Coupling of the Orientations of Thermotropic Liquid Crystals to Protein

Nov 24, 2006 - hydrolysis of lipids by phospholipase A2, reflect the lateral organization (micrometer-sized domains) of the proteins and lipids, respe...
0 downloads 15 Views 960KB Size
Download PDF
Langmuir 2007, 23, 8497-8507

8497

Coupling of the Orientations of Thermotropic Liquid Crystals to Protein Binding Events at Lipid-Decorated Interfaces Jeffrey M. Brake and Nicholas L. Abbott* Department of Chemical Engineering, UniVersity of Wisconsin-Madison, 1415 Engineering DriVe, Madison, Wisconsin 53706 ReceiVed NoVember 24, 2006. In Final Form: April 23, 2007 We report a study of the interactions of proteins with monolayers of phospholipids (D/L-R-dipalmitoyl phosphatidylcholine and L-R-dilauroyl phosphatidylcholine) spontaneously assembled at an interface between an aqueous phase and a 20-µm-thick film of a nematic liquid crystal (4′-pentyl-4-cyanobiphenyl). Because the orientation of the liquid crystal is coupled to the organization of the lipids, specific interactions between phospholipase A2 and the lipids (binding and/or hydrolysis) that lead to reorganization of the lipids are optically reported (using polarized light) as dynamic orientational transitions in the liquid crystal. In contrast, nonspecific interactions between proteins such as albumin, lysozyme, and cytochrome-c and the lipid-laden interface of the liquid crystal are not reported as orientational transitions in the liquid crystals. Concurrent epifluorescence and polarized light imaging of labeled lipids and proteins at the aqueous-liquid crystal interface demonstrate that spatially patterned orientations of the liquid crystals observed during specific binding of phospholipase A2 to the interface, as well as during the subsequent hydrolysis of lipids by phospholipase A2, reflect the lateral organization (micrometer-sized domains) of the proteins and lipids, respectively, at the aqueous-liquid crystal interface.

Introduction Recently, we reported the spontaneous formation of monolayers of phospholipids by fusion of lipid vesicles and micelles with the interface between an aqueous phase and a water-immiscible thermotropic liquid crystal (LC).1,2 The lipids hosted at this interface were shown to possess a lateral mobility (0.2 × 10-1215 × 10-12 m2/s) characteristic of biological membranes.1,3 A central conclusion of our past study was that the organization of the monolayer of lipid at the interface was coupled to the orientation of the liquid crystal located within 20 µm of the interface. This phenomenon permitted the optical reporting (by transmission of polarized light) of the lateral organization of lipid at the interface via the patterned orientation of the LC. We also reported that the interactions of proteins with these lipidladen interfaces of LCs can trigger orientational transitions in the LC.2 In this paper, we expand upon these past observations to report in some detail on the nature of the coupling between protein binding events at the interfaces and the resulting orientational transitions in the LCs. The general approach reported in this paper builds upon a number of past studies that have exploited the reorganization of fluid lipid-laden interfaces to report binding events between proteins and lipids.4-6 For example, Cornell et al. used bilayer membranes tethered to an electrode to monitor receptor-ligand binding through the selective opening and closing of gated ion channels (gramicidin).4 Ligand-induced changes in the lateral mobility of the gramicidin in the outer leaf of the bilayer led to * To whom correspondence should be addressed. E-mail: abbott@ engr.wisc.edu. Fax: 608-262-5434. (1) Brake, J. M.; Daschner, M. K.; Abbott, N. L. Science 2003, 302, 20942098. (2) Brake, J. M.; Daschner, M. K.; Abbott, N. L. Langmuir 2005, 21, 22182228. (3) Stryer, L. Biochemistry, 4th Ed.; W. H. Freeman and Company: New York, 1995. (4) Cornell, B. A.; Braach-Maksvytis, V. L. B.; King, L. G.; Osman, P. D. J.; Raguse, B.; Wieczorek, L.; Pace, R. J. Nature, 1997, 387, 580. (5) Okada, S. Y.; Jelinek, R.; Charych, D. Angew. Chem., Int. Ed. 1999, 38, 655. (6) Song, X.; Swanson, B. I. Anal. Chem. 1999, 71, 2097.

changes in the electrical conductance of the assembly.4 A second approach has exploited vesicles formed from mixtures of L-Rdimyristoyl phosphatidylcholine (DMPC) and polydiacetylene (PDA).5 When exposed to phospholipase A2, hydrolysis of the DMPC led to changes in color that were attributed to conformational changes of the PDA backbone.5 A third set of approaches have exploited changes in the self-quenching and resonant-energy transfer of fluorescently labeled lipids and receptors to report binding of cholera toxin to ganglioside GM1.6 The changes in self-quenching and resonant-energy transfer were driven by changes in the lateral distribution of lipids and receptors in lipid bilayers caused by multivalent interaction of the cholera toxin with GM1.6 In this paper, we elaborate on a new approach that reports binding events at lipid-laden interfaces. As mentioned above, the approach exploits the use of LCs to both amplify and transduce the reorganization of the lipid induced by the binding event (Figure 1). This approach permits spatial imaging of the interface and does not require the use of labeled lipids or analytes.7-12 Our use of LCs to report protein-lipid binding events at interfaces also builds from past studies that have used LCs to image ligands bound to receptors covalently immobilized on solid substrates.13-15 The specific binding of proteins13,14 and small molecules15 to receptor-laden solid substrates led to changes in the orientation of LCs.14 One mechanism by which the LC transduced the presence of the bound protein was through masking of the nanometer-scale topography present in the solid sub(7) Stahelin, R. V.; Cho, W. Biochem. J. 2001, 359, 679. (8) Keller, C. A.; Kasemo, B. Biophys. J. 1998, 75, 1397. (9) Dietrich, C.; Bagatolli, L. A.; Volovyk, Z. N.; Thompson, N. L.; Levi, M.; Jacobson, K.; Gratton, E. Biophys. J. 2001, 80, 1417. (10) Rao, N. M.; Plant, A. L.; Silin, V.; Wight, S.; Hui, S. W. Biophys. J. 1997, 73, 3066. (11) Plant, A. L.; Brigham-Burke, M.; Petrella, E. C.; O’Shannessy, D. J. Anal. Biochem. 1995, 226, 342. (12) Yang, T.; Jung, S.-Y.; Mao, H.; Cremer, P. S. Anal. Chem. 2001, 73, 165. (13) Gupta, V. K.; Skaife, J. J.; Dubrovsky, T. B.; Abbott, N. L. Science 1998, 279, 2077. (14) Skaife, J. J.; Abbott, N. L. Langmuir 2000, 16, 3529. (15) Shah, R. R.; Abbott, N. L. Science 2001, 293, 1296.

