Article pubs.acs.org/Langmuir
Surface Charge Dependent Nanoparticle Disruption and Deposition of Lipid Bilayer Assemblies Xiaoyin Xiao,† Gabriel A. Montaño,*,‡ Thayne L. Edwards,† Amy Allen,† Komandoor E. Achyuthan,† Ronen Polsky,† David R. Wheeler,† and Susan M. Brozik*,† †
Biosensors and Nanomaterials Department, Sandia National Laboratories, Albuquerque, New Mexico 87185, United States Center for Integrated Nanotechnologies, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, United States
‡
S Supporting Information *
ABSTRACT: Electrostatic interaction plays a leading role in nanoparticle interactions with membrane architectures and can lead to effects such as nanoparticle binding and membrane disruption. In this work, the effects of nanoparticles (NPs) interacting with mixed lipid systems were investigated, indicating an ability to tune both NP binding to membranes and membrane disruption. Lipid membrane assemblies (LBAs) were created using a combination of charged, neutral, and gel-phase lipids. Depending on the lipid composition, nanostructured networks could be observed using in situ atomic force microscopy representing an asymmetrical distribution of lipids that rendered varying effects on NP interaction and membrane disruption that were domainspecific. LBA charge could be localized to fluidic domains that were selectively disrupted when interacting with negatively charged Au nanoparticles or quantum dots. Disruption was observed to be related to the charge density of the membrane, with a maximum amount of disruption occurring at ∼40% positively charged lipid membrane concentration. Conversely, particle deposition was determined to begin at charged lipid concentrations greater than 40% and increased with charge density. The results demonstrate that the modulation of NP and membrane charge distribution can play a pivitol role in determining NPinduced membrane disruption and NP surface assembly.
S
of NP surfaces needs to be precisely designed and carefully evaluated before any in vivo medical applications are attempted. Cellular membranes are complex, dynamic structures that contain not only lipids, but a mixture of proteins, polysaccharides, etc. that can make determining specifics of NP interactions difficult.12 Thus, many investigations seeking to understand NP−lipid membrane interactions use lipid bilayer assemblies (LBAs) as biological mimics to understand this most basic of NP−membrane interactions.13−15 LBAs are also used as biomaterial platforms for developing biosensors and diagnostic systems. As such, the ability to control material assemblies using LBAs is a focus of many studies. For example, it was recently shown that fluidic LBAs were capable of organizing gold nanoparticles (AuNPs) into nanoring ordered structures by addition of ZrCl4.16 This study indicated that it
ynthetic organic or inorganic nanoparticles (NPs) have been widely used in biomedical applications, such as drug delivery, gene therapy, cell imaging, and nanomedicines.1−5 Thus, investigating NP interactions with biological cell membranes has become a fundamentally important research topic. In many cases, the NP needs to bind, disrupt, and penetrate the cell membrane to induce a response, which is greatly dependent on the size, geometric shape, surface charge, and surface functionality of the NPs.6−9 NPs having dimensions of less than 2 nm can typically penetrate the cell membranes, while the larger particles are largely found to be membrane disruptors.1,3−11 Membrane disruption can greatly alter the membrane permeability, membrane potential, and thus membrane function. Contrarily, on occasions where NP interaction without disruption is desired, it is necessary to eliminate toxic effects of NP interactions such as membrane disruption with biological membranes. Thus, depending on the desired membrane interaction and response, the functionality © 2012 American Chemical Society
Received: August 14, 2012 Revised: October 10, 2012 Published: November 19, 2012 17396
dx.doi.org/10.1021/la303300b | Langmuir 2012, 28, 17396−17403
Langmuir
Article
Figure 1. Schematics of partially and fully charged AuNPs, pH-tunable negatively charged CdSe/ZnS core shell QDs, and the three lipids used to assemble positively charged artificial bilayers.
