Change of Line Tension in Phase-Separated Vesicles upon Protein

Feb 15, 2012 - The results show how domain boundaries can lead to an accumulation of ..... the large energy cost of changing V (which requires work...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/Langmuir

Change of Line Tension in Phase-Separated Vesicles upon Protein Binding Jaime B. Hutchison,† Robert M. Weis,‡ and Anthony D. Dinsmore*,† †

Department of Physics and ‡Department of Chemistry, University of Massachusetts, Amherst, Massachusetts, United States S Supporting Information *

ABSTRACT: We measured the effect of a model membranebinding protein on line tension and morphology of phaseseparated lipid-bilayer vesicles. We studied giant unilamellar vesicles composed of a cholesterol/dioleoylphosphatidylcholine/palmitoylsphingomyelin mixture and a controlled mole fraction of a Ni-chelating lipid. These vesicles exhibited two coexisting fluid-phase domains at room temperature. Owing to the line tension, σ, between the two phases, the boundary between them was pulled like a purse string so that the smaller domain formed a bud. While observing the vesicles in a microscope, histidine-tagged green fluorescent protein was added, which bound to the Ni-chelating lipid. As protein bound, the vesicle shape changed and the length of the phase boundary increased. The change in morphology was attributed to a reduction of σ between the two phases because of preferential accumulation of histidine-tagged green fluorescent protein−Ni-chelating lipid clusters at the domain boundary. Greater reductions of σ were found in samples with higher concentrations of Ni-chelating lipid; this trend provided an estimate of the binding energy at the boundary, approximately kBT. The results show how domain boundaries can lead to an accumulation of membrane-binding proteins at their boundaries and, in turn, how proteins can alter line tension and vesicle morphology.

■

INTRODUCTION The structural arrangement of proteins in cell membranes plays a key role in their function, but remains a poorly understood problem because of the complex interplay of multiple lipids and proteins. While specific interactions among proteins determine their short-range interactions, there is growing evidence that the thermodynamic properties of the membrane itself also play a major role in arranging membrane proteins for proper function. Experiments showed that the membranes of giant plasma membrane vesicles of rat basophilic leukemia cells separate into two coexisting liquid phases.1,2 These phases correspond to the liquid-ordered and liquid-disordered (Lo and Ld) phases observed in vesicles formed with a ternary mixture of lipids,1−5 indicating that phase separation in model membranes can provide important insights for living cells. Because many proteins exhibit a preference for either the Lo or Ld phase, phase separationor even concentration fluctuations that arise under single-phase conditions2can provide a mechanism to sort particular lipids and proteins into domains in the membrane (i.e., rafts).6−10 The one-dimensional boundary between two phases can also play a role in sorting or directing proteins. Dumas et al. showed that bacteriorhodopsin proteins accumulate at the boundary between Ld and gel lipid phases in model membranes.11,12 They proposed that the energy of the boundary, which arises from the mismatch of the lengths of the acyl chains of the two lipid species, is reduced by the presence of the protein. Since then, other examples of © 2012 American Chemical Society

proteins that localize at domain boundaries have been reported (and reviewed13), including ion channels,13−15 antimicrobial peptides,16 N-Ras,17,18 pore-forming proteins Equinatoxin II19 and Bax,20 and (according to computer simulations) transmembrane WALP proteins that are functionalized with a sufficient number of saturated lipids.21 The tendency of these and other proteins to accumulate at domain boundaries may enhance their function, as proposed for the ion channels15 and pore-forming Equinatoxin II.19 Viewing this phenomenon at a continuum level rather than a molecular-scale level, the spontaneous adsorption of proteins at domain boundaries should accompany a reduction of the line tension of the boundary, σ. Here, σ (with units of J/m) may arise in part from a difference in the acyl-chain length of the lipids in the two phases.12 The magnitude of σ at the Lo/Ld interface can influence the dynamics of phase separation and the size, number, and stability of domains.22 A recent study of living human HaCaT cells showed micrometer-sized Ld-type domains formed as buds; those authors proposed that the line tension between the buds and the surrounding membrane favors the formation of budded domains, rather than coplanar domains.23 Hence, it may be that the line tension plays a key role in living cells and that domain-boundary partitioning of membrane Received: November 2, 2011 Revised: February 3, 2012 Published: February 15, 2012 5176

