NANOPARTICLE-CELL MEMBRANE INTERACTIONS: ADSORPTION KINETICS AND THE MONOLAYER RESPONSE

The fast-growing production and utilization of nanomaterials in diverse applications will undoubtedly lead to the release of these materials into the environment. As nanomaterials enter the environment, determining their interaction with biological systems is a key aspect to understanding their impact on environmental health and safety. It has been shown that engineered nanoparticles (ENPs) can interact with cell membranes by adhering onto their surface and compromising their integrity, permeability, and function. The interfacial and biophysical forces that drive these processes can be examined using lipid monolayers or bilayers as model cell membranes. Interfacial interactions between NPs and cell membranes have been proven to be affected by various parameters such as the physicochemical properties of the NPs, cell membrane composition, and the extent of exposure. This study focuses on the effects of NP charge, surface functional groups and interfacial activity on the response of lipid monolayers. Dynamic surface pressure measurements were used to examine the kinetics of nanoparticle adsorption and the monolayer response. Fluorescence and real-time in situ Brewster angle microscopy (BAM) imaging were employed to characterize the morphology and structure of the monolayers. Bulk concentrations of NP and phosphorus were examined to determine the extent of NP binding and lipid extraction. The results of this study will contribute to further understanding of the membrane’s role in ENP cytotoxicity and cellular uptake and aid the design of biocompatible nanomaterials with minimal or controlled membrane activity.


INTRODUCTION
The production and utilization of engineered nanoparticles (ENPs) in technology and medicine is constantly expanding; 1 however, there are still many uncertainties associated with the potential risks that they pose to environmental health and safety (EHS). 2,3 Fundamental studies that assess the hazard of ENPs are necessary in order to promote safe use and limit risks, and to guide the design of environmentally and biologically compatible materials. 4 Due to their high specific surface area and nanoscale size (<100 nm), ENPs display novel physical and chemical properties that are substantially different from those observed in the bulk materials. 5,6 Hence, ENPs are suitable candidates for a broad variety of commercial applications. For instance, metal NPs such as gold 7 or silver 9,10 exhibit unique optical, electronic and catalytic properties, primarily due to their localized surface plasmon resonance (LSPR) characteristics, 11 and they have been used for environmental remediation, (bio)chemical sensing, and drug delivery. 6,12 However, nanoparticles have been shown to bioaccumulate and exhibit various levels of toxicity. [13][14][15] This can be attributed to their size, shape, surface chemistry, and surface reactivity, which may allow them to penetrate tissues, enter cells, and interact with the compartments of the cell membrane. 16,17 This process can lead to a range of nanoparticle-induced biophysical and/or biochemical changes with the degree of change dependent on a variety of parameters such as cell membrane composition, NPs concentration and physicochemical properties, and the extent of exposure. [18][19][20][21][22] As a result, the safe use of ENPs in biological systems requires evaluation of their possible cytotoxicity.
Recent toxicological studies conducted in vitro and vivo have demonstrated that both carbon-based 13,14 and inorganic 23,24 NPs can strongly interact with cell membranes, and cause cytotoxicity through a variety of disruptive mechanisms including (1) adherence of the NPs to the membrane, (2) aggregation around the membrane, (3) removal of lipids from the membrane, and (4) permanently embedding into the membrane. 24 Adhesive forces between nanoparticles and cell surfaces driven by surface interactions, notably electrostatic, hydrophobic, and van der Waals, govern the timescale for nanoparticle-cell association, membrane disruption, and the extent of cellular uptake. [25][26][27] This behavior is independent of well-known cytotoxicity mechanisms related to chemical stability by which inorganic ENPs can release ions into solution or generate reactive oxygen species. 28 Dawson et al. 29 described how the scientific community generally views nanoparticle-cell interactions as occurring through "classical biological processes," but emphasized the importance of physical interactions (thus far neglected), such as those occurring between nanoparticles and membrane barriers. This is further emphasized by observations that greater nanoparticle-lipid interactions correlate with greater cellular uptake. 30,31 Hence, understanding nanoparticle-membrane interactions at the biophysical level will provide new insight into how nanoparticles affect cell function and viability. Understanding these interactions will elucidate the membrane's role in ENP cytotoxicity and cellular uptake and aid the design of biocompatible nanomaterials. This increased understanding may also provide new routes for designing nanoscale assemblies for biomedical applications.
