Arraying of Intact Liposomes into Chemically Functionalized Microwells

Charles Maldarelli,†,§ Alexander Couzis,† and Lane Gilchrist*,† ... of the wells, allows the liposomes to remain intact and to not unravel, rup...
0 downloads 0 Views 1MB Size
Download PDF
Langmuir 2006, 22, 5403-5411

5403

Arraying of Intact Liposomes into Chemically Functionalized Microwells Nikhil D. Kalyankar,† Manoj K. Sharma,†,| Shyam V. Vaidya,† David Calhoun,‡ Charles Maldarelli,†,§ Alexander Couzis,† and Lane Gilchrist*,† Department of Chemical Engineering, Department of Chemistry, and the LeVich Institute, and The City College of The City UniVersity of New York, New York ReceiVed January 27, 2006. In Final Form: March 16, 2006

Here, we describe a protocol to bind individual, intact phospholipid bilayer liposomes, which are on the order of 1 µm in diameter, in microwells etched in a regular array on a silicon oxide substrate. The diameter of the wells is on the order of the liposome diameter, so only one liposome is located in each well. The background of the silicon oxide surface is functionalized with a PEG oligomer using the contact printing of a PEG silane to present a surface that resists the adsorption of proteins, lipid material, and liposomes. The interiors of the wells are functionalized with an aminosilane to facilitate the conjugation of biotin, which is then bound to Neutravidin. The avidin-coated well interiors bind the liposomes whose surfaces contain biotinylated lipids. The specific binding of the liposomes to the surface using the biotin-avidin linkage, together with the resistant nature of the background and the physical confinement of the wells, allows the liposomes to remain intact and to not unravel, rupture, and fuse onto the surface. We demonstrate this intact arraying using confocal laser scanning microscopy of fluorophores specifically tagging the microwells, the lipid bilayer, and the aqueous interior of the liposome.

Introduction This article outlines a strategy for arraying intact, micrometersized unilamellar phospholipid bilayer liposomes on a surface by situating and immobilizing the liposomes individually in microwells that have been etched onto the surface in a regular pattern. The potential application of this liposome array is its use as a platform for sequestering and displaying, in the liposomes, integral and peripheral membrane proteins to form a microarray for multiplexed screening applications. The highly selective binding capabilities of surface membrane proteins make them ideal probe molecules for a protein array, yet their strict requirement for a lipid environment to preserve their conformation, mobility, and binding functionality has made them difficult to display. Conventional protein arraying techniques have been used to array aqueous soluble proteins (e.g., spot dispensing and drying,1-3 microcontact printing,4 and microfluidic patterning5). Reconstituting membrane proteins in a lipid environment of intact liposomes and arraying these liposomes on a surface can prove to be a flexible platform for the development of high-density membrane protein arrays. A further advantage is that the intact liposome format can be geared toward probing membrane proteinmediated transport or pore formation in vitro. The main problem with situating intact liposomes onto a surface is their general tendency to unravel and subsequently rupture * Corresponding author. E-mail: [email protected]. † Department of Chemical Engineering. ‡ Department of Chemistry. § Levich Institute. | Present address: Schering-Plough Research Institute, Kenilworth, NJ 07033. (1) Walter, G.; Bussow, K.; Cahill, D.; Lueking, A.; Lehrach, H. Curr. Opin. Microbiol. 2000, 3, 298-302. (2) Kodadek, T. Chem. Biol. 2001, 8, 105-115. (3) Lal, S.; Christopherson, R.; dos Remedios, C. Drug DiscoVery Today 2002, 7, S143-S149. (4) Blawas, A. S.; Reichert, W. M. Biomaterials 1998, 19, 595-609. (5) Bernard, A.; Renault, J. P.; Michel, B.; Bosshard, H. R.; Delamarche, E. AdV. Mater. 2000, 12, 1067-1070.

and fuse with other adsorbing liposomes. On a hydrophilic surface such as glass or silicon oxide, adsorbing liposomes fuse to form a continuous supported planar phospholipid bilayer (SPB).6-18 On a hydrophobic surface such as a methyl-terminated selfassembled monolayer (SAM) or a Langmuir-Blodgett deposited monolayer, liposomes adsorb and add a lipid monolayer to the hydrophobic surface to form a planar hybrid bilayer membrane (HBM).19,20 Intact liposomes have been successfully attached to a surface using a strategy in which the surface is chemically functionalized to present two chemistriessone to prevent adsorption and unraveling of liposomes and one to anchor the liposomes to the surface. In this strategy, the surface is modified uniformly with a passivating chemistry that is designed to block the nonspecific adsorption of lipids, liposomes, and proteins (e.g., passivation with SPBs and HBMs,21,22 bovine serum albumin (BSA),23 poly(ethylene glycol) (PEG) coatings including grafted (6) Lasic, D. D.; Papahadjopoulos, D. Science 1995, 267, 1275-1276. (7) Iwasaki, Y.; Tanaka, S.; Hara, M.; Ishihara, K.; Nakabayashi, N. J. Colloid Interface Sci. 1997, 192, 432-439. (8) Keller, C. A.; Kasemo, B. Biophys J. 1998, 75, 1397-1402. (9) Egawa, H.; Furusawa, K. Langmuir 1999, 15, 1660-1666. (10) Keller, C. A.; Glasmastar, K.; Zhdanov, V. P.; Kasemo, B. Phys. ReV. Lett. 2000, 84, 5443-6. (11) Thomson, N. H.; Collin, I.; C.Davies, M.; Palin, K.; Parkins, D.; Roberts, C. J.; J. B.Tendler, S.; Williams, P. M. Langmuir 2000, 16, 4813-4818. (12) Kumar, S.; Hoh, J. H. Langmuir 2000, 16, 9936-9940. (13) Jass, J.; Tjarnhage, T.; Puu, G. Biophys. J. 2000, 79, 3153-3163. (14) Ahmed, K.; Gribbon, P.; Jones, M. N. J. Liposome Res. 2002, 12, 285300. (15) Reimhult, E.; Hook, F.; Kasemo, B. Langmuir 2003, 19, 1681-1691. (16) Tokumasu, F.; Jin, A. J.; Feigenson, G. W.; Dvorak, J. A. Ultramicroscopy 2003, 97, 217-227. (17) Schonherr, H.; Johnson, J. M.; Lenz, P.; Frank, C. W.; Boxer, S. G. Langmuir 2004, 20, 11600-11606. (18) Liang, X.; Mao, G.; Ng, K. Y. S. Colloid Interface Sci. 2004, 278, 53-62. (19) Kalb, E.; Frey, S.; Tamm, L. Biochim. Biophys. Acta 1992, 1103, 307316. (20) Silin, V.; Wieder, H.; Woodward, J.; Valincius, G.; Offenhausser, A.; Plant, A. J. Am. Chem. Soc. 2002, 124, 14676-14683. (21) Chapman, D. Langmuir 1993, 9, 39-45. (22) Glasmastar, K.; Larsson, C.; Hook, F.; Kasemo, B. J. Colloid Interface Sci. 2002, 246, 40-47. (23) Martinez, K.; Meyer, B.; Hoviius, R.; Lundstrom, K.; Vogel, H. Langmuir 2003, 19, 10925-10929.