dx.doi.org/10.1021/la204225a | Langmuir 2012, 28, 5176−5181

Langmuir

Article

the Lo/Ld area ratio is comparable to that of a published study of vesicles with this composition.30 Microscopy of GUVs. GUVs were observed with differential interference contrast (DIC) and epifluorescence microscopy using a Zeiss Axiovert 200 inverted microscope equipped with a PlanNEOFLUAR 100× oil objective and a Hamamatsu CCD camera. To observe the fluorescence of GFP, a filter set with a 480 nm short-pass filter, a 505 nm dichroic mirror, and a 535 nm long-pass filter (Chroma Technology Corp., Bellows Falls, VT) was used with a 100

proteins may be important in biological function. The question then becomes: how much does σ change as a result of protein binding to the domain interface? In the case of Bax poreforming protein, the line tension at a membrane pore (distinct from a domain boundary) was reduced by 2.2 pN (or 40%) according to AFM measurements. 20 For liquid−liquid boundaries, the line tension is smaller and the effect of proteins may be correspondingly smaller and more difficult to measure. In the N-Ras studies,17,18 the authors indirectly inferred a reduction of σ based on the change in height of the Lo phase, but a direct measurement of the magnitude of the change of σ and an understanding of how it depends on the nature of the proteins and their concentration remains an open problem. In this article, we show that a model membrane-binding protein binds to lipid in the Ld phase and induces a shape change in phase-separated giant unilamellar vesicles (GUVs). We use 6×-histidine-tagged green fluorescent protein (histagged GFP) as this model protein. The GUVs are composed of a commonly used mixture of saturated and unsaturated lipids plus cholesterol to form phase-separated membranes, plus a controlled mole fraction of a nickel-chelating lipid (nickel lipid) to peripherally bind his-tagged GFP. Binding of his-tagged GFP has a clearly discernible effect on the morphology of the vesicles, which indicates a reduction in σ. We computed the fractional change in σ by quantifying a change of vesicle shape after protein bound. When the concentration of protein binding sites in the membrane was increased (i.e., an increase in the nickel lipid mole fraction), the reduction in σ became greater, reaching an average reduction of 46 ± 13% in membranes containing 10 mol % nickel lipid. We also found that poly(Lhistidine) was more effective than his-tagged GFP in reducing σ, which was attributed to the larger number of nickel lipids that poly(L-histidine) could bind. Based on the results, we present and discuss a mechanism for the reduction of σ. This mechanism is a two-dimensional version of the well-established tendency of particles to bind at liquid−liquid interfaces in Pickering emulsions.24−27 These results provide evidence that the preferential partitioning of proteins at domain boundaries can serve to organize proteins and also strongly reduce the line tension. These phenomena might have a function in living cells.

■

Figure 1. Microscope images of a phase-separated vesicle with 2.5 mol % nickel lipid during in a GFP-binding experiment. (a,b) DIC images of the vesicle before and after adding GFP, respectively. (c) Vesicle outlines from (a,b), rotated and superimposed to show the shape change. (d) Epifluorescence image of the same vesicle after adding GFP, demonstrating that the GFP partitions to the Ld phase. Brownian motion and convection produced the different orientations of the GUV in the three images. Scale bar: 10 μm.

MATERIALS AND METHODS

W mercury lamp (Figure 1). Video was recorded using S-VHS format (30 frames per second) and subsequently digitized. In separate fluorescence imaging experiments, we also investigated the nonuniformity of the GFP fluorescence intensity near the Lo/Ld domain boundary (see Supporting Information for details). To visualize GUVs and observe shape changes, an aliquot of sample was placed into a 120 μL sample chamber and imaged under DIC optics to confirm phase separation. Once a suitable vesicle was located, a mock experiment was usually first performed: an aliquot of (an isoosmolar) buffer (2.5−3.0 μL of 12 mM Na2HPO4, 70 mM NaCl, pH 6 to 7) was introduced to the sample chamber to verify that the change in shape did not occur as a result of osmotic shock, flow-induced stress, or other nonspecific effects. Then, an aliquot of his-tagged GFP was added (2.5−3.0 μL 40 μM GFP in buffer). His-tagged GFP was in excess under experimental conditions (the molar ratio of accessible nickel lipid to GFP varied between 0.3 and 0.03). Following injection, the vesicle was imaged with DIC (for shape analysis) and epifluorescence optics (to verify his-tagged GFP binding). The sequence of events before and after the addition of GFP was captured on videotape. Images selected for digitization were those video frames in which the vesicle was in focus and the radii were at their largest, conditions under which the focal plane corresponded most closely to the