NP uptake initiates with an attachment of the particle to the cell and subsequent interactions with the lipids and other components of the cell membrane. The interfacial and biophysical interactions that modulate this process can be examined using lipid bilayers or monolayers as model cell membranes. 27,[32][33][34][35][36][37][38][39][40] Cellular membranes are complex, multicomponent systems that contain a variety of charged and uncharged lipids with different degrees of tail saturation. In model cell membranes, attempts to mimic the complexity of real membranes involve adding multiple lipids to achieve a net surface charge and/or coexisting membrane domains (e.g., ordered and disordered). Two main advantages of model membranes are that the lipid composition can be varied, and that membrane organization and disruption can be measured directly using techniques that are not amenable to living cells. These simplified structures can be considered as first step approaches to investigate real systems due to their ability to mimic some of the most relevant physicochemical features of the real cell membrane. 27,34 The overall objectives of this dissertation were (1) to develop experimental approaches to capture the key parameters that control the duration and extent of nanoparticle adhesion to model cell membranes, and (2)

INTRODUCTION
Physical interactions between engineered nanoparticles and lipid membranes play an important role in nanotoxicology and nanomedicine. [1][2][3] Adhesive forces between nanoparticles and cell surfaces driven by surface interactions, notably electrostatic, hydrophobic, and van der Waals interactions, govern the timescale for nanoparticle-cell association, 4 changes in nanoparticle organization at the membrane/water interface, [5][6] membrane disruption, 7 and the extent of and cellular uptake. [8][9] The interfacial and biophysical interactions that drive these processes can be examined using lipid bilayers or monolayers as model cell membranes. [10][11][12][13][14][15][16][17][18] The main advantages of model membranes are that the lipid composition can be varied and that membrane organization and disruption can be measured directly using techniques that are not amenable to living cells. Model membranes have been used extensively to examine the adsorption of, and in some cases the resulting disruption caused by, carbonaceous, [19][20][21][22] metal oxide, [23][24][25][26][27][28][29] metallic, 11, 30-32 and polymeric 28,33 nanoparticles. Recent studies have also been conducted to determine how proteins or natural organic matter, adsorbed onto the nanoparticle surface, influence membrane interactions. 20,34 Cellular membranes are complex, multicomponent systems that contain a variety of charged and uncharged lipids with varying degrees of tail saturation. In model cell membranes, attempts to mimic this complexity involve adding multiple lipids to achieve a net surface charge and/or co-existing membrane domains (e.g. ordered and disordered). In context of nanoparticle-membrane interactions, Ha et al. 19 have shown that fullerene partitioning to lipid bilayers composed of biologically relevant ternary lipid mixtures that can form liquid ordered 'lipid raft' domains is lower below the phase transition temperature than above the transition temperature when the rafts are present. shown to bind quickly and strongly to PC/PG membranes leaving them intact, but causing an increase in membrane rigidity. 36 Cationic nanoparticle binding to PC and PC/PG membranes also lead to membrane protrusions and pore formation due to 'steric crowding' within the membrane as the nanoparticles pack on the surface and consume excess area between the lipids. 12 Steric crowding caused the lipids to pack more tightly or compress, which increased the surface tension of the membrane. Finally, we have also shown that anionic and cationic silver nanoparticles (AgNPs) bind to PC/PG membranes (bilayer vesicles) without membrane rupture. 11 However, AgNP binding did lead to membrane deformation and vesicle aggregation due to membrane-AgNPmembrane bridging. 11 Lipid monolayers have been successfully used to examine nanoparticle-lipid interactions based on changes in interlipid interactions that affect the degree of lipid packing and the monolayer phase behavior, and on lipid extraction from the air/water interface due to nanoparticle-lipid binding. 23-31, 33, 37-38 This study focuses on the effects of AgNP charge, provided by anionic and cationic polymer coatings ( Fig. 2-1), on the response of PC/PG monolayers (3:1 mol). Dynamic surface pressure measurements were used to examine the duration and extent of nanoparticle adsorption and the monolayer response. Sub-phase Ag and phosphorus (P) concentrations were examined to confirm AgNP binding and the extent of lipid extraction. AgNPs, referred to as Ag-COOH, were coated with a carboxylated amphiphilic polymer formed by hydrolyzing poly-(maleic anhydride-alt-1-octadecane). [39][40] Cationic AgNPs, referred to as Ag-NH, were prepared by coating Ag-COOH nanoparticles with polyethyleneimine. Sterile, ultra-filtered deionized water was obtained from Millipore Direct-3Q purification system and adjusted to pH 7.   DPPG and DOPG are sodium salts and the concentration of Na + counterions within the subphase was equivalent to 3×10 -5 mM. Isotherms were generated for a single compression/expansion cycle at a barrier rate of 10 cm 2 min -1 and π was measured using paper Wilhelmy plates. The total area of the trough during this cycle ranged from roughly 20-70 cm 2 .