phenomena such as these becomes simpler when using a model LBA system and provides a basis for interpretation of cellular study results. In this work, we probe NP−LBA interactions while varying the NP and LBA surface charge densities and membrane composition. The artificial LBAs were assembled using mixtures of three lipids: cationic dioleoyltrimethylammonium propane (DOTAP), zwitterionic 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), and zwitterionic distearoylphosphatidylcholine (DSPC). DOTAP carries a net positive charge, and POPC and DSPC are neutral. The variation of the DOTAP molar ratio in DOTAP−POPC, DOTAP−DSPC, and DOTAP−DSPC−POPC two- or three-lipid systems provides surface charge density (i.e., charge per surface area) changes. The net positive charge is proportional to the DOTAP concentration and molar ratio. A varying amount of DSPC, which has a phase transition temperature well above room temperature, and above those of the other two lipids, which are fluid at room temperature, allows for generation of gel-like domains of controlled size and proportions. The AuNPs were capped by a monolayer of 11-mercaptoundecanoic acid (MUA), resulting in a fully charged negative surface at pH above 8. The partially charged AuNPs were prepared through a ligand exchange reaction between 11-mercaptoundecanoic acid and dodecanethiol. The surface charge density (i.e., charge per particle) was then varied by the molecular ratio between these two ligands adsorbed at the NP surface. The CdSe/ZnS core shell quantum dots (QDs) were functionalized with carboxylic acid terminals, and their surface charge density was varied by the pH value of the buffer solutions through dissociation of carboxylic acid. Atomic force microscopy (AFM) was used to determine the NP−bilayer interactions of various mixedcomposition lipid bilayers before and after interacting with the NPs.
was the dynamic ability of the LBAs to reorganize due to their intrinsic fluidity and the membrane response to low pH that resulted in the nanoring structure formation. Furthermore, only AuNPs that were only partially charged exhibited the ability to form nanorings. This study suggests the ability to tune membrane−NP interactions by tuning either the NPs or the composition of the lipid membrane to obtain a desired effect. The surface charge of NPs has been proposed to play a leading role in cell membrane disruption.17−20 Positively charged NPs disrupt cell membranes to the greatest extent, while neutral, negative, and zwitterionic NPs have negligible effects.21,22 Since mammalian cell membranes, such as those of cancer cells and human airway epithelial cells, typically carry net negative charges, it is proposed that the electrostatic attraction of positively charged NPs initiates membrane disruption. Computational models support such findings.23,24 However, when using artificial supported lipid bilayers, Banaszak Holl’s group concluded that a wide variety of positively charged NPs (AuNPs, dendrites, etc.) also disrupt neutral lipid bilayers, such as the dimyristoylphosphatidylcholine (DMPC)-containing bilayers.25−27 While it is understandable that variations between biological membranes and LBAs exist due to the difference in composition, it is much simpler to begin understanding interactions with either system by beginning to look at the simpler LBA system in which it is possible to systematically control composition. For example, it is well-known that surface defects (ion channels, pits, cavities, etc.) can form during bilayer phase transition.28,29 When one considers that the fluidto-gel phase transition of DMPC bilayers is right around room temperature, it is possible that such surface defects in bilayers might play a critical role in membrane disruption. Using LBAs, it is possible to localize such an effect by composition and temperature. Furthermore, understanding the role of electrostatic interactions of NPs with lipids required for NP-induced cell membrane disruption can be investigated without the influence of charged protein constituents. It is also possible to investigate the NP surface charge and charge density in relation to membrane interaction and disruption. Investigating
■
RESULTS AND DISCUSSION Figure 1 illustrates the structures of partially and fully charged AuNPs, CdSe/ZnS core shell QDs, and the three lipids used to 17397
dx.doi.org/10.1021/la303300b | Langmuir 2012, 28, 17396−17403
Langmuir
Article