10.1021/la0602719 CCC: $33.50 © 2006 American Chemical Society Published on Web 05/11/2006

5404 Langmuir, Vol. 22, No. 12, 2006

PEG polymers24,25 or poly (L-lysine) - poly(ethylene glycol) copolymers (PLL-g-PEG)26-28 and PEG oligomers attached to a surface using SAMs.29-32) The matrix of the passivating layer is functionalized with chemical groups designed to couple to binding partners localized on the liposome exterior. Liposomes are brought to the surface and attach only at the anchoring points of the second functionalization and are prevented from rupturing and fusing by the surrounding passivating layer. To bind the liposome to the surface, early studies used antibody coupling33 or disulfide linkages,34 but recent techniques have used one of two methods. In the first,35-38 ssDNA is attached to the surface of the liposomes (usually by covalent binding to the lipid headgroups of the bilayer) with the strands extending out from the liposome face. A surface is prepared that consists of an SPB, and DNA strands complimentary to those extending from the liposome are attached to the bilayer by covalent linkages to the headgroups or to molecules (e.g., cholesterol) that embed themselves inside the bilayer. The liposomes are then attached to the surface by hybridization to form a double-stranded tether; because the surface bilayer is mobile, the tethered liposomes can laterally translate over the surface. In the second method,39,40 the receptor-ligand pair biotin-streptavidin is used to link the liposome to the surface. Streptavidin-biotin coupling proceeds by conjugating biotin to the headgroup of a phospholipid and incorporating this lipid in a few percent mole fraction into the liposome. A passivated surface is prepared and biotinylated, linked with streptavidin, and then conjugated to the liposomes with biotin on its surface. (See refs 39 and 40 using HBMs and SPBs, respectively, as passivating layers, refs 41 and 42 using a poly(ethylenimine) background, and refs 43 and 44 using avidin or BSA as the background.) By examining the surface using principally a quartz crystal microbalance, scanning force microscopy, and fluorescence techniques, these studies found the attached liposomes to be intact. Platforms for the large-scale microarraying of different liposomes were fabricated using photolithography to etch square 50 µm × 50 µm wells in glass, with the wells arranged in a (24) Merrill, E. W. In Poly(ethylene glycol): Biotechnical and Biomedical Applications; Harris, J., Ed.; Plenum Press: New York, 1992; pp 199-220. (25) Du, H.; Chandaroy, P.; Hui, S. Biochim. Biophys. Acta 1997, 1326, 236248. (26) Huang, N. P.; Voros, J.; DePaul, S.; Textor, M.; Spencer, N. Langmuir 2002, 18, 220-230. (27) Huang, N. P.; Michel, R.; Voros, J.; Textor, M.; Hofer, R.; Rossi, A.; Elbert, D.; Hubbell, J.; Spencer, N. Langmuir 2001, 17, 489-498. (28) Lussi, J. W.; Michel, R.; Reviakine, I.; Falconnet, D.; Goessl, A.; Csucs, G.; Hubbell, J. A.; Textor, M. Prog. Surf. Sci. 2004, 76, 55-69. (29) Prime, K. L.; Whitesides, G. M. Science 1991, 252, 1164-1167. (30) Prime, K.; Whitesides, G. J. Am. Chem. Soc. 1993, 115, 10714. (31) Harder, P.; Grunze, M.; Dahint, R.; Whitesides, G. M.; Laibinis, P. E. J. Phys. Chem. B 1998, 102, 426-436. (32) Chapman, R. G.; Ostuni, E.; Takayama, S.; Holmlin, R. E.; Yan, L.; Whitesides, G. M. J. Am. Chem. Soc. 2000, 122, 8303-8304. (33) Shibata-Seki, T.; Masai, J.; Tagawa, T.; Sorin, T.; Kondo, S. Thin Solid Films 1996, 273, 297-303. (34) Stanish, I.; Santos, J.; Singh, A. J. Am. Chem. Soc. 2001, 123, 10081009. (35) Patolsky, F.; Lichtenstein, A.; Willner, I. Angew. Chem., Int. Ed. 2000, 39, 940-941. (36) Yoshina-Ishii, C.; Boxer, S. J. Am. Chem. Soc. 2003, 125, 3696-3697. (37) Yoshina-Ishii, C.; Miller, G.; Kraft, M.; Tool, E.; Boxer, S. J. Am. Chem. Soc. 2005, 127, 1356-1357. (38) Benkoski, J. J.; Hook, F. J. Phys. Chem. B 2005, 109, 9773-9779. (39) Jung, L. S.; Shumaker-Parry, J. S.; Campbell, C.; Lee, S.; Geib, M. J. Am. Chem. Soc. 2000, 122, 4177-4184. (40) Boukobza, E.; Sonnenfeld, A.; Haran, G. J. Phys. Chem. B 2001, 105, 12165-12170. (41) Vermette, P.; L, M.; Gagnon, E.; Griesser, H.; Doillon, C. J. Controlled Release 2002, 80, 179-185. (42) Vermette, P.; Griesser, H.; Kambouris, P.; Meagher, L. Biomacromolecules 2004, 5, 1496-1502. (43) Pignataro, B.; Steinem, C.; Galla, H.-J.; Fuchs, H.; Janshoff, A. Biophys. J. 2000, 78, 487-498. (44) Okumus, B. Biophys. J. 2004, 87, 2798.