Materials
Step (2) was used to determine the change in monolayer surface pressure in the presence of AgNPs as a function of time. To measure dynamic changes in surface pressure (∆π) the trough was initially set to maintain a constant surface pressure (π0 = 10, 20, or 30 mN m -1 ) after the compression/expansion isotherms (step 1). Once the monolayer stabilized and π0 remained constant, the barrier positions were fixed at the corresponding interfacial area or charge density. AgNPs were added to the water subphase by injecting them behind the barriers using a syringe to avoid disrupting the monolayer. The volume and concentration of the AgNP solution that was injected was 100 uL and 5 mg mL -1 , respectively. The AgNPs were mixed within the subphase by recycling the solution using a peristaltic pump. Control experiments confirmed that the pumping action did not disturb the monolayers and that water evaporation did not alter the ∆π measurements. The initial AgNP concentration in the subphase was 3.6 mg L -1 or 33.4 M, which was estimated to provide excess surface coverage based on the AgNP cross sectional area at a monolayer surface area of 70 cm 2 .

AgNP Characterization.
AgNPs were characterized prior to the monolayer experiments to confirm their physicochemical properties and to determine the extent of AgNP dissolution. The average rc was 6 ± 2 nm based on TEM analysis and was common to both Ag-COOH and Ag-NH (TEM, Fig. 2-3A). The polymer coatings surrounding the AgNPs were not observed in the micrographs. Ag-COOH had a hydrodynamic radius, rh, of 14 ± 2 nm (0.02 PDI) and a zeta potential, ζ, of -63 ± 3 mV.
Ag-NH had a rh = 20 ± 3 nm (0.02 PDI) and a ζ = +46 ± 2 mV. The average coating thicknesses based on the difference between rh and rc were 8 nm for Ag-COOH and 14 nm for Ag-NH. The increase in coating thickness from Ag-COOH to Ag-NH is consistent with PEI coating of Ag-COOH.
The maximum absorbance due to AgNP surface plasmon resonance (SPR) was observed at a wavelength of 410 nm ( Fig. 2-3B). The SPR absorbance was measured over 3 months to confirm the stability of the AgNPs and determine the extent of dissolution. There was no shift in the SPR wavelength, indicating that the AgNPs were stable. A slight reduction in SPR absorbance was observed over 3 months consistent with a ~3% decrease in the AgNP concentration. Given that the monolayer studies were conducted within 1 month of receiving the samples, we did not account for AgNP dissolution in our analyses.
Finally, the surface activity of the native AgNPs was examined in the absence of a lipid monolayer ( Fig. 2-3C). The π-A isotherm for Ag-COOH and Ag-NH showed a π of 16.9 mN m -1 and 6.3 mN m -1 , respectively, with 74% compression (70 to 18 cm 2 ) indicating that polymer coatings rendered the nanoparticles surface active due to hydrophobic interactions at the air/water interface.

Dynamic changes in monolayer surface pressure due to AgNP adsorption.