Lipids, GFP, and the Preparation of GUVs. Dioleoyl-sn-glycero3-phosphocholine (DOPC), N-palmitoyl-D-erythro-sphingosylphosphoryl-choline (PSM), cholesterol (Chol), and the nickel salt of 1,2dioleoyl-sn-glycero-3-[(N-(5-amino-1-carboxypentyl)iminodiacetic acid)succinyl] (nickel lipid) were purchased from Avanti Polar Lipids Inc. (Alabaster, AL). Multiple batches of lipids were used over the course of the experiments. Superfolder his-tagged GFP was isolated with nickel-nitrilotriacetic acid affinity chromatography as described in the literature.28 The superfolder variant of GFP essentially eliminates dimerization, even in concentrated solutions.28 Poly(L-histidine) (polyhistidine), used in some experiments (Sigma-Aldrich, St. Louis MO, cat. # P9386), had a 5000−25 000 MW range and an average degree of polymerization of 86. GUVs were prepared by electroformation in 175 mM sucrose solution, using the technique of Angelova et al.29 In all experiments, the DOPC:PSM:Chol ratio was 2:2:1; this proportion was maintained as the amount of nickel lipid was varied (1, 2.5, 5.0, or 10 mol %). Vesicles composed of 2:2:1 DOPC:PSM:Chol at room temperature (25 °C) are known to exhibit coexistence of two distinct liquid phases,4 in which the Ld phase is rich in DOPC and the Lo phase is rich in PSM and Chol. We found that the area ratio of the Lo phase to the Ld phase ranged from 0.3 to 1 (0.6 ± 0.2). Our average value for 5177

dx.doi.org/10.1021/la204225a | Langmuir 2012, 28, 5176−5181

Langmuir

Article

equatorial planes of the two phases. Using this procedure, vesicle morphology was determined both before GFP addition and following the addition of GFP, after the vesicles reached steady state. Shape Analysis of GUVs. All of the vesicles chosen for analysis had shapes that were well-described by two spherical caps. This selection ensured that lateral tensions were sufficiently large that bending stiffness could be ignored31,32 and a simple, geometric analysis of the vesicles could also be performed. Using a method developed by Tian et al.,33 we start with eq 1, the force balance in the plane of the domain boundary. σ To cos(α) + Td cos(β) = Rb (1)

of his-tagged GFP. The overall shape of a vesicle containing an Lo and an Ld domain resembles a sphere with a bud. (The external shape of these two-phase vesicles resembles two merged soap bubbles but differs in an important respect: the merged soap bubbles have an internal membrane that separates the small and large bubble; the vesicle does not.) The “pinched waistband” corresponds to the Lo−Ld circular interface. After his-tagged GFP was introduced, over a time interval ranging from a few tens of seconds to two minutes, we observed that GFP bound to the vesicle (by epifluorescence) and the vesicle morphology changed and reached a new steady state (DIC image, Figure 1b). Figure 1c illustrates the shape change more clearly by superimposing the outlines of the vesicle shown in (a) and (b). Figure 1d shows GFP fluorescence from the Ld domain. Figure 2 shows a DIC image of a phase-separated GUV focused at the equatorial plane, and (in Figure 2b) overlaid with the geometric constructions used to measure the three radii: Rd (from the red-dashed circle), Ro (from the green-dashed circle), and Rb (from a line drawn between the points of intersection of the red and green circles), as well as α and β and the tensions (Td, To, σ) involved in force balance. From a collection of images like these, the σ/P were computed using eq 3. The numerical values fell in the range of (2−10) × 10−11 m2. To compare this result to the literature, we assume a line tension of 1 pN,12,22,33,34 and thereby obtain P = 0.01−0.05 Pa, and estimates for To and Td of ∼ 10−4 mN/m (via eq 2). These values are similar to tensions and pressures reported previously for GUVs.34 Figure 3 shows a plot of the percentage change of σ/P as a function of nickel lipid concentration. The plot shows data for

The quantities To and Td are the lateral membrane tensions in the ordered and disordered domains, respectively, α and β are the tangent angles made by the ordered and disordered domains with respect to the domain boundary, σ is the line tension at the boundary, and Rb is the radius of the boundary (Figure 2).