Kalyankar et al.

checkerboard pattern.45 Direct pipetting of picoliter solutions of liposomes into the wells formed SPBs on the well floors; this platform can then be used to pipet and tether proteoliposomes in the wells to form SPBs with embedded membrane protein or tether liposomes to previously assembled SPBs using DNA tethering.46 Similarly, Boxer et al.36,37,47-52 used microcontact printing to assemble a square grid of fibronectin (100 µm × 100 µm) on a glass coverslip. In a microfluidic cell, adjoining streams of two liposome solutions, with each solution containing liposomes with the same ssDNA tether (although the two liposome solutions had different tethers), were passed in a parallel flow over the coverslip to fill in the bare square grid with SPBs. The two-stream microfluidic flow patterned the squares of the grid with SPBs, with squares in a row completely underneath one or the other stream forming an SPB with only that stream’s tether while some squares underneath the interface where the stream join forming SPBs with a mixture of tethers. The fibronectin grids resist adsorption and function as barriers between the square domains. The patterned surface was then exposed to a solution of liposomes. Each liposome had on its surface strands of DNA complementary to one of the two types of strands extending from the SPB surface squares. The liposomes hybridize to the complementary strands extending from the surface, forming two rows of squares with a single type of liposome and DNA strand and separated by a row of squares with a mixture of the two kinds of liposomes. Similarly, another study,53 using photolithography, fabricated a matrix of square regions (200 µm × 200 µm) consisting of a biotinylated PLL-g-PEG polymer grafted to the surface. DNA tethers were attached to these regions by streptavidin-biotin binding, and the squares were surrounded by a passivating layer of PLL-g-PEG. Liposomes with complementary strands of DNA were linked to the tethers in the square regions to create a liposomal array displaying one kind of liposome. Individual liposomes have been arrayed using two methods. In the first approach, a square array of micrometer-sized dots of BSA conjugated to biotin was microcontact printed onto a glass surface, and the remainder of the surface was then backfilled with BSA.54 Surface biotinylated liposomes on the order of 100 nm in diameter were coupled from liposomal solutions to the dots by streptavidin with approximately one liposome to an island under very dilute conditions. In a different approach,55 photolithography was used to pattern a silicon oxide surface with a square array of TiO2 nanopillars approximately 100 nm in diameter. The surface of the pillars was biotinylated while the surroundings were passivated with PLL-g-PEG; biotinylated liposomes were then bound onto the pillars with streptavidin to array the liposomes individually. In this article, we provide a new format (Figure 1) for arraying individual liposomes by situating micrometer-sized liposomes individually in wells (with diameters of the same order as the (45) Cremer, P. S.; Yang, T. J. Am. Chem. Soc. 1999, 121, 8130-8131. (46) Groves, J. T.; Mahal, L. K.; Bertozzi, C. R. Langmuir 2001, 17, 51295133. (47) Hovis, J. S.; Boxer, S. G. Langmuir 2000, 16, 894-897. (48) Yoshinobu, T.; Ecken, H.; Ismail, A. B. M.; Iwasaki, H.; Luth, H.; Schoning, M. J. Electrochim. Acta 2001, 47, 259-263. (49) Yoshinobu, T.; Suzuki, J.; Kurooka, H.; Moon, W. C.; Iwasaki, H. Electrochim. Acta 2003, 48, 3131-3135. (50) Groves, J.; Boxer, S. Acc. Chem. Res. 2002, 35, 149-157. (51) Kam, L.; Boxer, S. J. Am. Chem. Soc. 2000, 2000, 12901-12902. (52) Grove, J. T.; Boxer, S. Acc. Chem. Res. 2002, 35, 149-157. (53) Stadler, B.; Falconnet, D.; Pfeiffer, I.; Hook, F.; Voros, J. Langmuir 2004, 20, 11348-11354. (54) Stamou, D.; Duschl, C.; Delamarche, E.; Vogel, H. Angew. Chem., Int. Ed. 2003, 42, 5580-5583. (55) Michel, B.; Bernard, A.; Bietsch, A.; Delamarche, E.; Geissler, M.; Juncker, D.; Kind, H.; Renault, J. P.; Rothuizen, H.; Schmid, H.; Schmidt-Winkel, P.; Stutz, R.; Wolf, H. Chimia 2002, 56, 527-542.

Chemically Functionalized Microwell Surfaces

Langmuir, Vol. 22, No. 12, 2006 5405

Experimental Section

Figure 1. Schematic of the intact liposome immobilization on patterned surfaces. (I) Passivation of background using PEG-silane contact printing. (II) Selective attachment of biotin inside microwells. (III) Selective attachment of Neutravidin into biotin microwells. (IV) Intact arraying of liposomes inside chemically functionalized microwells.

diameter of the liposomes). The wells are inscribed on a silicon oxide surface in a square pattern by using photolithography techniques. The area surrounding the wells is passivated using a self-assembled monolayer (SAM) of a PEG oligomer. The interior of the wells is functionalized with an amine SAM that allows conjugation to biotin. Biotinylated liposomes are then linked to the well interiors using avidin. We use confocal laser scanning microscopy to verify a one-to-one placement of the liposomes in the wells and to establish that they remain intact. Recessing the liposomes in wells rather than openly displaying them on tethers has the advantage that it further isolates the liposomes from each other, preventing fusion.