Dynamic changes in monolayer surface pressure, ∆π, were determined as Δπ = π(t) -π0 = 0 - (t), where π(t) is the dynamic surface pressure after AgNP addition and π0 is the initial surface pressure of the air/lipid/water interface. The relationship between ∆π and the initial air/lipid/water interfacial tension, 0, and the dynamic interfacial tension, (t), shows that an increase in ∆π would result from a decrease in (t) due to AgNP-lipid monolayer interactions (and vice versa). Changes in ∆π are depicted in Fig. 2 for the proposed AgNP-lipid monolayer interaction mechanisms. Hädicke and Blume 43 have shown that dynamic surface measurements with cationic peptides and anionic DPPG monolayers can be used to differentiate between peptide insertion into the monolayer (increasing Δπ) and lipid condensation due to peptide-lipid binding (decreasing Δπ). This approach has also been used to examine the insertion of gold nanoparticles into DPPC monolayers. 30 Increased initial packing (based on π0) prevented nanoparticle insertion and the decrease in Δπ indicates that Ag-COOH led to lipid condensation ( Fig. 2-2B2). This behavior was independent of phase state. A linear fit of Δπ as a function π0 at t = 180 min was used to estimate the minimum insertion pressure (MIP) of Ag-COOH, which corresponds to the condition Δπ = 0 (Figures 2-5A2 and 5B2). We refer to insertion as meaning that the nanoparticles breach the plane of the monolayer and occupy area at the air/water interface with or without an adsorbed lipid coating. The MIPs for that below this surface pressure the nanoparticles are capable of inserting into the monolayer. Above the MIP inter-lipid interactions within the monolayer resist nanoparticle insertion. The MIPs determined for Ag-COOH are considerably lower than those reported for 10 and 15 nm diameter anionic gold nanoparticles and zwitterionic DPPC monolayers. 30 It should be noted that the gold nanoparticle concentration was more than order of magnitude higher than what was used in this work. hydrophobic interactions with lipids tails, counterion-mediated (Na + ) binding to PGs, and electrostatic and charge-dipole interactions with PCs. The surface activity of Ag-COOH supports the assertion that Ag-COOH penetrated into loosely packed monolayers at π0 = 10 mN m -1 and resided at the air/water interface. It should be noted that the ability for Ag-COOH to insert into the monolayer might also stem from the nanoparticles being rendered partially hydrophobic due to the adsorption of lipids at the air/water interface and the formation of nanoparticle-lipid complexes. 24 Hydrophobic interactions do not, however, explain the reductions in surface pressure at 20 or 30 mN m -1 .
With regards to counterion-mediated binding, the Na + counterions associated with PGs may have facilitated the adsorption of Ag-COOH. This mode of adsorption has been proposed for anionic citrate-coated gold nanoparticles and DPPG monolayers, which caused an increase in surface pressure (or monolayer expansion). 31 Given that PGs comprised only 25 mol% of the monolayers examined herein, and significant decreases in surface pressure were observed consistent with lipid condensation, it is unlikely that counterion-mediating adsorption played a dominant role.
Electrostatic and charge-dipole interactions with PCs, which were present at 75 mol% in the monolayers, appear to be a main driving force for Ag-COOH adsorption.
At π0 = 20 and 30 mN m -1 the reductions in surface pressure suggest that the nanoparticles did not penetrate the monolayer, but rather remained bound to the monolayer below the interface and caused lipid condensation (i.e. a reduction in the effective area per lipid). It has been shown that anionic nanoparticles can bind to zwitterionic lipids 44 and pulmonary surfactant monolayers 28 through attractive interactions with the positive choline group of zwitterionic lipids. [23][24][25] Zwitterionic lipids have a dipole moment extending into the aqueous phase that can also lead to attractive short-range ion-dipole interactions. Anionic nanoparticles can reorient the headgroup dipoles of zwitterionic lipids, causing the dipole to orient perpendicular to the lipid/water interface and reducing the area per lipid. 44 Hence, lipid condensation in the monolayers appears to be attributed to the dipole reorientation of DPPC and DOPC. The ability for Ag-COOH to adsorb onto DPPC/DPPG monolayers is consistent with our previous work showing Ag-COOH adsorption onto DPPC/DPPG bilayer vesicles. 11 The role of lipid condensation was examined further using monolayers containing equimolar mixtures of PC and PG lipids (data not shown). Reducing the concentration of DPPC or DOPC from 75 mol% to 50 mol% reduced the magnitude of the Δπ decrease. With less PC lipid there was less lipid condensation.