10.1021/la0634286 CCC: $37.00 © 2007 American Chemical Society Published on Web 06/27/2007

8498 Langmuir, Vol. 23, No. 16, 2007

Brake and Abbott

fatty acid molecule at the sn-2 position of the glycerol backbone of the phospholipids.21,22 Physiologically, pancreatic PLA2s assist in digestion of phospholipids and cell surface remodeling, and PLA2s from venom are indicated in cell lysis and delivery of toxins to tissues.19,20 The active site of all PLA2s chelates Ca2+ which, in turn, promotes binding to and hydrolysis of phospholipids.21,23-27 The presence of Ca2+ (or another suitable divalent cation) is required for PLA2 to bind to phospholipids under most conditions (Figure 1A).24-27 Upon binding to the phospholipids in the presence of Ca2+, PLA2 will stereoselectively hydrolyze L-phospholipids.28,29 The hydrolysis products may then either form phase separated domains at the interface for longchained (n > 12) phospholipids or desorb from the interface for short-chained (n < 12) phospholipids (Figure 1B).30,31 Binding of PLA2 to D-phospholipids can still occur in the presence of Ca2+, but hydrolysis will not proceed (Figure 1C).32 The bound enzyme tends to accumulate at discontinuities in the lipid layer such as the boundaries between different lipid phases.33-35 The binding of PLA2 to the lipid also perturbs the organization of the acyl chains of the lipid, causing them to tilt toward the normal of the interface.35 We report below the results of a study that involved contacting PLA2 (in the presence and absence of Ca2+) with monolayers of either D or L enantiomers of phospholipids assembled at the aqueous-LC interface. Materials and Methods

Figure 1. Schematic illustration of proposed mechanisms by which interactions between PLA2 and phospholipids influence the anchoring of the LC. (A) In the presence of Ca2+, PLA2 binds to the monolayer of phospholipids at the aqueous-LC interface at defects in the lipid layer. In the absence of Ca2+, this interaction does not occur. (B) Bound PLA2 hydrolyzes L-phospholipids forming single-tailed lysophospholipids and fatty acids. Once formed, the products either phase-separate from the phospholipids or desorb from the interface, thus disrupting the anchoring of the LC. (C) PLA2 bound to D-phospholipids forms protein aggregates causing a disruption in the structure of the lipid monolayer which perturbs the anchoring of the underlying LC.

strate.13,16,17 In contrast to these past studies, the approach reported in this paper permits real-time reporting of the binding of proteins to fluid interfaces that mimic the mobility of biological membranes.18 The system addressed in this paper involves the interactions of lipids and phospholipase A2 (PLA2). This system was selected because it is well characterized and the interactions are tightly regulated by calcium ions (see below).19-21 PLA2 is a small (13-15 kDa) water-soluble protein that catalyzes the hydrolysis of L-phospholipids into an acyl lysopholipid molecule and a (16) Skaife, J. J.; Abbott, N. L. Chem. Mater. 1999, 11, 612. (17) Skaife, J. J.; Brake, J. M.; Abbott, N. L. Langmuir 2001, 17, 5448. (18) Brake, J. M.; Mezera, A. D.; Abbott, N. L. Langmuir 2003, 19, 86298637. (19) Yedgar, S.; Lichtenberg, D.; Schnitzer, E. Biochim. Biophys. Acta 2000, 1488, 182. (20) Van den Bosch, H. Phospholipases in Phospholipids; Hawthorne, J. N., Ansell, G. B., Eds.; Elsevier Biomedical Press: Amsterdam, 1982. (21) Slotboom, A. J.; Verheij, H. M.; de Haas, G. H. On the Mechanisms of Phospholipase A2 in Phospholipids; Hawthorne, J. N., Ansell, G. B., Eds.; Elsevier Biomedical Press: Amsterdam, 1982.