Figure 2. (a) Optical microscopy (DIC) image of a vesicle, with brightness and contrast enhanced. (b) The same image overlaid with circles and squares for obtaining radii (Rd, Ro, Rb) and angles (α, β). Also depicted are the tensions (To, Td, σ) that participate in the force balance. The definition of Laplace pressure, P (excess pressure inside the vesicle), and the requirement that it be the same throughout the interior of the vesicle gives P=

2To 2T = d Ro Rd

(2)

where Ro and Rd are the radii of the spherical caps of the ordered and disordered domains, respectively. Combining eqs 2 and 3 to eliminate the lateral tensions results in R σ = b × [R o cos(α) R d cos(β)] P 2

(3)

Equation 3 allowed us to measure the ratio of line tension to the Laplace pressure directly from the shape of the vesicle. Uncertainties were estimated from standard deviations computed for Ro, Rd, α, and β from at least two measurements. Error propagation was then used to calculate the uncertainties of the areas, volumes, and σ/P. The analysis was performed before and after the addition of his-tagged GFP to obtain two independent measurements of σ/P, and thereby the change of σ/P upon GFP binding.

Figure 3. Plot of the measured percent change in line tension over Laplace pressure (σ/P) due to GFP binding in vesicles with different nickel lipid concentrations. The gray-filled columns correspond to changes in σ/P of vesicles in which the interior volume remained unchanged, within experimental uncertainty. The unfilled columns correspond to vesicles with a discernible change in interior volume.

RESULTS AND DISCUSSION The nickel lipid partitions to the Ld phase, as might be expected because it has the same oleic acid acyl chains as the DOPC. By varying the amount of nickel lipid during vesicle preparation (between 1 and 10 mol %), we controlled the amount of histagged GFP bound to the Ld-phase. Three images from a typical experiment are shown in Figure 1. Figure 1a shows a DIC image of a GUV before the addition

all vesicles that could be described as sections of spheres (as described above). Each column in the plot corresponds to an experiment conducted on a single vesicle, and the error bars are estimated from two measurements of the same vesicle. Despite the scatter in the data, there is a clear trend of larger reductions in σ/P with increasing nickel lipid concentration. In some cases, the effect is dramatic, with a relative decrease in σ/P of approximately 100%. These cases corresponded to final

■

5178

dx.doi.org/10.1021/la204225a | Langmuir 2012, 28, 5176−5181

Langmuir

Article

stretching the membrane). While changing the area ratio Ao/Ad to satisfy the V and A constraints may cost energy (because the stoichiometry of each phase must be shifted in this closed system), we assume that this energy cost is small and neglect it from our discussion below. To explain the scatter in the data shown in Figure 3, we looked for correlations of σ/P or the change in σ/P with the vesicle geometry. We found no statistically significant correlations with the initial vesicle surface area (A, Ao, Ad, or changes in Ao/Ad), initial vesicle volume V, or initial vesicle “bulginess” (α + β). To assess whether the change in σ/P arises generally from a binding interaction between the nickel lipid and a histidine tag, we performed the same procedure using polyhistidine instead of his-tagged GFP. (2.5 μL of ∼40 μM polyhistidine in pH 5.5. buffer was added.) In these experiments, the vesicles contained either 2.5 mol % or 5 mol % nickel lipid. As with the his-tagged GFP, we identified the vesicles that maintained V (and hence P), and found their average change in σ/P. For 2.5 and 5.0 mol %, we found an average decrease of 27 ± 10% (12 vesicles) and 40 ± 9% (9 vesicles), respectively. (See Supporting Information for the full plot of data values.) These values are shown by the filled data points in Figure 4. These data are consistent with the idea that it is the chelation of nickel lipid with the hexahistidine tag (independent of GFP) that generates the decrease in σ/P, and that the larger valency of polyhistidine produces a greater change in σ/P (Figure 4). Model for the Change in σ. We now discuss the mechanism of the change in vesicle morphology and in σ/P. First, we note that vesicles that had constant V should also have constant pressure P. Because membranes are permeable to water but not to larger molecules such as sucrose, there exists an osmotic pressure, π, across the membrane ≈ kBT(N/V − c0), where kBT is Boltzmann’s constant, T is the absolute temperature, N is the number of solute molecules inside the vesicle volume V, and co is the concentration of solute outside the vesicle. In equilibrium, P (the difference in pressure across the membrane) balances π. Assuming that the vesicles did not lyse during GFP binding, N remained constant, so a constant V corresponds to constant π and P. Therefore, in these cases we equate a change of σ/P with a change of σ only. Two possible mechanisms were initially considered to explain the change of line tension. The first posits that GFP binding changes the compositions of the two phases and thus the chemical potentials of the species and, as a result, the line tension. While we cannot absolutely rule out this mechanism, we expect that a change in the chemical potentials that is sufficient to reduce σ by up to 46% would also change the relative areas of the Lo and Ld phases systematically. While the relative areas either increased or decreased in 34% of the cases, these changes were not systematic and did not depend upon the nickel lipid concentration. Another possibility is that GFP binding causes an increase in membrane lateral tension which, in turn, alters the phase boundary. To examine this possibility, we monitored the area of single-phase vesicles before and after GFP binding and found no change in area (see Supporting Information). Instead, we propose that σ is reduced by the accumulation of GFP-bound nickel lipid at the one-dimensional boundary between the Lo and Ld domains. Direct evidence of this accumulation is provided by digital-camera fluorescence images of GUVs having 1 mol % nickel lipid. The results varied from vesicle to vesicle, yet we found evidence of enhanced GFP