Materials. One-side-polished single-crystal silicon wafers were purchased from Montco Silicon Technologies Ltd. 3-Aminopropyltrimethoxysilane (APS), (heptadecafluoro-1,1,2,2-tetrahydrodecyl)trichlorosilane (fluorosilane), and 2-[methoxy(polyethyleneoxy)propyl]trimethoxysilane (PEG-silane, 90%, n ) 6-9) were purchased from Gelest Inc. HPLC-grade chloroform, sodium hydroxide (ACS reagent grade), sodium phosphate (anhydrous, 99%), EDTA (disodium salt, dihydrate, 99%), and sephadex G-75 were obtained from Sigma Aldrich Co. Nochromix was purchased from Godax Laboratories. Sulfuric acid (98%) and sodium chloride were purchased from Fischer Scientific. NHS-PEO4-biotin and fluoresceinconjugated Neutravidin (Neutravidin-FITC) were obtained from Pierce Biotechnology Inc. L-R-Phosphatidylcholine (POPC, egg/ chicken), cholesterol (98%), 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[biotinyl(polyethylene glycol)2000](ammonium salt) (Biotin-PEG-DSPE) were obtained from Avanti Polar Lipids Inc. Dextran-Texas red and 2-(4,4-difluoro-5,7-diphenyl-4-bora-3a,4adiaza-s-indacene-3-pentanoyl)-1-hexadecanoyl-sn-glycero-3-phosphocholine (β-BODIPY 530/550 C5-HPC) (D-3815 lipid probe) were purchased from Molecular Probes. A Sylgard 184 silicone elastomer kit was obtained from Dow Corning. HEPES (sodium salt) was obtained from Acros Organics. Negatively charged 1 µm carboxylfunctionalized Fluoresbrite yellow-orange (excitation maximum, 529 nm; emission maximum, 546 nm) particles were obtained from Polysciences, Inc. Deionized (DI) water with a resistivity greater than 18 MΩ‚cm was obtained from a Millipore system. All chemicals were used as obtained without further purification. Methods. Micropatterned surfaces were fabricated in the Cornell NanoScale Science and Technology Facility (CNF), Ithaca, New York, using standard photolithography techniques56 on silicon wafers with 1-µm-thick silicon dioxide layers that were thermally grown. Briefly, a layer of positive photoresist was spin coated onto the silicon wafers. Using a chrome-on-quartz mask with the required features, a 5X stepper was used to transfer the features onto the photoresist. Following the development of the photoresist, reactive ion etching was used to etch the features onto the silicon wafer. Finally, the resist was removed, resulting in surfaces that have a 5 mm × 5 mm active area containing 1.2-µm-diameter wells with a depth of 0.55 µm and with a center-to-center pitch of 3 µm. These surfaces were cleaned using Nochromix and sulfuric acid in a Branson 5200 sonicator for 30 min, followed by sonication (40 kHz) with DI water for 30 min and drying under a stream of nitrogen. The surfaces were then imaged under a Nanoscope III (Veeco Instruments, Inc.) atomic force microscope (AFM) in contact mode in air using silicon nitride tips with a force constant of 0.58 N/m (Figure 2). Polished silicon wafers were cut into approximately 7 mm × 7 mm squares and cleaned by sonicating (Branson 5200 at 40 kHz) in a mixture of Nochromix and 98% sulfuric acid for about 30 min, followed by sonication in DI water for 30 min. They were then dried in a stream of nitrogen, exposed to 15 mL of 0.5 mM fluorosilane solution in chloroform for 1 h, rinsed with pure chloroform, and dried. Highly viscous poly(dimethylsiloxane) (PDMS, Sylgard 184, DuPont) was mixed with the curing agent, poured onto these flat surfaces, and cured overnight at 60 °C. Flat blocks of PDMS were then peeled off of the surfaces after being cooled to room temperature. These blocks were cleaned using chloroform, dried in a stream of nitrogen, and impregnated with 15 mL of a 1 mM PEG-silane/ chloroform solution for 2 h to make stamp pads. These stamp pads were dried under a stream of nitrogen and were used for printing via conformal contact on the micropatterned surfaces for 3 min. Thus, PEG-silane gets transferred onto the plane of the micropatterned surfaces, leaving the wells uncoated. The micropatterned surfaces were then immediately dipped into 15 mL of a 5 mM APS/chloroform solution for 30 min, rinsed with chloroform, and dried. The micropatterned surfaces with amine wells were then dipped into 15 mL of a 1 mM sodium hydroxide solution to deprotonate (56) Rai-Choudhury, P. Handbook of Microlithography, Micromachining, and Microfabrication; SPIE Optical Engineering Press: Institution of Electrical Engineers: Bellingham, WA, 1997.