Cationic Ag-NH. In contrast to Ag-COOH, the monolayers responded differently to oppositely charged Ag-NH and MIP values could not be determined (Fig. 2-6). A significant increase in π was observed for DPPC/DPPG monolayers at π0 = 10 mN ma decrease in π was observed suggesting that lipid condensation occurred, and at π0 = 30 mN m -1 a two-state response was observed where π increased rapidly up to 10 min (insertion) and then decreased exponentially (condensation). The rapid increase in π observed initially at π0 = 10 and 30 mN m -1 was due to electrostatic attraction between the monolayers and Ag-NH that drove adsorption and insertion. Electrostatic attraction was also present at π0 = 20 mN m -1 , however, the surface pressure response reflected competition between lipid condensation and Ag-NH insertion, where at this initial surface pressure, lipid condensation had the greatest impact on π. For DOPC/DOPG, π was unchanged (10 mN m -1 ) or reduced (20 and 30 mN m -1 ) and there was no evidence of Ag-NH insertion. Only lipid condensation was observed at high initial surface pressures. Lipid condensation caused by cationic Ag-NH was driven by electrostatic attraction with anionic DPPG or DOPG lipids and inter-lipid charge neutralization. This differs from anionic Ag-COOH, which interacted with the zwitterionic lipids. Previous work has shown that cationic gold nanoparticles have a minimal effect on the surface pressure isotherms of DPPC, 31 which further supports the assertion that PGs were responsible for Ag-NH adsorption. The concentration of AgNPs in the sub-phase (Fig. 2-7) provides a number of insights into the monolayer response. First, there is generally little difference in AgNP concentrations between the two monolayers; the exception being Ag-COOH at the highest monolayer charge density where the standard errors were large. This suggests that AgNP adsorption was primarily driven by lipid headgroup interactions and that the monolayer response was driven by the lipid tail saturation and phase behavior. Second, the sub-phase concentration of Ag-NH is less than Ag-COOH, which means that more    Nano ZSX for their core radius, and hydrodynamic radius and zeta potentials, respectively. The average core radius (rc) of Ag-PEG was determined by analyzing multiple TEM images with the ImageJ software (n > 50). 35 To measure the average zeta potentials (ζ) and hydrodynamic radius (rh) of Ag-PEG, the as-received particles were diluted ten-fold in deionized water and analyzed at 25 °C. The values reported are based on triplicate measurements of three different samples. trough had a fully opened area of ∼80 cm 2 and a width of 7 cm (Fig. 3-2).

Ag-PEG characterization.
Ag-PEG nanoparticles were characterized prior to the monolayer experiments for their size, zeta potentials, stability and extent of dissolution.
As shown in Fig. 3-3A, the average core radius (rc) was 6 ± 2 nm based on analysis of TEM images. The polymer coatings surrounding Ag-PEG were not observed in the micrographs. The mean hydrodynamic radius (rh) and zeta potential (ζ) were measured to be 15 ± 2 nm (0.04 polydispersity index) and -10.6 ± 0.1 mV, respectively. The average coating thickness based on the difference between rh and rc was 9 nm. The maximum surface plasmon resonance (SPR) absorbance was observed at a wavelength of 425 nm (Fig. 3-3B). Ag-PEG SPR absorbance was measured by UV−vis spectroscopy over 3 months, and there was no significant shift and reduction in the SPR wavelength indicating that the nanoparticles were stable. Similar to our previous study on anionic (COOH) and cationic (NH)-coated AgNPs, 26 considering that the monolayer experiments were conducted within 3 months of receiving the samples, we did not account for NP dissolution in our analysis. As shown in Fig. 3 In the presence of DOPC/DOPG monolayers at an initial surface pressure of 10 mN m -1 (Fig. 3-4B) Ag-PEG remained surface activity and the lipid monolayer did not prevent Ag-PEG adsorption at the interface ( Fig. 3-4D). Considering that both Ag-PEG and DOPC/DOPG monolayers exhibit a net negative charge, adsorption can be attributed to hydrophobic interactions. Xi et al. 21 have also demonstrated that Ag-PEG similar to those used in this study bind to DOPC/DOPG bilayer vesicles. In their work, it was proposed that the surface activity of the PEG-polymer coating may have facilitated membrane penetration through hydrophobic interactions despite electrostatic repulsion.  monolayer were similar, suggesting that lipid extraction was not a significant factor ( Fig. 3-5). Therefore, we conclude that Ag-PEG did not extract lipids from the monolayers and that the lipids remained at the interface to form a mixed Ag-PEG + lipid film.

