Assembly of Lipid Bilayers on Silica and Modified Silica Colloids by

ARTICLE pubs.acs.org/Langmuir

Assembly of Lipid Bilayers on Silica and Modified Silica Colloids by Reconstitution of Dried Lipid Films Eric E. Ross,* Sze-Wing Mok, and Steven R. Bugni Department of Chemistry & Biochemistry, Gonzaga University, 502 E. Boone Ave, Spokane, Washington 99258, United States

bS Supporting Information ABSTRACT: A method is presented for the assembly of lipid bilayers on silica colloids via reconstitution of dried lipid films solvent-cast from chloroform within packed beds of colloids ranging from 100 nm to 10 μm in diameter. Rapid solvent evaporation from the packed bed void volume results in uniform distribution of dried lipid throughout the colloidal bed. Fluorescence measurements indicate that significant, if not quantitative, retention of DOPC or DPPC films cast between subbilayer and multilayer quantities occurs when the colloids are redispersed in aqueous solution. Phospholipid bilayers assembled in this manner are shown to effectively passivate the surface of 250 nm colloids to nonspecific adsorption of bovine serum albumin. The method is shown to be capable of preparing supported bilayers on colloid surfaces that do not generally support vesicle fusion such as poly(ethylene glycol) (PEG) modified silica colloids. Bilayers of lipids that have not been reported to self-assemble by vesicle fusion, including gel-phase lipids and single-chain diacetylene amphiphiles, can also be formed by this method. The utility of the solid-core support is demonstrated by the facile assembly of supported lipid bilayers within fused silica capillaries to generate materials that are potentially suitable for the analysis of membrane interactions in a microchannel format.

’ INTRODUCTION Lipid bilayers have been assembled on a broad array of colloidal supports to provide biomembrane interfacial properties to chemical sensors and delivery vehicles.1 Colloidal supports may serve different roles depending on the application, but often provide control over the diameter, size dispersity, and stability of the assembled lipid bilayer system. For example, early reports examined phospholipid bilayers on glass microspheres ranging from 0.5 to 3 μm to study lipid biophysical properties by NMR2 and to determine enzymatic binding to phospholipids.3 The decreased rotational diffusion of supported bilayers with a relatively large, stable diameter decreased motional averaging of spectra in the former example, while in the latter, solid spheres provided a heterogeneous platform for binding assays that reduced vesicle aggregation, increased lipid density, and facilitated removal of unbound proteins. Many reports describing colloid-coupled lipids have followed.1 Selected examples include those describing phospholipid monoor bilayer formation on colloids composed of other materials such as silanized silica,4 quantum dots,5,6 polyelectrolyte layers,7
9 polymeric,10,11 gold,8,12 hydrogel,13,14 and magnetic particles,15 involving thin polymer cushions designed to insulate lipids or bilayer-embedded biomolecules from unwanted interactions with the colloidal core material16,17 or using porous silica colloids and beads for the bilayer-mediated entrapment and release of encapsulated solutes.18
21 A fundamental process in the development of any lipid-colloidal system is the elucidation of assembly conditions and methodology that r 2011 American Chemical Society

produces uniform coverage of stable, low-defect, single or multilamellar bilayers. Self-assembly of lipid bilayers via vesicle fusion is the most common methodology employed for both porous and nonporous silica colloids. Vesicle fusion involves adsorption, rupturing, and spreading of vesicles to form continuous single lamellae on a substrate. As with planar silica or titania surfaces,22
26 this process is pH, ionic strength, and lipid dependent,27,28 yet it is very effective for many neutral and charged lipid formulations on silica colloids ranging from tens of nanometers to many micrometers.1,2,29
32 Fusion has also been reported to occur with vesicles bound to silica or modified silica colloids through avidin
and streptavidin
biotin linkages.33,34 Formation of bilayers on polymer cushions or hydrogel particles can be driven by alkyl groups synthetically incorporated into the target polymer13,14 or by fusion to mixed monolayers of lipid and alkylated PEG films.35 On planar silica surfaces, it has been shown that a poly(ethylene glycol) (PEG) cushion layer can result from the fusion of liposomes containing a minor PEGylated lipid component,36 but this approach has not been utilized on colloids to our knowledge. Lipid bilayer formation on nonsilica and nonfusion promoting colloids can often be driven by electrostatic interactions between charged polymer surfaces