Materials. Dodecyltrimethylammonium bromide (DTAB), tris(hydroxymethyl)aminomethane (Tris), ethylenediaminetetraacetic acid (EDTA), calcium chloride, hydrochloric acid, bovine serum albumin (BSA), lysozyme from chicken egg white, cytochrome-c from bovine heart, phospholipase A2 from naja mossambica mossambica, L-R-dipalmitoyl phosphatidylcholine (L-DPPC), D-Rdipalmitoyl phosphatidylcholine (D-DPPC), and L-R-dilauroyl phosphatidylcholine (L-DLPC) were obtained from Sigma-Aldrich (St. Louis, MO). Octadecyltrichlorosilane (OTS) and sodium chloride were obtained from Fisher Scientific (Pittsburgh, PA). Texas Red1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine (TR-DPPE) and Alexa Fluor 488 conjugated albumin from bovine serum (BSA-488) were purchased from Molecular Probes (Eugene, OR). All chemicals were used as obtained without further purification. Deionization of a distilled water source was performed using a Milli-Q system (Millipore, Bedford, MA) to give water with a resistivity of 18.2 MΩ cm. 4′-Pentyl-4-cyanobiphenyl (5CB) was obtained from EM Sciences (New York). The glass microscope slides were Fisher’s Finest Premium Grade obtained from Fisher. Gold specimen grids (20 µm thickness, 283 µm grid spacing, and 50 µm bar width) were (22) Joubert, F. J. Biochim. Biophys. Acta 1977, 493, 216. (23) De Haas, G. H.; Bonsen, P. P. M.; Pieterson, W. A.; van Deenen, L. L. M. Biochim. Biophys. Acta 1971, 239, 252. (24) Menashe, M.; Romero, G.; Biltonen, R. L.; Lichtenberg, D. J. Biol. Chem. 1986, 261, 5328. (25) Lichtenberg, D.; Romero, G.; Menashe, M.; Biltonen, R. L. J. Biol. Chem. 1986, 261, 5334. (26) Yu, B.-Z.; Ghomashchi, F.; Cajal, Y.; Annand, R. R.; Berg, O. G.; Gelb, M. H.; Jain, M. K. Biochemistry 1997, 36, 3870. (27) Yu, B.-Z.; Rogers, J.; Nicol, G. R.; Theopold, K. H.; Seshadri, K.; Vishweshwara, S.; Jain, M. K. Biochemistry 1998, 37, 12576. (28) De Haas, G. H.; Postema, N. M.; Nieuwenhuizen, W.; van Deenen, L. L. M. Biochim. Biophys. Acta 1968, 159, 103. (29) Van Deenen, L. L. M.; de Haas, G. H. Biochim. Biophys. Acta 1963, 70, 538. (30) Zografi, G.; Verger, R.; de Haas, G. H. Chem. Phys. Lipids 1971, 7, 185. (31) Ransac, S.; Ivanova, M.; Verger, R.; Panaiotov, I. Methods Enzymol. 1997, 286, 263. (32) Bonsen, P. P. M.; de Haas, G. H.; Pieterson, W. A.; van Deenen, L. L. M. Biochim. Biophys. Acta 1972, 270, 364. (33) Grainger, D. W.; Reichert, A.; Ringsdorf, H.; Salesse, C. Biochim. Biophys. Acta 1990, 1023, 365. (34) Reichert, A.; Ringsdorf, H.; Wagenknecht, A. Biochim. Biophys. Acta 1992, 1106, 178. (35) Dahmen-Levison, U.; Brezesinski, G.; Mohwald, H. Thin Solid Films 1998, 327-329, 616.

Orientations of Liquid Crystals and Protein Binding obtained from Electon Microscopy Sciences (Fort Washington, PA). Sylgard 182 elastomer and curing agent (PDMS) were obtained from Dow Corning Corp. (Midland, MI), and Norland Optical Adhesive 61 (PU) was obtained from Norland Products, Inc. (Cranbury, NJ). Preparation of Supported Films of 5CB. Clean glass slides16 were functionalized with OTS according to published procedures.36 A flow cell consisting of an OTS-coated slide (0.3 cm × 1 cm) placed in a well (volume ≈ 150 µL) formed of an underlying glass support attached by PU to a thin (∼0.2 cm) slab (4 cm × 2.5 cm) of PDMS with a hole cut in its center (∼0.5 cm × 1.5 cm) was assembled according to previously reported methods.1 Films (∼20 µm) of LC were then formed on the OTS-coated glass slide by impregnating gold grids placed on the slide with ∼1 µL of 5CB. Excess 5CB was removed by contacting a capillary tube with the droplet of 5CB.36 After contacting the 5CB with various aqueous phases of interest (see below), an optical cell was formed by placing a second slide over the open well while being careful not to trap air bubbles. A water-tight seal between the cover slide and the PDMS well was achieved by fastening all four sides of the flow cell with binder clips. Exchange of solutions within the closed well was accomplished by inserting needles through opposite sides of the PDMS well and by using a syringe pump to inject controlled amounts of aqueous solutions through one of the needles while draining solution through the second needle. Preparation of Aqueous Dispersions of Phospholipids. Preparation of vesicle dispersions of lipids followed published methods.1,37-39 Briefly, the lipids were dissolved in chloroform and dispensed into glass vials. The chloroform was then evaporated under a stream of N2 for ∼1 min, and the vial containing the lipid was immediately placed under vacuum for at least 2 h. The dried lipid was resuspended in an aqueous solution consisting of 10 mM Tris, 100 mM NaCl adjusted with hydrochloric acid to a pH of 8.9 (TBS) and sonicated for 3 cycles of 5 min at 15 W. The resulting lipid suspension was twice-filtered using a 0.22 µm filter (Millipore), and the presence of vesicles (36 ( 5nm diameter) was confirmed by quasi-elastic light scattering. Solutions containing mixed micelles of lipids and DTAB (∼3.4 nm diameter by quasi-elastic light scattering) were formed by resuspending the dried lipid in TBS containing 3 mM DTAB at a molar ratio of 30:1. All lipid solutions were used within 48 h of preparation. Formation of Phospholipid Layers on 5CB. Formation of a lipid monolayer at the aqueous-5CB interface was accomplished by immersing the interface of the LC under ∼100-250 µL of either aqueous dispersions of lipid vesicles for 2 h or aqueous solutions of mixed micelles for 30 min. At the end of the equilibration period, in order to remove the surfactant, the aqueous phase was flushed with ∼10 mL of a lipid-free solution consisting of either TBS containing 5 mM calcium chloride (TBS-Ca2+) or TBS containing 5 mM EDTA (TBS-EDTA). Throughout the equilibration and rinsing procedures, the lipid-laden 5CB interface remained in continuous contact with the aqueous phase. Formation of a continuous lipid monolayer was confirmed by examining the optical texture of the 5CB and, when the lipid layer contained TR-DPPE, by examining the fluorescence of the lipid-laden aqueous-5CB interface (imaging techniques described below). Samples which did not show uniform homeotropic anchoring of the 5CB and a continuous distribution of TR-DPPE were discarded. Interactions of Proteins with Phospholipid-Laden Interfaces of 5CB. After formation of lipid-laden aqueous-5CB interfaces, the lipid was exposed to aqueous protein solutions by rapidly exchanging the aqueous phase with ∼1-10 mL of a prescribed protein solution in either TBS-Ca2+ or TBS-EDTA. The optical texture of 5CB and, when using lipid layers containing TR-DPPE, fluorescence of the lipid layer were then continuously imaged according to the methods described below. Proteins solutions (36) Brake, J. M.; Abbott, N. L. Langmuir 2002, 18, 6101. (37) Lee, S.; Kim, D. H.; Needham, D. Langmuir 2001, 17, 5544. (38) Keller, C. A.; Kasemo, B. Biophys. J., 1998, 75, 1397. (39) Deems, R. A. Anal. Biochem. 2000, 287, 1.