image of the fluidic LBA area, three distinct heights can be observed. The lowest height represents surface defects and inconsistencies within the LBA. The next level of height structures represents DOTAP bilayer areas, and the highest level areas represent areas of asymmetric distribution of a DOTAP and DSPC mixed LBA. The asymmetric distribution of charged DOTAP lipids minimizes the electrostatic repulsion among the highly charged individual domains and thus is thermodynamically favorable. The mixed leaflet domains were still observable after the bilayer was heated to 60 °C and then cooled back to room temperature in the absence of lipid vesicles (Figure S1, Supporting Information); however, such thermoannealing disrupted the LBAs and led to the formation of large or small cavities near the gel domains. When POPC was mixed with DSPC (Figure S2D, Supporting Information) or was added to the DOTAP−DSPC mixtures (Figure S2C), the mixed leaflet domains were no longer visible. In addition to larger gel domain sizes, the fluidic phases were featureless (Figure S2C,D). Since POPC is miscible with DOTAP, POPC decreased the surface charge density in the fluid-phase regions by mixing with charged DOTAP. Such dilution of surface charges results in a decrease of the electrostatic repulsion necessary for the formation of mixed leaflet domains of DSPC− DOTAP. The same mechanism is also responsible for the decrease of gel domain sizes in the case of DSPC−DOTAP bilayer mixtures. At the same percentage of DSPC, the domain sizes in DOTAP−DSPC are about 10 times smaller than those in POPC−DSPC and POPC−DOTAP−DSPC bilayers (Figure S2). The decrease of domain size with increased surface charge density would also be associated with electrostatic repulsion that must be minimized among the charged fluidic domains.33−37 Dependence of the Bilayer Surface Charge Density on Membrane Disruption and Particle Deposition. The twoor three-lipid LBAs were stable in Tris buffer (0.05 M, pH 8.2) when the surfaces were imaged using AFM. The DOTAP may assist in LBA formation and stability on negatively charged mica substrates, leading to bilayer formation without observable pits or cavities. Formation of pit-free bilayers was essential in our experiments to clarify the consequence of NP interactions. After the addition of ∼1 pM negatively charged AuNPs, four typical changes were plausible: (1) no changes; (2) bilayer disruption; (3) NP deposition on the bilayer or within the bilayer; (4) disruption of the bilayer and deposition of NPs on the surface. The effect of the surface charge density of both NPs and the lipid bilayers was investigated. We did not observe any changes in AFM imaging of POPC LBAs after interacting with either fully or partially charged AuNPs. However, in the presence of ≥10% DOTAP, the fully charged AuNPs disrupted the LBA, leaving large particle aggregates on the mica surface (data not shown). With partially charged AuNPs, the membrane disruption could be resolved as a function of the DOTAP surface concentration (Figure 3). At DOTAP concentration below 20%, random pits or holes were observed after partially charged AuNPs were introduced (Figure 3A). The pits were ∼5 nm in depth corresponding to the LBA thickness. The disruption seemed to follow a nucleation and growth mechanism, and the growth of large cavities could be followed over time using AFM. The pits can grow into cavities on the order of a few micrometers within 1 h (Figure S3, Supporting Information). Our interpretation is that the AuNPs first peel the oppositely charged DOTAP molecules
assemble the artificial LBAs. The average core size of fully and partially charged AuNPs is around 3.5 nm. The QDs are packed with carboxylic acid terminals at their outer surfaces with a hydrodynamic size of ∼20 nm. We used buffer solutions with pH values of 5.0−8.2 to vary the amount of surface charges at the QD surfaces. At room temperature, DOTAP and POPC form miscible fluid LBAs, while DSPC has a phase transition temperature well above room temperature (∼52 °C). Nanostructured Lipid Bilayers. Since both DOTAP and POPC form fluidic LBAs at room temperature, their miscible mixtures, regardless of the molecular ratios, were also observed to be fluidic. The AFM images were nearly featureless and appeared the same for different DOTAP surface concentrations (images not shown). Contrarily, LBAs containing DSPC with fluid lipids often showed two phases with the domain sizes varying with the DSPC concentration.30−32 Of particular interest, we observed nanostructured networks (i.e., mixed leaflet bilayers) within the so-called fluidic phases when DSPC was present. Figure 2 shows AFM images of DOTAP−DSPC
Figure 2. AFM images of mixed DOTAP−DSPC lipid bilayers: (A, B) 60% DOTAP−40% DSPC, (C, D) 20% DOTAP−80% DSPC. The inset in (C) indicates mixed fluid-phase domains. The inset in (D) is a cross section profile of the line drawn in the image. Lighter areas indicate “raised” gel-like domain patches that typically are thicker than fluid-like domains.
mixed-phase LBAs. At a DOTAP molar ratio of 60%, as shown in Figure 2A,B, the fluidic phase occupies much more than 60% of the surface, indicating that some of the fluidic phase also contains DSPC. Upon closer inspection, it can be observed that, within the fluidic domains, two lipid phases actually exist (Figure 2B,C, inset). The mixed leaflet bilayers could be observed over the entire fluidic domain up to the edges of the gel domains (Figure 2C, inset). The height variation (