5406 Langmuir, Vol. 22, No. 12, 2006

Figure 2. Contact mode AFM images of a micropatterned surface. Three-dimensional surface plot of a micropatterned surface (average depth of wells ∼550 nm). the amine terminal functional groups. They were then exposed to 2 mL of a 1 mg/mL solution of NHS-PEO4-biotin in a filtered (0.2 µm pore size filter) pH 10 phosphate buffer solution for 1 h. They were then thoroughly washed with DI water, dried, and dipped in 2 mL of a Neutravidin-FITC solution (filtered using a 0.2 µm filter) of 40 µg/mL concentration in a pH 10 phosphate buffer for 1 h. The micropatterned surfaces were then cleaned using buffer to remove unattached Neutravidin and stored in a pH 10 phosphate buffer solution. Lipid films were prepared with two different formulations: (I) POPC/cholesterol/biotin-PEG-DSPE in 0.63:0.31:0.06 molar proportions. (II)POPC/cholesterol/biotin-PEG-DSPE/D-3815 in 0.62:0.30: 0.06:0.02 molar proportions. These constituents were mixed using chloroform and dried overnight under vacuum, resulting in approximately 2 mg of lipid films in glass vials. These vials were sealed under nitrogen using Parafilm (Fisher Scientific) and used as needed. HEPES buffer (1 mL; 20 mM HEPES, 100 mM sodium chloride, 1 mM EDTA) of pH 7.5 containing 1 mg of dextran-Texas red dye was used to hydrate these films to form multilamellar vesicles. After three freezethaw cycles each lasting 15 min, this solution was extruded using a 1 µm pore size membrane in a miniextruder by Avanti Polar Lipids to form unilamellar vesicles (liposomes) of the same size range as the pore size of the membrane. Particle size analysis of the extruded liposomes using a Nano-series Zetasizer (Malvern Instruments) confirmed the presence of liposomes of approximately 0.85 µm average diameter. To exclude the dye (dextran-Texas red) outside the liposomes, the solution was passed through a column packed with Sephadex G75 gel using a BioLogic HR workstation and a BioFrac Fraction collector setup (BIO-RAD). After separation, the purified fraction was checked for the correct particle size using the Zetasizer and the absence of any background dextran-Texas red fluorophore using epifluorescense microscopy (Nikon Eclipse TE 200 inverted microscope). The surfaces with the Neutravidin-FITC grid were glued to previously cleaned glass microscope slides (obtained from Fischer scientific) using Permabond industrial-grade elastomer bonding adhesive. Press-to-seal silicone isolators were used to prepare the sample cell around these surfaces, and 500 µL of liposome solution was exposed to them for 3 h. Multiple rinsing steps using prefiltered pH 7.5 HEPES buffer were used to wash off liposomes that were not attached to the surface and excess dye. Coverslips (clean and with a monolayer of PEG-silane) were used to cover the surfaces under pH 7.5 HEPES buffer for confocal laser scanning microscopy.

Kalyankar et al. Confocal laser scanning microscopy (CLSM) was used to image the fluorescently labeled microwells and liposomes using the Leica TCS SP2 AOBS confocal microscope. This CLSM is equipped with an acousto-optical beam splitter and a prism spectrophotometer detector,61 which negates the need for dichroic mirrors and bandpass filters; furthermore, this microscope is based on a single pinhole design that ensures accurate co-localization over the entire visible range. All images were acquired using an 8-bit pixel depth setting. The microwell grid labeled with FITC-Neutravidin in the well interiors was imaged using the 488 nm laser line with a 510-535 nm detection window. To image the immobilized liposomes prepared using formulation I, the laser lines at 488 and 594 nm were used to excite the FITC tag on the Neutravidin-functionalized microwell grid and the dextran-Texas red cargo inside the liposomes, respectively. Imaging was done at several locations on the surface, and the detection windows were 500-560 and 610-670 nm for FITC and Texas red, respectively. To image the immobilized liposomes prepared using formulation II, the laser lines at 488, 514, and 594 nm were used to excite the FITC tags on the Neutravidinfunctionalized microwell grid, the BODIPY tag incorporated into the lipid bilayer of the liposomes, and the dextran-Texas red cargo inside the liposomes. Imaging was done at several locations on the surface, and data was collected for the three probes using sequential scanning mode. In this mode, emitted photons comprising the fluorescence intensity in the images are collected in a sequential fashion in which only one excitation laser and detection channel are turned on at any one time. In this way, we minimize bleed-through, often termed detection channel-to-channel crossover or crosstalk, due to the overlap of emission spectra. The fluorescence detection windows for FITC, BODIPY, and Texas red were 510-535, 550580, and 610-690 nm, respectively.

Results An AFM image of the microwell-patterned surface before chemical functionalization is shown in Figure 2 as a 3-D surface plot. This imaging is done in AFM contact mode. The average diameter of the wells is 1.2 µm with a pitch (center to center) of 3 µm. The average depth of the wells is 550 nm. Figure 3 shows confocal laser scanning microscopy images of Neutravidin-FITC fluorescence collected from the chemically modified microwell-patterned surfaces for two different detection regions (z sections), as described in the Experimental Section, before liposome attachment. This analysis was performed using a 488 nm laser to excite the FITC fluorophore. The detection window for fluorescence was 510-535 nm. Figure 3a is the fluorescence image collected at a detection region close to the top of the micropatterned surface. Figure 3b is an enlarged image of Figure 3a, which indicates the presence of negligible background fluorescence. Hence there is no indication of Neutravidin present in the background of the wells, in agreement with the fact that PEG-terminated monolayers are resistant to nonspecific protein adsorption.32 Furthermore, the aminoterminated silane during the backfilling step reacted only with the bare silica in the wells of the microwell-patterned surface. Figure 3c, which is the fluorescence image of the micropatterned surface focused on the bottom of the wells, shows that Neutravidin-FITC was present both along the walls of the wells and also at the bottom of the wells. Different z sections were taken along the depth of the wells starting from the top of the surface and our results show that the intensity distribution in the (57) Kang, M. C.; Yu, S. F.; Li, N. C.; Martin, C. R. Langmuir 2005, 21, 8429-8438. (58) Yap, F. L.; Zhang, Y. Langmuir 2005, 21, 5233-5236. (59) Michael, K.; Taylor, L.; Schultz, S.; Walt, D. Anal. Chem. 1998, 70, 1242-1248. (60) Epstein, J.; Leung, A.; Lee, K. B.; Walt, D. Biosens. Bioelectron. 2003, 18, 541-546. (61) http://www.leica-microsystems.com/Confocal_Microscopes.