morphologies that were nearly spherical, with little or no bulge in the ordered domain. In most cases, the interior volumes of the vesicles were found to be the same before and after addition of the his-tagged GFP solution, within measurement uncertainty. These data are represented by the gray-filled columns in Figure 3. However, in 8 out of 40 cases, there was a distinguishable change in interior volume; these latter data are represented by open columns. The average responses as a function of the incorporated nickel lipid concentration (mol %) are shown in Figure 4. The

Figure 4. Average percent decrease in σ/P as a function of the nickel lipid concentration. The unfilled data points correspond to his-tagged GFP and were derived from the constant-volume vesicles of Figure 3. The curve is the best fit of the data to the model described in the text, and the dashed curves show the 95% confidence range for the fitted parameters. The filled circles show the decrease in σ/P produced by poly(L-histidine).

percent change in σ/P increased monotonically as the concentration of protein binding sites increased. In this analysis, only those vesicles whose interior volume remained unchanged were considered. (As discussed below, these vesicles have constant pressure P, so that the change in σ/P corresponds only to σ.) For nickel lipid concentrations of 1, 2.5, 5, and 10 mol %, the average change in σ/P was −1 ± 4%, −16 ± 5%, −32 ± 8%, and −46 ± 13%, where the uncertainties are standard errors of the mean. The systematic dependence on the nickel lipid concentration indicates that the concentration of bound GFP determines the magnitude of the shape change. Geometric constraints may play a role in restricting the possible morphologies, or at least in determining what shapes will minimize the total free energy. In general, a change of shape from one set of [Ro, Rd, Rb] to another can change total interior volume V, total surface area A, and the relative areas of the two domains, Ao/Ad. Given an initial configuration [Ro, Rd, Rb], we cannot in general expect to find another configuration that conserves all three quantities A, V, and Ao/Ad. (This can be seen by counting constraints and was verified by direct geometric calculations.) In the data, we find that, among the 53 measured vesicle changes that preserved V, the majority (80%) also conserved A within our estimated uncertainty and a further 13% nearly conserved A (i.e., within twice our estimated uncertainty). Among those that conserved V, 34% underwent a measurable change in Ao/Ad; these were nearly evenly split between increasing or decreasing Ao/Ad. The fact that most of the vesicles maintained constant V and A is plausible because of the large energy cost of changing V (which requires work against the osmotic pressure) and of changing A (which require 5179

dx.doi.org/10.1021/la204225a | Langmuir 2012, 28, 5176−5181

Langmuir

Article

contains ∼80 histidine residues (on average), while each GFP histidine tag has only six residues (Figure 4). As a further test of this domain-boundary affinity model, the data were fit to a simple adsorption equation,36,37 corresponding to a Langmuir adsorption equilibrium between a twodimensional ideal gas of particles (clusters of GFP-bound nickel lipids in the Ld phase) and particles bound to the onedimensional Ld-Lo boundary. Bound clusters of diameter l0 were modeled as objects whose centers of mass are confined to within a distance h of the domain boundary. Bound clusters have an internal energy lower than the clusters in the surrounding two-dimensional gas by the amount ε. (This binding energy ε should be proportional to the reduction of the line energy, l0σ.) We then approximate Δσ, which is the change in energy per length, as the product of ε and the number of bound clusters per unit length. Assuming for simplicity that h = l0, the model leads to

