structure from the 200 ns simulations of the POPC system, resulting in the POPC + PS complex.The final system consists of POPC, PS, water, and ions.PS was initially positioned away from the membrane surface, and some water molecules were removed to make room for the PS molecule.The concentration of NaCl in the system was kept at 0.1 M, representing a high salt concentration.

Details of the MD simulations
All MD simulations were performed using GROMACS 10 2021 program with CHARMM36 11 force field.TIP3P 12 water model was used in all simulations.The cutoff distance for the van der Waals and the coulombic interactions was 1.2 nm.The particle mesh Ewald summation (PME) 13 method was used to calculate the long-range electrostatic interactions.A 5000 step of energy minimization was done using the steepest descent algorithm to minimize the system.Subsequently, the system was equilibrated for 1 ns under the NPT ensemble conditions, employing a semi-isotropic Parrinello-Rahman barostat 14,15 set to maintain a pressure of 1 bar.The barostat was configured with a coupling constant of 5 ps and a compressibility of 4.5 × 10−5 bar−1, applied in both the lateral and membrane normal directions.Nose-Hoover thermostat 16,17 was used to maintain the temperature at 310 K.A production simulation of 1 µs was carried out from the last step of the equilibration.Only the last 800 ns simulations from the production run are used for analysis.
We also performed steered molecular dynamics (SMD) simulations to pull polystyrene from the bulk water to the center of the membrane.The center of mass of the PS was subjected to an external force with a spring constant of 1000 kJ mol −1 nm −2 , facilitating its movement across the membrane at a constant velocity of 0.1 nm ns −1 .The lateral motion of the PS was not constrained.A constraint was applied to the phosphorous atoms of the membrane during the SMD simulations.

Calculation of potential of mean force (PMF)
We calculated the PMF as a function of the distance between the center of mass (COM) of PS and COM of the membrane, using Umbrella Sampling (US) 18 technique and the weighted histogram analysis method (WHAM) 19 implemented in GROMACS.First, initial conformations were stored at intervals of approximately 0.1 nm along the SMD simulation trajectory.They were subsequently used to initiate independent MD simulations in each umbrella sampling window for 100 ns.The center of mass varies within a Δz-wide sampling window, while the z-location of the PS was fixed in each window.Finally, the PMF was obtained by integrating the force over z after it had been averaged over time and distance in each window.
We calculate the PMF of PS as a function of the distance (∆z) between COM of PS and COM of POPC, shown in Fig. S1(a).We set the minimum PMF value for PS in bulk water as the reference

Analysis Calculation of number density
We calculate the number density, which is defined by  We further extend our analysis by computing density profiles for PO 4 − and NMe 3 + ions, as well as sodium and chloride ions, along the z-axis of the simulation box across three distinct umbrella sampling windows, R1 (Fig. S11A and S11B), R2 (Fig. S11C and S11D), and R3 (Fig. S11E and S11F).The density profiles for the headgroups and the ions do not show any significant difference between any of the windows compared to the density profiles obtained from equilibrium simulations.This suggests that neither the position nor the peak of the density profiles for the ions or the headgroups is impacted by the location of PS.

Figure S1 :
Figure S1: Schematic of the TIRF microscope and the microfluidic sample.

Figure S2 :
Figure S2: Control experiment of PS bead with blank coverslip (No SLB) in buffer and 1000 µM NaCl solution.Representative frames from the raw data have been plotted for different conditions.The scale bar is 2 μm.

Figure S4 :
Figure S4: Cumulative distribution of SRTs for PS particles in buffer solution (green), 10 μM salt (blue), 100 μM salt (yellow), 1000 μM salt (dark red), and after washing with buffer from exposure to 1000 μM salt (pink).Three and two component fittings are represented as dashed and dashed-dotted lines, respectively.Three exponential fit provides a better fit.

Figure S8 :Figure S9 :
Figure S8: Uniform coverage of NR on SLB, representing the formation of a homogeneous single layer of lipid bilayer on plasma-cleaned coverslip.
PMF = 0).The PMF shows a peak at ∆z = 17.4 Å.The barrier height, defined as the PMF difference between the peak and the tail of the PMF profile, is observed to be ~29.5 kJ/mol.The PMF shows that PS is highly stable inside the POPC due to the hydrophobic interactions between the PS and the lipid tails.We show three snapshots of the systems, one at R1, another at R2, and the third at R3, shown in Figs.S1 (b)-(d), respectively.

Fig. S10 .
Fig. S10.(A) PMF of PS insertion as a function of the distance (∆z) between the COM of PS and COM of POPC.Three different regions are marked in circles in the PMF profile: one at the minimum of the PMF (inside POPC), shown in green circle (R1), another at the barrier, shown in cyan circle (R2), and the third at the tail (bulk water), shown in purple circle (R3).The Snapshots corresponding to these three regions are shown in Figs.(B)-(D).Water molecules are not shown due to clarity.Polystyrene is illustrated in red with vdw representation.The lipids are shown in grey with stick representation, and phosphorous atoms are shown in tan spheres.

.
Here, n is the number of atoms, and V is the box volume corresponding to a particular bin of 1 Å width.The density is calculated along the z direction of the box.We calculate the number density of PO 4 − , NMe 3 + , Na + , and Cl − ions from the last 800 ns of the equilibration MD simulations.We also calculate the number density for the ions from the three different umbrella sampling windows.The three windows represent three different regions in the PMF: one chosen near the minimum (R1), one at the barrier (R2), and one from the tail (R3) of the PMF.Note that the PS is in the bulk water at the R3 window.

Table S1 :
Number of particles on the surface.All data collected from the same region of the lipid bilayer surface at different NaCl salt concentrations.A gradual increase in the number PS particles interacting with the lipid surface is confirmed from the average number of particles.

Table S2 :
Average and standard deviations of short and long displacement populations calculated from Gaussian sampling of single frame displacement graphs.

Table S4 :
Average and standard deviation of α2 values of particles exposed to 10 μM, 100 μM, and 1000 μM salt.

Table S5 :
Average diffusion coefficients of PS under different conditions

Table S6 .
Average number of Na + ions present within 5.0 A of PO 4 − headgroup, and the number of Cl − ions present within 5.0 Å of NMe 3 + headgroup.This number is averaged over the