Chemically Functionalized Microwell Surfaces

Figure 3. Neutravidin-FITC fluorescence images of the micropatterned surfaces using confocal laser scanning microscopy. (a) Image focused on the upper slice (top). (b) Enlarged image focused on the upper slice (top). (c) Enlarged image focused on the lower slice (bottom).

wells becomes more uniform as the bottom of the well is approached (data not shown). Solutions containing fluorescently labeled liposomes were prepared using the two formulations as described in the Materials and Methods section. We characterized the particle size distribution of these solutions using light scattering (Malvern Zetasizer), which confirmed the presence of liposomes with 850 nm average diameter. In addition, a visual inspection of the solution using epifluorescence showed that the highly soluble dextran-Texas red cargo was localized within intensely bright ∼1 µm domains that randomly moved in the field of view, indicative of intact liposomes. Figure 4 shows fluorescence data taken after exposing loaded liposomes (with dextran-Texas red cargo) prepared using formulation I to Neutravidin-FITC inside the wells on the microwell patterned surfaces. The two fluorophores were chosen on the basis of their well-separated emission spectra in order to minimize crosstalk between detection channels. Appropriate spectral detection regions were chosen while imaging FITC and Texas red fluorescence simultaneously using confocal microscopy. Figure 4a shows the fluorescence (excitation, 488 nm; emission, 500-560 nm) from the green FITC tag on Neutravidin molecules present inside the microwells on the micropatterned surfaces. Figure 4b shows the fluorescence (excitation, 594 nm; emission, 610-710 nm) from dextran-Texas red cargo encapsulated inside liposomes. Figure 4c, which is an overlay image of a and b, clearly shows the co-localization of the fluorescence from the two dyes. This suggests that liposomes are residing inside the chemically functionalized microwells as conceptualized in Figure 1 and are intact. The occupancy level of the microwells is remarkable and was obtained in essentially quantitative yield, with fewer than 5 out of over 1500 sites remaining unoccupied.

Langmuir, Vol. 22, No. 12, 2006 5407

Figure 4. Confocal laser scanning microscopy images of the micropatterned surfaces after exposure to liposomes prepared using formulation I. (a) Neutravidin-FITC grid. (b) Dextran-Texas red inside liposomes situated on the grid. (c) Overlay image of a and b. (d) Line intensity profile incorporating one of the empty wells.

On the basis of both the 0.85 µm average diameter of the liposomes measured using a Malvern Zetasizer, which is on the order of the microwell dimensions (1.2 µm), and the uniform intensity of the dextran-Texas red fluorescence at each site over a moderate sample size, we conclude that there is only one liposome in each microwell. The circled wells in Figure 4b show the absence of red fluorescence at places where corresponding FITC fluorescence is present. This is an indication of an empty well. To check this hypothesis, we have plotted the intensity line profiles of the green and red fluorescence incorporating one of the empty wells in the analysis as shown in Figure 4d. It shows the marked decrease in dextran-Texas red fluorescence intensity corresponding to an individual microwell. This result appears to be attributed to the case of a well occupied by a liposome that subsequently releases its contents or a well that was never occupied by a liposome. The above evidence does not allow us to visualize the fluorescence of the bilayer of the liposome directly, only the cargo inside the liposome. As such, the cargo could be localized onto the grid by adsorption without the need for the lipid bilayer. To directly confirm the presence of the liposome coexisting with the cargo, we utilized formulation II, which places a dye in the lipid bilayer. Figures 5 and 6 show fluorescence data taken after exposing loaded liposomes (with dextran-Texas red) prepared using formulation II, which has BODIPY-fluorophore-tagged lipids (Molecular Probes D-3815). This dye was selected on the basis of an emission profile and maximum that lies between that of FITC and Texas red. Appropriate spectral detection regions were chosen while imaging FITC, BODIPY, and Texas red fluorescence in sequential scanning mode in which each dye is sequentially excited to suppress signal cross talk or bleed-through. Figure 5 shows the fluorescence data obtained from a detection region close to the bottom of the microwells as depicted in the

5408 Langmuir, Vol. 22, No. 12, 2006

Figure 5. Confocal laser scanning microscopy images of the micropatterned surfaces after exposure to liposomes prepared using formulation II taken in a detection region close to the bottom of the microwells. (a) Neutravidin-FITC grid. (b) BODIPY (D-3815) lipid in the lipid bilayer of the liposomes. (c) Dextran-Texas red cargo encapsulated inside liposomes situated on the grid. (d) Overlay of the three images. (e) Relative location of the detection window close to the bottom of the wells. (f) Line intensity profile taken across the section shown in d.

schematic model in Figure 5e. Figure 5a is the fluorescence from the FITC (excitation, 488 nm; detection, 510-535 nm) tag on Neutravidin molecules present inside the microwells on the micropatterned surfaces. Figure 5b shows the fluorescence from the BODIPY (excitation, 514 nm; detection, 550-580 nm) tag of D-3815 present in the lipid bilayer of the liposomes. Figure 5c shows the fluorescence from dextran-Texas red (excitation, 594 nm; detection, 610-690 nm) cargo encapsulated inside liposomes. Figure 5d, which is an overlay image of a-c, shows the microwell co-localization of fluorescence from the three dyes with negligible background fluorescence. The co-localization of the lipid and cargo dyes provides strong evidence that intact liposomes and not just the cargo are situated inside the chemically functionalized microwells. Figure 5f shows the line intensity profile taken along a section shown in Figure 5d. The green (FITC) and yellow (BODIPY) curves are relatively flat across the cross section of the wells because fluorescence data that is collected from the bottom and walls of the wells is excluded. The red (dextran-Texas red) line intensity profile curve shows a centrally symmetric distribution of the cargo present inside the

Kalyankar et al.