fluorescence at the domain boundary in six images of vesicles. However, the same deconvolution process applied to four vesicles containing 5 mol % nickel lipid showed no fluorescence enhancement at the boundary. This result provided evidence that image deconvolution did not produce artifacts, yet the result was unexpected because the larger nickel lipid concentration should have yielded greater accumulation at the boundary. We speculate that the fluorescence enhancement at the boundary is less apparent because the difference in fluorescence intensity between the boundary and the (bulk) Ld phase is smaller despite the greater intensity overallwhich is to be expected as the boundary is driven to saturation at the larger concentration of nickel lipid. The methods and results of this analysis, including a comparison to computer-generated control images, are provided in the Supporting Information. The localization of proteins at one-dimensional domain boundaries is analogous to the adsorption of surfactants or colloidal particles at two-dimensional liquid boundaries, like those found in oil−water emulsions.13,17,25−27 In threedimensional suspensions of colloidal particles, it is wellknown that adsorption at a liquid boundary can lower the total interfacial energy and reduce the measured interfacial tension.24,25 In these cases, the particles need not have an amphiphilic structure. Instead, adsorption to the boundary is driven by the reduction in the contact area between the two liquid phases, provided that the particle does not have too strong a preference for one phase over the other. A similar phenomenon may contribute to the reduction in σ at the domain interfaces in fluid membranes. In the case of bacteriorhodopsin, the affinity for the domain boundary was based on the thickness of the hydrophobic region of the protein.11 In our case, the affinity for the boundary may arise from the fact that both poly(L-histidine) and 6 × his-tagged GFP may bind multiple nickel lipids, thereby creating small lipid−protein clusters. In a previous study of his-tagged GFP, Nye and Groves provided evidence that 10 × his-tagged GFP bound with mono- and polyvalent interactions to nickel lipid in supported lipid bilayers (i.e., one GFP to multiple lipids), with polyvalent binding remaining stable for hours.35 The changes in structure and/or orientation of the nickel lipid acyl chains generated upon binding histidine are, to our knowledge, not known in detail. However, it is reasonable to speculate that GFP binding could change the splay in the acyl chains, which would in turn modify the hydrophobic length. According to the model of Kuzmin et al.,12 this modification would reduce σ, which depends on the square of the hydrophobic height mismatch. Alternatively, the separation between the nickel lipid headgroups may be altered when more than one lipid binds to a single histidine tag. In this case, the GFP-nickel lipid cluster may act as an entity that is distinct from the unbound nickel lipid and has reduced solubility in the Ld phase. Finally, we note that, even if the packing of nickel lipid were unaffected by multivalent binding to a histidine tag or polyhistidine, there could be an entropic drive to bind clusters to the interface between domains. A cluster of N particles would lose approximately the same amount of translational entropy as a single lipid, but it would cover a larger length at the boundary and more strongly reduce the line energy. We expect the affinity of a protein−lipid cluster for the boundary to increase with its size, because a larger cluster covers a greater length of boundary. This expectation is qualitatively confirmed by the greater reduction of σ/P produced by poly(L-histidine) compared to his-tagged GFP: each poly(L-histidine) molecule

Δσ/σ = ε/(l0σ)(x·exp[ −ε/ kBT ])/(1 + x·exp[ −ε/ kBT ])

(4)

where l0 is the cluster diameter and x is the cluster mole fraction, taken to be equal to the nickel-lipid mole fraction. In principle, each hexahistidine tag can bind between one and six nickel lipids (although the steric hindrance among the GFP might prevent this); we therefore set l0 equal to two lipid diameters. If the line tension is assumed to be ∼1 pN, we obtain l0σ ≈ 1.6 pN·nm, which is ∼0.3 kBT at 298 K. For simplicity, we set l0σ = 0.3 kBT, which then allows us to use eq 4 to fit the data for his-tagged GFP to a single parameter, ε/kBT. The fit was done using the nonlinear least-squares method (Origin Sof tware, OriginLab Corp., Northampton MA). The quality of the fit is decent (solid line in Figure 4) indicating that the functional form of the model is reasonable. The best-fit value for ε/kBT is −0.80 ± 0.04. Varying the value of l0σ by a factor of 3 did not significantly change the quality of fit and yielded ε values in the range of −0.4 to −1.5 kBT. Hence, we find ε values that are on the order of ∼l0σ, providing evidence that the model is self-consistent. While this model was developed in the present context of a peripherally bound protein, it might also apply to integral membrane proteins, though in these cases there could be additional mechanisms at play.