Langmuir, Vol. 23, No. 16, 2007 8499 consisting of PLA2 labeled with Alexa Fluor 488 (PLA2-488) were prepared using a commercially available Alexa Fluor 488 protein labeling kit (Molecular Probes). The labeling efficiency was calculated to be ∼1 fluorophore per PLA2 molecule, and the labeled enzyme was purified by extensive dialysis against water to remove unbound Alexa Fluor 488 and phosphate buffer. Fluorescence imaging of labeled proteins (PLA2-488 and BSA-488) was performed by first exchanging the aqueous phase for one containing no protein. This additional rinsing step was required to remove the background fluorescence of labeled proteins in the bulk solution. Imaging of the Optical Textures of 5CB. The orientation of 5CB was examined by using plane-polarized light in transmission mode on an Olympus BX60 microscope with crossed polarizers. The optical cell containing the film of 5CB was placed on a rotating stage located between the polarizers and imaged with the aqueous5CB interface either facing up or down. The focus was adjusted to either the aqueous-5CB interface (facing up) or the OTS-5CB interface (facing down). Identical textures in the 5CB were observed regardless of the orientation of the sample. Orthoscopic examinations were performed with the source light intensity set to 50% of full illumination, and the aperture set to 10% in order to collimate the incident light. Homeotropic alignments were determined by first observing the absence of transmitted light during a 360° rotation of the sample. Insertion of a condenser below the stage and a Bertrand lens above the stage allowed conoscopic examination of the cell. An interference pattern consisting of two crossed isogyres indicated homeotropic alignment.40 In-plane birefringence was indicated by a bright, colored appearance of 5CB and the presence of brush textures, typically four-brush textures emanating from a line defect, when the sample was viewed between crossed polarizers.41 All images were captured using a digital camera (Olympus C-2040 Zoom) mounted on the microscope and set to an f-stop of 2.6 and a shutter speed of 1/320 s. Epifluorescence Imaging of Aqueous-5CB Interfaces. Lipids and proteins adsorbed at the aqueous-5CB interface were imaged by transmitted light and epifluorescence microscopy using an Olympus IX71 inverted microscope equipped with a 100 W mercury lamp. A fluorescence filter cube with an excitation filter of 560 nm, and an emission filter of 645 nm was used to visualize Texas Red fluorescence. Alexa Fluor 488 fluorescence was visualized using a fluorescence filter cube with an excitation filter of 470-490 nm and an emission filter of 515 nm. Images were collected with a Hamamatsu 1394 ORCA-ER CCD camera (Bridgewater, NJ) connected to a PC. Throughout the imaging, the sample was oriented with the aqueous5CB interface facing down (toward the objective) to allow concurrent focus of the aqueous-5CB interface and the grid surface.

Results and Discussion Specific Binding of Phospholipase A2 to D-DPPC. First, we characterized the response of 5CB to specific binding of PLA2 to monolayers of D-DPPC formed from mixed surfactant-lipid micelles (see Methods) at the aqueous-5CB interface. As reported previously, specific binding of PLA2 (100 nM in TBS with 5 mM Ca2+) to monolayers of D-DPPC supported at the aqueous5CB interface causes an orientational (and thus optical) response in the LC.2 Here we expand on our past observations by examining the dynamics of the binding event and the extent to which the orientational response of 5CB reflects the lateral distribution and redistribution of lipid and protein on the interface. Initially, we examined the response of the LC to low concentrations of PLA2. In these experiments, we used experimental procedures that led to the deposition of a densely packed monolayer of D-DPPC at the interface (delivered using mixed micelles consisting of 3 mM DTAB and 0.1 mM D-DPPC).1 In (40) Bloss, F. D. An Introduction to the Methods of Optical Crystallograpy; Holt, Rinehart and Winston: New York, 1961. (41) Sonin, A. A. Freely Suspended Liquid Crystalline Films; John Wiley and Sons: New York, 1998. (42) Dennis, E. A. J. Lipid Res. 1973, 14, 152.

8500 Langmuir, Vol. 23, No. 16, 2007

Figure 2. PLA2 interaction with monolayers of D-DPPC assembled at an interface between nematic 5CB and aqueous solutions. The D-DPPC monolayers were formed by either (A-C) fusion of vesicles with the aqueous-5CB interface or (D) adsorption from mixed micelles containing 3 mM DTAB and 0.1 mM D-DPPC. (A) Optical image of 5CB after 16 h of exposure of the D-DPPC monolayer to 1 nM PLA2 in TBS-Ca2+. (B) Optical image of 5CB after 16 h of exposure of the D-DPPC monolayer to 100 nM PLA2 in TBS-Ca2+. (C) Optical image of 5CB after 16 h of exposure of the D-DPPC monolayer to 100 nM PLA2 in TBS-EDTA. (D) Optical image of 5CB after 16 h of exposure of the D-DPPC monolayer to 100 nM PLA2 in TBS-EDTA. All optical images were obtained using polarized white light with the sample located between crossed polars. Scale bar, 300 µm.

the presence of Ca2+ and at concentrations of PLA2 less than 100 nM, we observed the anchoring of 5CB at the D-DPPC-laden interface to remain homeotropic (dark optical texture) for >1 week (Figure 2A). The homeotropic orientation of the LC is identical to the orientation observed when a monolayer of lipid is hosted at the aqueous-5CB interface in the absence of PLA2.1,2 When the bulk concentration of PLA2 was increased to 100 nM in the presence of Ca2+, however, we observed the optical texture of the LC to become bright over the course of >2 h. After the initial brightening of the optical texture, we observed the appearance of the interface to evolve slowly over ∼16 h (Figure 2B). The prominent brushlike textures and interference color of the film of 5CB observed in Figure 2B indicate near-planar anchoring of the LC at the D-DPPC-laden interface of the LC.41,43 The change in appearance of 5CB upon binding of DPPC was irreversible even upon exchange of the bulk aqueous solution containing PLA2 for one containing neither protein nor Ca2+. In contrast, the optical texture of 5CB remained dark after >1 week of exposure of the D-DPPC-laden aqueous-5CB interface to 100 nM PLA2 in the absence of Ca2+ (Figure 2C). Because the binding of PLA2 to D-DPPC can be regulated by Ca2+, we concluded that the difference in appearance of the LC between Figure 2B (PLA2 with Ca2+) and Figure 2C (PLA2 without Ca2+) reflects the specific binding of PLA2 to D-DPPC in the presence of Ca2+. We make two comments regarding the above observations. First, the results above indicate that a bulk concentration of PLA2 of ∼100 nM is required to trigger a change in orientation of the (43) Brake, J. M.; Mezera, A. D.; Abbott, N. L. Langmuir 2003, 19, 64366442.