Received: March 14, 2011 Revised: June 2, 2011 Published: June 02, 2011 8634

dx.doi.org/10.1021/la200952c | Langmuir 2011, 27, 8634–8644

Langmuir and ionic vesicle constituents such as cationic alkylated ammonium lipids27,28,37 or anionic phosphatidic acid.7 Despite the utility of vesicle fusion, it is not without some limitations. Among these are the general requirements that lipids be in the fluid phase (LR), the need to prepare single unilamellar vesicles (SUVs) prior to colloidal incubation, the inability to generate multilayers, the fractional incorporation of the vesicle lipids into the supported film (excess vesicles are typically used to ensure surface saturation), as well as the incompatibility with many surface chemistries. Methods overcoming some of these limitations have been reported, including reports by Carmona-Ribeiro and co-workers describing both the cationic-lipid assisted rupturing of adsorbed vesicles of the gelphase lipid DPPC28 and the adsorption of cationic bilayer fragments to charged colloids,38,39 and by Cho et al. demonstrating peptide induced rupturing of vesicles to assemble bilayers on gold and titania surfaces.40 Alternatives to vesicle fusion include methods utilizing detergent depletion,41 solventexchange,42
44 and spin-coating;45
47 these approaches are significantly less characterized than vesicle fusion and have largely involved planar supports. Spin-coating is most closely related to the solvent-casting method described herein, in that it distributes dried lipid films to be reconstituted in place to form supported lipid bilayers. Planar surfaces are wetted with dissolved lipids, then rapidly spun to remove excess solvent and generate uniform stacks of bilayers (multilayers) on the surface. The excess lipid layers can be desorbed in some cases to leave a single bilayer on the substrate.45 Lipid multilayer films resulting directly from solvent evaporation (no spinning) have been formed on IR active substrates such as silicon or zinc selenide for spectral studies of lipid phase transitions.48,49 Dubertret et al. demonstrated that alkylated quantum dots encapsulated in micelles of pegylated lipids could be achieved by reconstitution of an evaporated mixture.5 Finally, a variety of formulations composed of lipids and, for example, drugs, DNA, or amino acids, have been spraydried to form colloidal particles for improved performance of dry powder inhalers.50
52 Reconstitution of the dried colloids with water has in some cases been shown to form liposomal particles with encapsulated solutes. To our knowledge, lipid bilayers supported on solid-core particles have not been prepared by reconstitution of dried lipid films on colloids or by spray-drying. Previously, we have shown that dried lipid films are uniformly deposited throughout close-packed beds of silica colloids by solvent evaporation from chloroform or ethanol solutions, and that data supports the conformity of lipid bilayers to individual colloids upon rehydration.53 In that work, the packed beds were in the form of colloidal crystals that were stabilized to prevent disruption upon rehydration. In the absence of thermal or chemical sintering, the colloids of the crystal are readily dissolved by polar solvents with mild agitation.54 Here, we examine the efficacy of assembling and retaining lipid bilayers on dissolved colloids by resuspension of a packed colloidal bed in which dried lipid films have been deposited by solvent-casting. Figure 1 schematically illustrates the process. Previously, colloidal crystals,53,55,56 as well as a number of other nano- and mesoporous thin metal oxide films,57
60 have been shown to support vesicle fusion to form superficially supported bilayers with the most cited goal being the generation of a supported membrane addressable from both sides of the substrate. To form bilayers on the constituent colloids of a colloidal crystal by fusion, vesicles smaller than the mesopores of the crystal were

ARTICLE

Figure 1. Cartoon depicting (i) the solvent-casting of lipids within packed beds of colloids, (ii) evaporation of solvent to form dried lipid films, and (iii) reconstitution of the films and resuspension of colloids to form colloid-supported bilayers.

required. This places a lower practical limit on the diameter of colloid that could be derivitized by vesicle fusion within a packed bed. Furthermore, the solvent casting method described within does not depend on a complex interaction between hydrated vesicles and the surface.22
24 Consequently, colloidal surfaces that do not support vesicle fusion can be derivatized with lipid bilayers by this method. The process is compatible with lipids in both gel and liquid phases, and is highly lipid-efficient; in many cases, all lipid dried within the packed bed is retained on colloids after resuspension. These factors suggest that the method may be of use in applications utilizing colloids with diverse surface chemistry or with lipids or amphiphiles that are expensive or thermally labile, or simply do not form continuous bilayers by vesicle fusion. Our interest in this process originates from efforts to assemble a support for lipid bilayers similar to that provided by colloidal crystals within microchannels. Materials presenting a high density of supported lipid bilayers connected with the superlative pore network of a crystal structure have potential applications in flowthrough assays or chromatographic analysis of membrane partition or adsorption events. In conclusion, we demonstrate a simple method for assembling lipid-coated colloids in fused silica capillaries as a basis for such studies.