55
The collapse pressure (πc, mN m -1 ) and collapse area (Ac, cm 2 ) were determined from − A isotherms of Ag-PEG at high nanoparticle concentrations (0.71 to 3.55 mg L -1 ) (Fig.3-7). The collapse pressure was directly proportional to Ag-PEG concentration. Based on Ac, and assuming 2D hexagonal packing, an effective Ag-PEG radius of 12.5 ± 3.9 nm was calculated at the interface. The calculated 'interface radius' of the nanoparticles is consistent with the measured hydrodynamic radius. Hence, Ag-PEG assembled as densely packed monolayers at the air/water interface at high concentrations, and the monolayers collapsed once they exceeded hexagonal packing. A comparison between compression/expansion isotherms of Ag-PEG at air/water and air/lipid/water interfaces are shown in Fig. 3-6. At low Ag-PEG concentrations ([Ag-PEG] ≤ 0.35 mg L -1 ), the isotherm shifted to smaller area with respect to the isotherm of the lipid mixture alone, noting that more compression was necessary for the Ag-PEG + lipid films to attain the same arbitrary surface pressure compared to pure lipid film. This behavior is not attributed to the extraction of lipid molecules (Fig. 3-5

INTRODUCTION
The environmental concentration of polymeric particles is constantly increasing due to the significant amount of plastic waste that is being disposed in the oceans and soil. [1][2][3] Recent studies on the size distribution of the plastic debris have shown that millimeter-size plastics can be fragmented to even smaller particles, referred to as micro-and nano-plastics, 4-7 which may pose a significant threat both to the environment and human health. [6][7][8][9][10][11][12][13][14][15] The small size of these particles (<1µm) makes them a susceptible of ingestion by organisms that are at the base of the food-chain. 1 The potential adverse effects associated with interactions between these materials and biological systems could be comparable to those observed with engineered nanoparticles (ENPs). [16][17][18] Toxicological studies conducted in vitro and vivo have demonstrated that polymeric ENPs can translocate across living cells to the lymphatic and/or circulatory system, 19,20 accumulate in secondary organs, 21 and impact the immune system and cell health. [22][23][24] NP cellular uptake begins with an initial adhesion of the particle to the cell and subsequent interactions with the lipids and other components of the cell membrane. The interfacial and biophysical forces that modulate this process can be examined using lipid bilayers or monolayers as model cell membranes. [25][26][27][28][29][30][31][32][33][34] Two main advantages of model membranes are that: (1) the lipid composition and structure can be precisely controlled, thereby capturing the essential aspects of the real cell membranes, and (2) the membrane organization and disruption can be measured directly using techniques that are not amenable to living cells. 18 Model membranes have been used extensively to study the adhesion of, and in some cases the resulting disruption caused by both carbon-based and inorganic ENPs. 35 In the work discussed below, we have examined the response of human red blood cell model membranes to the adhesion of polystyrene (PS) nanoparticles with a particular emphasis on the effect of NP surface chemistry on this process.
Physicochemical properties of NPs, such as size, charge and surface chemistry are the main factors modulating NP durability and solubility in biological media as well as their biocompatibility and membrane interactions. 36 Upon encountering biological fluids (e.g. blood, lymph, cytoplasm, cell culture media) nanoparticles are covered by biomoleculesof which proteins have received the most attention, forming what is described as a "corona". 37,38 Recent research has revealed that in many cases it is the biomolecular corona that interacts with biological systems and thereby constitutes a major element of the biological identity of the nanoparticle. [39][40][41][42][43][44] In particular, the corona is composed of a tightly, but not completely irreversibly, adsorbed layer of biomolecules ("hard" corona), which is surrounded by a more loosely associated and rapidly exchanging layer of biomolecules ("soft" corona). 45 The formation of a corona has been reported for several nanoparticles, including polystyrene, 46 silica, 47 carbon nanotubes, 48 silver, 39 and gold. 49 The amount, composition, and orientation of biomolecules present in the corona strongly influence  HSA (5% in PBS) was added to the microcentrifuge tubes, and the tubes were incubated at 37 °C for one hour. The tubes were subsequently centrifuged three times (18000 rcf, 4 ℃) with a PBS solution wash between each centrifugation step. Finally, the sedimented NPs were re-dispersed in PBS to isolate the NPs and associated complexed proteins.