Brake and Abbott

LC. By assuming an area per lipid at the aqueous-5CB interface of ∼0.5 nm2 and a surface dissociation constant, KD, of 10-410-3 M (reflecting both binding to the interface and binding of lipid to the active site of the enzyme), we calculate the interfacial area per bound PLA2 at equilibrium to be ∼500-5000 nm2 (∼1340 nm separation between bound PLA2 molecules).39,42,44 This density of bound PLA2 is similar to the areal densities of proteins observed to trigger a response of the LC to receptor-ligand binding on solid surfaces.14 Second, we note that the time required for the specific binding of PLA2 to D-DPPC to be transduced into an optical signal by the LC (∼2 h) exceeds the characteristic time-scale for diffusion of the enzyme to the interface (∼10 s).45 This observation suggests that other processes involving reorganization of the surface-bound PLA2 likely occur before the binding event is reported by the LC. The observation of irreversible binding of the PLA2 further supports the conclusion that the PLA2 is involved in processes of reorganization at the interface following binding. Past studies of nonspecific binding of proteins to lipid-laden surfaces have shown that, after the initial association with the interface, proteins can undergo a second kinetic process that may involve reorganization, including penetration and denaturation of the protein.46,47 When using lipidladen interfaces of LCs, we note that specific molecular recognition of the lipid by PLA2 was required for the binding of the protein at the interface to be reported by the LC. Whereas the experiments described above were performed using an interface of the LC that was decorated with a densely packed monolayer of D-DPPC, we also investigated whether the surface concentration and organization of D-DPPC influenced the response of 5CB to specific and nonspecific interactions of the lipid with PLA2. We have shown previously that by contacting aqueous dispersions of vesicles of D-DPPC with 5CB at temperatures below the melting temperature of the D-DPPC bilayers (Tm ) 41 °C), surface coverages of D-DPPC that are less than a densely packed monolayer can be obtained (∼72% ( 10% of a monolayer versus ∼94% ( 18% of a monolayer when using mixed surfactant-lipid micelles).2 We first measured the response of the LC to the binding of 1-100 nM PLA2 at the D-DPPC-laden interface prepared from vesicles. In the presence of Ca2+ and 1-100 nM PLA2, the optical responses of 5CB were identical to those observed for monolayers of D-DPPC at the aqueous-5CB interface formed from mixed surfactant-lipid micelles (see Figure 2A,B). That is, the orientation of 5CB remained homeotropic at bulk concentrations of PLA2 35 °C), the appearance of the 5CB became uniformly dark between crossed polarizers (Figure 4C) and uniformly bright in the absence of polarizers (Figure 4D). A variety of imperfections not spatially correlated to the textures in the LC were also observed in Figure 4B,D. These imperfections arise from defects in the glass substrates and do not indicate the anchoring of the LC or the structure of the lipid-laden interface. These results lead to two conclusions regarding our experimental system. First, the longrange ordering of the 5CB associated with the nematic phase is required to obtain an optical image of the lateral organization of proteins and lipids at the aqueous-LC interface. Second, abrupt changes in the director profile and defects within the LC near the lipid and protein-laden interface of the LC leads to scattering of light and imaging of the domain boundaries without use of polarized light. Nonspecific Binding of Proteins to Phosphatidylcholines. The results reported above demonstrated that the orientational response of the LC was triggered by calcium-mediated interactions between PLA2 and densely packed monolayers of D-DPPC. Because the ultimate utility of these lipid-decorated interfaces to report specific binding events will be determined, in part, by their tolerance to nonspecific interactions, here we report on the response of 5CB to nonspecific adsorption of other proteins (BSA, cytochrome-c, lysozyme) to densely packed monolayers of lipids with phosphatidylcholine headgroups (D-DPPC and L-DLPC) supported at the aqueous-5CB interface. All of these proteins