■

CONCLUSIONS In summary, we have investigated the binding of a model membrane-associating protein, histidine-tagged GFP, to a nickel lipid in Lo/Ld phase-separated giant unilamellar vesicles. The peripheral binding by his-tagged GFP reduced the line tension at the boundary between Lo and Ld phases by up to 46%, depending on the mole fraction of nickel lipid. To our knowledge, this is the first time that such measurements have been made. An adsorption model with a single adjustable parameter provides a reasonable fit to the data with a best-fit binding energy of approximately −1 kBT for a lipid−protein cluster at the boundary between phases. Our results suggest that proteins that peripherally bind to multiple lipids or cause local reorganization of the lipid composition should exhibit an affinity for the boundaries of domains in membranes. This affinity reduces the line tension and thereby lowers the energy cost of forming a domain. Finally, the localization of proteins at the domain interface may further enhance their clustering, which may be important for their function. 5180

dx.doi.org/10.1021/la204225a | Langmuir 2012, 28, 5176−5181

Langmuir

■

Article

(16) Guo, L.; Smith-Dupont, K. B.; Gai, F. Diffusion as a Probe of Peptide-Induced Membrane Domain Formation. Biochemistry 2011, 50, 2291. (17) Nicolini, C.; Baranski, J.; Schlummer, S.; Palomo, J.; Lumbierres-Burgues, M.; Kahms, M.; Kuhlmann, J.; Sanchez, S.; Gratton, E.; Waldmann, H.; Winter, R. Visualizing association of NRas in lipid microdomains: Influence of domain structure and interfacial adsorption. J. Am. Chem. Soc. 2006, 128, 192. (18) Weise, K.; Triola, G.; Janosch, S.; Waldmann, H.; Winter, R. Visualizing association of lipidated signaling proteins in heterogeneous membranes-Partitioning into subdomains, lipid sorting, interfacial adsorption, and protein association. Biochim. Biophys. Acta 2010, 1798, 1409. (19) Schon, P.; Garcia-Saez, A. J.; Malovrh, P.; Bacia, K.; Anderluh, G.; Schwille, P. Equinatoxin II permeabilizing activity depends on the presence of sphingomyelin and lipid phase coexistence. Biophys. J. 2008, 95, 691. (20) Garcia-Saez, A. J.; Chiantia, S.; Salgado, J.; Schwille, P. Pore formation by a bax-derived peptide: Effect on the line tension of the membrane probed by AFM. Biophys. J. 2007, 93, 103. (21) Schafer, L. V.; de Jong, D. H.; Holt, A.; Rzepiela, A. J.; de Vries, A. H.; Poolman, B.; Killian, J. A.; Marrink, S. J. Lipid packing drives the segregation of transmembrane helices into disordered lipid domains in model membranes. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 1343. (22) Garcia-Saez, A. J.; Chiantia, S.; Schwille, P. Effect of line tension on the lateral organization of lipid membranes. J. Biol. Chem. 2007, 282, 33537. (23) Vind-Kezunovic, D.; Nielsen, C. H.; Wojewodzka, U.; Gniadecki, R. Line tension at lipid phase boundaries regulates formation of membrane vesicles in living cells. Biochim. Biophys. Acta 2008, 1778, 2480. (24) Pickering, S. U. Emulsions. J. Chem. Soc. 1907, 91, 2001. (25) Pieranski, P. Two-Dimensional Interfacial Colloidal Crystals. Phys. Rev. Lett. 1980, 45, 569. (26) Kutuzov, S.; He, J.; Tangirala, R.; Emrick, T.; Russell, T. P.; Boker, A. On the kinetics of nanoparticle self-assembly at liquid/liquid interfaces. Phys. Chem. Chem. Phys. 2007, 9, 6351−6358. (27) Du, K.; Glogowski, E.; Emrick, T.; Russell, T. P.; Dinsmore, A. D. Adsorption energy of nano- and microparticles at liquid-liquid interfaces. Langmuir 2010, 26, 12518. (28) Pedelacq, J. D.; Cabantous, S.; Tran, T.; Terwilliger, T. C.; Waldo, G. S. Engineering and characterization of a superfolder green fluorescent protein. Nat. Biotechnol. 2006, 24, 79. (29) Angelova, M. I.; Soleau, S.; Meleard, P.; Faucon, J. F.; Bothorel, P. Preparation of giant vesicles by external AC electric fields. Kinetics and applications. Prog. Colloid Polym. Sci. 1992, 89, 127. (30) Veatch, S. L.; Keller, S. L. Seeing spots: Complex phase behavior in simple membranes. Biochim. Biophys. Acta 2005, 1746, 172−185. (31) Baumgart, T.; Das, S.; Webb, W. W.; Jenkins, J. T. Membrane elasticity in giant vesicles with fluid phase coexistence. Biophys. J. 2005, 89, 1067. (32) Allain, J. M.; Ben Amar, M. Budding and fission of a multiphase vesicle. Eur. Phys. J. E 2006, 20, 409. (33) Tian, A. W.; Johnson, C.; Wang, W.; Baumgart, T. Line tension at fluid membrane domain boundaries measured by micropipette aspiration. Phys. Rev. Lett. 2007, 98, 208102. (34) Baumgart, T.; Hess, S. T.; Webb, W. W. Imaging coexisting fluid domains in biomembrane models coupling curvature and line tension. Nature 2003, 425, 821. (35) Nye, J. A.; Groves, J. T. Kinetic control of histidine-tagged protein surface density on supported lipid bilayers. Langmuir 2008, 24, 4145−4149. (36) Langmuir, I. The adsorption of gases on plane surfaces of glass, mica and platinum. J. Am. Chem. Soc. 1918, 40, 1361. (37) Adamson, A. W. Physical Chemistry of Surfaces; 3rd ed.; WileyInterscience: Hoboken, NJ, 1976.