Characterization of NPs and NP-HC complexes. NPs and NP-HC complexes
were characterized using transmission electron microscopy (TEM; JEOL JEM-2100F) operating at 200 kV and a Malvern Zetasizer Nano ZSX for their core radius, and hydrodynamic radius and zeta (ζ) potentials, respectively. The average size of PS NPs was determined by analyzing multiple TEM images with ImageJ software (n > 50). 70 To measure the average ζ-potentials and hydrodynamic diameter (dh) of NPs, the as- Exposure of PS NPs to protein led to changes in their hydrodynamic properties.
We incubated the PS NPs in human serum albumin (HSA) solution for 60 min to allow NP-HSA complexes to form and separated these complexes from free and weakly complexed HSA via a series of centrifugation and washing steps comparable to those previously used to operationally define the hard corona on nanoparticles (Fig. 4-2A). 1,50,76 The changes in ζ-potential and dh of the particles (Fig. 4 M HSA, which reached a plateau at dh = 123 nm for 300 μM HSA. We infer that HSA concentration of 300 μM is sufficient to saturate the NP surface and form a close-packed monolayer of protein corona. 71 The increase in dh due to corona formation was about 20 nm and was common to all of them, corresponding to a hydrodynamic-shell thickness of 10 nm (Fig. 4-2C). Negative-staining TEM of NP-HC complexes confirmed that the HSA shell thickness on the NPs was 7 ± 1 nm ( Fig. 4-2B Dynamic changes in surface tension (DST) for unmodified, carboxylate-modified, and amine modified PS and PS-HC complexes are depicted in Fig. 4-3A-C, respectively.
In general, as NPs diffuse from the bulk and adsorb to the interface, they effectively reduce . Early in this process, decreases relatively slowly due to the adsorption of single particles to a pristine interface. When the surface concentration of NPs increases, drops more rapidly. At long times ( → ∞), where the interface approaches maximum coverage, the rate of NP surface adsorption decreases due to a steric barrier and approaches a plateau reflecting a pseudo-equilibrium condition.
As shown in Fig. 4-3A-C, bare NPs were not inherently surface active. Although Brewster angle microscopy images showed the adherence of particles at the air-water interface ( Fig. 4-3D-F), the reduction in interfacial tensions due to their attachment, functional group.
HSA corona complexation rendered the NPs surface active due to hydrophobic interactions at the air-water interface, which led to a lower equilibrium surface tension  Results are shown in Fig. 4

Adsorption kinetics at the air-water interface. Dynamic interfacial tension data
can be further analyzed using the classical model of Ward and Tordai 81 to quantitatively describe the kinetics of NP adsorption. The following asymptotic equations have been employed to interpret data from the early ( → 0) and late ( → ∞) times of nanoparticle adsorption.
At early times (first-stage), an individual NP that is adsorbing to the interface encounters a bare interface. Assuming there is no barrier to adsorption at this stage, the rate of particle diffusion through the bulk is the rate-limiting factor and the diffusioncontrolled Ward and Tordai mechanism can be applied. 81 Bizmark et al. 82 modified the Ward and Tordai model to account for NPs larger than 10 nm with adsorption trapping energy exceeding 10 3 kBT: Here, is Avogadro's number, ∆E is the trapping energy of a single particle at the interface, is its diffusion coefficient, and 0 is the molar concentration. The number of NPs adsorbed at the interface is significantly less than that remaining in the bulk and C0 is assumed to be constant throughout the adsorption process.
Surface coverage at any time during the adsorption process can be calculated from the measured surface tension: 82 where ∞ is the maximum fraction of surface coverage, which is 0.91 for hexagonal close packing of spheres 83 , 0 is the pristine interfacial tension of water, and ∞ is the equilibrium interfacial tension. For native NPs, considering that they were not surface active, ∞ was determined based on calculated excess PS surface concentrations at the end of adsorption process and was less than 0.5 for all three types of them. We note that for NP-HC complexes, ∞ = 0.91 as they were surface active and assembled as denselypacked monolayers at the air-water interface. , in which is the hydrodynamic radii of the particles and is the viscosity of water at room temperature. As summarized in Table   1, the values of 1 and are within the same order of magnitude, indicating that equation (1) is valid during the early times adsorption of particles from the bulk to the air-water interface.