Figure 6. Confocal laser scanning microscopy images of the micropatterned surfaces after exposure to liposomes prepared using formulation II taken in a detection region close to the top of the microwells. (a) Neutravidin-FITC grid. (b) BODIPY (D-3815) lipid in the lipid bilayer of the liposomes. (c) Dextran-Texas red cargo encapsulated inside liposomes situated on the grid. (d) Overlay of the three images. (e) Relative location of the detection window close to the top of the wells. (f) Line intensity profile taken across the section shown in d.

liposomes, which is a co-localization finding consistent with the presence of intact liposomes within the microwells. Another set of data was collected in a detection region close to the top of the microwells offset by 800 nm in the z direction, and images from this optical section are shown in Figure 6. Detection windows identical to those for the data in Figure 5 were used to collect the fluorescence from the three dyes. Figure 6a is the fluorescence from the FITC tag on Neutravidin molecules present inside the microwells on the micropatterned surfaces. Figure 6b shows the fluorescence from the BODIPY tag present in the lipid bilayer of the liposomes. Figure 6c shows the fluorescence from dextran-Texas red cargo encapsulated inside liposomes. Figure 6d, which is an overlay image of a-c, further supports the claim that liposome attachment was selective to the inside of the microwells on the chemically functionalized micropatterned surfaces. Figure 6f shows the line intensity profile taken along a section shown in Figure 6d. The green (FITC) and yellow (BODIPY) curves have minima close to the center of the microwells because the majority of the fluorescence collected in this region is from the walls of the microwells and the lipid bilayers of liposomes present inside them, respectively. The red (Texas red) curve shows a centrally symmetric distribution of

Chemically Functionalized Microwell Surfaces

Figure 7. Three-dimensional reconstruction from z-section data acquired using the CLSM of the immobilized liposomes tagged with β-BODIPY 530/550 in the lipid layer and carrying dextran-Texas red cargo.

Figure 8. Confocal laser scanning microscopy images taken after exposing liposomes prepared using formulation II to the NeutravidinFITC layer on a flat silicon wafer. (a) Uniform Neutravidin-FITC layer. (b) BODIPY (D-3815) lipid from the broken liposomes. (c) Dextran-Texas red cargo smeared over the surface. (d) Overlay of a-c.

the cargo inside the liposomes, as expected, with two shoulders present from an optical artifact caused by multiple reflections of the emitted fluorescence within the microwell. (See Figure 9 and the explanation below.) This centrally symmetric distribution for this highly water-soluble cargo could not have been realized if the liposome had ruptured, as is shown below in Figure 8. In addition to the two sections shown in Figures 5 and 6, a series of four additional z sections were taken in order to reconstruct the 3-D image shown in Figure 7. The optical sections start near the top of the microwell (tentative midsection of the liposome) and extend down into the silicon substrate over 0.87 µm with a step size of 0.28 µm. Figure 7 was obtained from

Langmuir, Vol. 22, No. 12, 2006 5409

the view directly above the array. This data was processed to intersect the center of a set of microwells along the rightmost and bottommost edges to display the co-localization of dyes within the wells in 3-D. This Figure clearly shows the encapsulation of cargo by the BODIPY-tagged lipid and is the clearest demonstration of the immobilized liposomes being intact. Surfaces with intact liposomes were exposed to a 1 M NaCl solution for 1 h in order to lyse the liposomes. The surfaces were then washed with prefiltered pH 7.5 HEPES buffer and imaged using confocal laser scanning microscopy to check for the presence of the dyes. It was found that the Neutravidin-FITC fluorescent grid could still be observed, but there was no fluorescence signal obtained from the BODIPY fluorescent tag. Texas red fluorescence was also systematically absent from the microwells, but in some areas, it appeared to be randomly smeared over the surface. This picture contrasts sharply with the organized co-localization of the three dyes for the case of the intact liposome arraying. To demonstrate the importance of the microwells in confining and preventing the unraveling of the liposomes, liposomal solutions were brought into contact with uniformly functionalized surfaces. For pure silicon surfaces, the liposomes ruptured upon exposure to these untreated surfaces, as concluded from the fact that the dextran-Texas red fluorescence signal was smeared over the entire surface and was not localized in a regular pattern. Liposomes prepared using formulation II and loaded with dextran-Texas red cargo were also exposed to uniform Neutravidin-FITC-coated flat silicon wafers. Confocal laser scanning microscopy results (Figure 8) showed remaining uniform FITC fluorescence (Figure 8a) and that liposomes ruptured and released their fluorescent cargo. Dextran-Texas red was smeared over the Neutravidin-FITC layer on the flat silicon surface (Figure 8c), and lipid dye was present in small amounts, in the form of islands, over the surface (Figure 8b). This result further supports the use of patterned and chemically modified microwell surfaces to array intact liposomes. Figure 6c, which is the fluorescence data of dextran-Texas red present inside the liposomes arrayed on the microwell patterned surfaces taken close to the top of the microwells, shows the presence of two shoulders in the line intensity profile. We attribute this result to the complex optical reflections of the signal from the fluorescent cargo inside the microwells. To support this explanation, negatively charged 1-µm-diameter carboxyl-functionalized fluorescent particles (excitation maxima, 529 nm; emission maxima, 546 nm) were arrayed, instead of the liposomes, in the microwells of our patterned surfaces. The same microwellpatterned surfaces, functionalized with an amine SAM at the bottom of the well and a PEG SAM in the space between the wells, were exposed to a 0.0025 wt % solution of particles for 24 h and then rinsed with DI water and imaged using the confocal laser scanning microscope. The negatively charged particles were electrostatically coordinated with the positive amine termination of the wells.57,58 The passivating PEG functionality surrounding the wells allowed for the easy removal of the noncoordinated particles. Figure 9a shows confocal laser scanning microscopy data collected in a detection region close to the top of the microwells using the 514 nm laser for excitation and a 535-615 nm detection window. Figure 9b is a vertical slice (z slice) that confirms the presence of the particles inside the microwells. An examination of Figure 9a shows that the fluorescent pattern is identical to the pattern observed for the dextran cargo from the liposome immobilization experiments on these patterned surfaces (Figure 6c). In particular, both Figures exhibit a fluorescent

5410 Langmuir, Vol. 22, No. 12, 2006

Figure 9. (a) Confocal laser scanning microscopy images of electrostatically arrayed microparticles inside microwells. (b) ysection (xz plane) image from line as indicated in 9a.

concentric pattern consisting of a bright ring, a dark region, and a bright center. This pattern can be easily explained by assuming that the fluorescence of the centrally located particle (or liposome) is reflected off of the cylindrical inside surfaces of the wells. This reflection pattern is clearly shown in the y-section (xz plane) image shown in Figure 9b.