Brake and Abbott

are known to interact with phosphatidylcholines through a variety of electrostatic and hydrophobic interactions.46-50 We note here that we sought to test the generality of our conclusions by using two lipids, D-DPPC and L-DLPC. The melting temperature of bilayers of L-DLPC is 4 °C, and our past studies demonstrate that it transfers spontaneously from vesicles to the interface of the LC to yield a densely packed lipid-laden interface (monolayer coverage) and homeotropic anchoring of 5CB. The results described below were obtained with L-DLPC. Similar results were obtained with D-DPPC. We first investigated the optical textures of L-DLPC-laden interfaces of 5CB after contact (>24 h) with 1 µM BSA, cytochrome-c, or lysozyme. The aqueous buffer was TBS-Ca2+ at pH 8.9; the same conditions used in experiments described above with PLA2. Under these conditions, the net charge on these proteins varies from negative (BSA, pI 4.8) to positive (cytochrome-c, pI 10.5; lysozyme, pI 11). Inspection of Figure 5 reveals that the optical textures of 5CB remained uniformly dark upon contact with all of the proteins (Figure 5A,C,D) indicating homeotropic anchoring of the LC. These images were identical to those obtained after formation of the L-DLPC monolayer (prior to exposure to the proteins). Two additional comments are in order. First, close inspection of the optical appearance of the LC within the metal grids shown in Figure 5 reveals that the LC is bright near the metal surface. These bright rims are due to the interaction of the LC with the grid surface. The distance over which the edges influence of the LC is of the order of the thickness of the film of the LC. Variation in brightness of the rims between samples is due to small variations in the amount of LC in each grid. Second, a number of small bright domains (containing black crosses) are apparent in Figure 5A,C,D. These bright domains are small water droplets that are observed to form spontaneously within the LC upon prolonged contact of the LC with some aqueous solutions. We do not yet fully understand the factors that determine the appearance of these droplets. They are not caused by the presence of the proteins in the solutions. By using fluorescently labeled BSA (BSA-488), we sought to obtain evidence of the association of BSA-488 with a monolayer of L-DLPC after exchanging the aqueous phase so that no BSA488 remained in the bulk (Figure 5B). The epifluorescence micrograph showed a uniformly dark aqueous-5CB interface, and quantitative analysis revealed that no measurable (and irreversible) association of BSA-488 with the L-DLPC-laden interface had occurred. These experiments were also conducted with monolayers of either L-DLPC or D-DPPC formed by adsorption from mixed surfactant-lipid micelles (3 mM DTAB, 0.1 mM lipid). In all cases, identical results to those presented above were obtained. These results are consistent with the known ability of densely packed phosphatidylcholines supported at a variety of interfaces to resist penetration and irreversible adsorption of these proteins.48,49 These results also provide further support to our conclusion that when using densely packed monolayers of lipids on LCs, the response of the LC is selective, only reporting specific protein-lipid interactions. Hydrolysis of L-DPPC-Laden Interfaces of 5CB by Phospholipase A2. We next examined the hydrolysis of the lipids by PLA2. The enantiospecificity of PLA2 was examined using monolayers of L-DPPC formed using mixed surfactant-lipid micelles. Unlike D-DPPC, L-DPPC can be hydrolyzed by PLA2 in the presence of Ca2+.23,29 Although our past studies have reported that the hydrolysis of L-DPPC by PLA2 (1 nM in TBS (50) Sankaram, M. B.; de Kruijff, B.; Marsh, D. Biochim. Biophys. Acta 1989, 986, 315.

Orientations of Liquid Crystals and Protein Binding

Langmuir, Vol. 23, No. 16, 2007 8503

Figure 5. Interaction of monolayers of L-DLPC assembled at the aqueous-5CB interface with solutions (TBS-Ca2+) containing (A,B) 1000 nM fluorescently labeled BSA (BSA-488), (C) 1000 nM cytochrome-c, and (D) 1000 nM lysozyme for 16 h. (A,C,D) Optical images of 5CB were obtained by polarized light microscopy with the sample located between crossed polars. (B) The epifluorescence micrograph of surface-associated BSA-488 was obtained after exchanging the bulk solution containing BSA-488. Scale bar, 300 µm.

with Ca2+) can trigger an anchoring transition in the LC,2 here we go beyond our past observations by reporting how the lateral distributions of lipid on the interface during the hydrolysis reaction influence the spatial and temporal response of the LC. We first contacted a L-DPPC-laden interface of the 5CB with 1 nM PLA2 in TBS-Ca2+. We observed bright domains (10-100 µm) within the LC to form after ∼1 h of contact of L-DPPC with the PLA2 (Figure 6A). These domains were observed to grow slowly with time. However, they never encompassed the entire surface. This observation contrasts to the uniformly dark optical images obtained when D-DPPC-laden aqueous-5CB interfaces were contacted with 1 nM PLA2 (see Figure 2A). We interpret the difference in optical textures as evidence that the enantiospecific hydrolysis of L-DPPC can trigger orientational transitions in the LC. Transmission of white light through the LC in the absence of polarizers showed distinct defect lines in the LC (Figure 6B), similar to those reported in Figure 4B. When the L-DPPC was doped with 1% TR-DPPE, the distribution of TR-DPPE was inhomogeneous and similar in pattern to the patterned orientational response of the LC (Figure 6C). This observation contrasts to the specific binding of PLA2 to D-DPPC where the distribution of lipids was not measurably perturbed. Past studies have reported that hydrolysis of L-DPPC causes the formation of phase-separated domains of the acid product and L-DPPC.33,34,51 We speculate that the inhomogeneous distribution of the TR-DPPE and the orientation of the LC reflect the (51) Burack, W. R.; Yuan, Q.; Biltonen, R. L. Biochem. 1993, 32, 583.

accumulation of the hydrolysis products of L-DPPC, especially palmitic acid, in spatially localized regions of the interface. Past studies have suggested that PLA2 accumulates at the acid domains through electrostatic interactions.33,34 We sought to find evident of this association by using fluorescently labeled PLA2 (PLA2-488). However, we observed no significant fluorescence activity when using 1 nM of PLA2. We also sought to determine the influence of the phase state of the 5CB on the stability of the lipid/protein domains by heating the sample shown in Figure 6A-C above 41 °C. Examination of the 5CB between crossed polarizers yielded a uniformly dark optical texture consistent with the appearance of 5CB in its isotropic phase (Figure 6D). In the absence of polarizers, no defect lines were apparent within the film of 5CB (Figure 6E). The epifluorescence micrograph of the L-DPPC at the aqueous-5CB interface, however, was indistinguishable from Figure 6C, showing a persistent and heterogeneous distribution of TR-DPPE. Upon cooling into the nematic phase, the optical texture of 5CB remained similar to that shown in Figure 6A. These results indicate that the domains are not lost upon heating the 5CB into the isotropic phase and then cooling back into the nematic phase (for less than 1 h). We next examined interactions of L-DPPC with PLA2 at bulk concentrations (100 nM) where specific binding to monolayers of D-DPPC was observed (see above) in an attempt to differentiate between the binding of PLA2 to the interface and the subsequent hydrolysis of the lipids. Small domains (∼10-50 µm) were observed to form in the LC after ∼1-2 h of contact of the

8504 Langmuir, Vol. 23, No. 16, 2007

Brake and Abbott

Figure 6. Interaction of monolayers of L-DPPC (doped with 1% TR-DPPE) assembled at the aqueous-5CB interface with aqueous solutions containing PLA2. Transmission of (A) polarized (crossed polars) and (B) unpolarized white light through nematic 5CB and (C) epifluorescence micrograph of lipid monolayer after >60 min of contact of L-DPPC with 1 nM PLA2 in TBS-Ca2+. Transmission of (D) polarized (crossed polars) and (E) unpolarized white light after heating the 5CB imaged in (A,B) into its isotropic phase. Transmission of polarized light (crossed polars) through nematic 5CB supporting L-DPPC which had been contacted with 100 nM PLA2 in TBS-Ca2+ for (F) 2 and (G) 36 h. (H) Transmission of polarized light (crossed polars) though nematic 5CB supporting L-DPPC and (I) the corresponding epifluorescence micrograph of the lipid after contact of the lipid with 1 nM PLA2 in TBS-EDTA for 16 h. Scale bar, 150 µm.