Using values, we were able to extract the first-stage adsorption energy, |∆ 1 |, by fitting the slope of early time DST data against 0.5 . As shown in Table1, there was a clear correlation between the adsorption energy and the ζ-potential of the NPs. Anionic unmodified and carboxylate-modified PS had similar adsorption energy, while greater values were observed for cationic amine-modified PS. Anionic PS NPs were electrostatically repelled from the interface, since the ζ-potential at the air-water interface has been shown to be negative. 84,85  In the first-stage approximation ( < 0.3) proposed by Bizmark et al., 82 only one slope was observed when DST was plotted against 0.5 . We observed similar behavior for NP adsorption at early times. However, for NP-HC complexes, two distinct stages with clearly different slopes were noted in a plot of early time DST over 0.5 (Fig. 4-5B) consistent with the results of recent work by Tian et al. 86 using poly(ethylene oxide) (PEO)-modified polystyrene NPs to study the adsorption kinetics at the air-water interface. 86 The presence of two distinct stages at the early time adsorption were comparable when NP-HC complexes were employed. As shown in Fig. 4 (Table 1). The observed two-stage transition for NP-HC complexes is attributed to protein denaturation at an interface. 81 HSA has hydrophilic groups on its surface that make it water-soluble, but hydrophobic peptide residues in the core.
Proteins denature at a hydrophobic interface wherein the hydrophobic core peptides unfold at the interface, while the hydrophilic peptides orient toward the aqueous phase.
The extent of increase in adsorption energy due to HSA denaturation at the air-water interface was consistent with the extent of HSA associated with NPs. The greater increase in |∆ | was observed for PS-NH-HC, while unmodified and carboxylatemodified PS NPS showed similar values.    It has been shown that anionic nanoparticles can bind to zwitterionic lipid monolayers and bilayers through attractive interactions with the positive group of zwitterionic lipids (e.g. choline group of POPC and ethanolamine group of POPE). [88][89][90] Moreover, zwitterionic lipids have dipole moments extending into the aqueous phase that can lead to attractive short-range ion-dipole interactions. Both anionic and cationic nanoparticles can reorient the headgroup dipoles of zwitterionic lipids, causing the dipole to orient perpendicularly to the air-water interface and reducing the area per lipid. 90 Hence, lipid condensation in the RBC monolayers can be attributed to the dipoles reorientation of zwitterionic POPC and POPE. We observed similar behaviour in our previous work using carboxylate-and amine-modified silver NPs and PC/PG monolayers. 25 The morphology of the monolayer was visualized in situ using Brewster angle microscopy (BAM) technique. As shown in Fig. 4-8A, the extent of lipid condensation was greater for PS-NH compared to unmodified PS and PS-COOH, suggesting that inclusion of cationic nanoparticles within a monolayer induces more modification in the monolayer lipid packing. The extent of increase in RBC monolayer DST due to the adsorption of NP-HC complexes was smaller compared to that for bare NPs (Fig. 4-7), indicating that NP-HC complexes induced less lipid condensation. These results are consistent with our previous work using cationic and anionic silver nanoparticles and show that hydrophobic interactions were responsible for NP insertion, while electrostatic and charge-dipole interactions were responsible for lipid condensation. Moreover, real time BAM imaging of the film displayed lipid condensations at early time NP-HC complexes adsorption, and the formation of homogenous densely packed monolayer at equilibrium ( Fig. 4-8B1&2). Hence, we infer that at early times, NP-HC complexes penetrate into monolayers through attractive short-range ion-dipole interactions, bind to zwitterionic lipids and cause lipid condensation (increasing ) similar to what we observed for bare NPs adsorption. This process follows by the protein corona partitioning between coexisting membrane domains via attractive hydrophobic interactions (increasing ) and unfolding at the air-water interface. 91,92 This leads to the formation of homogenous densely packed RBC+HSA film at the interface, in which NPs are an integral part of the mixed film. 93 This behaviour was common to all three NP-HC complexes.

Excess NP and NP-HC concentrations at the air-lipid-water interface.
To further quantify the extent of NP and NP-HC complex adsorption at the air-lipid-water interface, the subphase concentrations of PS were analyzed by UV-vis spectroscopy.