Discussion and Conclusions In this article, we describe a highly reproducible method of preparing arrays of individual intact liposomes approximately 1 µm in diameter and with a center-to-center spacing on the order of micrometers. Liposomes are deposited from solution into an array of microwells that have been etched in a 3 µm pitch on a surface, with the wells of the same size as the liposomes so that only one liposome is inserted into a well. The liposomes are attached to the bottom of the wells by a Neutravidin adapter linkage between biotin bound to the liposome surface and biotin conjugated to an amine self-assembled monolayer functionalizing the bottom of the well. The areas surrounding the wells are passivated with a PEG self-assembled monolayer to prevent the adsorption of liposomes and lipid matter in the space on the surface between the wells. Fluorescence imaging, using confocal laser scanning microscopy of an FITC dye attached to Neutravidin conjugated to the bottom of the wells, a water-soluble dextranTexas red dye in the liposome interior, and a lipid dye embedded in the bilayer, confirms the avidin functionalization of the wells, the confinement and intactness of the liposomes in the wells, and the effectiveness of the passivation of the surrounding area with PEG.

Kalyankar et al.

As mentioned in the Introduction, we envision the use of these liposome arrays as a platform for sequestering and displaying surface cell membrane proteins in the liposomes to form a membrane protein microarray. Several challenges remain before these proteoliposome arrays can be employed as membrane protein microarrays. First, the design should be able to display in the liposome positioned at each microwell registry a different protein. Second, the registries should be spatially indexed so that the identity of the membrane protein sequestered in the liposome at the well registry is known. Configuring the array so that at each microwell a different protein is displayed is easily undertaken by first preparing batches of liposomes with membrane proteins, with each batch containing the same membrane protein in the liposome bilayer. To form the array, the batches are pooled together, and the mixture is deposited onto the surface to allow random insertion of the liposomes into the microwells. Although this loads the microarray with different proteins, the identity of the protein at a given registry is unknown. The most facile way to index the array spatially using this pooling and deposition procedure is to encode the liposomes. Fluorescent optical encoding can be implemented in such a way that when a liposome batch is prepared with a particular protein, fluorescent dyes are dissolved in the aqueous interior with particular concentrations and emission profiles to form an intensity/color code unique to the prepared batch. After the liposome batches, each with unique codes that identify the sequestered protein, are mixed together and deposited onto the surface, fluorescence imaging of the microarray can be used to read the code at each well registry and identify the protein residing at that location. This type of encoding scheme has been suggested for a similar microarray format by Walt et al. 59,60 In their case, micrometer-sized polystyrene beads, functionalized with probe molecules such as antibodies, are encoded with fluorescent dyes partitioned into the beads. The beads are then mixed and deposited onto a surface consisting of a cross section of a fiber optic bundle in which the optical fibers have been etched into microwells to situate the beads. Instead of a widefield microscopy examination of the microarray to read the fluorescence code as suggested in our approach, the optical fibers themselves can be used to address the beads spatially and read their fluorescence codes. The membrane protein array format based on individually arrayed micrometer-sized proteoliposomes has certain advantages when compared to the larger-scale formats discussed earlier in the Introduction. In these formats, a tile mosaic is constructed in which supported phospholipids bilayers (SPBs) are either placed in recessed squares on a surface in wells or assembled contiguously on a surface and separated by protein corrals. The size of these tiles is on the order of tens of micrometers on a side. In the case of the recessed squares, the format is addressed as a traditional titer plate, and different membrane proteins can be directly displayed by pipetting different proteoliposomes into each of the wells and allowing the liposomes to form a bilayer with the embedded protein. Although easily spatially indexed by robotic deposition, the format is on a much larger scale than the micrometer scale of our display. In the case of the corralled bilayers, microfluidic flow, as described in the Introduction, is used to form tiles of SPBs such that along a row of tiles a common DNA tether extends from the SPB surface and adjoining tile rows have different tethers. Mixtures of proteoliposomes containing DNA that is complementary to the DNA of the ones on the tile surface are deposited onto the tiles and attach to the tiles containing their complementary DNA to form the proteoliposome array. Aside from the fact that the format is again large, the microfluidic flow necessary to define rows of tiles

Chemically Functionalized Microwell Surfaces

with different DNA tethers on their surfaces is less easily implemented and less amenable to micrometer-scale miniaturization than the pool and deposition method used when single liposomes are arrayed. Acknowledgment. This work was performed in part at the Cornell NanoScale Science and Technology Facility (CNF), a member of the National Nanotechnology Infrastructure Network, which is supported by the National Science Foundation (grant

Langmuir, Vol. 22, No. 12, 2006 5411

ECS 03-35765). We also thank the National Science Foundation (NSF) for funding the project (grant CTS 0428673). Additional financial support for this project from the Army Research Office is gratefully acknowledged (45349-LS-HSI; W911NF-04-1-0013) through the Department of Defense Hispanic Serving Institutions Research and Instrumentation Program. This project was supported in part by an NIH RCMI grant (RCMI RR03060). LA0602719