L-DPPC with PLA2 (Figure 6F). The domains had a qualitative appearance similar to those observed as a result of the hydrolysis of L-DPPC by PLA2 at a bulk concentration of 1 nM (see Figure 6A). Subsequently, the optical texture of the 5CB became bright and contained brushlike textures indicating a meandering azimuthal orientation of the LC (Figure 6G). These features of the texture were similar to those observed to accompany specific binding of PLA2 to D-DPPC (see Figure 2B). However, in contrast to the optical images of 5CB upon specific binding of PLA2 to D-DPPC, the small domains that result from the hydrolysis of L-DPPC are still evident in Figure 6G. We concluded, therefore, that specific binding of the PLA2 to the interface and the subsequent hydrolysis of the lipid give rise to distinguishable optical features in the LC. We also confirmed that the above optical responses reflected the enzymatic activity of PLA2 by depriving the enzyme of Ca2+. When the system did not contain Ca2+ (TBS-EDTA), the anchoring of 5CB remained uniformly dark after >16 h of contact of the L-DPPC-laden aqueous-5CB interface with aqueous solutions of PLA2 (1-1000 nM) (Figure 6H). Epifluorescence imaging of the aqueous-5CB interface showed that the TR-

DPPE remained uniformly distributed within the interface, also consistent with the anchoring of 5CB (Figure 6I). Hydrolysis of L-DLPC by Phospholipase A2. We next examined the orientational response of the 5CB to the hydrolytic activity of PLA2 at interfaces laden with L-DLPC.2 Whereas the hydrolysis products of L-DPPC are sparingly water-soluble and persist on the interface, the products of hydrolysis of L-DLPC are readily dissolved in water and thus were expected to desorb from the interface.30,31 The dynamic response of 5CB to hydrolysis of monolayers of L-DLPC containing 1% TR-DPPE supported at the aqueous5CB interface by PLA2 is shown in Figure 7. The optical texture of 5CB was uniformly dark after both adsorption of L-DLPC to the aqueous-5CB interface from an aqueous vesicle dispersion and subsequent exchange of the aqueous phase for TBS-Ca2+ which did not contain lipid (Figure 7A). The corresponding epifluorescence micrograph showed a uniform distribution of lipids at the aqueous-5CB interface (Figure 7B). PLA2 at a concentration of 1nM was then introduced into the calcium containing buffer while the optical texture of 5CB was monitored. After 45 min, bright domains (∼10-50 µm) were observed to

Orientations of Liquid Crystals and Protein Binding

Langmuir, Vol. 23, No. 16, 2007 8505

the lipid layer (Figure 7H) were achieved. The optical texture of 5CB (Figure 7I) was observed to remain uniformly dark and the epifluorescence micrograph of the lipid layer showed a uniform distribution of TR-DPPE (Figure 7J) when monolayers of L-DLPC were prepared in an identical fashion but exposed to 1 nM PLA2 in the absence of calcium. We interpreted the change in anchoring of the LC reported above to be due to the calcium-dependent hydrolysis of L-DLPC by PLA2 at the aqueous-5CB interface. As mentioned above, for L-DLPC, the water-soluble hydrolysis products are known to readily desorb from the interface rather than form phase separated domains.30,31 We, therefore, interpreted the dark regions in the epifluorescence micrographs of the aqueous-5CB interface (Figure 7D,F) to correspond to a clean (no lipid or products) interface. Our interpretation is supported by quantitative comparison of the fluorescence intensities of the domains in Figure 7D,F with the fluorescence intensities of the aqueous-5CB interface before (Figure 7B) and after (Figure 7H) hydrolysis of the lipid by PLA2. The lipid-rich domains had the same average fluorescence intensity as the initial lipid monolayer. Likewise, the lipid-depleted domains, the final aqueous-5CB interface, and the background all had similar fluorescence intensities. We note that the presence of the domains may be influenced by the tendency of PLA2 to aggregate at defect or phase boundaries in lipid layers.33-35,52,53 The results described above indicate that 1 nM PLA2 leads to easily distinguished changes in the optical appearance of the LC (e.g., domains) after 15-45 min and that the changes are complete after 90-120 min. We sought to determine if these time-scales are consistent with a primitive kinetic model for hydrolysis of L-DLPC by PLA2.31 Assuming steady-state conditions with excess substrate (lipid) present, the hydrolysis rate, v (units of molecules/ (cm2·s)), can be expressed as31

V ) QmE0ΓS

(1)

where Qm is the global kinetic constant of hydrolysis of lipid (∼10-15 cm3/(molecules·s),31 E0 is the total enzyme concentration in bulk solution (∼6 × 1011 molecules/cm3), and ΓS is the surface concentration of DLPC (units of molecules/cm2). The global kinetic constant lumps together many effects, including the association of PLA2 with the lipid monolayer and the binding of lipid to the active site of PLA2. The integrated form of the reaction rate in eq 1 can be arranged to provide an estimate of the reaction time, t, as a function of the fractional conversion, X, of the lipid layer into products,

t ) (QmE0)-1 ln

(1 -1 X)

(2)

Figure 7. Interaction of monolayers of L-DLPC (doped with 1% TR-DPPE) assembled at the aqueous-5CB interface with aqueous solutions containing 1 nM PLA2 imaged concurrently by polarized light microscopy (A,C,E,G,I) and epifluorescence microscopy (B,D,F,H,J). Contact of L-DLPC with an aqueous solution containing 1 nM PLA2 in TBS-Ca2+ for (A,B) 0, (C,D) 45, (E,F) 90, and (G,H) 300 min. (I,J) Contact of L-DLPC with an aqueous solution containing 1 nM PLA2 in TBS-EDTA for 300 min. All L-DLPC monolayers were formed by fusion of vesicles with the aqueous-5CB interface. Polarized light microscopy was performed using a white light source and crossed polars. Scale bar, 150 µm.