ASSOCIATED CONTENT

S Supporting Information *

The method of image deconvolution, the estimates of GFP accumulation at the domain boundary, the data for singlephase-vesicle control experiments, and the data for polyhistidine are described in supplemental methods and in 11 supplemental figures. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Aruni Karunanayake Mudiyanselage for help with the purification of GFP, and the UMass Biomedical Innovation Initiative, the IGERT Program in Nanotechnology Innovation (NSF DGE-0504485) and NSF DMR-0907195 for funding.

■

REFERENCES

(1) Baumgart, T.; Hammond, A. T.; Sengupta, P.; Hess, S. T.; Holowka, D. A.; Baird, B. A.; Webb, W. W. Large-scale fluid/fluid phase separation of proteins and lipids in giant plasma membrane vesicles. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 3165. (2) Veatch, S. L.; Cicuta, P.; Sengupta, P.; Honerkamp-Smith, A.; Holowka, D.; Baird, B. Critical fluctuations in plasma membrane vesicles. ACS Chem. Biol. 2008, 3, 287. (3) London, E. Insights into lipid raft structure and formation from experiments in model membranes. Curr. Opin. Struct. Biol. 2002, 12, 480. (4) Veatch, S. L.; Keller, S. L. Miscibility phase diagrams of giant vesicles containing sphingomyelin. Phys. Rev. Lett. 2005, 94, 148101. (5) Dietrich, C.; Bagatolli, L. A.; Volovyk, Z. N.; Thompson, N. L.; Levi, M.; Jacobson, K.; Gratton, E. Lipid Rafts Reconstituted in Model Membranes. Biophys. J. 2001, 80, 1417. (6) Simons, K.; Ikonen, E. Functional rafts in cell membranes. Nature 1997, 387, 569. (7) Sprong, H.; van der Sluijs, P.; van Meer, G. How proteins move lipids and lipids move proteins. Nat. Rev. Mol. Cell Biol. 2001, 2, 504. (8) Edidin, M. The state of lipid rafts: From model membranes to cells. Annu. Rev. Biophys. Biomol. Struct. 2003, 32, 257. (9) Simons, K.; Vaz, W. L. C. Model systems, lipid rafts, and cell membranes. Annu. Rev. Biophys. Biomol. Struct. 2004, 33, 269−295. (10) van Meer, G.; Voelker, D. R.; Feigenson, G. W. Membrane lipids: where they are and how they behave. Nat. Rev. Mol. Cell Biol. 2008, 9, 112. (11) Dumas, F.; Sperotto, M. M.; Lebrun, M. C.; Tocanne, J. F.; Mouritsen, O. G. Molecular sorting of lipids by bacteriorhodopsin in dilauroylphophatidylcholine/distearoylphosphatidylcholine lipid bilayers. Biophys. J. 1997, 73, 1940. (12) Kuzmin, P. I.; Akimov, S. A.; Chizmadzhev, Y. A.; Zimmerberg, J.; Cohen, F. S. Line tension and interaction energies of membrane rafts calculated from lipid splay and tilt. Biophys. J. 2005, 88, 1120. (13) Alessandrini, A.; Facci, P. Unraveling lipid/protein interaction in model lipid bilayers by Atomic Force Microscopy. J. Mol. Recognit. 2011, 24, 387. (14) Seeger, H. M.; Bortolotti, C. A.; Alessandrini, A.; Facci, P. Phase-Transition-Induced Protein Redistribution in Lipid Bilayers. J. Phys. Chem. B 2009, 113, 16654. (15) Cannon, B.; Hermansson, M.; Gyorke, S.; Somerharju, P.; Virtanen, J. A.; Cheng, K. H. Regulation of calcium channel activity by lipid domain formation in planar lipid bilayers. Biophys. J. 2003, 85, 933. 5181

dx.doi.org/10.1021/la204225a | Langmuir 2012, 28, 5176−5181