’ METHODS AND MATERIALS Materials. With the exception of PSIDA (a pyrene derivitized metal-chelating iminodiacetate lipid)61 which was a kind gift from Dr. Darryl Sasaki (Sandia Biomolecular Materials and Interfaces Department, Sandia National Laboratories, Albuquerque, New Mexico), all lipids were acquired from Avanti Polar Lipids (Alabaster, AL) in powder form including 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1, 2-distearoyl-sn-glycero-3-phosphocholine (DSPC), 1,2-dimyristoylsn-glycero-3-phospho-L-serine (sodium salt) (DMPS), L-R-phosphatidylcholine from chicken egg (eggPC), and 1,2-dipalmitoylsn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (ammonium salt) (Rh-DPPE). 10,12-Tricosadiynoic acid was obtained from Sigma-Aldrich (St. Louis, MO). Silica colloids from 0.1 to 10 μm in diameter (particle size standard deviation 99.8%, ACS reagent grade), chloroform (>99%), and acetonitrile (anhydrous, 99.8%) were from Sigma-Aldrich. Octadecyltrichlorosilane (OTS, 90þ%) and silicon tetrachloride (99%) were purchased from Sigma-Aldrich and 2-[methoxy(polyethylenoxy)npropyl]trichlorosilane (PEG-TS, n = 6
9; 90%) was purchased from Gelest, Inc. (Morrisville, PA). The proteins bovine serum albumin (BSA, Cohn fraction V, 96%) and melittin from honeybee venom (91.8%) were from Sigma-Aldrich, as were the compounds fluorescein 5(6)-isothiocyanate (FITC, mixed isomers), ethylenediaminetetraacetic acid (EDTA), 8635

dx.doi.org/10.1021/la200952c |Langmuir 2011, 27, 8634–8644

Langmuir

ARTICLE

Figure 2. Fluorescence microscopy images of a dried lipid pattern observed on (a) a nonporous glass slide, (b) a colloidal crystal composed of 250 nm silica colloids, and (c) cross section of the crystal to demonstrate uniformity of dried lipid films through the crystal depth. copper(II) chloride (97%), potassium iodide (99%), and buffer compounds tris(hydroxymethyl)aminomethane hydrochloride (TRIS-HCl) and tris(hydroxymethyl)aminomethane (TRIZMA-base). Water was deionized and supplied from a building system monitored at 17.6 MΩ. Fused silica capillaries (360 μm o.d., 100 or 75 μm i.d.) were from Polymicro Technologies (Pheonix, AZ). Supelco glass vials (Bellefonte, PA) with 2 mL or 300 μL (interlock vials with fused glass inserts) volumes and screw tops were used with PTFE lined caps and purchased from Sigma-Aldrich. Fisher brand 4 mL clear vials with PTFE septum autosampler caps used for loading colloids into capillaries were from Fisher Scientific. Lipid solutions were transferred using Hamilton Company glass syringes with Teflon-lined plungers and removable needles purchased from Sigma-Aldrich. Compressed nitrogen was from AL Compressed Gases (Spokane, WA). Methods. The key to the approach described in this work is the uniform distribution of dried lipid films throughout a packed bed of colloids that results from the wicking properties and rapid evaporation of solvent from the void volume (less than 2 μL) of micrometer and submicrometer colloidal beds. This phenomenon is widely utilized in spotting of porous substrates for microarrays and thin layer chromatography. Microarrays using lipid-spotted polyvinyldifluoride substrates were reported by Kanter and co-workers62 and are commercially available.63 To our knowledge, solute spotting on colloidal crystals has not been reported. The fluorescence images in Figure 2 graphically demonstrate the lateral heterogeneity differences in solute distribution arising from the evaporation of a 2 μL drop of chloroform containing DPPC (with 1% Rh-DPPE) on either a nonporous glass slide or a 20-μmthick colloidal crystal composed of 0.25 μm silica colloids. The meniscus evaporation assembly method used to form the colloidal crystals originally described by Colvin et al.64 creates well-packed colloidal beds that can vary considerably in thickness along the length of the substrate as the colloidal slurry concentrates during crystal formation. There are no visible differences in lipid distribution in thin (1 μm) or thick (100 μm) regions of the colloid film. If the total surface area of the colloids comprising the bed is known, a mass of lipid corresponding to a given number of bilayers can be deposited throughout the bed of colloids. In a packed bed, the volume fraction of spheres may range from 0.74 for highly crystalline arrangements to a minimum of 0.56 for random loose packing.65 Although the exact volume fraction attained in the work here has not been determined, it is most likely an intermediate value between the two extremes. Even without specific attempts to crystallize the colloids, the bed formed from evaporation of an ethanolic slurry of monodisperse colloids likely has better than random packing due to the self-established order in concentrated slurries.66 A 20 μL drop of 10 wt %. silica/ethanol dried directly on a glass coverslip produces a colloidal film that exhibits discernible diffraction of visible light arising from

crystalline order, though not as strongly as well-crystallized films. The beds of colloids used in this work are formed by evaporation of this slurry in small vials and have a colored hue when the colloids are