Using eq 2, we calculate that a conversion of 42% corresponds to a reaction time of 15 min (onset of response in LC) and that a conversion of 96% corresponds to 90 min. Although our estimates of reaction times fail to account for an induction period, substrate depletion, product accumulation, and other effects which would cause Qm to vary with reaction time,31,39,51,52 the relationship calculated between conversion and reaction times is generally consistent with the response of the LC. We also note that past studies have demonstrated that PLA2 partially extracts lipids from an interface prior to cleavage.54-59 It is possible that the

form in the LC (Figure 7C). These domains were spatially correlated to regions depleted in TR-DPPE (Figure 7D). The bright LC domains (Figure 7E) and lipid-depleted domains (Figure 7F) continued to grow and coalesce until a uniformly bright LC texture (Figure 7G), and a dark epifluorescence micrograph of

(52) Jain, M. K.; de Haas, G. H. Biochim. Biophys. Acta 1983, 736, 157. (53) Jain, M. K.; Jahagirdar, D. V. Biochim. Biophys. Acta 1985, 814, 313. (54) Kaasgaard, T.; Leidy, C.; Ipsen, J. H.; Mouritsen, O. G.; Jorgensen, K. Single Molecules 2001, 2, 105-108. (55) Mouritsen, O. G.; et. al J. Phys.: Condens. Matter 2006, 18, S1293S1304.

8506 Langmuir, Vol. 23, No. 16, 2007

Brake and Abbott

Figure 8. Interaction of a monolayer of L-DLPC (doped with 1% TR-DPPE) assembled at the aqueous-5CB interface with an aqueous solution containing 1 nM PLA2 in TBS-Ca2+. The monolayer of L-DLPC was formed by adsorption from mixed micelles containing DTAB (3 mM) and L-DLPC (0.1 mM). (A) The optical image of 5CB by polarized light microscopy (crossed polars) and (B) corresponding epifluorescence micrograph of the lipid layer were taken 20 min after exposure to PLA2. Scale bar, 150 µm.

Figure 9. (A) Optical image of 5CB after 30 min of contact with aqueous solutions containing 1:1 lauroyl lysoPC and lauric acid in TBS-Ca2+. (B) Optical image of 5CB after exchange (>5 min) of the aqueous solution containing lauroyl lysophosphatidylcholine and lauric acid for TBS-Ca2+ buffer containing neither surfactant. Both optical images were obtained using polarized white light with the sample located between crossed polars. Scale bar, 300 µm.

LC may influence the kinetics of PLA2 hydrolysis through its effect on such processes. The characteristic time for diffusion of PLA2 to the lipid-laden interface (t ≈ Γ02/E02D ≈ ΓS2/KD2D where Γ0 is the maximum surface excess concentration of PLA2, ΓS ≈ 2 × 106 molec/µm2, KD ≈ 6 × 104 µm-3, and D ≈ 100 µm2/s), by comparison, is only 11 s.39,42,44,45 The optical response of 5CB caused by interactions between DPPC and PLA2 was observed above to be a function of the method of preparation of the monolayer of DPPC (i.e., adsorption from vesicles versus mixed surfactant-lipid micelles). The difference in response was explained by noting the difference in relative surface coverages and lipid organization obtained using these two preparation methods.1 Unlike DPPC, L-DLPC can be transferred to the aqueous-5CB interface via vesicles and mixed surfactant-lipid micelles with equal efficiency.1 We hypothesized, therefore, that hydrolysis of monolayers of L-DLPC formed by both methods should cause similar optical responses in the LC. After formation of a monolayer of L-DLPC at the aqueous-5CB interface by adsorption from mixed surfactant(56) Pedersen, S.; Jorgensen, K.; Baekmark, T. R.; Mouritsen, O. G. Biophys. J. 1996, 71, 554. (57) Honger, T.; Jorgensen, K.; Stokes, D.; Biltonen, R. L.; Mouritsen, O. G. Methods Enzymol. 1997, 286, 168. (58) Zhou, F.; Schulten, K. Proteins: Struct., Funct., Genet. 1996, 25, 12. (59) Stepaniants, S.; Izrailev, S.; Schulten, K. J. Mol. Model. 1997, 3, 473.

lipid micelles, the introduction of 1 nM PLA2 into the aqueous phase in the presence of Ca2+ caused the formation of bright domains of LC after ∼15 min (Figure 8A). The shapes of these domains were elongated relative to the ellipsoidal domains observed during hydrolysis of monolayers of L-DLPC formed by adsorption from vesicles (see Figure 7C). After ∼90 min, the optical texture of the LC became completely bright with an appearance similar to those observed after 90-300 min for L-DLPC formed by adsorption from vesicles (see Figure 7G). The bright domains of 5CB were found to correlate with lipiddepleted regions of the interface (Figure 8B) using epifluorescence microscopy. The lipid-depleted domains were also observed to be elongated relative to the ellipsoidal lipid-depleted regions imaged during hydrolysis of monolayers of L-DLPC formed by adsorption from vesicles (see Figure 7D). The differences that we observed between the shapes of domains during hydrolysis of L-DLPC suggest that the L-DLPC-laden interfaces prepared by the two methods are not identical. The nature of these differences is not known to us but could include factors such as density of defects or residual surfactant (DTAB) within the monolayers. The results described above are consistent with a mechanism in which the LC reported hydrolysis of L-DLPC through desorption of the hydrolysis products of L-DLPC from the

Orientations of Liquid Crystals and Protein Binding

aqueous-5CB interface.30,31 We tested this proposition by forming mixed monolayers consisting of a 1:1 mixture of lauroyl lysophosphatidylcholine and lauric acid (the products of hydrolysis of L-DLPC by PLA2). In the presence of a bulk aqueous phase containing these hydrolysis products, the anchoring of 5CB was homeotropic resulting in a uniformly dark optical texture within each of the grid squares (Figure 9A). However, upon exchange of the aqueous phase for one that did not contain the reaction products or L-DLPC, the optical texture of 5CB rapidly (